We caught some heat around here when Dave Nadig called the Nasdaq-100 the worst index in the world.
There's no denying that substantial assets track the NDX, and specifically, that the PowerShares QQQ Trust (NasdaqGM:QQQ) is one of the most popular trading vehicles in the world. It's not often we get an object lesson in how the pros manage a major ETF portfolio, so we followed with significant interest the handling of the rebalance by the trustee of the Q's, BNY Mellon.
But first, a little background on how ETFs really manage their portfolios.
Every morning, before the open, ETFs provide at least one basket to the National Securities Clearing Corporation-a creation basket. That basket is a list of shares that the ETF issuer will accept at the close of that day in exchange for new ETF shares. ETF issuers can also provide additional information to help authorized participants manage their risk. That additional information comes in the form of a completion file-everything you'd need to add to the creation basket in order to mirror the intraday net asset value of the fund. Collectively, this set of securities and cash positions is often called the "calculation basket." (Occasionally, an issuer will also supply another basket-the redemption basket, if the securities the firm wants to eject to redeemers varies at all from the creation basket.)
Most of the time, none of this is particularly interesting. APs use the creation basket to model the intraday fair value of an ETF, so that they can be sure of having their positions in order at the end of the day.
But don't issuers already provide this value as intraday net asset value (iNAV?)
Yes and no.
Here's a little-known fact about mutual fund accounting:If an ETF or mutual fund buys 100 shares of IBM today at noon, the new position for your IBM trade, for example-that you and I would recognize immediately in our Schwab accounts if we were buying it on our own account-is NOT included in tonight's net asset value. Instead, it's included in tomorrow's NAV (T+1). That means that the net asset value and portfolio you're looking at for most funds is actually a day lagged.
Most of the time, this doesn't matter much. After all, the composition of most indexes doesn't change that often, or even that much, and the tracking error induced by reporting whatever was in the portfolio instead of IBM is limited to how different that old asset is from the new asset.
But things get more interesting in ETFs. Let's take a look at the Nasdaq-100 rebalance that occurred on April 29, and the trading that the "Q's" were forced to engage in.
As we covered here two weeks ago, the Q's (and any NDX tracking investment) were forced to sell Apple (Nasdaq:AAPL) down from 20 percent of their portfolio to 12 percent, and increase their Microsoft (Nasdaq:MSFT) position by 5 percent, and increase a large number of other positions along the way.
The timing of the trade is important. Nasdaq announced that the new weights would be effective on May 2. That means portfolios tracking the index needed to have a 12 percent position in Apple on the open of trading on that day. As a portfolio manager, that means you needed to change your portfolio, essentially in an instant, from the close on April 29.
In this case, T+1 accounting works in the Q's favor. They had the ability to make their trades any time they liked on April 29, knowing that the NAV of the fund wouldn't show the results of that trading until Monday the 2-precisely when the new closing level of the NDX would be live.
But how, exactly, would they make this trade? Would the Q's still be taking orders for creation units loaded up with 20 percent Apple on April 29, knowing that the fund would need to unload that Apple on the close? Or would the team at BNY Mellon that manages the Q's portfolio put forth a creation basket devoid of Apple, knowing they would already have too much Apple at the end of the day?
In the end, the creation basket on April 29 was essentially the full portfolio you'd need to calculate that day's Apple-heavy NAV. And a not-insignificant number of shares were in fact created-over 27 million, for a value of $1.6 billion. Of that $1.6 billion that was delivered, roughly $320 million was Apple stock, or about 1 million shares.
The team at BNY Mellon that handles the portfolio management of the Q's for PowerShares chose to deal with this trade internally, rather than trying to game the creation basket. However, they did give themselves room to work the trade. On April 29, the Q's closed creations at 1 p.m.-2 hours earlier than the regular 3 p.m. cutoff-so that the ETF's portfolio team could work the rebalancing trades. Because the underlying is so liquid, there was minimal risk that the Q's would trade at too much of a premium, given that everyone assumed (correctly) that normal creations and redemptions would resume on Monday.
So what happened at the close? iNAV was only a penny off the actual closing NAV of the Q's at the end of the day-fully representing the 20 percent Apple position, while the last-traded price on the consolidated tape of $59.13 diverged slightly. The 15 basis points premium, although larger than usual for the Q's, was nothing to write home about. So what's the report card for the Q's on the trade? Did the extra work they made for themselves cost them anything meaningful?
And finally, what about market impact? What can we learn from the tape to answer the question we posed last Friday-how much money is really indexed to the NDX?
Based on trading volumes, we know what the largest outside number could be. If the average trading volume in AAPL was roughly 12 million a day going into the trade, and the actual shares traded were roughly 36 million, we could say that the NDX trade was the difference-some 24 million shares. Of course, this is a gross oversimplification. Not everyone trading Apple last Friday was an indexer, and the fact that investors added $8 billion of new volume on Friday suggests a $100 billion asset base (with apple receiving an 8 percent haircut).
Another way of looking at it is from market impact. Last week, we used various common market impact models to estimate the downward movement that different sell orders coming into the market would have on AAPL. If we picked 1 p.m. as the "game-on" starting point-when the Q's officially closed creations and told the market they were getting to work-Apple came down just under 1 percent.
And finally, what has the market impact on AAPL been since the announcement of the rebalance on April 5?
Over the full period, nothing. AAPL's earnings announcement dwarfed any meaningful impact. At worst case, if we attributed the entire market move from the date of announcement to the pre-earnings low, AAPL lost roughly $10 billion in market cap; which, again, reverse-engineering from the 8 percent reduction Apple faced in the rebalance, puts the index-effect at perhaps just north of $100 billion. Such an assumption, however, seems substantially overblown, and we mention it mostly out of interest.
The end result of the analysis, however, leaves us unconvinced that $300 billion in NDX assets affected a trade in the underlying stocks in and around the close.
Our conclusion here is this:Regardless of whether NDX is a good index or a bad index, the team at BNY Mellon that manages the Q's portfolio for PowerShares handled the extraordinary rebalance trade efficiently and with skill, providing investors with precisely the experience they would expect over any period-from intraday to the long term.
Copyright ® 2011 Index Publications LLC . All Rights Reserved.