Recent Posts

Sunday, August 17, 2014

Understanding Central Bank Control Of Interest Rates

One topic that periodically comes up is the issue of whether bond yields are "controlled" by the central bank, or whether they are set by "market forces". The typical context of the discussion is whether the bond markets can force governments to follow certain policies. I am in the camp that the central bank does "control" bond yields, but there are some subtleties in understanding how that control is defined.

Operation Twist

I have written this article partially in reaction to another by Philip Pilkington, "Does the Central Bank Control Long-Term Interest Rates?: A Glance at Operation Twist".  In this article, he looked at the experience of Operation Twist during the 1960's. (In Operation Twist, the idea was that the Fed would purchase long-term bonds so that it could reduce borrowing costs, yet keep short rates higher.)

I do not want to comment too much on the details of Operation Twist, but I recommend the analysis by Bill Mitchell in this article. In it, he cites a BIS Quarterly Review which pointed out:
This policy experiment is often thought to have been a failure. In fact, the experiment never
happened. The Treasury’s extension of maturities overwhelmed the Federal Reserve sale of bills and purchase of bonds. 
The Treasury did not coordinate with the Fed; they extended the maturity of their debt issuance when the Fed was lengthening the maturity of their purchases. This supply change overwhelmed whatever demand the Fed created. (And guess what the Treasury did when the Fed launched QE this time?)

Instead, I want to focus on why the type of analysis Philip Pilkington undertakes in his article is inherently so difficult.

The Fed Cannot Even Set The Fed Funds Rate Exactly

Chart: Fed Funds Target Rate - Target Versus Market
Pretty well everyone would agree that the Fed "controls" the fed funds rate. However, most people are only familiar with the policy rate (which has been redefined in the current zero rate environment), and not the market rate - the effective fed funds rate. The effective rate shown above is an average value within the inter-bank market for excess reserves at the Fed (fed funds). During the 1990s,  there was often a fair amount of oscillation around the target rate. (In earlier eras, monetary policy was less transparent, and traders had to guess what the target rate was.)

And stepping back further, we realise that we are not really interested in the fed funds rate; we are interested in the floating rates borrowers face in the real economy. There are a variety of varying term and credit premia embedded within those rates, creating yet another wedge between the policy rate and the rates seen by market participants. It is only with a fairly heroic assumption that those premia are stable that we can use the fed funds target rate as a proxy for short-term rates in the real economy.

Difficulties With Controlling Bonds

When we are looking at the current bond market environment, there is no reason to expect that the Fed would have better "control" of bond-yields than they would the fed funds effective rate. And since they do not announce a yield target that market participants could use as an anchor, the "control" would have to be worse.

Instead when people such as myself, Philip Pilkington or Modern Monetary Theory (MMT) authors speak of central bank "controlling" bond yields, it is with the understanding that yields do vary day-to-day, and possibly can drift some distance from where policymakers would prefer them to be. However, there are limits - we do not see spreads of thousands of basis points (a basis point is 1/100 of 1%) from the overnight policy rate to bond yields. And the spreads that we do see can be largely explained by expected central bank policy.

Pretty well all modern approaches to finance and economics agree that bond yields represent the expected average of the overnight rate until the bond maturity - which is controlled by the central bank - plus a term premium. The magnitude of the term premium is an open question; I argue that it should be treated as small and stable, but I recognise that not everyone will agree with that assessment. (See my "theme" article for more background on the role of rate expectations.) In any event, the level of bond yields is largely set by what the central bank is expected to do; and so if bond yields are out of line, they can signal to markets that the priced expectations are incorrect.

Therefore, it is somewhat surprising that the belief that "markets" determine bond yields is so widespread. Some of it represents the desire of bond market participants to feel important - who doesn't want to be a "Master of the Universe"? Some of it represents inadequate older economic models that could not represent time; there was "money" and "bonds", and they floated in mathematical space with no connection between them. But most of it probably represents an ideological short-cut by fiscal conservatives - if they don't like a policy, they intone in a deep voice that the "bond vigilantes" will do something about it.

One could debate whether market expectations are fully under the control of the central bank. If the central bank sets interest rates "too low", the economy will presumably overheat, and so the central bank would be forced to reverse policy later. That is the bedrock assumption of mainstream economic models. Additionally, it is unclear whether policy makers set the overnight rate in a vacuum; if the market is signalling that rate hikes are expected, it is presumably difficult for the committee to cut rates. One justification for the existence of the government bond market is that it provides a signal about the needed direction of interest rates, and this signal is not corrupted by the well-known problem of groupthink that hits entrenched bureaucratic committees.

Another area of debate revolves around the term premium - how large can it get? Most reasonable analysis is that it does not get too large, at least based on things like fiscal risk premia (at least in countries where the central government controls the currency; but even in the euro area government bond yields were stabilised when the ECB started acting responsibly). But even if term premia rise - so what? All that is implied that the yield curve steepens, and that this would slow growth - all else equal. But just as the central bank policy rule cancels out fiscal policy in modern economic models, it should also cancel this out. All that should happen is that short rates end up lower, and long end rates are higher. Why policymakers should care about this outcome is a mystery to me.

As an aside, Dynamic Stochastic General Equilibrium (DSGE) models cannot tell us anything about the effect of the term premium. Those models are built around a governmental budget constraint that turns into mathematical nonsense if a term premium is introduced. Creating a wedge between the discount rate and the expected growth rate of debt creates an error term within the "budget constraint" that can be arbitrarily large.

In summary, in the current environment, I believe that it is safe to say that central bank controls bond yields, but it does so imprecisely and that there is scope over disagreement about how expectations fit within this notion of control. However, it is clear that market participants have limited ability to force policy choices upon governments that know what they are doing, which is what really matters in this context.

Finer Control - Interest Rate Pegs

We can envisage tighter control of bond yields by central banks - yield pegging. In this operating environment, the central bank could set target yields across the yield curve.

Keynes analysed this possibility. In Philip Pilkington's article, he discusses a 1933 open letter by Keynes to President Roosevelt on the subject. Herein, I comment on what Keynes wrote in Chapter 15 of the General Theory. He argued:

Perhaps a complex offer by the central bank to buy and sell at stated prices gilt-edged bonds of all maturities, in place of the single bank rate for short-term bills, is the most important practical improvement which can be made in the technique of monetary management.
He also noted the problems such a framework faces:
(1) There are those limitations which arise out of the monetary authority's own practices in limiting its willingness to deal to debts of a certain type.
(2) There is the possibility, for the reasons discussed above, that, after the rate of interest has fallen to a certain level, liquidity-preference may become virtually absolute in the sense that almost everyone prefers cash to holding debt which yields so low a rate of interest...
The second point is what is referred to as the liquidity trap: if bond yields are too close to 0%, there is no incentive to hold them. You earn almost no interest, yet are exposed to potential capital losses if yields rise. I have made this argument about JGB yields below 0.6% awhile ago, but apparently nobody listened to me (yet). As the Japan case shows, this lower limit for bond yields is pretty low, and so this limitation is not a major concern.

But the first point is more important: in order to set bond yields at levels that changes over time, the central bank would likely end up buying a good portion of the market. No savvy market participant will want to have capital losses inflicted upon it by changes in central bank yield settings, and so they will dump their holdings to the central bank if they sense a change of policy. If they are wrong, they will be able to buy back their bonds from the central bank at almost the same price shortly thereafter.

The Fed would routinely lose tens of billions each year if it attempted to manage bond yields in a predictable fashion. It would be impossible to distinguish between incompetence and competence, or even corrupt collusion with its counterparties. For this reason, such a framework would be politically untenable.

A more sensible approach would be to keep yields within a range for an extended period. This is what happened historically in the United States for a decade after its entry into World War II. As the chart below shows, the Fed capped bond yields at 2.5%.

This policy was more sustainable, but it hobbles the ability of monetary policy to slow the economy. The Fed chafed under the restrictions, and the policy was abandoned in the Treasury-Fed Accord of 1951.

If the ceiling is set too low, the central bank will end up owning the entire government curve. The disappearance of the government curve then means that private sector yield curves can no longer be adjusted by the central bank.

The MMT ZIRP Option

One policy recommendation of Modern Monetary Theory is a permanent zero interest rate policy. The way around the problems of setting bond yields discussed here is the elimination of bonds. The operating procedures for the central government would be adjusted to allow the government to spend solely by expanding the monetary base (although Treasury Bills yielding 0% might be issued to provide an alternative to insured bank deposits).

Such a policy would be achievable, and not very different than the status quo in the United States and Japan. It would cause panic amongst believers in the Quantity Theory of Money, but those fears would be groundless. Instead, I think there may be other unintended consequences of such a policy, but I would have to discuss them in a later article due to length considerations.

(c) Brian Romanchuk 2014


  1. This comment has been removed by the author.

  2. Brian, you've highlighted the subtleties very nicely. Excellent post.

    As for the premise, I can't think of anyone with a decent grasp of economic policy who says that CBs can't control nominal interest rates or short term real rates. Isn't this debate really about CB control of long term real interest rates?

    1. The part about short rates (the fed fund effective rate) is to show that the "control" is not perfect, even at the front of the curve. (Obviously, purely administered rates like the discount rate are controlled precisely.) The "errors" are presumably greater as they move to further maturities, We therefore cannot have too demanding a definition of "control", as we would end up with the silly result that they do not "control" short rates, which as you note, nobody believes. This was maybe too arcane a point, but as an ex-applied mathematician, I sometimes worry about definitions.

      I would have to think about the "real" part of "long-term real rates" (where I assume the real yield is the yield on an index-linked bond). If you had a divergence between inflation outlooks between the central bank and market pricing, there would be difficulties.

      The other issue with regards to real rates is whether the central bank can hold them away from some "natural rate" (however defined) for an extended period. Within models, the standard result would say that is impossible. If those models are a good guide to reality, this would indicate that market expectations should price a reversion towards a "natural rate", limiting central banks' control if they need to hit an inflation target. However, since we are assuming that they are trying to hit an inflation target, their control has to consist of putting the bond yield at a level consistent with that target,

      My reading of the market is that they do not really seem to be priced on the basis that real yields will revert to a natural rate (on a 10-year horizon at least). This means that an inflation target can be respected, and the central bank has some ability to change the level of nominal bond yields.

    2. Thanks. I have no problem with your definition of 'control'. As you, I am using 'control' to mean that the CB can keep the rate within a fairly narrow band. So we're on the same page!

      Thanks for the response on the issue of real rates. It's not an area I know well but think it's key to understanding some of the limitations of CBs.

  3. Just remembered that, on the issue of control, the early part of 1967 is considered a counterexample, as it is said the Fed "lost control" of the long term yields. The short term rate was cut but the long term rates kept rising. The interpretation at the time was that increasing deficits were at fault...I can't recall which article I got this from but will try to locate. Do you have any insight into this?

    1. I am not aware of that example, but the case which is closer historically is the panic in Canada about the rise in bond yields and fall inthe currency in the 1990s due to fiscal worries.

      My feeling is that this was just a panic. But it is clear that there are limits to control, and that it is possible to issue "too many" bonds that destabilise the bond market.

  4. This comment has been removed by a blog administrator.

  5. This comment has been removed by a blog administrator.

  6. This comment has been removed by a blog administrator.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.