Bloomberg fought and won a lengthy case against the Federal Reserve to obtain the details of the various US bailout programmes. That was back in the summer. FT Alphaville salutes them for it, as have many others.
But when Bloomberg came out with their latest story using the data, we wondered how they calculated the $13bn of “profits” banks earned on bailout transactions.
Headlines using aggregated numbers warrant scrutiny. And we’d like to give a second hat tip to Bloomberg for kindly, and transparently, sharing their methodology with us.
First, here are the Fed programmes that their figures included:
– Discount window
– Commercial Paper Funding Facility
– Primary Dealer Credit Facility
– Term Auction Facility
– Term Securities Lending Facility
– Single-tranche open market operations
The Asset-Backed Commercial Paper Money Market Mutual Fund Liquidity Facility wasn’t counted on the basis that the banks didn’t tend to “use” the money, but rather passed it on to funds. (We suspect it was also because no-one could say it without taking a second breath.)
Now, obviously the amount that any one bank borrowed from the collection of facilities is going to vary over the period of study — that being from August 2007 to April 2010.
Indeed, Felix Salmon has a chart of how much it varied for Morgan Stanley, in his brilliant post about what it looks like to have a lender of last resort, i.e. to not be in Europe:
Felix on the above chart:
The orange line is the amount that Morgan Stanley owed to the Federal Reserve on any given day — an amount which peaked at $107 billion on September 29, 2008. And the red line is the ratio between the two: Morgan Stanley’s debt to the Federal Reserve, expressed as a percentage of its market value. That ratio, it turns out, peaked at some point in October, at somewhere north of 750%.
The tricky thing about the Bloomberg calculation is how to go from amounts borrowed to profits made.
To arrive at a guesstimate, the analysts dug up the “net interest margin” for banks. For some, this was available quarterly, for some semiannually, and others just annually. They always used the highest frequency data available.
To take an example, let’s look at how the data looks for Bank of America, using the latest 10-Q (click to expand):
The net interest margin takes the interest that can be earned on investments (hence presumably a huge range of various activities, some risky, some less so), subtracts the firm’s cost of funds (presumably a blended rate across different types of debt at various maturities), and then divides by interest earning assets.
The bank cites the low interest environment for the recent decline in the above table, and also gave the following explanation:
The decrease was primarily due to lower consumer loan balances and yields and decreased investment yields, including the acceleration of purchase premium amortization from an increase in modeled prepayment expectations and increased hedge ineffectiveness. Also negatively impacting net interest income was lower trading−related net interest income. Net interest income benefited from ongoing reductions in long−term debt balances and lower rates paid on deposits.
Probably not the best thing that Bank of America is reducing its longer-term funding, but back to the task at hand, the point is that net interest margin is an average of an average of an average, and so on. We’re pretty sure that there’s an analogy of a mix-tape in there, but we wouldn’t want to lose our younger readers over it.
The analysts at Bloomberg then match the net interest margin with a calculation for the daily average amount borrowed from the Fed for the period in question. Taking such an average is also a simplification, of course.
Furthermore, the cost of using funding from the Fed’s various programmes will differ from the bank’s own cost of funds, which is baked into the net interest margin. A further operation that an analyst could perform would be to back out an estimate of the bank’s cost of funds and apply an estimate for the blended cost of funds of the Fed facilities. But then, the analysts probably have families to go home to.
Net-net, the Fed will quite likely have been cheaper
than bankruptcy over the period under study, hence the Bloomberg guesstimate is probably on the conservative, i.e. low, side. That said, it is impossible to know without having more information of what the funds were used for.
FT Alphaville doesn’t mean to take methodological pot-shots at the Bloomberg analysts. As we said, their fight to obtain this information was admirable and they are transparent about the way they produced the estimate.
When estimates go in headlines though, they sometimes are in danger of being confused with facts. Headlines with footnotes don’t fly. We admit, however, that “Fed Loans Could Have Theoretically Earned Banks $13B* in Net Income (*Rough Estimate)” isn’t nearly as catchy as “Secret Fed Loans Gave Banks Undisclosed $13B”.
And the story of how these facilities came to be, who used them, under what terms, and so on, is something the public should know about. Go read about that in the Bloomberg story, because there’s plenty of that in there.
Here are the top 15 banks when ranked by the amount of theoretical money they could have made according to the analysts’ estimates:
The presence of foreign banks is notable, isn’t it? Perhaps something worth keeping in mind when the US asks Europe to please get their house in order?
Secret Fed Loans Gave Banks Undisclosed $13B – Bloomberg
Bloomberg Report Exposes the Federal Reserve Doing Its Job – The Atlantic
Chart of the day, Morgan Stanley bailout edition – Felix Salmon, Reuters
Break ‘Em Up Now, We’ve Seen Enough – The Reformed Broker
REVEALED: More Details On The Fed’s Breathtaking $7.7 Trillion In Loans To Large Banks – Business Insider
Multi-Trillion Bank Bailout Leads to Multi-Billion Bank Profit Bloomberg Finds – Zero Hedge