Trading Patterns Point to Leaks Ahead of Federal Reserve Announcements

(Update at 5 p.m.: Adds comment from the Federal Reserve and changes tense throughout to emphasize that the period under study was before the Fed’s enhancement of security.)

A research paper has found “robust evidence” that for years some traders got early news of U.S. Federal Reserve rate announcements and then traded on it during the Fed’s media lockup.

The paper, covering September 1997 through June 2013, detected abnormally large price movements and imbalances in buy and sell orders that were “statistically significant and in the direction of the subsequent policy surprise.” The moves occurred during the window between when Fed announcements were supplied to the news media and when they were permitted to be released to the public, the authors write.

“Back-of-the-envelope calculations indicate that the aggregate dollar profits … range between $14 [million] and $256 million across the four markets that we examine,” the authors write.

In response, Federal Reserve spokesman Joe Pavel said that the Fed “enhanced its media release security procedures” last October “to better protect the information against premature release.” The Fed used to ring a cowbell to announce when news media could send out reports over open lines. Now it seals reporters off from the outside world during the “lockup” and flips a switch to open the communications circuit when the news embargo is lifted.

The paper is titled “Can information be locked-up? Informed trading ahead of macro-news announcements” and was written by Gennaro Bernile, Jianfeng Hu, and Yuehua Tang of Singapore Management University. It has been posted online on the Social Science Research Network. Bernile, the lead author, said he plans to start submitting it to academic journals in about a month after getting feedback on it from colleagues.

The results could indicate that someone who received the information during the media lockup was leaking it. In an interview, Bernile mentioned a further possibility: Someone who got the information by some other means could have been trading on it during the lockup to throw investigators off the scent. “From a rational perspective, it would make sense to behave that way,” Bernile said.

The trading anomalies that Bernile and colleagues spotted began about 15 minutes before the news embargo was lifted and continued at a fairly even pace. That distinguished them from the burst of trading that could sometimes occur at the moment the news embargo was lifted. Nanex, a firm that analyzes high-frequency trading, reported last year that on Sept. 18, trading in gold futures and exchange-traded funds linked to gold intensified within 1 millisecond of 2 p.m. East Coast time, when the Fed released a statement saying it would refrain from tapering its $85 billion in monthly bond purchases.

As Bloomberg News reported last year:

The quick response is unusual because information takes seven milliseconds to travel to Chicago from Washington, where the Fed statement is released, according to Nanex. The time between Washington and New York is two milliseconds, Nanex said. The trading wasn’t possible unless the statement had been available in those financial centers before the 2 p.m. embargo time in Washington, Nanex said.

The Federal Reserve announced in October that it would tighten its procedures governing the release of Federal Open Market Committee policy statements.

“What is really interesting about our work is that it shows the problem is not a problem of milliseconds,” Bernile.

In addition to the E-mini S&P 500 futures contract, the authors studied the E-mini Nasdaq 100 futures, the SPDR S&P 500 ETF, and the PowerShares QQQ ETF tracking the Nasdaq 100 index. They were unable to get data on over-the-counter markets. Bernile said in the interview that he suspects including them would have boosted estimates of trading profits.

The researchers found no evidence of trading abnormalities during media lockups ahead of the Bureau of Labor Statistics’ release of the monthly employment and inflation reports, or the Bureau of Economic Analysis’ release of the gross domestic product report.

Bernile, 39, received a Ph.D. in finance from the University of Rochester in 2006. He taught at the University of Miami from 2006 to 2013, except for August 2008 to February 2010, during the financial crisis, when he was a visiting scholar at the Securities and Exchange Commission in Washington. Last year he moved to Singapore Management University. A paper he co-wrote on the backdating of stock options was published in 2009 in the Journal of Accounting and Economics.

    Before it's here, it's on the Bloomberg Terminal.