Introduction

From summer 2007 to June 2009, US financial markets flirted with disaster. The housing market had collapsed and, because many major financial institutions were heavily invested in mortgages, they lost hundreds of billions of dollars. The economy went into recession in December 2007 and contracted a full 5.1 percent in 18 months. Millions lost their jobs; other millions lost their homes. At the height of the crisis, experts and ordinary citizens alike wondered whether the nation’s banking system would survive.

The US Federal Reserve System, the nation’s central bank, stepped in with a variety of programs to stabilize the banks. While the public knew that the “Fed,” as it was called, was taking action, the agency released few details about individual transactions. At Bloomberg News , investigative reporter Mark Pittman became convinced that banks were using Fed loans to disguise financial weakness. In May 2008, he filed a Freedom of Information Act (FOIA) request with the Fed to learn how much it had loaned each bank, and on what terms. [1] When after six months the Fed failed to respond, Bloomberg L.P. (the parent company) in November 2008 filed a lawsuit to obtain detailed records about its massive lending activities.

Over the next two years, the two parties fought a legal battle. Bloomberg won at both the district and appeals court levels, but the Fed or its proxy continued to appeal the decision until, in March 2011, the Supreme Court refused to hear the case. The lower court ruling—favorable to Bloomberg—stood. The Fed had already released a portion of the requested documents in December 2010. In March 2011, it released the rest—29,000 pages of PDFs. [2] The second tranche included key details about the amounts the Fed had lent through its so-called “discount window” to Morgan Stanley, Citigroup, Bank of America and other financial institutions from August 2007 to April 2010.


Mark Pittman
© Bloomberg

A team of Bloomberg editors and reporters had to sift through the data (Pittman had died in November 2009).The process was technologically complex, intellectually overwhelming and a logistical nightmare. They had to make sense first of the acronyms and jumbled numbers in the Excel spreadsheets that the Fed released in 2010; then “scrape” data off the March 2011 PDFs. But by summer 2011, the reporters had extracted a treasure trove of stories from the vast wealth of information the Fed provided.

In spring 2011, differences arose about how to add up the numbers. Numbers pose a consistent conundrum for business journalists; their interpretation can put a story on the front page, or bury it deep in the business section. At the same time, business reporters—like all beat journalists—are dependent to some degree on maintaining good relations with the institutions they cover. Unjustified sensationalism can come back to haunt them the next time they need a reliable source inside the financial world.

Veteran reporter Bob Ivry argued that it was only fair, and reflected the ongoing gravity of the situation, to report the cumulative borrowings of each bank over the months of the financial crisis. Especially for headline and lede paragraphs, he wanted to see a large number that would attract wide attention to a story of vital public interest. Ivry’s colleagues, however, argued that the cumulative number was deceptive if in fact a bank had merely borrowed and repaid the same sum repeatedly over those same months.

The question was: what would be the most accurate calculation? There would be multiple stories based on the Fed data; they had to agree on a consistent approach to the numbers. How could Bloomberg best tell the story of how the Fed—unbeknownst to the public or Congress—had floated some of the nation’s largest and most influential banks? In July, as Bloomberg prepared a series based on the Fed documents, Ivry once again raised the issue: cumulative or discrete numbers?


[1] FOIA requires federal agencies to make government documents, with some exceptions, available to the public and the press.

[2] PDFs are electronically locked documents that cannot be altered without considerable effort, and which are difficult for software programs to parse for data.