By: Frank Armstrong
*As seen on Forbes
Suppose I gave advanced notice that on a particular day three stocks were going to have huge buy orders while three others would have enormous sell orders. What might happen?
We might reasonably expect that within milliseconds of that announcement, traders would begin to position themselves for an easy profit.
That’s exactly what happens whenever an index reconstitutes. Staggering numbers of trades must take place on a particular day. These trades will be far in excess of normal volume for those stocks and won’t reflect any fundamental change in the value of the underlying company.
In this case the Dow 30 Index will drop Alcoa (AA), Bank of America (BAC) and Hewlett-Packard (HPQ), while adding Goldman Sachs (GS), Visa (V) and Nike (NKD) on the close of trading September 20.
Because pure index funds must exactly mirror the holdings of their index they don’t have any option but to trade on that day. Their marching orders are to have zero tracking error with the index.
The expectation that an index fund will get EXACTLY the return of the target index sets the stage for an interesting problem: Zero tracking error is achievable, but it comes at a very high price. In order to achieve zero tracking error, the fund manager must hold exactly the index for every second of the day. The managers must do trades NOW, which deprive them of any kind of negotiation power. The cumulative costs of these many trades are not trivial, and will be born by the fund’s shareholders.
Stocks that are leaving the index will be under enormous sell pressure, while stocks being added to the index suddenly become widely sought after. The whole world knows the exact time the substitution must take place. Prices around the trading date go strangely haywire as not only fund managers but speculators place huge orders. In advance of the trade date traders will buy the shares to be added while selling short shares to be deleted. Reversing those trades on the reconstitution date should yield a juicy arbitrage profit.
Stocks on the sell list will be depressed while stocks on the buy list will bounce. A few days later those stock prices will reflect normal volumes and return to the consensus price for their firms. None of this is good for the shareholder. They are almost guaranteed to lose money on shares leaving the index, while they must buy shares at high prices to replace them.
The index fund manager’s problem is that anything less than perfect index replication will result in random tracking errors. The demand from consultants and clients for zero tracking error drives him to a sub-optimal solution. Zero tracking error trumps total return as the performance standard. While it meets the standard of mindless simplicity, in the greater scheme of things this isn’t a particularly rational approach.
This reconstitution effect is widely documented, and a glaring exception to an otherwise efficient market. Speculators routinely make huge virtually riskless profits at the expense of index fund shareholders. While the Dow isn’t as widely followed by institutions or funds as for instance the S&P 500, reconstitution will certainly have an impact. It happens in indexes around the world, sometimes to even greater cost.
Relaxing the definition
So, what’s the alternative, and what might the benefits be? Suppose we relaxed the zero tracking error requirements in favor of avoiding reconstitution drag? A simple solution would be to execute the trades a week after reconstitution date. By that time the dust would have settled. Over time the cost savings will swamp any small tracking error. A number of fund families like Dimensional Funds Advisors (DFA) have adopted similar tactics to enhance performance. For institutions and individuals that value performance over zero tracking error that might be a very good choice.