As of today, if I’m reading the various numbers right, the Fed discount rate is 0.75%, and 30-year T-bills are at 3.37%. Presumably this arbitrage is still happening, then, right?
Has there been any point in the history of the Fed when this set of transactions could not have been run at a profit for the banks; i.e., when the discount rate was equal to or greater than a long bond? If not, why is this only controversial now?
Is it controversial now? AFAICT, it is merely being pointed out. And Hellestal detailed upthread that Stiglitz didn’t mention the discount window and was probably referring to an array of emergency programs carried out under Bush/Obama/Bernanke.
I’ve been trying to keep the answers in this thread simple so far, but this next is going to have to be one step more complicated.
First, I want to emphasize again that the discount window is not especially relevant. Stiglitz did not mention the discount window. He mentioned general borrowing from the Fed. The OP falsely assumed that borrowing from the Fed automatically meant using the discount window, presumably because of a lack of familiarity with these other options. But as I already cited in Post 4, the Fed during the crisis had a large number of lending facilities to rely on in order to prop up the markets. If we want a specific example, we can do that. This is the interactive Fed lending graph from Bloomberg I already cited. You can click on any bank to get a breakdown of the different Fed facilities the banks used. We can look at JPMorgan for one example and see that they were in hock to the Fed for 525 days with an average outstanding balance of 12 billion. There are little buttons you can see, lists of acronyms, where you can break down the most important lending facilities. The most important here were the TAF and the AMLF. DW (discount window) hardly shows up at all. Borrowing for cheap.
Now I say “for cheap”, but what does that really mean? You’re asking why banks wouldn’t do the same sorts of trades today that they did in 2008-09, and the answer is that the markets are much different today.
I’ve already used the example of someone from the government coming to you and lending you a million dollars at near 0%, and then the next today someone else from the government coming to borrow money from you at 2%. What I didn’t do (because I was trying to keep things easy to follow) was to calculate the marginal benefit of this government largesse. The total spread is 2% on a million dollars, or 20k. But that doesn’t quite tell the whole story because banks might have some alternate sources of funding. What matters is the cheapest alternative source of funding with basically identical conditions. If public borrowing is 1% under the same terms, then my marginal benefit from uncle sugar’s visit is the additional 10k.
So now we can answer that question. Why aren’t banks borrowing from the discount window at 0.75% in order to lend at 3.77%? Well, why would they do that when they can borrow on the repo markets (another form of collateralized borrowing) for around 0.13%? We are past crunch time in the financial markets, and private lending is competitive again. So I have no doubt that there are at least some people out there doing exactly what you suggest: borrowing short on the overnight repo market (and then rolling over the borrowing indefinitely) in order to invest long in 10- or 30-year bonds.
(If you’re wondering why more investors aren’t doing this, it’s because of the risk. If interest rates rise, then the overnight rate will head up while simultaneously the price of these long bonds will plummet. You get screwed at both ends if you time it wrong. Eventually investors reach a point where the risk no longer seems worthwhile given the interest rate spread between the assets.)
We should see now that this isn’t quite the correct calculation. We’re not just looking at the spread between the discount window (or whatever) versus the long bond. These are different kinds of loans. It’s also a question of risk, and we have to look at other similar short term forms of collateralized borrowing like the repo market. When comparing different short-term borrowing options, it’s not enough to compare the rate in isolation. We have to compare the conditions in borrowing. I’ve already linked this, but now it’s worth revisiting in more detail:
The Fed loosened the conditions for borrowing during the financial crisis.
Of course they did.
It’s not.
It’s not only controversial now. This complaint has been circling from the very beginning when people learned what the Fed was doing. The only thing that’s special about this thread is that the OP has only now, for the first time, read about this issue. In contrast, anyone who was following the business news should already be completely familiar with these complaints. Here’s a nice article from 2011: “Wall Street Aristocracy Got $1.2 Trillion In Secret Loans”. (More generally, Walter Bagehot was complaining about overly easy lending from central banks in the 19th century. This is a recurring problem.) To emphasize that it’s the marginal difference in rates that’s important, I’ll quote this selection.
This says “28 day loans”, but I want to emphasize again that loans can be and were rolled over. This means we have to be careful when we’re talking about the term of the loan. Just because a loan is overnight doesn’t mean it will be paid back tomorrow. It might be rolled over. Just because a loan is four weeks doesn’t mean it will be repaid after a month. It might be rolled over. I’ve said this several times before in this thread, but here’s an example of why this is important:
So let’s take a breather…
Whew! Okay.
In all of my previous examples, I was trying to keep calculations simple to get the basic point across. I was using examples of borrowing at 0% and lending at 2% because that’s the easiest example to follow. The reality is that we have to look at a more complicated calculation. Naturally, people have tried to guesstimate the benefit to banks of all this Fed lending.
As they say, this is not a perfect calculation. If I had to go through the numbers myself, I might’ve used a slightly different technique than they used here. But if I were in Stiglitz’s place? If I were writing a popular piece to describe what happened? I would do exactly what he did. It’s fully accurate in the essentials: banks were borrowing low from one part of the government, and then lending back to the government at higher rates. That’s not all they were doing, of course. But any bank that had satisfied its day-to-day liquidity problems would be looking for a place to earn a nice spread, and lending back to the government was the safest option at the time. This description avoids most of the technicalities, but it gets the big idea literally correct.
Stiglitz’s description is just… innocuous. It’s unobjectionable. What he described literally happened. Short and non-technical and to the point, and I’d write the same thing in his position.
Why recapitalize banks “on the sly”? Why not just give them cash and save the paper-pushing costs? We were already bailing them out so it’s not like this secrecy avoided a public backlash or anything.
Assuming Citi did what you say they did and the above quote agrees with your methodology Citi earned back only $.18 billion of the over $13 billion interest they paid on their TARP loan. We know the loan programs have ended now or in case of the discount window nearly non-existent.
So Stiglitz is saying banks like Citi (the largest borrower of the nine TBTF) recouped the government’s “investment” via some Fed program. Using the quote you provided this is clearly not true.
Because it is not happening “on the sly”. It happened like this:
Buffett cleaned Goldman’s clock.
Why on earth would banks borrow at 6% and higher (which they did) when they could get “free money” from the Fed?
… I’m not really understanding how that is relevant (I am not a financial savant.)
I repeat – Why on earth would banks borrow at 6% and higher (which they did) when they could get “free money” from the Fed?
The answer is that Stiglitz is wrong. There is no free money from the Fed unless you want to repay it the next day (or in 90 days back in 2008-09).
The TARP banks were public knowledge. The Fed loans were different in the beginning.
The bank names were entirely hidden until the courts decided to open things up. So each individual bank not involved in something public like TARP could claim that they had not taken any “bailout” cash, when in fact they were borrowing secretly from the Fed at below market rates. This was a huge part of the scandal when the Fed released its info. Banks that had claimed to be perfectly safe were still getting funding.
Two options here: 1) They were lying about how safe they were. 2) They were taking advantage of the gift. Of course, different banks were in different situations, but even more interesting is that those options are not mutually exclusive. A bank that needed emergency funding to stay alive during a brief crisis could easily see how lax standards were, and so continue its borrowing for an extra profit after it was out of danger. I would do the same thing in their position. It’s just good business. (Poor public policy outside of a crisis, but good banking.) Obviously some of the lending was used for emergency measures, but how much? That’s a harder question to answer.
One other perceived problem about Fed lending was the size. The Fed lending was around 1.2 trillion dollars. There were no statutory limits, so it could have gone even higher. Compare again to TARP, approved by Congress, which was originally 700 billion.
Those programs ended long ago and were also short term.
“The Term Auction Facility (TAF) is a temporary program managed by the United States Federal Reserve designed to “address elevated pressures in short-term funding markets.”[1] Under the program the Fed auctions collateralized loans with terms of 28 and 84 days to depository institutions that are “in generally sound financial condition” and “are expected to remain so over the terms of TAF loans.” Eligible collateral is the same as that accepted for discount window loans and includes a wide range of financial assets.”
Yes, like the discount window extension of 90 days, theoretically banks could have played your 3-card monte game with TAF loans and shuffled those short loans into Treasury long bonds to capture a single coupon.
It would seem to me that these events would not necessarily be mutually exclusive due to funding limits. I am making this point out of context, but it seems your logic is flawed here. Only if the funds from the fed were limitless would this be a logically cohesive assertion as far as I can tell.
JPMorgan had TAF loans outstanding for six months. They had AMLF loans outstanding for most of nine months, with a brief interruption. This information is in the interactive graph I already cited. You can click on different banks to get their loan information. Citi had TAF loans for about nine months. Bank of America had TAF loans for about nine months. Wachovia had TAF loans for almost a year, with one interruption (not that it helped them).
I mentioned previously in this thread how banks might continue to enjoy borrowing privileges after the first loan term expired. Post 2:
Post 4:
Post 23:
The data that confirms this is right there in the graphs. Banks were borrowing under TAF far longer than three months.
This is similar to the repo markets on which Lehman was relying for its short-term financing. Repo loans are technically overnight loans, but banks might need funding more than 24 hours. As long as both parties agree to renew the loan, then the short-term funding can extend indefinitely. The risk with this kind of short-term arrangement is that the lender can renegotiate terms, or demand full repayment, or demand that the borrower provide more collateral to continue the arrangement, on any day they wish. This is exactly what happened to Lehman in 2008.
After the shock, the government’s sort of unofficial motto was “No more Lehmans”. The Fed was not going to cut off its funding so suddenly as Lehman’s creditors did.
This is mostly agreeable with my understanding of TAF. However, there are two problems.
1- There is no evidence TAF loans were used to buy Treasuries.
2- The interest rate was typically too high to buy Treasuries.
See:
http://publicintelligence.net/term-auction-facility/
The use of TAF does not help Stiglitz’s claim in the least bit.
You just copy&pasted the result of a single auction from March 2008 (before Lehman) that they were using as an example of the auction process.
The page has a lot of additional information about subsequent auctions.
If we were to scroll down, we would see a small table that shows interest rates from the TAF near the end of 2007 and the beginning of 2008. The table clearly shows that interest rates for these loans were dropping as the housing collapse became more severe. Specifically, the minimum bid was linked to the OIS (overnight indexed swap). As the OIS declined, so did the minimum possible bid for the auctions. Obviously, the interest rate was not a constant. It was going down during the early stages of the crisis, exactly as we should expect. But that’s not the entirety of the information from this page. There is more.
If we continue to scroll down, we will see the results from the subsequent year, the 2009 auctions, which is to say the interest rates for TAF loans when Obama was first in office. As far as I can tell, your cite gives the interest rate for every single TAF auction held in the entire year.
The interest rate for every single auction listed in your cite for 2009 is 0.25%.
Like Stiglitz, I would call this rate close to zero. I was actually prepared to link the Excel file from the Fed itself to demonstrate that the vast majority of TAF loans had been issued at a mere quarter point, but I have to admit, my contribution was not quite as good as this page. The Excel file is much harder to navigate and to read. The page you offered is a much clearer demonstration.