ETF transaction costs and tracking difference

For investors outside the US. Personal investments, personal finance, investing news and theory.
Sister forums: Canada, Spain (en español)
---------------
Post Reply
Topic Author
a55
Posts: 3
Joined: Sat Oct 12, 2019 3:13 pm

ETF transaction costs and tracking difference

Post by a55 » Sat Oct 12, 2019 3:29 pm

Hi everyone,

I have been looking into the annual reports of iShares, where it is explicitly mentioned that the funds transaction costs are not included in the TER (so far so good).

The annual reports state the exact amount of transaction costs incurred for the fiscal year for each fund and presumably the impact on return can be easily calculated from that total amount to proportionally per oustanding share (which in the case of iShares seems to be so small that could be practically ignored ,I've roughly calculated something like 0.0000019% for their fund that incurs the highest transaction costs).

In the tracking difference calculations, they take into consideration the TER (they show the tracking difference gross as well as net of TER), as expected. Since the fund transaction costs though are not part of the TER, I wonder if they are typically included in the tracking difference or not. In the case of iShare funds this may be anyway irrelevant (due to the tiny amounts of transaction costs), but I am still curious if transaction costs are typically calculated as part of the tracking difference or not in general (what do fund providers usually do?)

Cheers

stlutz
Posts: 5422
Joined: Fri Jan 02, 2009 1:08 am

Re: ETF transaction costs and tracking difference

Post by stlutz » Sat Oct 12, 2019 4:26 pm

Yes, transaction costs are reflected in the tracking error. Commissions aren't the big cost as much as spreads and market impact are.

To pick one popular iShares ETF: USMV. It has always had an ER of .15%. Since inception it has underperformed it's bogey by .19%. So, it loses about another .04%/yr. through trading costs.

This tends to be bigger factor is less liquid sections of the bond market.

alex_686
Posts: 5097
Joined: Mon Feb 09, 2015 2:39 pm

Re: ETF transaction costs and tracking difference

Post by alex_686 » Sun Oct 13, 2019 6:41 am

a55 wrote:
Sat Oct 12, 2019 3:29 pm
... if transaction costs are typically calculated as part of the tracking difference or not in general (what do fund providers usually do?)
Technically no. The calculations are just the difference between the fund and the index. The index being maintained in a spreadsheet of pure abstraction. Once you have done the calculation you can start to guess and hypothesis what caused the difference. However, all of these calculations are going to indirect. Direct, indirect, and quality of the trades are the obvious place to start.

User avatar
BeBH65
Posts: 1555
Joined: Sat Jul 04, 2015 7:28 am

Re: ETF transaction costs and tracking difference

Post by BeBH65 » Sun Oct 13, 2019 6:57 am

BeBH65. (only an investment enthusiast, not a financial adviser, perform your due diligence). | Have a look at https://www.bogleheads.org/wiki/Outline_of_Non-US_domiciles

Topic Author
a55
Posts: 3
Joined: Sat Oct 12, 2019 3:13 pm

Re: ETF transaction costs and tracking difference

Post by a55 » Sun Oct 13, 2019 8:13 am

stlutz wrote:
Sat Oct 12, 2019 4:26 pm
Yes, transaction costs are reflected in the tracking error. Commissions aren't the big cost as much as spreads and market impact are.
So if they are indeed reflected in the tracking error then they should be also part of the tracking difference.

I am not so concerned about commissions but rather long-term impact on returns due to transaction costs. I assume spreads/market impact are more interesting for the entry/exit points and less important for very long term buy-and-hold investment. By transaction costs here I mean for example index fund portfolio reallocation and other such miscellaneous internal fund costs that are indirect and not part of the TER/OCF.

One of the EU brokers I'm using somehow estimates that the ongoing transaction costs are almost as high as the TER for a particular fund I'm looking into and thus impacting the returns, which would be worrysome for long term investments. Their estimation makes no sense to me as the fund annual reports suggest otherwise (the ETF has been outperforming the index since inception and offsetting the TER, and their transaction costs are rather negligible).

Incidentally, I've found the following paragraph in the iShares 2018 annual report, which appears to suggest that the tracking difference does indeed include the transaction costs (of course that's not to say that any other fund provider does the same):
Cash management, efficient portfolio management techniques including securities lending, transaction costs from rebalancing and currency hedging can have an impact on tracking difference and tracking error.

alex_686
Posts: 5097
Joined: Mon Feb 09, 2015 2:39 pm

Re: ETF transaction costs and tracking difference

Post by alex_686 » Sun Oct 13, 2019 1:27 pm

a55 wrote:
Sun Oct 13, 2019 8:13 am
stlutz wrote:
Sat Oct 12, 2019 4:26 pm
Yes, transaction costs are reflected in the tracking error. Commissions aren't the big cost as much as spreads and market impact are.
So if they are indeed reflected in the tracking error then they should be also part of the tracking difference. ...
One of the EU brokers I'm using somehow estimates that the ongoing transaction costs are almost as high as the TER for a particular fund I'm looking into and thus impacting the returns, which would be worrysome for long term investments.
Your first sentence is a bit of nonsense. "Tracking Error" is the industry's generic standardized method for "tracking differences". The words are functionally synonymous in this case.

The point that I was trying to make in my above post is that it is very easy to identify explicit transaction costs like commissions. It is very hard to track implicit costs like the bid/ask spread. By the very act of trading you change what the bid/ask spread it. As such it is impossible to replicate the index. It is a little like the old joke where a physicist claims he can solve all of a dairyman's problem if he can just assume perfectly elastic spherical cows.

As a counterpoint, a skilled (i.e. expensive) trader can actually beat the bid/ask spread by being a passive trader and trading with information traders.

Lastly, why would it be worrisome if the indirect costs of trading were greater than the explicit cost of trading? Does it matter if the tracking error comes from explicit costs or from other implicit costs? Some assets just have a higher trading costs. Small cap, value, certain bonds, EM, etc. What really matters if the asset meets the risk / return profile needed to meet your goals. Look at how the asset works in totality.

Topic Author
a55
Posts: 3
Joined: Sat Oct 12, 2019 3:13 pm

Re: ETF transaction costs and tracking difference

Post by a55 » Sun Oct 13, 2019 1:51 pm

alex_686 wrote:
Sun Oct 13, 2019 1:27 pm
Lastly, why would it be worrisome if the indirect costs of trading were greater than the explicit cost of trading? Does it matter if the tracking error comes from explicit costs or from other implicit costs? Some assets just have a higher trading costs. Small cap, value, certain bonds, EM, etc. What really matters if the asset meets the risk / return profile needed to meet your goals. Look at how the asset works in totality.
Agreed, but part of meeting goals and figuring out how the asset works is also understanding indirect costs that impact returns. I am basically faced with contradicting info: a financial report which shows that the recurring management costs are essentially eliminated due to the fund consistently overperfoming. On the other hand the broker company suggests that there are certain recurring fund costs (completely attributed to the fund provider) that will be eating into the annual returns. So I was basically trying to figure out if the tracking diff/error is inclusive of all that and can be used as a (more) reliable estimator than the broker company (which I have no idea where they are getting their figures from, and aren't very helpful about them, suggesting that I talk to the funds provider since the numbers come from the fund).

I suppose the "truth" is in the audited annual statements and it could be that the broker system is simply showing wrong estimations, but I was wondering if I am overlooking something in the financial reports (thus my question was if the TD is covering all that).

alex_686
Posts: 5097
Joined: Mon Feb 09, 2015 2:39 pm

Re: ETF transaction costs and tracking difference

Post by alex_686 » Sun Oct 13, 2019 3:15 pm

Lots of points, I am reordering you response.
a55 wrote:
Sun Oct 13, 2019 1:51 pm
So I was basically trying to figure out if the tracking diff/error is inclusive of all that and can be used as a (more) reliable estimator than the broker company (which I have no idea where they are getting their figures from, and aren't very helpful about them, suggesting that I talk to the funds provider since the numbers come from the fund).
I have had this conversation before, and as I broker I would have to direct you to the fund as it is very much the fund's responsibility. And I have also worked on the fund side, and we are not going to tell you the exact methodology. That being said, there is the generic implementation which the SEC requires everybody to follow.

https://www.investopedia.com/terms/t/trackingerror.asp

A couple of items here. The inputs of tracking error are normally the end of day price of the fund and the index value. So, it is calculated after expenses - both implicit and explicit. You don't need the annual report. In fact, that is one of its virtues - It does not rely on self reported accounting values but uses directly observable and easily gathered inputs. You just need the last 250 days worth of closing prices for the index and the fund.

Notice that the tracking error is a square root. As such it is always positive. So a tracking error does not tell you if the fund under/over preformed the index - it is just the difference from the index.
a55 wrote:
Sun Oct 13, 2019 1:51 pm
I am basically faced with contradicting info: a financial report which shows that the recurring management costs are essentially eliminated due to the fund consistently overperfoming. On the other hand the broker company suggests that there are certain recurring fund costs (completely attributed to the fund provider) that will be eating into the annual returns.
A couple of points. I am a bit skeptical of what you are saying. The statements are problematic and could get the broker into hot water. Maybe the broker was using the term incorrectly, maybe he threw in some weasel words and implied stuff, maybe you misunderstood. For context I used to have to write and review marketing material, and well as calculate these numbers.

Passive funds can't claim that they are beating the index. They would emphasis a low tracking error suggesting they closely tracked the index. Active funds can - sort of - and would emphasis a high tracking error. After all, they should be making large "active share" bets which are different then index and should generate higher returns.

To extend on the bit above that Tracking Error is always positive, it is a very poor metric to determine if the fund is beating the index inclusive of trading and management expenses.

alex_686
Posts: 5097
Joined: Mon Feb 09, 2015 2:39 pm

Re: ETF transaction costs and tracking difference

Post by alex_686 » Sun Oct 13, 2019 3:27 pm

Oh - one last thing. The drag caused by trading is very low. It is low for passive funds, it is low of ETFs, it is very low for passive ETFs. A few bps (.01%) per year. You have to go very deep in the weeds to figure out what the impact is. Even then it is kind of a guessing game.

Post Reply