With time, the disparate pieces of information contained in the Fed datafiltered through the spreadsheet and entered into the graphical analyzercombined to reveal significant new facts about the pace, level and scale of borrowing by banks, both individually and in total. At different times and in a variety of conversations, Editors Friedman, Voskuhl and Bilodeau plus reporters Ivry, Kuntz, and Keoun debated how best to present the new information. The data visualization tool would document day-by-day and peak borrowing by individual banks across all seven lending programs. But what about the accompanying stories? What would be their emphasis? Kuntz says:

What everybody was doing when they got the data was just basically adding up all the loans that were given. Brad [Keoun] and I decided very early on that the number that really mattered was not how much a bank borrowed in aggregate, but how much they owed every day during the crisis. If you borrow a million dollars a day and pay that million dollars back every day [and do that] for a month, some people would say youd borrowed $30 million. I think thats ridiculous. Thats just a rolling line of credit. Youve only borrowed a million dollars. Youve been in debt to the tune of a million dollars for 30 days, but that doesnt mean youve borrowed $30 million.

Ivry disagreed. He wanted to present the cumulative number. He thought the data visualization tool was a great idea, and acknowledged that of course any story would include individual bank borrowing numbers. But he hoped that the headline, at least, would take advantage of the magnitude of the aggregate number to attract public attention to the grim story about how close to collapse several banks had come. At first, Editor Friedman agreed. I wanted to go with the bigger number! he says.

Because bigger numbers are sexier. Trillions are better than billions. Thats just the instinct. But its important to think about numbers. I have in my career spent a lot of time thinking about the use of numbers and how theyre misused or used correctly Numbers are not always your friends. They have to be beaten into shape.

Friedman on how reporters and editors should think carefully about how to use numbers.

But Keoun, using an argument similar to Kuntzs, persuaded Friedman that daily outstanding balances were more accurate. Others agreed. Ivry found himself alone in his view. There the matter rested.

Discount window data . By March 2011, Bloomberg was close to publishing a preliminary series of stories about individual banks and their borrowing records. They continued, however, to hope for the discount window information, which would complete the picture and make it more accurate. On March 21, 2011, the Supreme Court rejected the October 2010 appeal from the Clearing House Association and refused to hear the case. That left intact the March 2010 federal court decision, which gave the Fed five days to release the data that Bloomberg had requested nearly three years earlier. The order, noted Ivry and colleague Greg Stohr in a piece for Bloomberg News , marks the first time a court has forced the Fed to reveal the names of banks that borrowed from its oldest lending program, the 98-year-old discount window. ^[39] After the decision, Bloomberg Editor-in-chief Winkler said in a statement:

The Federal Reserve forgot that it is the central bank for the people of the United States and not a private academy where decisions of great importance may be withheld from public scrutiny. As only Congress has the constitutional power to coin money, Congress delegates that power to the Fedand the Fedmust be accountable to Congress, especially in disclosing what it does with the peoples money. ^[40]

Kuntz says they decided to hold off publishing the stories in development in order to include the discount window data. But there was a catch: the Fed had released the initial December 2010 information in Excel spreadsheets, which made it relatively easy to manipulate the data. When it finally released 29,346 pages of discount window data on March 31, they were in PDF format. Whats more, the Fed provided data for the duration of the financial crisis, from August 2007 through April 2010. Recalls Ivry:

We couldnt add up the numbers. We couldnt sort them. We couldnt crunch them. We couldnt separate what we wanted. We couldnt say Goldman Sachs did this. In PDF form, they were basically worthless to us.

So Kuntz, as before, sent the PDFs to Bloombergs data office in Princeton. The data team scanned the documents, and within a day managed to scrape the data off them using character recognition software. But as with the original files, recalls Kuntz, that presented us with a whole other level of inscrutable bank names, inscrutable subsidiaries, and inscrutable data that we then had to fold into what I had already built in the spreadsheet to again add those totals to the totals that we already had. He adds:

I was thinking, when is this going to end? I thought that constantly. I mean, I knew we had a good story and it was a no brainer that we had to wait and get the discount window stuff in there because it would be, like, why are we telling people everything but the discount window, which is one of the biggest indicators of a banks poor health?

Many of the teams early stories were about the release itself. On April 2, for example, Ivrys reported on the FOIA process and the document release, and concluded that making the information public had not in fact compromised the reputation of the banks involved. ^[41]

Cumulative used . At least two stories raised anew the question of cumulative versus discrete numbers, and Ivrys view that aggregate numbers carried a unique punch prevailed. In the March 21 story about the impending discount window release, Ivry and and Stohr noted that the new data would give taxpayers insight into the Feds unprecedented $3.5 trillion effort to stem the 2008 financial panic. ^[42] Then on April 1, Ivry and another colleague published a story about Fed lending to Arab Banking Corp., then 29 percent owned by the Central Bank of Libya (under the rule of strongman Muammar Quaddafi. ^[43] The article reported that Arab Bank had aggregate borrowings of $35 billion. It contrasted that number with the banks largest single outstanding loan amount of $1.2 billion, in July 2009.

Ivry filed another article on May 26 that was typical of the kind of coverage Kuntzs evolving master spreadsheet and the Fed data releases were able to generate. The story described lending under Single Tranche Open Market Operations (STOMO), an $80-billion program whose details werent revealed to shareholders, members of Congress or the public. ^[44] The Fed charged interest on STOMO loans as low as 0.01 percent, although its main lending facility interest rate stood at 0.5 percent.

The existence of STOMO had come as a surprise to the Bloomberg team, which affectionately dubbed it St. Omo. As it turned out, they were not alone. Congress had overlooked the program when, under Dodd-Frank, it required the Fed to publish emergency lending data. In preparing the May 26 story, Ivry phoned Congressman Frank, co-sponsor of the Dodd-Frank act and former chair of the House Financial Services Committee, to ask about it. I said, Representative Frank, have you ever heard of Single Tranche Open Market Operations? recalls Ivry.

Nope, never heard of it. Did you know that it was used as an emergency lending facility for 21 banks? Nope. That was my lead quote when I wrote about it. Not even the chairman knew. So you see, I mean, who knows what else is there?