Thursday, March 21, 2013

Open mic night on interest rate spreads

Ok, readers. Here's a chance for you to flex your muscles. The following chart shows various short-term interest rates:


Why are these rates all so different? Can the differentials between them be arbitraged away? What sorts of institutional rigidities might be preventing arbitrage? For instance, we know certain institutions like Fannie Mae and Freddie Mac can't get interest on reserves held at the Fed. What other sorts of fine details might be important? Or are the differentials between these various rates not currently open to arbitrage? Can they be explained by term risk? How much do other sorts of risk, like liquidity risk, counterparty risk, default risk etc drive spreads?  A few specific questions:

a) The DTCC Treasury General Financial Collateral (GCF) repo rate used to trade at or below the fed funds rate. The Treasury GCF repo rate is a collateralized rate. Since collateral reduces risk, it makes sense it would trade below the fed funds rate. But why is the riskier rate now below the safer rate?

b) Why does the Fed funds rate generally trade above the t-bill rate? There's presumably less term risk in the FF rate, which would imply a lower Fed funds rate. Does interbank risk account for the higher fed funds rate?

c) Is it a risk-free trade to fund oneself in the fed funds market and invest the proceeds overnight at the interest rate?

d) Why would banks hold t-bills at all if they can simply keep reserves at the Fed for a superior return of 0.25%? The credit risk seems similiar: as a bank, you're exposed to the Fed in the case of reserves, and the Treasury in the case of bills.

e) Sometimes the 4-week t-bill yield crosses over the 3-month. Why? The credit risk is the same, and presumably bills are equally liquid. Are these inversions purely related to expected changes in yields?

Data, ideas, links,etc all much appreciated. I'm sure I'll have more questions in the comments, or if you have other interesting observations, go for it.

[Update 22/03/2013: I reinputted the Treasury GCF rate since my original data was off]

15 comments:

  1. Perhaps, for some effectively arbitrary reason, certain kinds of common contracts in the relevant markets require (or, less likely, prohibit) the use of certain bonds as collateral (say, 3 months but not 4 weeks) and I think even in general equilibrium that should introduce some difference in the rates.

    Also - if the fed funds rate were lower than the T-bill rate wouldn't you just borrow fed funds and buy T-bills?

    ReplyDelete
    Replies
    1. I'd buy the idea that some bills are more collateralizable (ie. liquid) and this introduces liquidity premia that drives the rate differentials in the charts. I don't know enough about these markets to be sure how large the effect is.

      "Also - if the fed funds rate were lower than the T-bill rate wouldn't you just borrow fed funds and buy T-bills?"

      I agree. But using that argument, if the T-bill rate is below IOR, wouldn't you (assuming you were a bank) just sell all your t-bills and buy reserves? ... and keep borrowing bills and selling them short in order to hold reserves?

      Delete
  2. Sorry, off-topic, but I wonder if you've heard about the soon-to-be-launced Amazon coins. Could this be a real world implementation of McDonald's chartalism?

    http://www.economist.com/blogs/democracyinamerica/2013/02/value-and-virtual-world

    ReplyDelete
    Replies
    1. AMZN coins wouldn't provide the negative incentive for their use (ie "burger tax"), but eventually they might provide a strong positive incentive (preferential pricing vis a vis USD).

      If Amazon lifted some of the announced restrictions (like prohibitions on cashing out the coins and transfers among customers), it seems like AMZN coins could potentially steal some of bitcoin's thunder.

      Delete
    2. Interesting. Could be, I'll have to follow that story. I'm still waiting for airmiles to become more exchangeable.

      Delete
  3. What about transactions costs.

    The T-bill markets are dealers markets. The investor buys at the ask and then maybe holds to maturity and has to roll over.

    The Fed Funds market is a brokers market. A broker has to find a seller (borrower) every day. A wire transaction goes back and forth every day.

    So, if banks are investing in T-bills and selling Federal funds as earning assets, then T-bills have less transactions costs and so lower yields.

    Under "normal" conditions, banks have lots of better earning assets, and the banks, particularly in a zero interest on reserves environment are just managing temporary fluctuations in cash. T-bills won't be held to maturity and will be sold. The bid-ask spread is important. The constant Fed funds fees of relending over and over is less important because you probably will soon stop lending and instead end up borrowing.

    There are no 4 week T-bills, of course. There are various T-bills that are coming due in 4 weeks. Perhaps the transactions costs relate to these as well. If you don't plan hold for some months, having to roll over is costly. If you think you must sell in 4 or 4 weeks, avoid the selling at the bid price.

    That is what I would add to your questions.

    ReplyDelete
    Replies
    1. Hi Bill. Transactions costs is a good theory. I'd buy that.

      But why would any bank hold a t-bill yielding, say, 0.1%, when it could be holding reserves at 0.25%? Presumably these instruments are near substitutes from a credit-risk perspective. Reserves at the Fed aren't brokered, which should mean low transaction costs.

      Why would the spread between IOR and t-bills display so much variance? Let's say that the spread is related to transaction costs (say it is more costly to buy and sell reserves), then a widely fluctuating spread must be related to widely fluctuating transaction costs. I would find it hard to believe that costs are that volatile. Why doesn't the flat-ish IOR force the rates below it to be flat too?

      Delete
    2. On a side note, according to their website, the Treasury does issue 4-week t-bills.

      Delete
  4. Aren't T bill rates lower than interest on reserves because non-banks can not make use of the interest on reserves facility? For non-banks the only safe way to hold money is as T bills. That is why T bill rates sometimes go negative even when banks are getting interest on reserves. I guess Apple or Berkshire Hathaway have no choice but to use T bills as a way to hold billions of dollars in total safety. How would a bank arbitrage between T bills and IOR? All a bank can do is to move over to entirely using IOR and no longer hold T bills. Apple is not going to lend its T bills to a bank for the bank to short T bills versus IOR.
    Hope this isn't drivel, I'm no expert.

    ReplyDelete
    Replies
    1. "All a bank can do is to move over to entirely using IOR and no longer hold T bills. Apple is not going to lend its T bills to a bank for the bank to short T bills versus IOR. Hope this isn't drivel, I'm no expert."

      Lol, don't worry, I'm no expert either.

      You make a good point that a bank can only sell off its existing t-bill portfolio in order to exploit the gap. Why wouldn't shorting t-bills close the gap? A bank can borrow t-bills from AAPL and sell them in order to get IOR. Better yet, it can sell all its existing assets for treasuries, and then repo these treasuries and invest the proceeds at IOR. Why wouldn't banks compete to make these trades until the gap between t-bill/repo rates and IOR has shrunk considerably? They should be arbing this trade until their sole asset is reserves.

      Delete
  5. My impression was that any attempt by banks to buy up the tbills it would just push up the price and so increase the gap. What can the bank exchange for tbills from say AAPL? Bank deposits are worth less than tbills because it would be a massive risk to have a multi-billion dollar bank deposit. Basically tbills are worth a lot to non-banks but for banks are worth no more than bank reserves. But bank reserves are simply not available for non-banks. In principle bank reserves could be converted into shipping containers full of paper cash and that mountain of paper could be used to buy tbills from AAPL but negative interest rates on tbills would need to get huge to cover the logistics cost of that.

    ReplyDelete
  6. On a general note, here's a conversation I had on this topic with twitter phenom Alea that's worth reading:

    Storify link.

    Alea gives plenty of interesting details for the gap between the Fed funds rate and IOR. These include frictions like the fact that foreign banks can hold overnight reserves but don't have to pay FDIC fees, whereas all other banks must pay these fees. He also mentions Fed funds transaction costs, as does Bill W above.

    If Alea is right, I don't see why foreign banks don't currently own every single reserve deposit at the Fed right now. Domestic banks are earning 0.25% minus the 8/9bps FDIC, fee, so if they can lend at a FF rate of 0.22%, say, they'll earn more than if they kept reserves at the Fed. Foreign banks could borrow all these funds at 0.22% and invest them at the overnight rate of 0.25%, earning the spread. Repeat the trade until the spread disappears, or every single reserve deposit has been transferred from domestic banks to foreign banks.

    ReplyDelete
  7. a) The DTCC Treasury General Financial Collateral (GCF) repo rate used to trade at or below the fed funds rate. The Treasury GCF repo rate is a collateralized rate. Since collateral reduces risk, it makes sense it would trade below the fed funds rate. But why is the riskier rate now below the safer rate?
    ****** all balance sheet, funding is expensive now at dealers, regulatory etc.. it costs dealers more to fund so collateral now more expensive.

    b) Why does the Fed funds rate generally trade above the t-bill rate? There's presumably less term risk in the FF rate, which would imply a lower Fed funds rate. Does interbank risk account for the higher fed funds rate?
    **** Not everyone has access to the fed funds market, it's primarily a bank traded market unless you are talking about fed fund futures, which actually really don't yield anything and aren't terribly liquid. but for cash parking purposes t-bills are the only riskless game in town.

    c) Is it a risk-free trade to fund oneself in the fed funds market and invest the proceeds overnight at the interest rate?
    ****not sure what you mean by this one? but lets say you could access cash and you would pay 16 bps or wherever effective traded. could you park that at the fed and pick up IOER? i guess so, but don't think there is alot of demand for cash from banking sector.. everyone long it generally. thats why effective trades below IOER.

    d) Why would banks hold t-bills at all if they can simply keep reserves at the Fed for a superior return of 0.25%? The credit risk seems similiar: as a bank, you're exposed to the Fed in the case of reserves, and the Treasury in the case of bills.
    ***other than inventory to sell to customers i don't think they would

    e) Sometimes the 4-week t-bill yield crosses over the 3-month. Why? The credit risk is the same, and presumably bills are equally liquid. Are these inversions purely related to expected changes in yields?

    **** that curve is pretty flat, but my guess is one of the bills just disappears from the market, some guy is short it and can't get it back... bills are definitely not all equally liquid... HTH!! great site..

    ReplyDelete
    Replies
    1. Thanks. Very interesting.

      " but don't think there is alot of demand for cash from banking sector.. everyone long it generally"

      Do you mean to say that there isn't a lot of supply of cash from banking sector? If everyone is long then that would imply plenty of demand, no?

      Delete
    2. So i guess i'm saying it really isn't a true arb b/c it has a limited amount of participants that are always naturally long reserves. Also i *think* operationally it would look weird if you were constantly bidding in the fed funds market taking in funds and then shifting the funds into the excess reserve account? i'm guessing this practice is generally frowned upon?

      Delete