People working for the U.S. Securities and Exchange Commission who owned stock in companies under investigation were more likely to sell shares than other investors in the months before the agency announced it was taking enforcement actions, according to a new academic paper.

SEC employees holding shares of five firms including JPMorgan Chase & Co. and General Electric Co. in 2010 and 2011 sold stock in 62 percent of the trades they initiated, compared with 50 percent among all the investors who traded those shares in that period, Emory University accounting professor Shivaram Rajgopal reports in the paper.

Rajgopal, who plans to present the work today at a University of Virginia accounting seminar, said in a telephone interview that while the analysis doesn’t prove misconduct it points out a suspicious pattern.

“It does suggest it is likely, or probable, that something is going on,” he said.

The records, obtained from the SEC under a Freedom of Information Act request by Rajgopal and his co-author, Roger M. White, a doctoral student at Georgia State University, don’t identify individuals.

The limitation means the researchers couldn’t tell if an individual trader made or lost money in a transaction. They also couldn’t discern if those trading worked in jobs where they might have advance knowledge of actions that could push stock prices lower.

Rajgopal and White analyzed records of 7,200 trades from 2009 to 2011. Since 2009, most of the SEC’s 4,000 employees have had to report their investments and trades to the agency.

‘Circumstantial Smoke’

“All we can do is show some statistical, circumstantial smoke, and we don’t know if there is really a fire,” Rajgopal said. Still, he added, “It’s not clear why we should find this pattern by sheer chance.”

John Nester, an SEC spokesman, declined to comment on the study.

Brad Barber, a finance professor at the University of California-Davis School of Management, cautioned that the study was in preliminary form and hadn’t yet been through the academic review process.

“It nonetheless raises interesting questions if proven to be accurate,” Barber said.

Beginning in August 2010, the SEC’s ethics rules prohibited employees from buying or selling shares of companies under investigation and generally required them to obtain permission before trading. The rules forbid them from trading in any financial company directly regulated by the SEC, such as a bank-owned broker-dealer. The rules also generally require workers to hold any stock they buy while working at the SEC for six months before selling the shares.

Enforcement Actions

The agency said in January it is reviewing the holdings of about 3,400 employees after some of its New York staff was found to own securities prohibited by ethics rules.

The trades involving the five companies were one part of Rajgopal’s and White’s study. According to the data, SEC employees made 87 trades in shares of JPMorgan, General Electric, Bank of America Corp., Citigroup Inc. and Johnson & Johnson in the 90 days before the SEC announced the companies had paid to settle enforcement claims.

In the case of Bank of America, for example, the researchers examined trades in the three months before the firm agreed to pay $150 million on Feb. 4, 2010, to settle claims it misled shareholders about bonuses and losses while acquiring Merrill Lynch & Co.

In the month before the cases became public, more than 70 percent of the employee trades were sell orders. Buy and sell orders from all investors in those companies were evenly divided in those time periods, the researchers found.

Beating Market

Some of the employees’ sales could have been motivated by a need to comply with the ethics rules on prohibited stocks, Rajgopal said. “But if that is the case, you still have to wonder about the timing,” he said.

For the broader analysis of the 7,200 trades, the researchers modeled a portfolio based on the SEC employees’ investment decisions and found their stock trades beat a market index by as much as 16 percent from August 2009 to December 2011. They measured the one-year return for each transaction.

The profits generally didn’t come from picking stocks, with the researchers finding that SEC employees “appear unable to capture gains in their buy portfolios.” Instead, almost all of the excess return stemmed from selling shares that later declined in value relative to the broader market, they wrote.

That finding indicates employees may have sold shares in companies they knew were under a non-public SEC investigation, the authors write.

“They do manage to get out before bad news hits the market,” Rajgopal said.