Connect the Stocks

It's December. The share prices of two companies — say, Coca-Cola (KO) and Walt Disney (DIS) — have both dipped in trading. Why? If there's no apparent change to the underlying value, the answer may be found in stocks' list of institutional holders. If KO and DIS have many of the same owners, the December dip may be explained by mutual funds' sell-offs.

Perhaps it's time to buy KO and/or DIS to reap gains later in the New Year?

A study by IESE's Miguel Antón and Christopher Polk of the London School of Economics builds on and advances the existing evidence that mutual fund flows put significant price pressure on stocks.

In their article for the Journal of Finance, the authors connect stocks through their active mutual fund ownership. Pairs of connected stocks are then tracked in trading and found to have abnormally high correlations in their movements.

The authors propose a trading strategy based on these high correlations. By buying connected stocks whose partners have recently fallen in trading and selling connected stocks whose partners have recently risen in trading, the authors calculate returns in excess of nine percent per year.

Connectedness and Comovement

Using U.S. stock market data from 1980 to 2008, Antón and Polk focus on larger stocks — i.e., those with market capitalizations above the New York Stock Exchange (NYSE) median value — for their sample. They then look at the mutual institutional owners to connect thousands of pairs. A typical pair in their sample has about nine funds in common.

The authors control for stocks' similarity in industry, size, book-to-market and momentum characteristics as well as the degree of common analyst coverage. With these controls, the authors aim to isolate shared mutual fund ownership as the cause of comovement and not, for example, the fact that two stocks are both in the banking industry during a credit crunch.

But what about the possibility that fund managers have common criteria for their portfolios? For example, might it be that the manager of a large-cap growth fund chooses to invest in companies that have similar fundamentals?

To rule out the possibility that fund managers' criteria is behind comovement, the authors turn to the 2003 mutual fund trading scandal.

The 2003 Mutual Fund Trading Scandal

In 2003, allegations of illegal trading hit the world of mutual funds hard. New York Attorney General Eliot Spitzer went after those that were believed to be engaging in market timing and late trading. (Late trading involves placing orders to buy fund shares after the market closes while still getting that day's closing price instead of the next day's opening price. Late trading can be a lucrative short-term strategy when funds' share prices move on after-hours news.)

In the aftermath of the scandal, 25 fund families settled allegations that they had traded illegally. Investors shunned those 25 fund families. In fact, the 25 lost an estimated 14 percent of their capital within one year and more than 24 percent within two years. At the same time, funds not implicated in any wrongdoing gained almost 12 percent in capital. The common ownership of stocks in the authors' sample was impacted significantly by the scandal.

The scandal's aftermath presented a natural experiment to test the strength of the link between common ownership and comovement.

Antón and Polk find the evidence is there to prove their point: common ownership is responsible for correlations in trading. They also find that comovement is stronger when the stocks' common owners are experiencing heavy net outflows or inflows and when the stocks in the pair have relatively low floats.

A Connected-Stock Trading Strategy

If, to a certain degree, movements of connected stocks are predictable, locking in some trading gains should be, too.

To show this, the authors calculate what happens if they buy or sell stocks that have been temporarily pushed away from their true value by mutual fund trading. They find trading deviations from true value last for the next six months. Controlling for market size, momentum and other factors, the authors report returns in excess of nine percent per year.

Note that calculated returns do not factor in trading costs, which might be significant for a relatively short-term trading strategy.