Your Complete Guide to Index Investing with Dan Bortolotti

ETF Liquidity and Trading Volume

2012-12-03T11:56:50+00:00September 10th, 2012|Categories: ETFs|23 Comments

One of the most common misunderstandings about ETFs is that their daily trading volume has a large effect on their liquidity. Some investors avoid ETFs that don’t trade very often because they are concerned that making a purchase or sale will significantly affect the price: they worry, for example, that if they place a $10,000 buy order on a low-volume ETF, they will pay too much. Or conversely, if they place a $10,000 sell order, they will receive too little.

On the surface that’s a reasonable concern, because this is what happens with individual stocks. When a small investor buys or sells large-cap stocks, the market impact is zero. But if you place an order for a micro-cap company that is infrequently traded, your order may well move the price because of simple supply and demand. You can see this liquidity effect in the stock’s bid-ask spread: the difference between what you’d pay for Royal Bank and what you’d receive when you sell might be 0.05%. On a junior mining company it might be 2% to 4% or even more.

With ETFs, however, this isn’t the case. ETFs are open-end funds, which means new units can be created or redeemed as necessary, so supply and demand has little effect. Say a new ETF launches this week with 200,000 shares, each trading at $20, for a net asset value of $4 million. If an institutional investor wanted to put $1 million in the fund, their order would not drive the price sky high, as it would with a small-cap company with a finite number of shares. The ETF provider would likely just create 50,000 new units and deliver them to the institutional investor. The net asset value of the fund would now be $5 million and there would be 250,000 shares, each one still valued at $20.

What lies beneath

What really governs an ETF’s liquidity is the underlying securities. An ETF that holds large-cap stocks or government bonds is highly liquid regardless of its trading volume. You can see this when you compare the iShares S&P/TSX 60 (XIU) and the BMO Dow Jones Canada Titans 60 (ZCN). The former is the most frequently traded ETF in Canada, with millions of units changing hands each day. The latter trades only a few thousand. But both ETFs usually have a bid-ask spread of just one cent, which works out to about 0.05%. The reason is they have virtually the same holdings: the largest and most liquid companies in Canada, which also trade with tight spreads.

Compare that with the BMO Junior Oil Index ETF (ZJO) or the iShares S&P/TSX Venture (XVX), which trade with bid-ask spreads in the range of 0.30% to 1% or more. Yes, they happen to have low trading volumes, but that’s not the reason. Indeed, some US-listed ETFs that hold international small-caps do have high trading volumes but still carry wide spreads because of the illiquidity of their underlying stocks.

That said, there can still be some concerns with low-volume ETFs. While you don’t necessarily need to avoid them, you do need to be careful when you place your orders. I’ll explain more in my next post.


  1. Jon Evan September 10, 2012 at 5:18 pm

    The difficulty I have in comparing similiar etfs is why their market unit price is different. ZCN is about a dollar cheaper than XIU but they have virtually the same holdings as you say. Is ZCN then a better buy based on price/unit? The market unit price difference is even harder for me to understand when looking at Vanguard etfs which in Canada have gone on sale at $25/unit which are often twice as expensive as similar competitor etf products.

  2. Canadian Couch Potato September 10, 2012 at 6:16 pm

    @Jon: That’s a good question, and it has tripped up others as well. The unit price of an ETF tells you virtually nothing. It’s like saying this restaurant charges $2 for a slice of pizza and that one charges $3. Unless you know how large the slices are, the comparison is meaningless. It reminds me of the Yogi Berra joke where the waitress asks if he wants his pizza cut into six or eight slices, and he says, “You’d better make it six. I’m not hungry enough to eat eight.” :)

    Imagine an ETF tracking the TSX 60 has $100 million in assets. It might have 5 million outstanding shares, which would mean a net asset value (NAV) of $20 per share. Now imagine a second ETF tracking the same index with $150 million in assets. It might also slice that pie into 5 million shares, in which case the NAV per share would be $30.

    If you had $6,000 to invest, you could buy 300 shares of the first ETF or 200 shares of the second. Both would have an identical value and there is no reason to prefer one over the other.

    ETF providers generally try to keep the share price manageable: about $20 seems ideal. After many years of steady increases they often split them so they aren’t trading at $80 or $90 a share. That makes it easier to buy smaller amounts and reinvest dividends etc. In August 2008, iShares split seven ETFs, including XIU four-for-one. (Then a month later the market split the price in half again!)

    Hope this makes sense now.

  3. Jon Evan September 10, 2012 at 6:54 pm

    Thanks Dan. One must do the arithmetic!

  4. Nick September 11, 2012 at 12:28 pm

    So for a new fund, you have one player (the fund provider) and his proxy (i.e. your brokerage) providing the liquidity.

    Can they not manipulate the market price of the ETF since you only know the NAV from the previous day on their website (assuming they calculated it correctly)? For thousands of shares your losses (and their gains) on a trade will be significant, far outstripping the $10 trading fee.

    It would be very hard to manipulate XIU or RBC (RY) since the “float” on any day is enormous. Although, the brokerages can always bet against you from their inventory.

    I just like it when there are lots and lots of players in the market for a stock/ETF. You more likely to get a better price. Like one car dealer versus twenty in your town.

    Very good information. Thanks.

  5. Canadian Couch Potato September 11, 2012 at 12:39 pm

    @Nick: The ETF provider cannot manipulate the price. You are correct that it is difficult or impossible for you or me to check the NAV in real time, but market makers can do it because an ETFs holdings are transparent. Your brokerage does not really provide the liquidity: that is supplied by a number of market makers who are hired specifically to ensure that the ETF always trades close to NAV even if there is little volume.

    I agree there is certainly an advantage to using ETFs that have high trading volume, but they are not huge advantages. I want to make sure people don’t think low trading volume opens the doors for price manipulation, very high transaction costs, or other major issues. I’ll have more to say about this in my next post.

  6. Paul G. September 11, 2012 at 4:14 pm

    My main worry about low transaction volume is that it means the ETF isn’t very popular, and as such may shut down in the short or medium term.

    Maybe in 5 years ETFs aren’t as popular for some reason, and the market contracts by 50%, and todays moderately popular ETFs close. I don’t want to find myself forced to change my portfolio and think it all through once again.

    I want to buy and hold, which is my way of saying “buy and forget”.

  7. David M September 11, 2012 at 6:08 pm

    I hope you can comment on some ETFs that don’t appear to trade much at all, such as CJP, and whether we should avoid them, or rather looking at the holdings to make our decisions. Perhaps in the next post you might give tips for novices on how to buy them, e.g., do we need to use stop loss orders, etc.

  8. Canadian Couch Potato September 11, 2012 at 6:34 pm

    @Paul G: That’s a reasonable concern, especially in a taxable account. If the ETF has a large unrealized gain and you are forced to liquidate it you could take a big tax hit. ETF providers seem to take this responsibility seriously: iShares, for example, has never shut down an ETF in its history (with the exception of one speculative product after they bought Claymore). In fact, Mary Anne Wiley of iShares discusses this idea in her recent blog, which quotes yours truly:

    @David M: Tune in on Thursday for my next post, which addresses your question directly. For now let me say never use stop loss orders! But you should use limit orders any time you place an ETF order, regardless of its trading volume. More details to come.

  9. Que September 12, 2012 at 1:57 pm

    Can you please explain this example?

    XWD consists of :
    IVV – ISHARES S&P 500 FUND 53.58%

    XWD is the cheapest, yet it always seems to have a much larger spread than IVV, EFA, & EWC (all three usually have a spread that is tight as possible – 1 cent apart, but XWD is never that tight).


  10. Canadian Couch Potato September 12, 2012 at 2:25 pm

    @Que: I brought your question to my contact at BlackRock, who responded as follows:

    While the spread for some stocks / ETFs may appear to be 1 cent, this may not actually be the case if you want to trade more than 100 shares. We always advise investors to look at level 2 market data (showing the depth of the market rather than just the top order, which may be very small in size (and typically is in the US as the high-frequency trading community is active in most stocks, but only for very small trades). For most ETFs the spread is very tight, but it’s not always possible to get large trades fully filled at the best bid / offer.

    Once a market maker can get access to the underlying ETFs (perhaps for a spread of 2 cents) he / she will then need to cover any costs they may have offering markets in these funds by perhaps adding another 2 cents to their spread (depending on the complexity of the hedge required, the number of underlying names (because there will be a ticketing and custodian cost associated with each related trade). The additional 2 cents we see for XWD covers these additional costs and provides a small profit margin for the market maker.

    Liquidity for an ETF should always be viewed as the liquidity of the underlying marketplace, while spread should be viewed the same way, but with the additional costs that market participants need to take on as a result.

  11. Best of Blogs - Traveling is overrated September 14, 2012 at 2:02 am

    […] This week the Canadian Couch Potato discussed ETF Liquidity and Trading Volume. […]

  12. Que September 14, 2012 at 3:52 pm

    Thanks so much for following up Dan, its very much appreciated.

    This would be a question for the brokerage, but do you think it is common to have to pay for Level 2 market data?

    Thanks again,

  13. Canadian Couch Potato September 14, 2012 at 4:06 pm

    @Que: My pleasure, and glad it was helpful.

    I know that Scotia iTrade offers Level 2 quotes for free, though you have to specifically request it. Probably best to just call your brokerage and ask. If other readers have experience to share, please do so.

  14. Amir October 26, 2012 at 9:01 pm

    With low volume ETFs is there a danger of an order being partially filled. For example if I put in a limit order to purchase 10,000 shares of an ETF that typically only trades 1,500 shares is there a risk that only part of my order will be filled? I don’t believe that in Canada all-or-none orders are permitted.

  15. Canadian Couch Potato October 27, 2012 at 10:29 am

    @Amir: That should not be an issue, as ETFs can create new units whenever necessary. But if you are placing a very large order like that, you should consider contacting the ETF provider directly. They should be able to ensure that the transaction is handled properly.

  16. Oldie February 8, 2013 at 1:06 am

    @CCP: Sorry to ask a follow up question so late, but do shares of DLR/DLR.U follow the same general rules you listed above, i.e. that low background trading volume does not magnify the impact of superimposed large trades because the ETFs hold baskets of shares with large market capitalizations? The reason I ask is because I checked the trading volume of DLR and it was 6000 today, and generally trades at low volumes, i.e. under 20,000. In fact there was only one spike this year over 50,000 occurring in mid January of about 130,000 on one day (my guess is that someone did a Norbert’s Gambit of $1.3M, because there is a corresponding spike of 130,000 shares of DLR.u about the same date). I think I would be nervous trying to trade that many shares, about 10 times the normal daily trading volume, but maybe I’m worrying too much about liquidity and predictable (low bid-ask) trading prices.

  17. Canadian Couch Potato February 8, 2013 at 7:47 am

    @Oldie: Horizons seems to be good at keeping DLR’s spread at two cents, which is all that matters. I have traded it several times and have not had a problem, but you definitely need to use limit orders to avoid surprises.

  18. […] note: The following article, which originally appeared at this source on a few months ago, highlights key misconceptions many swing traders have when trading ETFs. […]

  19. Dave August 23, 2013 at 5:54 pm

    @Oldie and @CCP I just put through a $1M order for DLR and it executed immediately. I put a limit on it at the quoted asking price just to avoid surprises (and just saw this post after that advised the same). The market makers must be doing their job since our purchase was about 5% of their entire fund. As you would guess this is for Norberts gambit

  20. Oldie August 23, 2013 at 7:23 pm

    @ Dave: I think you just demonstrated that the DLR gambit is just about as foolproof and predictable as you can get for whatever amount you want and is economical for moderately large sums (i.e. over 5-10k) where you are essentially paying 0.2% — 1 cent on 10 dollars twice — (plus 2 trades) for the transaction. I just walked my daughter through a DLR Gambit last month for about 5k, so the details are fresh in my mind. She wanted it stress free, so 0.2% was a very good deal with no hurry to turn over your newly purchased DLR shares. She did it on Questrade, so there was a small added perk that there was no commission on the purchase leg. For the record, in February, after all that stewing, obsessing and scheming, I finally decided to convert my own US$1.4M to CAD, and I was able to get my broker to agree to immediate journalling to the Canadian dollar side and selling after purchase of shares by special arrangement — apparently this organization doesn’t like to do it, and I didn’t want the hassle of changing accounts, to RBC Direct, for instance — and did a Norbert’s Gambit using TD. It was a little hair raising for me, being the driver of the proceedings with no prior experience, but nothing unexpected happened, and I did the deal for approximately 0.025% (1 cent on $80 twice plus 8 trades — I broke the $1.4M amount into 4 separate smaller Gambits for safety, which turned out in hindsight to be unnecessary).

  21. Dave August 23, 2013 at 11:11 pm

    @Oldie I did a similar $5k test at TD with no issues. For the benefit of others reading this, I calculated precisely what I saved. I opened two browser windows and at the exact moment I bought DLR I got a quote from TD to buy US dollars to see what I would have gotten. The quote for US dollars in a regular exchange would have gotten me $4959.49 and after the gambit and $20 spent in transaction fees I ended up with $5009.85. This was $50.36 saving, or 1.0% of my money. Without the trading fees I would have saved $70.34 or $1.4%. Seeing that it was worthwhile even on a small amount gave me the confidence to in the mechanics and I went ahead with a larger order where the trading fee becomes negligible.

  22. Jon March 6, 2015 at 12:44 pm

    Hi CCP, is this the case with bond ETF’s? A while ago I purchased VAB with a very generous limit order and only half of it got filled. Before reading this post, I had assumed it was because it was a new ETF and they simply didn’t have the trading volume.


  23. Canadian Couch Potato March 7, 2015 at 3:07 pm

    @Jon: ETF orders should always be filled quite close to the NAV unless something very unusual is going on, like this:

    Certainly VAB should not pose any problem. How large was the order, and how “generous” was the limit?

Leave A Comment