Index Fund Tracking Error Sources

Posted by Preet on Mar 24, 2010 | 2 comments

NOTE: I’m scrambling to write this before I get on a plane and my laptop battery is near death, so pardon any typos for the time being – I’ll edit it tomorrow, and may even re-write it! It’s good to be my own editor…. :)

Not all index funds are created equal. Some actually track their indices pretty well, and some do a poor job. Most people think that tracking an index would be a relatively simple thing but you know what they say, “in theory, theory and practice are the same but in practice they are not.”

Some sources of index fund tracking errors:

1. Resampling (Or Optimization)

If an index has 500 constituents, then it is impractical to replicate all the holdings when the fund has a small amount of assets. For example when an index fund first starts trading, it may only buy another manufacturer’s ETF to get market Beta until there are enough assets in the fund to actually go out and buy some or all the holdings itself. The index fund manager may also choose to hold a portion of the 500 holdings until the fund gets really big (to minimize transaction costs). How do they pick which stocks to hold and which they don’t? It’s up to them, but one method is to pick a combination of stocks that allow them to replicate the GICS sector allocations in the index (Global Industry Classification Standard). That means that if financials are 20% of the index and consumer discretionaries are 20% and so on, they will pick the combination of stocks that allow them to match those numbers – in this case they are seeking to match sector Betas.

2. Cash Flow timing

When money is added to a fund it must then be deployed into the holdings. In the case of ETFs, if not enough money is added to a fund to buy a creation unit, it might sit in cash until the next day. If the underlying stocks move between the positive cash flow and the cash deployment, this could affect the index fund’s performance. In the case of a mutual fund, the portfolio manager (yes, index funds have them too actually!) might get a small cash flow and not be able to deploy it into all the underlying constituents – they may choose to buy an ETF for market or sector beta, or buy a portion of the underlying constituents and make up the difference the next trading day when new money comes in, or if money leaves the fund for a redemption.

3. Proxies

Some index funds (with foreign exposure) may buy the foreign holdings on foreign exchanges, and some may buy ADRs or GDRs (American Depositary Receipts or Global Depositary Receipts). ADRs trade in the US but may trade at a premium or discount to the actual underlying stock.

4. Market Access

Again, index funds with foreign exposure may have stocks that trade in markets that are closed when domestic markets are open and vice-versa. If the index fund buys the direct stocks, someone has to deploy the cash overnight – it can be the fund custodian who sub-contracts out to a foreign prime broker, or the fund might have an office in that market. But if the fund operates in a different market, they can only receive the money during their hours of operation, so the underlying stocks can change in value between the time the cash comes to the fund and when it gets deployed.

If the fund buys ADRs then you still have the issue of the ADR lagging the movement of the underlying stock since money gets deployed right away, but in a security (the ADR) that can itself be moved by supply-demand issues on the market it trades even though the underlying security is not being traded. Again, this can introduce tracking error.

5. Dividend Drag

This really falls into the cash flow management arena, but instead of the cash flows being due to investors adding or subtracting money from the fund, with dividend drag it is due to the receipt of dividends earned on the underlying stocks being held. The fund receives cash which has to wait to be deployed.

6. Securities Lending Income

Same principle as with dividend drag, except the positive cash flow is due to the income generated from loaning out stocks in the fund to short sellers.

7. Brokerage commissions

The fund itself has to pay commissions to buy and sell stocks, so this will create a drag on returns too.

8. MER

Ah yes, can’t forget this one! The Management Expense Ratio is made up of the Management Fee and Operating Expenses, and of course these will drag down performance of the fund as well.

Conclusion

These are some of the areas which can introduce tracking error and I haven’t even talked about currency concerns. Different index companies tracking the same indices can have dramatically different tracking errors and its certainly something that doesn’t get enough attention. Note that some of these factors may generate positive or negative tracking errors and some (i.e. fees) can only generate negative tracking error. In its purest form, tracking error is the absolute magnitude of the deviation from the index and is not normally referred to as being positive or negative, but breaking it down this way is helpful.

Even More Clarity on Vanguard ETF Taxation

Posted by Preet on Sep 22, 2009 | 4 comments

A well informed authority on indexing emailed me to offer some further insight into the taxation nuances of Vanguard ETFs. Without revealing the source, I can assure you that what follows comes from a real expert on the subject matter. For those who have missed it, you may want to read these related posts first: Vanguard ETFs have Different Tax Considerations that other ETFs, and Tax Efficiency of Vanguard ETFs Follow-Up.

The claim that the activity of the investors in the Vanguard traditional index funds will cause adverse tax consequences for holders of the ETF class of shares has been raised before. However, this has yet to materialize and the proof is in the pudding. Out of the 39 ETFs, only the REIT ETF paid a small capital gains (in 2004, 2005, and 2006), and the Consumer Staples ETF paid a small gain in 2004. Not one of the large, broadly diversified ETFs, like VTI or VWO, has ever paid a gain. The same can’t be said for other ETF manufacturers’ commensurate products.

Vanguard’s ETF structure, which combines a conventional class of index fund shares with exchange traded shares, leverages the advantages provided by both classes of shares. I’d cite three primary advantages: cost efficiency, tax efficiency, and tracking efficiency.

First, the structure enables Vanguard to offer some of the lowest-cost ETFs in the market. Introducing a separate share class of an existing fund substantially reduces the start-up costs that can encumber the launch of a stand-alone ETF, and it leverages the economies of scale of an existing large pool of assets to minimize ongoing operating and trading costs. That’s why they are able to offer the Emerging Markets ETF (VWO) at one-third the cost of the most similar competitor product, and a Total Bond Market ETF (BND) at less than half the cost of competing products.

Second, unlike stand-alone ETFs, this share-class structure gives Vanguard additional ways to maximize after-tax ETF returns relative to competitors. All ETFs can minimize capital gains by distributing their lowest-cost shares to meet redemption requests — Vanguard does this too. However, they have an additional tool in their arsenal. As you noted, as cash flows into the funds’ conventional shares, they purchase stocks at a wide variety of tax lots, and when investors redeem shares of the conventional funds, or when there are index changes that require sale of a security, they sell the highest cost lots first, typically resulting in the realization of capital losses. These losses are stored in the fund for up to eight years to be used to offset capital gains that might be realized in the future, and they benefit both conventional share investors and ETF investors.

However, contrary to your post, conventional open-ended funds can “in-kind” shares to investors — if there was a large enough transaction that might cause a capital gain, an open-ended fund could do an in-kind redemption of securities instead of cash. It’s rarely utilized (although it has been done) and in Vanguard’s case, rarely necessary. In fact, Vanguard’s index funds are so large, and take in such steady cash flow, that they can typically “cross” incoming cash with outgoing redemptions, thereby avoiding the need to buy or sell stocks altogether.

For these reasons, the suggestion that if there were large redemptions from Vanguard’s traditional index funds the ETF shareholders would be impacted is a hypothetical situation that we haven’t seen manifest. We have just experienced the second worst stock market decline in history, and index funds experienced positive cash flows. If investors were going to redeem en masse, we’d have seen it in 2008. Vanguard routinely “stress tests” their portfolios to see how much of a fund could be redeemed before it realizes a capital gain. For instance, the largest stock index fund (and largest ETF) Vanguard Total Stock Market Index Fund (and its ETF share class VTI) could be redeemed in its entirety and still not realize a capital gain. That’s every investor selling every share, and still no capital gains realized. Even in bull markets, about 75% of the fund would need to be redeemed to trigger a gain. It is highly unlikely that 75% of a $105 billion mutual fund is going to be redeemed.

Could transactions in the traditional shares ever impact the ETF shares? Perhaps — in a fund with a very small asset base that doesn’t have steady cash flows, and if an in-kind redemption wasn’t practical in that case for one reason or another. Is that slim possibility the only factor that should be considered when comparing products? Clearly not. I think that the best way for investors to compare ETFs is not on the merits of one structure over another, or even on whether a fund paid a gain or didn’t. The better measure of success is after-tax return. If you select a fund that has terrible tracking relative to its index and is high-cost, but didn’t happen to have a capital gain, what does that tell you? Not much.

Finally, the share class structure. It gives them a leg up on benchmark tracking. With an established base of assets, exchanged-traded shares can track with a greater degree of precision because they own significantly more securities than ETFs that do not have a critical mass of assets. The funds can fill in around the creation basket, buying other securities with the cash flow from the conventional share class. Cash flows into the underlying fund also give the portfolio manager greater investment flexibility to adjust the portfolio for benchmark changes. Frequently, the fund can be re-targeted without having to sell any stocks, just by using incoming cash. On the other hand, stand-alone ETFs may have to sell securities in order to purchase new index entrants, and weight their portfolios appropriately because all of their cash flows are paid in kind, not with real cash.

Thank you kind stranger! Here’s hoping you weigh in again in the future, because you are always welcome to do so, as are all readers. Cheers!