A crisis of ETFs: what’s really going on?

A few months ago, a friend reached out to me with a few question on ETFs. I casually answered it and haven’t thought much since, but have subsequently realised it might be useful for others. Oftentimes, active managers say “ETFs are destroying our business so it must all be their fault!!!” And ETF managers say “research we’ve paid for says it has nothing to do with ETFs!!!”. The truth, as always, is somewhere in the middle.

Below I recreate his questions, and also my answers, for anyone interested in ETFs and their role in recent moves. I’ve chosen to do this without any pictures or edits - a simple copy and paste so you can see an honest set of concerns and answers, rather than a stylised presentation.

Hope this is useful - please share your thoughts.

Email in March:

Hope you are coping fine under the market volatility. There has been some comments on ETF lately, which overplays the mechanical nature of ETF, and its market impact. Will be good to hear your thoughts as I am keen to dispel some myths related to ETF. Internal portfolio management will always be a focus for us, but long-term, I think ETF should be part of our portfolio management tools. Will be keen if you have any related materials at your end. Few broad thoughts for your comments as well:

• When ETF exchanges hands, the pricing discovery is no different form a single-name stock. There is no broader market impact here, as no actual liquidation of underlying stocks are required. 
• Actual market impact will require ETF redemption, where investors convert their ETF holdings into underlying assets and sell. Is this fair? How does ETF redemption work? When will it be triggered typically and who are the players involved? 
• I will suspect that stratified sampling is used heavily in ETF management. So the comment on mechanical rules severely downplay the sophistication of the industry.

My response:

First, let me apologise and say that we don’t have any materials as we are simply too small to focus on broader ETF industry education ?? we focus more on educating our clients about the importance of allocation to Asia, picking lower cost strategies than active funds in the region, and hopefully also of course using our strategies ??

That said, I have to say that my view on ETFs has evolved somewhat since my early diehard days at iShares. I do think that most comments on ETFs overstate the market impact, but I also don’t agree w/ BLK that there is no impact whatsoever. Once you’ve read, happy to get on the phone and discuss with you if you like.  

So here are my thoughts:

• ETFs create 2 doors instead of 1 for liquidity: all this talk of ETFs creating a liquidity problem as investors rush for the exits is nonsense… before you could only trade the underlying stock/bond. Now you can trade either the ETF or stock/bond. Pessimistic investors will draw a parallel to swaps and the issues caused in 2008. ETFs, however, are markedly different. Swaps, CDS, CDOs, etc. – they all relied on investment bank pricing and investment bank inventory. So when liquidity dried up and their balance sheets got in trouble, pricing disappeared. ETFs, in contrast, very rarely sit on investment bank, or anyone else’s balance sheet, for the matter. They are owned by a very diversified set of investors – retail, sovereign, insurance, pension, etc. The only job an investment bank or other mkt participant has is to make a market in the ETF and to do this they don’t actually have to take any major risk… there is an indicative fair value published most of the time. They can even do it on an agency only basis and avoid risk altogether.
• Many ETF investors are not purely static participants. I’m finalizing an op-ed on this topic now. S&P 500 ETFs are only 15% of ETF assets in the US… investors use ETFs for ACTIVE asset allocation, not passive holdings. With faster and faster growth in value/quality/esg/thematic ETFs I’d actually argue that there is nothing passive about ETFs and the entire argument is moot. I don’t think there is necessarily a herd mentality at this point. Also, it’s important to note that while the media always focuses on ETFs and the rise of passive, many internal passive holdings are completely excluded from those calculations. Just as an example - US pensions, Norges Bank, HKMA, CIC/SAFE, GIC, GPIF - all of them have internal passive portfolios. Obviously retail and individual investors are now participating more and more in passive, but it can’t be a problem just because retail investors got in on the game and because we can see the numbers more easily.
• Where I do think there is an issue is on the liquidity impact of the underlying stocks/bonds. Bonds inside LQD/HYG are more liquid than bonds outside them. Stocks inside S&P 500 are more liquid than outside. But in FI that might simply be a development problem – in due course, there will be more FI ETFs including with tilts to quality/value/momentum, etc thereby creating diversified index views. Until then, bonds outside the blockbuster ETFs are punished in liquidity because they cannot be sold to the ETF. Their risk is harder to price. But again, that’s not the ETFs fault – the ETF is in fact making some bonds more liquid, not making others illiquid. But I’m convinced the distortion is there.
• The other question I worry about is whether or not passive holdings overall, not just ETFs, are starting to impact the betas of the most widely held positions. One thing I’ve been wanting to do for some time is to study the evolution of betas of say the top 10 stocks in the S&P 500 over the last 50 years. Look at rolling 3yr betas of the top 10 positions every few years to see if the betas are trending up, down, or are unaffected. My gut feel is that the biggest positions are now whipsawed more and more by market sentiment because they are more owned by passive investors of all kinds. Is that a problem? Debatable. For mkt cap ETFs – they are already the most liquid stocks. For non mkt cap ETFs – those products are unlikely big enough to make a dent. But I do believe the effect is there. Will make time to run the #s.

Sorry – that was stream of consciousness with one eye on mkt activity. Hope it all makes sense. Big picture – I don’t believe ETFs are creating liquidity problems, but I do believe that there is an impact on pricing when enough of a stock/bond is owned by passive investors (not just ETFs). We’ve all known about the S&P 500 index entry/exit effect – why would it be any different everywhere else?

Now, to your questions specifically:

• When ETF exchanges hands, the pricing discovery is no different form a single-name stock. There is no broader market impact here, as no actual liquidation of underlying stocks are required. 

Correct – that’s the 2 doors analogy I made above. When I buy or sell an ETF, I’m typically buying or selling from someone else who already owns the ETF. The underlying stocks are not touched. If they are, that’s the redemption/creation mechanism which we’ll talk about below, but that usually only occurs in small ETFs that don’t have much market impact anyway. It’s a very simple analogy from Mark Wiedman, former head of iShares – if you are in a burning house, wouldn’t you want it to have 2 doors instead of 1? If you own the security directly, or via a mandate/MF – your only choice is to sell that security. You are limited to the average turnover of that security and the investors who have a different view on that one security. With an ETF, you have a 2nd door and a 2nd investor pool – investors who don’t know that stock but do have a macro view on the exposure. 2 doors instead of 1.

• Actual market impact will require ETF redemption, where investors convert their ETF holdings into underlying assets and sell. Is this fair? How does ETF redemption work? When will it be triggered typically and who are the players involved? 

If there are only sellers, then what will happen is that the price of the ETF will start to drop below the fair value of all the stocks inside. Let’s say S&P500 stocks have an average price of 100. The ETF should trade at 100. But if there are many sellers, the ETF might drop to 99.9. That means a market maker can buy the ETF from sellers at 99.9 and redeem it. Mechanically what will happen is that the ETF will deliver the underlying 500 shares to the market maker. The market maker will then go and sell them at 100 (minus 3-4 bps for transaction costs), generating a 6-7 bps profit. Obviously in reality things are more complex. The 500 stocks will no longer be worth 100 – so for a redemption to occur the ETF price has to drop more than the fair value of all the stocks and the market makers transaction cost estimate to liquidate that portfolio. In addition, an institutional investor can simply sell the ETF to the market maker and tell the market maker not to liquidate the underlying positions but to simply deliver them to that institutional investor (only possible for in-kind markets of course). So then the market maker has no transaction costs (and no risk) and can make it even tighter. They are purely an agent. The institutional investor or the market maker can even decide to sell only a portion of the 500 stocks and keep the ones they like. All of a sudden, that redemption is no longer a passive redemption – it’s an active trade.

• Is this fair? 

Why not – you can buy/sell stocks. Same w/ ETFs.

• How does ETF redemption work? 

Basically as above – ETF will deliver underlying shares. MM can do whatever they want with them. Exchange happens at end of day close prices. That’s for in-kindable markets. For direct ID markets, the ETF or the MMs (depending on the market/product) will estimate their liquidation costs and if there is a profit to be made, creations/redemptions will occur.

• When will it be triggered?

Whenever the price of the ETF deviates from fair value +/- relevant transaction costs. Importantly, it’s the view of the MM (or GIC) that determines if a redemption/creation occurs. The ETF provider has no say/input. Also, it’s the MM/GIC view on fair value that determines it, and it’s a choice, not an obligation. In other words, Citi can be doing a creation while GS is doing a redemption due to different views of underlying fair value and transaction costs.

• Who are the players?

Anyone who has the right capital mkts licenses and is approved by the ETF provider can do it. There are specialist mkt makers (Jane Street, Flow Traders, Susquehanna, Cantor Fitzgerald, Virtu) and more traditional brokers (GS, Citi, BAML, MS, CS, KGI, Philip, DBS Vickers, etc.). Anyone can do it but past studies have found that in the US basically ~15 players make up the bulk of the create/redeem and market-making market for ETFs. This is one potential area of risk – what if they all fail… then the pricing of ETFs would widen out around fair value massively until new players come in. But they obviously won’t be as good and won’t have as good a tech setup. But for all of these guys to step away would mean that something went horribly wrong – remember – at most they will carry the risk 1 day until they can sell the underlying positions. They are not required to sit on inventory of ETFs of underlying stocks. So for all of them to fail, basically the entire ETF market and underlying stock market would have to be closed/collapsed. To my knowledge, there’s been only one major day when a broker fully stepped away – during some heavy EM sales in 2013 (I think), Citi bought from clients over 1B of EEM (MSCI EM) in NY time. They redeemed the entire amount but of course couldn’t sell the underlying at the same time as Asia was closed during NY time. So they stopped accepting redemptions on EEM and other EM ETFs for 12 hours… when Asia opened they sold the underlying stocks and the next day accepted redemptions again. Importantly: 1) Citi continued to make markets (no redemptions, but still offered 2-way quotes on exchange), 2) Citi continued to allow creations, 3) Citi did not change any of their activity for any other ETFs other than EM, 4) other market makers and brokers continue to make markets and offer redemptions in EEM, 5) EEM never stopped trading or allowing redemptions

• I suspect that stratified sampling is used heavily in ETF management. So the comment on mechanical rules severely downplay the sophistication of the industry.

Yes and no. You are right that the PM has the right to decide when to buy/sell, what to hold and not. So the less liquid stocks are sometimes sampled out of the ETF. Index rebalances are traded actively using studies on past behavior, index arb, etc. So there isn’t a mad rush for a sale at the same time a stock exits the index, etc. PMs can even manage their buy/selling/rebalancing through creations/redemptions – for an inflow, request that the market maker delivers more of a stock you’re underweight (so the market maker pays the transaction cost) and vice versa. So the sophistication is high. However, there are limits to this… ultimately, if AAPL is 5% of S&P 500 nearly every S&P 500 ETF will own 5%. The sampling happens at the bottom, not at the top. That’s why I wonder if the largest positions are seeing increased betas over time… You can’t sample your way out of an MSFT or AAPL exposure. But at the bottom, absolutely – even ETF PMs are active managers in this aspect – their job is not to outperform, but to lower the cost of replicating exposure and that means identifying the cheapest way to trade and the best time to do a trade. You can even hold a stock that exited your index for a month if you determine the expected TE impact to be lower than the realized transaction cost. However, again, you can only do this w/ small positions. If AAPL exits the index, going from 5% to 0%, even a 50 bps transaction cost will be lower than the TE risk you take on, so you have to sell pretty quickly right about the same time as everyone else.