Wednesday, November 26, 2014

Target the spread?

To send you off with some more Thanksgiving good cheer, here is another out of the box Neo-Fisherian idea.

Perhaps the Fed (or the Treasury) should target the spread between real and nominal interest rates.

Above, I plotted the real (TIPS) and nominal 5 year rates. By the usual relationship \[ i_t = r_t + E_t \left[ \pi_{t+1} \right] \] we typically interpret the difference between real (r) and nominal (i) rate as the expected inflation rate.

Now, the usual Neo-Fisherian idea says, peg the nominal rate (i), eventually the real rate (r) will settle down, and inflation will follow the nominal rate. It's contentious, among other reasons, because we're not quite sure how long it takes the real rate to settle down, and there is some fear that real rate movements induce a temporarily opposite move in inflation.

So why not target the spread? The Fed or Treasury could easily say that the yield difference between TIPS and Treasuries shall be 2%. (I prefer 0, but the level of the target is not the point.)  Bring us your Treasuries, say, and we will give you back 1.02 equivalent TIPS. Give us your TIPS, and we will give you back 0.98 Treasuries. (I'm simplifying, but you get the idea.) They could equivalently simply intervene in each market until market prices go where they want. Or offer nominal-for-indexed swaps at a fixed rate.

Now, I think, the Neo-Fisherian logic is even tighter. If the government targets the difference \( i_t - r_t  \), in a firmly committed way, \( E_t \left[ \pi_{t+1} \right] \) is going to have to adjust.  I plotted 5 years, because I'm attracted to the idea of nailing down 5 year inflation expectations, but the general idea works across the maturity spectrum.

They might have to buy and sell a lot, you say? Indeed.  $4 trillion is a lot already, and Japan is embarked on even larger QE.   This might have fiscal consequences, you say? Indeed. That is, actually a lot of the point. Neo-Fisherian ideas are wrapped up with fiscal theory of the price level, and the spread peg is pretty much a fiscal commitment. It's a way of committing that we're going to inflate away the nominal debt at 2%, no more, no less. It's almost a modern gold standard in that way.  TIPS are illiquid, you say? Indeed. The contract structure could be improved a lot. But most of all, they'll be a lot more liquid when the Fed starts trading them every day!

What about the level of interest rates? That's the best part of the idea. If you're a free-money-market type, you could advocate that the Fed no longer target the level of either rate. If you're of the view that raising the level of interest rates is an important policy for the Fed to stabilize the real economy and induce short-run inflation movements (dynamics here), then the Fed can also move the level of short rates around, and at the same time target the spread.

The Fed has long used the TIP-Treasury spread to measure inflation expectations. But the same equation suggests the Fed (and Treasury) can directly control those expectations.

And, I hate to mention it, if a government wants to raise inflation expectations, firmly targeting such a spread would be a way to do it.


  1. As someone watching the professional debate on inflation modelling from the outside, I've wondered if the central bank should instead look for a policy that doesn't rely on very accurate models. This seems to be it.

    I do hope that the free-money-market types gain more influence. Instead of trying to manage interest rates, I think it's more useful to do something about missing markets that lead to such wild swings in equilibrium rates.

  2. Reminds me of the message from Benigno and Woodford 2003 Macroannual paper, where the standard New Keynesian optimal monetary policy rule (flexible inflation targeting) was coupled with a requirement to stabilize expected inflation, which could be interpreted as the fiscal rule. But nothing prevented the central bank to adopt that rule, as long as the Treasury played its stabilization role accordingly.

  3. Closer. It's like inflation forecast targeting, which is what (modern) central banks do now, but using a market-based forecast instead of an internal forecast generated by their research department.

    But I think you need to work on what exactly they are buying and selling. Because the Fed prints US dollars, but it does not print TBills or TIPS. So your proposal ("Bring us your Treasuries, say, and we will give you back 1.02 equivalent TIPS.") would be like the Fed trying to peg the exchange rate between the UK pound and the Euro.

    1. Nick I think the key point Cochrane makes above is about the fiscal commitment to monetize the debt at precisely 2% per year. Fiscal policy is not in the purview of the Fed, but it can act in coordination with the rest of the government. Whereas a peg between the UK pound and the Euro is not something that the US can achieve unilaterally.

    2. Anwer: sure, but the way to ensure that is to modify John's proposal slightly, to ensure that base money expands automatically by an unlimited amount if there is an incipient narrowing of the spread, and contracts automatically by an unlimited amount if there is an incipient widening of the spread. Fiscal policy will (eventually) adjust to be consistent with that monetary rule, provided the central bank is expected to retain its independence and continue to follow that rule.

      It's like the gold standard, except: it's a crawling peg, it's targeting the forward price not the spot price, the CPI basket replaces gold.

    3. Your explanation is flying over my head, perhaps because I don't understand the gold standard very well. Prof. Cochrane notes his reliance on the Fiscal Theory of the Price Level. Any method (including QE) that moves risk back and forth between the Fed's balance sheet and the private sector could help implement the proposed policy, though Cochrane suggests a very direct implementation. You are talking about the expansion and contraction of base money to maintain the targeted spread. Perhaps you are using a different theory of inflation than him?

    4. I agree with Nick (and anon below). John, you need to be careful about making a market between bonds and TIPS.

      You can get around this problem by targeting the TIPS spread indirectly ie. by doing whatever amount of open markets operations in traditional securities is necessary to keep the spread on target. This is what the Fed has traditionally done: it has indirectly targeted consumer goods & services prices via Treasury purshases. But then you've got another problem on your hands. Part of your reasoning for operating directly in the TIPS market was to solve their illiquidity problem -- TIPS are notoriously less liquid than Treasuries.

      So when you indirectly target the TIPS spread, you won't know what portion of the differential is due to inflation expectations and what is due to liquidity premia. You could end up reacting to the former when you should only be reacting to the latter.

      One way to get around this would be to target inflation swaps, since they don't suffer from the same liquidity problems. Or at least, so says the Cleveland Fed:

    5. Interesting observations, JP. The Fed could continue its open market operations using traditional securities, but many people complain about the price distortions caused by these large-scale purchases. And some central banks are bearing agency costs through their holdings of securities. I'd much rather see them use the macro securities proposed by Robert Shiller for their open market operations. If the Fed made markets in these new securities, it would also be a new and potentially useful form of monetary plumbing.

  4. I think the simplest way to implement it is to create a "pure" synthetic bet on inflation, which is long one $100 TBill and short one $100 TIPS (or is it the other way around?), and stand ready to buy or sell it for a $2 bill. Something like that. (This finance stuff always muddles me.)

  5. Ah the Nick Rowe beat me to the punch. To pile on: Will the Fed really make a locked two way market of infinite depth at unchanging spread valuation, and commit credibly to always do so? In times of crisis when everybody wants the most liquid instruments, will it be able to proide infinite liquidity in the spread securities, in league with the Treasury? Would such a policy merely provide great confusion and gaming opportunities for the STRIPS market? Willthe Fed ignore exogenous inflation events like oil shocks, and just eep on substituting at the old policy spread?

    Well I guess if you like rules but are disenchanted with MV = pq as a policy tool and do not like targeting of nominal rates, you have to finds something new. I am presuming that the quantity theory is the primary dispute between marke moentarists and neo- Fisherians?

    Michael S.

    1. I think the dispute can be resolved, once we understand our (too often implicit) assumptions. Both sides agree that a 2 for 1 stock split causes the real value of a stock to halve. We call that the Quantity Theory of Money, when it is applied to the good that is used as medium of exchange and account. And we worry more about what happens if the stock price is sticky, so all monetary trades get disrupted. And we prefer to target a level path for NGDP. So Scott Sumner's proposal for pegging the price of NGDP futures is similar to John's proposal for pegging the spread between Tbills and TIPS.

    2. They are both using market information, rather than central bank models, but the goals seem to be different. Market Monetarists are managing aggregate demand and are willing to accept unpredictable inflation. Prof. Cochrane is aiming for predictable inflation, to improve the clarity of all other market signals.

  6. No need to post this since no one will care, but I think you're right. However, the names of various views confuses me for some reason. It's worth a shot.

  7. Thanks Nick and Anwer for your help. Now I understand a little better.

    One alternative to the Fed's offering either the firm spread price for the TIPS/ Tnote exchange-- or a side bet at a set price (per Nick above)--would be for the Fed to merely be best bid or best offer on the spread securities, to nudge the relative bond prices until the market took guidance and brought the market expected inflation level to the target. Other bonds would then join the expectations party through spreading and arb by the bond market. So maybe my objections are in fact trivial and easily overcome.

    I am still not sure how cramming off-market inflation expectations on the market can make them market expectations. Is the Fed that powerful?

    I apologize for taking so much space versus you genuine experts and will go away now.

    Michael S

    1. Michael,

      No, the Fed is not that powerful. By law, the central bank is not permitted to directly submit bids on government debt auctions. Meaning that before the central bank can even touch government securities, someone else (usually a private bank) submits a bid on them.

      The central bank can attempt to buy / sell those debt securities from the market. It remains an open question whether any market participant is legally obligated to honor a central bank's attempts at open market operations.

      Also, recognize that an increase of 1% on the TIPS real rate is a larger bond price drop than an increase of 1% on the nominal Treasury rate. This would complicate efforts by a central bank in trying to maintain a spread between TIPS and Treasuries in a rising rate environment.

  8. Sorry I missed in my post of a few minutes ago that JP Koning had aleady pointed out that the Fed could nudge bonds to policy spreads without the extreme measure of a fixed exchange mechanism.

  9. One possible problem is disagreement on inflation between financial and non financial firms. Survey data for Uruguay, Costa Rica and Perú show that disagreement exists and might be persistent. In that case, the firms that invest in physical capital and hire labor (mainly non financial firms) , which are the ones behind price formation might see your monetary stance as erroneous, and that might lead to an inflation bias.

  10. "They might have to buy and sell a lot, you say? Indeed. $4 trillion is a lot already, and Japan is embarked on even larger QE. This might have fiscal consequences, you say? Indeed. That is, actually a lot of the point."

    It's a hedge fund model for central bank policy. This strategy is a subset of that.

    (They're tinkering with this idea slightly by ramping up the RRP facility.)

    This can be easily separated from the question of how the monetary base should be managed, given liability management opportunities in the hedge fund mode.

    It's an aggressive model with deep fiscal/monetary effects.

  11. The spread targeting approach does seem compatible with your FTPL ‘monetary policy with interest on reserves’ paper.

    Although more generally it seems like an extension of central bank policy and activity from the short end of the yield curve to the long end.

    The central bank has to set the interest rate on the liabilities it issues. The market can’t set this rate on its own – it can only be a source of information that the central bank uses to set the rate. If the central bank ignored its own active role in short rate setting, that would amount to a booked interest rate of zero on bank reserves. And there is no doubt that in an environment of abundant liquidity banks would drive the short rate on Treasury bills to nominal zero or thereabouts. This is a matter of default arbitrage forces. My point being that even if the central bank targets long yield spreads as you suggest in the post, the central bank must continue to make the decision on what the short nominal rate should be for bank reserves. And this rate setting will propagate out to the rest of the market through an arbitrage process, as you point out in your paper.

    Also, the central bank has to issue reserves in some quantity in whatever monetary regime we have. There’s no getting around that - or it’s no longer a central bank.

    So whatever the FTPL says about it, the central bank has to make the decision on the nominal rate it will pay as interest on reserves – whether its zero as under the previous regime or potentially non-zero under an abundant liquidity regime. This holds whatever it decides to do in the long end of the market.

    Targeting the inflation rate through long rate spread management OMO as you suggest seems potentially complementary to the way in which the central bank sets the short rate, however it does the latter. So this can hold for an FTPL short rate model.

    The inflation expectation embedded in the long nominal Treasury rate is natural feedback for the inflation expectation that may be embedded in the short rate. That embedding could be made more explicit and targeted for the long rate with what you suggest here. But it will only be implicit in the short rate – unless the nominal short rate is set and announced as a precise formula with empirical real rate and inflation expectation inputs, which could be the case using your long rate spread targeting idea.

    At a general level, the idea of targeting long rate spreads looks like an extension of central bank activity out the yield curve, and in that way it seems similar to the case of pegging the long Treasury rate in the 1940’s, or even the soft commitment to keep short rates low for a prolonged time as has been implied by the central bank’s commitment to QE.


  12. .....

    On basics, it seems to me that your FTPL model is quite independent of the issue of any policy for monetary base expansion. The whole point of it is that interest paying reserves are debt. Therefore, the composition of debt as between interest paying reserves and Treasury debt is irrelevant for purposes of the FTPL model. (You even abandon currency, which I think is eminently and unusually sensible.) Specifically, there is nothing in the FTPL model for the interest rate level that says that the interest rate accrual on reserves must be retained as new reserves. This is the interesting question that Nick Rowe has raised here:

    In fact, Treasury cash management discipline means this is definitely not the case. Interest on reserves is paid by issuing more Treasury debt, other things equal, in the normal course. It is a marginal deficit funding requirement, other things equal. It is also a fact of operations and accounting as determined by the institutional configuration where Treasury has a deposit account at the central bank. This is all inherent in the institutional separation of Treasury and the central bank. It’s all well and good to consider alternative institutional arrangements, but Nick’s question should be answered first in the context of existing arrangements, as in fact it was put that way. That leaves reserve growth as a pure function of central bank OMO. I don’t think your FTPL model has much to do with “base money” at all, nor do I think you intend it to. It doesn’t seem to have a monetarist flavor as far as I can see. Deciding on the mix of bank reserves and Treasury debt is quite separate from the act of setting and paying the interest rate on reserves. This is all compatible with the rational separation of Treasury and central bank institutional roles, which is something I think you emphasize in your paper.

  13. John,

    "So why not target the spread? The Fed or Treasury could easily say that the yield difference between TIPS and Treasuries shall be 2%. Bring us your Treasuries, say, and we will give you back 1.02 equivalent TIPS. Give us your TIPS, and we will give you back 0.98 Treasuries."

    Are we also under the impression that all government securities are 1 year in duration or less? On a thirty year security offering coupon (non-compounded) interest, I would need at least 1.60 TIPs for every 1 nominal Treasury to make that swap at a 2% spread.

    Also, if realized inflation comes in higher than expected, this has some pretty dire fiscal consequences.

  14. John,

    1 / 1.02 = 0.98039216 not 0.98 assuming we are talking about frictionless swaps.

    Also, JP Koning above mentions that TIPs suffer liquidity problems that are not as severe in the nominal treasury market. TIPs also suffer higher price volatility for the same interest rate change compared to nominal treasuries.

    How does a central bank / Treasury maintain a constant spread between TIPs and nominal securities when the same amount of interest increase / decrease on the securities causes different bond price changes?

    For instance, suppose market rates for nominal treasuries are 4% and market rates for TIPs are 2%. Suppose at the next auction the rate for nominal treasuries jumps to 5% and the rate for TIPs jumps to 3% - spread stays the same. If I am holding nominal treasuries that fall in bond prices represents a 1% / 4% = 25% loss. If I am holding TIPs that fall in bond prices represents a 1% / 2% = 50% loss.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.