I am a proponent of passive index ETF investing. Historical performance supports that view with most active managers trailing the market. Management fees (MER) explain most of the underperformance and are easily found for most funds – including ETFs. However, there are other performance-drags that even passive-index-tracking funds encounter.
One of the arguments presented to me against passive index investing is that, when we account for these expenses and imprecisions, that ETFs don’t perform as well as expected. Their “tracking error” is bad and they lag indexes just like active funds.
We think we are buying an ETF that tracks an index minus the MER, but are we?
- Is it a serious problem or a nothingburger?
- How can it inform ETF selection for portfolio design?
What is ETF index tracking error?
An ETF is designed to track an index that contains a bunch of stocks or bonds. In reality, there are several obstacles to doing this in practice. This leads to fluctuations of the ETF above and below the value of the tracked index at any given time.
The magnitude of that oscillation (usually measured as standard deviation) is the tracking error.
What is the ETF index tracking difference?
The difference between the index value and ETF value is constantly changing due to the tracking error. So, spot measurements are not useful. However, over longer periods of time, the ETF will usually lag the index slightly. This is the tracking difference.
What causes the ETF tracking difference?
The biggest and most obvious contributor to the lag is the management expense ratio (MER). That is the management fee plus applicable sales tax which is constant and easily quantifiable. Hence, it is front and center in the materials that ETF providers promote and compete with each other over.
Funds also incur costs from trading, referred to as the trading expense ratio (TER). The MER and TER account for the bulk of tracking error, but not all of it.
How these factors and a few others interact results in the tracking difference. What are they?
The number of holdings.
An index that holds 60 large-cap stocks, like the TSX 60, is pretty easy to track. There are only 60 holdings. In contrast, a larger index, like the whole TSX has thousands of stocks. It would be impractical and expensive to constantly buy/sell all of them to mirror the whole market. This has a couple of impacts.
One is that ETFs meant to track indexes with a few highly-liquid holdings do so more tightly – like the TSX 60. This is also why we see many core ETFs for indices like the TSX 60, or FTSE Canadian All Cap Index, instead of the TSX in its entirety.
The FTSE Canadian All Cap Index takes a sampling of 200 Canadian large, mid, and small-cap companies. However, VCN.TO which tracks it, has 202 holdings. You will see these numbers don’t match for most ETFs and their tracked index. Why is that?
Index Holding Turnover.
These indexes are also constantly listing new stocks and de-listing others. When this happens, the ETF may or may not buy and sell their holdings immediately. There is a degree of management decision-making.
Some of the indexes tracked even have a component of human decision-making. For example, when looking at US small-cap stocks, the Russel 2000 is driven by math and the S&P 600 has a committee component. Mechanical models will result in more turn-over while committees introduce human decisions. This can affect their performance.
When you introduce management decisions, whether by a human or an algorithm, you have introduced an aspect of market-timing. This can have an influence on fund performance due to skill and luck. With mechanical processes and long time periods, random chance should attenuate this. However, the reality is that there can be over or under-performance for years.
As alluded to above, ETFs are not static. There are trading costs when the underlying holdings are bought and sold. That could happen due to turnover in the index being tracked as mentioned above. It can also occur from rebalancing.
Some ETFs are cap-weighted. That means they simply hold the equities according to their market capital and don’t need to be rebalanced. However, others are equal weight. For example, SPY tracks the S&P 500 by capital weight and RSP is an equal-weight ETF tracking the S&P 500.
The function effect of equal weighting is that the huge companies have the same allocation as the smaller ones. It gives more diversification to reduce specific risk and a smaller-cap tilt. Smaller companies may have more risk and more growth potential. However, as the different stocks grow at different rates, they will require rebalancing. That will trigger trading costs and there would also be tax consequences if there is a net capital gain.
Self-balancing asset allocation ETFs also have the potential to incur trading and tax costs due to rebalancing. However, that is unlikely if these popular ETFs continue to grow rapidly with new contributions. The new contributions could simply be used to buy more of the lagging asset classes.
Cash is for spending, not sitting in your ETF.
Even as we contribute money to ETFs and they don’t just grow instantly. That cash is used by the ETFs to purchase units of the index. That service is provided by the market makers, as described in my post about the legislative changes facing ETFs. The delay should be minimal if dealing with large efficient market makers.
The more common source of cash are the dividends collected from the holdings in the ETF. This cash is passed out periodically, such as monthy or quarterly. In the interim, it sits in the ETF or is short-term invested for interest. Alternatively, it can be invested in other index-trackers or futures contracts which would incur some cost.
Cash is usually invested to collect some short-term interest. That usually makes less compared to if it were fully invested. It can vary by market volatility, but on average, that lag causes some performance drag.
Market efficiency and holding liquidity.
Market makers that build the units of ETFs make money on the bid-ask spread. The bid-ask spread is the difference between what buyers want to pay and sellers want to sell for. If an ETF covers markets where some of the holdings are hard to come by or trade in small volumes, then the bid-ask spread will be larger. That costs money to the ETF provider which is taken from the fund’s assets.
ETFs can have side-hustles too.
It isn’t all bad news. Some ETFs can also make some extra money off of their holdings that boost their assets. For example, they may lend their stocks (for a fee) to those using options contracts.
Where to find the ETF Index Tracking Difference?
There are several methods to find the tracking difference.
For US-listed ETFs, ETF.com is easy and informative.
There are two places to look there.
One year snapshot.
In the efficiency section, it will show the tracking difference for the trailing year as the median and maximum deviation. It also gives useful information such as the MER and whether the fund has a side gig making money from lending securities.
Tracking difference can fluctuate significantly year-to-year depending on the balance of the various factors that influence it. So, I find looking over a longer time frame more useful. This is displayed in the performance section.
The tracking difference is the index performance minus the ETF’s Net Asset Value (NAV). For a well-run efficient ETF, it will often be the same as the MER or sometimes even less. As mentioned, index methodology can have an effect. Below are screenshots for IJR and VIOO which both track the S&P 600 small-cap index.
As seen above, there is minimal tracking difference for IJR or VIOO. Both are well-managed and efficient ETFs. IJR is much larger and has a slightly lower MER which likely explains its miniscule advantage. When you look at the 2018 annual management report, IJR had $24M income from lending its securities that almost totally offset the management fee of $30M. There is a good income-producing side-gig for these small cap ETFs and their tracking difference is actually less than you’d expect from their MER! A bacon and cheese double-patty nothingburger.
Past performance does not predict the future. However, I find five years of consistent management performance reassuring. If you want more data, you need to go to the source. For example, the iShares site has 10 years of data for IJR that has the same message.
For Canadian-listed ETFs, go to the source.
The ETF provider will often have the necessary information on their product information page. Again, the section to search is the performance area. Subtract the ETF NAV Performance from the Index Performance. Here is the data for XIU.TO. The grand-daddy of Canadian ETFs tracking the TSX 60.
Well, they don’t net income to offset their fees with a side gig, but there is very little hidden cost beyond the MER. In fact, the iShares website even says up to 0.02%/yr “other costs” may be incurred. Pretty transparent, and ugh, another nothingburger. I am getting pretty stuffed.
The tracking difference is a nothingburger for simple large-cap indices. Dare we test more ETFs?
What about funds that require more trading?
Let’s try a Canadian index that requires more fancy footwork, like rebalancing. An equal-weight version of TSX 60 ETF like I mentioned earlier in the post.
HEW also tracks the TSX 60 companies, but it holds them in equal weight. Not only does that result in a higher MER to manage, but there are also more of those hidden costs. That leads to a tracking difference of -0.73%! Not a nothingburger.
Check the regulatory documents for operating expenses.
The other place to look to understand this further, or if you can’t find performance data, is in the annual management report. If you can’t find that, it will sometimes be in the ETF’s regulatory fact sheet. Horizon has a nice fact sheet on their site for their ETFs. For HEW, it describes the MER as 0.61% and the trading costs as 0.6% for a total expense ratio of 0.67%. The rest of the tracking difference must come from the other factors mentioned earlier in the post.
What about ETFs that track less liquid markets?
Some markets are less liquid or more difficult to trade in. Preferred shares are a good example. Let’s look at the tracking difference for ZPR.TO which tracks the Solactive Laddered Canadian Preferred Shares Index. That is an index that will have lots of regular turnover and deals in a less liquid area of the market. There is also major dividend income to distribute that could cause “cash-drag”. If we are going to find something juicy – this is the place.
Extra costs beyond the MER of about 20 basis points and trailing the index by almost 70 basis points. Definitely not a nothingburger. However, some perspective is needed.
Passive Index Tracking Error vs. Active Management
Ok. So, passive index investing may not always match the market. There will be short term fluctuations (tracking error). There may even be longer-term consistent lag (tracking difference). However, it still pales in comparison to the drag of active management fees in the 1.5-3%/year range.
It would also be really hard for managers to make up their costs since the issues that cause tracking difference are all magnified by an active approach: timing, cash management, holding turnover and trading costs..
I am not convinced to change my approach to actively managed funds.
Other Implications of Tracking Error For Building Your Portfolio
Use large efficient ETFs as core holdings.
Large efficient ETFs trading in a liquid market should match the index very closely. Those with a good revenue stream from holding-loans may even have a tracking difference less than their management fees like I showed with IJR and XIU.TO. Past fund performance does not predict the future. However, I do think that good management is more predictable than the general market forces.
Consider which index you track.
I gave the example of choosing the S&P 600 over the Russel 2000 for US small-cap exposure as an example. If you want to target an area of the market, considering how it is tracked may make a difference. Even an index has some component of management to it. Of course, I can’t predict what method will work best moving forward. However, a consistent track record may suggest an important factor is in play.
Weigh weighting-strategy carefully.
Equal-weight may seem appealing for diversification and potential growth. For example, the big five banks make up a huge part of the TSX 60. Just 3 companies (Apple, Microsoft, and Amazon) make up 30% of the Nasdaq 100. However, the management and trading costs that get incurred may blunt or overwhelm the benefit.
If deviating from a simple cap-weighted ETF, take the time to weigh the decision carefully. Looking at the tracking difference can help.
Think twice before going after niche areas of the market.
There are many different areas of the investable market that we can target. We may do that because we want to target a specific “factor” like value or size that has historically outperformed the broader market over long timeframes. We may want to overweight or underweight an area that we feel has some other attractive characteristics. Some asset classes, like REITs or preferred shares, may (or may not) add some diversification or income advantages.
These strategies all deviate from tracking the market as a whole. There are potential risks and benefits of doing that. In addition to the usual dispersion of returns risk over our investing timeframe, we should also consider the practical costs of investing in smaller niches. That can be quantified by looking at the tracking difference of ETFs that cover those areas.
There is little point in picking an index or fund that you think should outperform by a few basis points if the tracking difference of the relevant ETFs to invest there is larger.