Friday, September 07, 2007

The Fed Funds Puzzle

Greg Mankiw has a post today about “the LIBOR puzzle” – the observation that LIBOR has recently moved up even while other interest rates (such as the 3-month T-bill rate and the expected federal funds rate) have moved down. (In fact, a lot of interest rates that businesses and consumers actually pay have moved up, but these typically are too idiosyncratic to produce the nice statistical series that economists like.)

As I said in Greg’s comments section, I don’t find this to be much of a puzzle: there is the T-bill rate, which is risk-free, and then there is the LIBOR spread – the price of LIBOR counterparty risk – which typically moves in the opposite direction except in response to unexpected news about inflation or Fed policy. The spread (typically <1%, though it is higher at the moment) is normally smaller than the T-bill rate (typically >3%), so the movements in the T-bill rate are usually larger in absolute terms; thus the sum of the two – namely LIBOR – usually moves in the same direction as the T-bill rate. The only difference now is that the spread has moved more than the T-bill rate in absolute terms; thus the sum – LIBOR – has moved contrary to the T-bill rate. It just happens that we’ve experienced a shock which has had more effect on credit spreads than on the level of risk-free interest rates. That’s actually what you might expect from a large shock to the financial sector: since it has only a small expected effect the real economy, it produces a relatively small change in the risk-free interest rate; since it has a large effect on financial conditions, it produces a large change in the risk spread.

You can read more of my ideas about this in the comments section of Greg’s post, but I want to bring up something that has puzzled me for a while. It’s an institutional puzzle. Since the Humphrey-Hawkins Act, the Fed’s primary responsibility has been the management of the US macroeconomy. That being the case, why does the Fed choose to target an interest rate which
  1. has little direct relevance to the US macroeconomy and
  2. is not under the Fed’s direct control?
Wouldn’t it make more sense to target, say, LIBOR, off of which many interest rates in the US are priced, and which in any case reflects the general cost of the type of risky credit available to private sector agents in the US? Or, alternatively, wouldn’t it make more sense to target, say, the 3-month Treasury bill rate, which the Fed could (if it chose to do so) control with great precision by making a market in T-bills at a temporarily fixed bid and offer.

I presume the Fed has its reasons for targeting the federal funds rate, but if I, in my ignorance, were asked (and I’m just as glad I won’t be) to design a monetary policy, I think I would start by coming up with a LIBOR target and then estimating the expected spread between LIBOR and T-bills, and then I would fix the T-bill rate such that the expected LIBOR would equal the target. Then I would monitor the spread and adjust the T-bill target as the spread changed over time.

In extreme circumstances (like right now, perhaps), it might be necessary to free up the T-bill rate to deal with issues in the banking system. Perhaps under today’s circumstances I would maintain the bid for T-bills but temporarily stop offering them, thus placing a ceiling on the T-bill rate. That’s actually roughly what the Fed has done with the federal funds rate, except that it can’t enforce the ceiling precisely, and it is making a vague pretense to putting a floor on the funds rate, even though the data seem to show zero as the only floor.

Labels: , , , ,

10 Comments:

Blogger Karl Smith said...

I would think that in the past the Funds rate acted as the LIBOR rate does now. Maybe thats wrong.

However, the prime rate used to be the reference rate I heard most often. Credit Cards and some car loans are still based on prime. I know that know prime is just funds + 3. However, even before it became that mechanical wasn't the prime rate and the funds rate linked?

Fri Sep 07, 06:52:00 PM EDT  
Blogger Unknown said...

The Fed doesn't want to, nor should it, interfere with the market mechanism for pricing credit risk, so targeting a risky rate like LIBOR is clearly out.

The Fed would have a problem fixing the T-Bill rate because they don't have the authority to issue Treasuries, so they couldn't prevent rates from falling once they'd sold through all of their T-Bill inventory.

Fri Sep 07, 08:17:00 PM EDT  
Blogger knzn said...

The federal funds rate also includes an element of risk compared to Treasuries. (At least from the data I've seen, the 3-month T-bill usually trades at a yield below federal funds, despite the longer maturity.) So the Fed is already (sort of) interfering with the market mechanism for pricing credit risk. But in any case, the Fed wouldn't really be interfering with the market mechanism; it would just be taking the market price of risk as a given and operating within that constraint. Arguably, the current procedure interferes with the market mechanism for pricing risk to a greater degree than the one I suggest, because the Fed actually tries to maintain constant control of the (slightly) risky rate. Under my procedure, the Fed would be reacting to changes in the LIBOR spread rather than trying to control LIBOR constantly.

As for the possibility of running out of T-Bills, I don't see how the current procedure avoids that problem. If the Fed wants to raise interest rates but has nothing left to sell, it's out of luck. That's why Alan Greenspan wanted the 2001 tax cut. Granted, the category of assets that the Fed can sell now is broader than the 3-month T-bill. If my hypothetical Fed ran out of 3-month T-bills, it would have to start using a different instrument, but that doesn't seem like a big problem to me.

Karl has a good point that I'm probably overreaching when I say that the federal funds rate "has little direct relevance to the US macroeconomy." I wasn't aware that banks set their prime rates automatically in that way, but the spread does seem to have been more-or-less constant since the late 1980s. (But what is supposed to happen under the conditions of the past month, in which the Fed has mostly failed to hit its target? In practice the average spread of prime over fed funds has been higher than usual.)

I still think LIBOR would be a better target. When the prime spread over LIBOR narrows (as recently, for example), it means that the prime rate isn't fully accounting for the price of risk, and we should expect more non-price rationing of credit (e.g., stricter credit standards), which the Fed should offset by easing at the source.

Fri Sep 07, 10:51:00 PM EDT  
Anonymous Anonymous said...

"That’s actually roughly what the Fed has done with the federal funds rate, except that it can’t enforce the ceiling precisely, and it is making a vague pretense to putting a floor on the funds rate, even though the data seem to show zero as the only floor."

As one who commented on this "zero floor" early on, allow me to respond. Remember that fundamentally we are talking about bank reserves here. Banks are constrained to hold reserves, but they are not rewarded for holding any more. So the price of reserves (the funds rate) acts like a shadow price. If the constraint binds, the Fed can adjust the level of reserves to get the price where they want it. But if the constraint doesn't bind (as when banks start holding excess reserves), the Fed starts to notice that they can pull, but not push on the string.

I strongly suspect that the low rates are occurring at the end of the day when banks with a lot of excess reserves are offering some of those reserves for anything they can get. And note that since the week following that first day of difficulty (and except for a reserve maintenance day on the 29th) the low end has been 2 or 3 and several days in the high 4s. In other words, the slack is coming out of the string.

All of this is basically what I pointed out before, and it is interesting and worthy of note. It's like watching an eclipse. It doesn't happen often, but it is worth studying because it can be explained. It's not a sign that something is dreadfully wrong with the universe.

Frankly, this is exactly the thing you would hope to have happen in this circumstance. Banks should start to hold some excess reserves. And I'm not concerned that a few low valued fed funds trades are going to cause banks to expand loans and juice the economy.

If the Fed paid interest on excess reserves (like the Bank of Canada does, I believe) then this wouldn't be an issue.

Now, the problem with targeting the T-bill (or the LIBOR or the spread) is, as Marc Shivers suggests, a problem of authority. The Fed has the authority to control bank reserves. That's all. Monetary policy intervention takes place exclusively through the reserve channel. That's why it works the way it does.

If you want to target the T-bill (or LIBOR or the spread) directly, then you are standing ready to buy/sell T-bills (or borrow/lend at the LIBOR rate). The mechanics of this would be similar to targeting an exchange rate. You simply couldn't target the T-bill rate AND maintain any meaningful control over reserves.

If you, as a central banker, are committed to being a residual buyer/seller of T-bills and the fiscal authority gets an appetite for spending, then.... uh oh!

Sat Sep 08, 02:11:00 AM EDT  
Blogger Unknown said...

Two follow-up comments:

First, I disagree that the Fed is interfering with credit risk pricing when it intervenes in the Fed Funds market. Fed Funds have an element of credit risk to them, since they are uncollateralized overnight loans between banks. As a result, some banks pay higher rates than others for Fed Funds. When the Fed wants to intervene in that market, it does so by either borrowing or lending in the Fed Funds market, but on a fully collateralized basis through repos, so there's no subsidy for riskier banks.

Second, it's not possible for the Fed to run out of assets to sell if it wants to permanently reduce the money supply, since the assets on its balance sheet were acquired by issuing currency (i.e. Federal Reserve notes). If it sold all its assets, there would be no currency left in circulation.

Sat Sep 08, 11:17:00 AM EDT  
Blogger knzn said...

The Fed has the authority to control bank reserves. That's all. Monetary policy intervention takes place exclusively through the reserve channel. That's why it works the way it does.

To some extent, the Fed controls bank reserves by buying and selling T-bills outright. It also does repos and discount window loans, etc., but obviously it has the authority to buy and sell T-bills. It certainly has the authority to buy or sell a specified quantity of T-bills. What would be wrong with defining its reserve activity in terms of yield rather than quantity?

If you, as a central banker, are committed to being a residual buyer/seller of T-bills and the fiscal authority gets an appetite for spending, then.... uh oh!

IIRC that's the way it was before the Fed-Treasury accord. But the whole point seems like a straw man to me. The Fed would have to adjust its target in response to events, just as it adjusts its fed funds target in response to events. In practice, the fiscal authority has much more inertia than the monetary authority, so the Fed would have a chance to react to any fiscal changes. And in any case, the same issue exists with the current policy. If the Fed is committed to a federal funds rate target, and Congress (with the cooperation of the President) suddenly decides to go on a spending binge, the Fed would have to create huge quantities of money anyhow (until it adjusts the target, of course).

When the Fed wants to intervene in that market, it does so by either borrowing or lending in the Fed Funds market, but on a fully collateralized basis through repos, so there's no subsidy for riskier banks.

There's no relative subsidy for riskier banks, but there is an absolute subsidy. Surely Fed intervention prevents some otherwise marginal banks from failing -- just by increasing the overall availability of reserves. I don't see how this is fundamentally different from targeting LIBOR. The Fed would not be subsidizing anyone, because it would not intervene directly in the euromarket; it would just alter its domestic open market operations in response to changes in LIBOR. How is that different from altering its open market operations in response to changes in the federal funds rate? Or, for that matter, from altering its funds rate target in response to changes in economic activity?

it's not possible for the Fed to run out of assets to sell if it wants to permanently reduce the money supply, since the assets on its balance sheet were acquired by issuing currency

If you want to take my suggestion literally, then let's say the Fed starts out by swapping all its other assets for 3-month T-bills. (I suppose technically it would have to do a new swap after every auction to maintain the maturity.) In that case, the only way it could subsequently run out of 3-month T-bills would be by reducing the money supply to zero. I think it's a safe bet that the interest rate would rise sufficiently before it got to that point.

In practice, if the Fed were running low on 3-month T-bills to sell, it could just change its instrument to, say, 1-month or 6-month bills. Or it might take the shortage as an indication that the target needed to be changed.

Sat Sep 08, 12:33:00 PM EDT  
Anonymous Anonymous said...

"What would be wrong with defining its reserve activity in terms of yield rather than quantity?"

The two approaches have vastly different implications. Put very simply, if you define monetary policy intervention in terms of T-bill yields, you lose control over reserves. Instead, you would be maintaining price controls on T-bills. When the market sells, you buy... and reserves necessarily increase. This is a recipe for instability of a serious order.

You're suggesting a monetary policy that causes the central bank to lose control over the one thing that the central bank is generally charged with controlling...bank reserves. That ought to answer the question, but...

"The Fed would have to adjust its target in response to events, just as it adjusts its fed funds target in response to events."

What are the criteria for making the adjustment? Suppose you are trying to manage the LIBOR spread, a variable that responds to changes in the environment for risk, the fiscal policy climate, etc. How do you propose determining--in real time--when to allow the spread to close or widen and when not to? How do you determine--in real time--what underlying movements in the spread reflect movements in the actual thing central banks want to control (like inflation, for example)?

Targeting the funds rate is certainly not the only policy and perhaps not the best. But we have a pretty good idea of how it works operationally. And we have an imperfect (but decent 90+% of the time) knowledge of how it interacts with variables the Fed wants to control.

There's a fairly high burden of proof you'd have to meet to convince me that targeting a measure of risk (the whole concept of that just seems fraught with danger) would provide more stability than the current system.

Sun Sep 09, 02:04:00 AM EDT  
Blogger knzn said...

When the market sells, you buy... and reserves necessarily increase. This is a recipe for instability of a serious order....You're suggesting a monetary policy that causes the central bank to lose control over the one thing that the central bank is generally charged with controlling...bank reserves.

I don't think the Fed really has control of bank reserves now, because it cannot control the rate at which currency is withdrawn from or added to the banking system by depositors. It has to observe the behavior of depositors and react to that. But it doesn't actually do that (usually); what it does is observe the federal funds rate and react to that. So in the end it isn't even really trying to control bank reserves, and the system is already unstable (as, indeed, we have observed in the present episode, where the Fed has been creating large quantities of reserves in response to excess demand at the target funds rate).

Suppose you are trying to manage the LIBOR spread....How do you propose determining--in real time--when to allow the spread to close or widen and when not to?

What I suggested is taking the spread as exogenous and managing the T-bill rate target to keep LIBOR roughly constant. That requires some judgment as to whether a given change in the spread will be persistent, but that type of problem already exists with the federal funds rate. How do you know when a change in the funds rate needs to be met with a reserve action and when it is just a temporary fluctuation that can be ignored?

Sun Sep 09, 11:45:00 AM EDT  
Blogger Warren Mosler said...

I start with the general statement that the liability side of the banking system's balance sheet is not the place for market discipline.

And whether you agree with me in theory or not, in practice all banks that are in 'good standing' with the regulators qualify for govt. insured deposits. That means all 'legal' bank assets are already funded or eligible for funding with what is functionally govt. funding.

So why should the fed not accept for collateral what the fslic and fdic already accept? No net risk is added to the govt. It's only disruptive not to do so. Shareholders remain in first loss position, and the banks remain subject to continuous regulatory scrutiny to make sure they are adequately capitalized and their assets are of 'legal' quality.

So with capital requirements and regulation taking care of the market discipline aspect of the member banks, why have an interbank market at all?

Why not simply have all banks borrow/lend to the fed at the fed's target rate(s) as set by the FOMC as a matter of 'appropriate monetary policy?' (whatever that actually means)

so my position is that the fed should lend at it's target rate vs any collateral the banks may have (and in unlimited quantities) and functionally eliminate the overnight interbank markets.

Let me here add that 'unlimited lending' will not result in increased net lending. loans still create deposits, and so the net funding needs for the banking system from the fed remains only that necessary to 'offset operating factors' like the demand for actual cash and changes in tsy balances at the fed, etc.

The fed could do the same for the entire term structure of rates, but let's just say it doesn't want to do this, for whatever reason, so the term markets will be left to interbank trading. but with the overnight markets 'guaranteed' there is unlikely to be any remaining spread between ff and libor (apart from setting of non member banks credit in $ affecting the bba settings, and that's another story, also very relevant today).

Warren Mosler
moslereconomics.com

Sat Apr 26, 01:19:00 PM EDT  
Blogger yanmaneee said...

yeezy shoes
adidas superstars
christian louboutin
christian louboutin outlet
hermes
cheap jordans
air max 2019
nike max
jordan shoes
nike flyknit

Mon Jul 15, 03:42:00 AM EDT  

Post a Comment

<< Home