NGDP futures–clarifications and extensions

Bill Woolsey is very familiar with index futures targeting, but also knows the details of how futures markets actually work much better than I do.  Thus I thought this letter from him would help clarify the nuts and bolts of the plan from the perspective of real world futures markets.  I would also like to mention that the commenter “123” noted that the Fed might have to take a large long or short position if there was a sudden change in money demand during a financial crisis.  Bill mentions that perhaps the Fed should be allowed to trade on its own account.

On further reflection, I would like to add to my previous suggestion that before each auction the Fed should adjust for predictable variations in base demand due to trend, seasonal and day of the week factors.  Now I think they should try to estimate the equilibrium base demand before each auction, so that (ex ante) their expected net long or short position is zero.  I think this would make the market work more efficiently, require less trading, and allow the Fed to take less risk.  The Fed would do the “heavy lifting” and the market would merely fine tune the Fed’s forecasts (as well as keep them honest of course.)  The rest of this post is Bill’s suggestions and clarifications:

Sumner proposes that the Fed target nominal GDP on May 17, 2010 to be $15.9 trillion.  He proposes that the Fed create and then buy and sell unlimited quantities of an index futures contract on NGDP at that value.  He proposed that the contracts be defined at one trillionth of NGDP, and so the Fed would buy and sell unlimited futures contracts at a price of $15.90.

Perhaps the most important element of the proposal, however, is that the Fed will make ordinary open market operations in parallel to the trades of these futures contracts, and these open market operations will impact base money.

Trades in the futures contract will be at the initiative of “the market,” specifically, futures speculators.  Because the “market” initiates the trades, it has sometimes been called “index futures convertibility.”

If a speculator chooses to buy a contract at the price of $15.90, the Fed will create and sell the contract for $15.90.   At the same time the Fed sells the future, it will simultaneously make an open market sale.  Presumably, it would sell treasury bills.  Sumner proposes that it sell five times the dollar value of the contract, so $79.50 worth of T-bills would be sold per futures contract the Fed sells.. (In my view, $15.90 is a very low value for a future contract.  Perhaps one hundred millionth of NGDP, or $159,000 would create a future contract more typical.)

If a speculator chooses to sell a contract at the price of $15.90, the Fed would create and buy the contract for $15.90.   At the same time the Fed buys the contract, it also makes an open market purchase, presumably buying Treasury bills worth $79.50.  (Sumner also lists other sorts of securities the Fed might buy.)

What exactly do these contracts mean?    Sometime after May 17th, 2010 the level of nominal income would be calculated for that particular date.  (Some weighted average of first quarter 2010 NGDP and second quarter 2010 NGDP.)    And then contracts would be settled.

If NGDP is exactly $15.9 trillion, then nothing will happen at all.    These are index futures and all settlements are with cash.  If settlement value is equal to the price then there is nothing to settle.

Suppose that NGDP is $16 trillion; $100 billion above the target.   Then the contracts will be settled.   The longs, who had bought the contracts, will be paid 10 cents for each contract.   The shorts, who had sold the contract, must pay 10 cents for each contract.
But what if NGDP is only $15.7 trillion; $200 billion below the target?  Then the settlement will require a payment to the shorts, who had sold the contract, of 20 cents, and the longs will be required to pay 20 cents.

If a speculator expects that NGDP will be above the target, then she will have an incentive to buy a contract in order to receive a profit at settlement.  However, if that speculator’s expectation was in error, and NGDP is below target, she will suffer a loss.

In the opposite situation, a speculator who expects that NGDP will be below target, has an incentive to sell a contract, to make a profit from settlement.   Similarly, he will risk losing money if NGDP unexpectedly rises above target.

From the Federal Reserve’s perspective, if speculators who expect NGDP to be above target (bulls) purchase contracts in an equal quantity to the number sold by speculators expecting NGDP below target (bears,) then at settlement time, the Fed acts as an exchange and transfers funds between the longs and the shorts depending on the actual value of NGDP.    In this situation, the market has provided no signal that NGDP is expected to deviate from target and provides no impetus to open market operations and changes in base money.

If the speculators expecting NGDP to be above target (bulls) purchase more contracts than those expecting it to be below target (bears) sell, then the Fed must make up the difference in the market and is a net seller of the contracts.   In Sumner’s proposal, this triggers an open market sale on T-bills equal to five times the value of the Fed’s short position.    The open market sales decrease base money today, this begins to decrease nominal GDP and so decrease the expected level of NGDP one year from today.

If the speculators expected NGDP to be below target (the bears) sell more contracts than the speculators expecting it to be above target (bulls) buy, then the Fed must be a net buyer of the contracts.   The Fed makes open market purchases of T-bills equal to five times the amount of the Fed’s long position.   The open market purchases increase base money today, and increase the expected level of NGDP one year from today.

Equilibrium requires a level of base money such that the market expectation of NGDP is on target.  Speculators may trade the contract, but with the market expectation of NGDP on target, that means that purchases and sales by speculators balance.   The Fed takes no net position on the contract and base money is unchanged.   If, on the other hand, the market expectation of NGDP is not on target, then speculators will take a short or long position on the contract, with the Fed taking the balancing long or short position on the contract, triggering open market operations in T-bills, and changes in base money.   If base money is “too low,” it will rise.  If base money is “too high,” it will fall.

In my view, rather than require the Federal Reserve to make open market operations equal to some multiple of its short or long position, the Fed should be given discretion to make open market operations as it chooses, subject to the general requirement that it seek to keep its position on the contract at zero.

Like in ordinary futures markets, Sumner explains, there would be margin accounts for the speculators in the contracts.  This will be 10% of the value of contract and the Fed is supposed to pay interest on these margin accounts.

It is helpful to remember that “margin accounts” for future contracts are also called a “performance bond.”   Its purpose is to make sure that those who suffer losses on the contract will make payment at settlement..   Therefore, it is really a requirement that investors pledge collateral.

In my view, while the Fed could create special deposit accounts for speculators and pay them interest, having speculators put up T-bills for collateral, much like banks must do for primary credit loans, would avoid those complications.     The relevant question is whether the value of the collateral will be sufficient to cover the loss generated by an unfavorable move in NGDP.



14 Responses to “NGDP futures–clarifications and extensions”

  1. Gravatar of JKH JKH
    12. May 2009 at 03:37

    Bill Woolsey’s explanation is very good.

    However, I’m not entirely clear on meaning where you’ve cross-referenced several aspects.

    Bill writes:

    “In my view, rather than require the Federal Reserve to make open market operations equal to some multiple of its short or long position, the Fed should be given discretion to make open market operations as it chooses, subject to the general requirement that it seek to keep its position on the contract at zero.”

    You write:

    “Bill mentions that perhaps the Fed should be allowed to trade on its own account.”

    I’m assuming your meaning of “trade on its own account” is applicable to OMO in the sense that Bill has written about it, which seems like an amount of discretion intended to respond to unusual swings in futures market directional pressures.

    “Now I think they should try to estimate the equilibrium base demand before each auction, so that (ex ante) their expected net long or short position is zero.”

    I’m assuming this is a separate idea from Bill’s. The way you have written it, it reads like a formulaic base case amount intended to offset a forecast of a “neutral” OMO requirement.

    So these seem to be two separate ideas, both adjustments to OMO response from what is otherwise required?

  2. Gravatar of DG DG
    13. May 2009 at 00:41

    Very neat explanation – thanks.

    Don’t JKH’s comments about the Fed having discretion over how to “interpret” the market signal undermine the point of the idea which I thought was to remove the potential for activism?

    Relatedly, where does the OMO of 5x the Fed’s position come from? That is, why 5?

    Following on from discussion on previous comment, where I pointed out that the targeting of GDP was pointless, you replied (Mr Sumner) that my criticism was incorrect because you’d be targeting NGDP, not GDP. Am I right here in my understanding of your argument here that since inflation has no innate trend, unlike real GDP, the targeting of NGDP would allow you to take up the slack left by our not knowing?

    If so, then I’m still not sure you get a different outcome. Doesn’t inflation have an innate “trend” rate of growth which is the reciprocal of trend growth, reflecting structural productivity improvements? Only money supply growth which are usually exogenous (CB fine tuning) but can be endogenous (the gold rush?) will generate deviations from that equilibrium path. And aren’t those deviations what ultimately cause distortionary inflation of some sort, whether in product markets or asset markets? Apologies if I’ve got this completely wrong or if I’m getting bogged down in a seperate discussion on the pros and cons of inflation targeting


  3. Gravatar of ssumner ssumner
    13. May 2009 at 05:14

    JKh, That’s a good point. I was thinking about ex ante trading, although not necessarily according to a mechanistic model, but rather to move the base to the expected equilibrium before futures trading even began. That expected equilibrium might reflected subjective estimates of the impact of events like Lehman.

    I knew that Bill contemplated Fed discretionary moves after trading started, but I am not quite sure about his proviso that they try to keep their net position near zero. If they keep the net position at zero, then I don’t see how the futures markets can send signals to the Fed about base demand (in other words there might be a circularity problem.) I’ll have to think about this a bit more.

    DG, The trend real growth rate is not impacted by monetary policy. Thus when the Fed targets NGDP growth, they are implicitly targeting long run inflation at 5% minus the real GDP trend growth rate. The short run is more complicated, as monetary policy affects both prices and output. Under NGDP targeting, short run fluctuations in inflation are “healthy” as they contribute to desired movements in real wage rates.

    The 5X ratio was arbitrary. It simply showed that you could set up the market in a way where traders did not have to commit huge amounts of capital to the NGDP futures market, in order to give useful signals to the Fed.

  4. Gravatar of Bill Woolsey Bill Woolsey
    13. May 2009 at 11:53


    As I see it, the Fed just offers to buy and sell at the target price. It’s position on the contract is always passive.

    However, I think it should be able to trade bonds as it chooses. By impacting base money, it will impact expected future nominal income and so impact the trades of the speculators, and therefore its net position on the contract.

    The reason specualtors trade is because they disagree with the rest of the market–not with the Fed. And not to cause the Fed to change base money in an optimal way.

    The Meltzerites are betting that those foolish Krugmanistas have caused base money to be too high. The Krgumanistas are betting that those foolish Meltzerites have keep base money from getting high enough.

    The Fed _is_ free riding on this activity. Now, you could let the Fed play this game too. Side with the Meltzerites or Krugmanistas. And hold them responsible for profits and losses.

    Or, you can require that they “free ride” on the speculators betting against one another, always keeping its position on the contract zero.

    I have no clue as to how to distinguish discretionary open market operations from the Fed setting base money according to trend and seasonal factors.

    For example, as I see it, the Fed would realize that the demand for base money is growing. If it does nothing, nominal income next year will almost certainly not rise 5% (or 3% as I prefer.) Speculators would realize this and sell futures. The Fed would buy them and by T-bills, raising base money and raising nominal income and expected nominal income. To me, it is this sort of thing that generates cicularity.

    I would think that instead the Fed would of course underake open market operations to raise base money to match the trend growth rate in demand. This would preempt the need to speculators to sell futures and get the Fed to buy futures to motivate the open market purchases.

    But what if the Fed got it wrong. Now, it is true that we treat “the market” as a single speculator, then if that speculator trades because he Fed got it wrong and the Fed changes base money until they no longer think the Fed got it wrong and reverses their position–what is the point?

    No,the reason to trade is the expectation that there are other speculators out there who will get it wrong.

    Anyway, I don’t think there is circularity. (It isn’t exactly circularity anyway.)

    By the way, I think that what would happen with season facotrs (that will be reversed during the perioid) is that banks reserves would need to flucutate to meet seasonal currency demands. There would need to be enough base money to cover Christmas currency demand, and during the rest of the year, the banks would hold it as excess reserves.

    At least, I don’t see how targetting nominal income one year from now is going to do anything about seasonal fluctuations in the demand for base money.

    Of course, private currency issue would solve that problem just fine. Make it so that fluctuations in the demand to hold currency are not fluctuations in the demand for base money.

    Speaking of private money, last time I worked on this approach (years ago,) it was in the context of a privatized system where there was no unique money issuer, Each money issuer could take positions on the contract or else hedge as they choose. A bit fanciful, perhaps. But a trully market driven policy has no “circularity” problem.

    I have always been concerned about the “lag” issue. Why one year in the future? Why not six months? Or one quarter ahead. Or 18 months?

  5. Gravatar of ssumner ssumner
    14. May 2009 at 06:25

    Bill, You did a good job explaining why people trade, even if “the market” on average, thinks the Fed did (or will) get it right.

    I don’t quite follow the seasonal point you make. If seasonal demand at Christmas is normally 10% higher, then the Fed simply raises the base by that amount, before the start of trading. I couldn’t quite tell whether you thought that would work.

  6. Gravatar of Nick Rowe Nick Rowe
    14. May 2009 at 07:36


    “I have always been concerned about the “lag” issue. Why one year in the future? Why not six months? Or one quarter ahead. Or 18 months?”

    I am not sure if this is the right answer. One danger of trying to target NGDP at too short a lag is if prices are sticky (they are), and if real output is also a bit sticky (it is), then trying to bring NGDP exactly back to target too quickly (if some shock had forced it away temporarily) might cause very large relative fluctuations in those goods whose price and outputs were less sticky. Also, asset prices and interest rates might be very volatile. This is exactly the same reason (in principle) why the Bank of Canada targets inflation at an 18 month to 2 year horizon, normally, rather than trying to bring it immediately back to target if it wanders off target.

    There is *some* truth in the “long and variable lags” story, Scott. Even if asset prices can adjust very quickly, a lot of the components of NGDP cannot. So trying to make NGDP adjust too quickly would cause a lot of relative volatility between different components. The quickly-adjusting components would yo-yo.

    Or did I misunderstand you Bill?

  7. Gravatar of 123 123
    14. May 2009 at 12:59

    I have the same concerns about 1 year targeting, and I would be more comfortable with 1.5-2 year targeting.

  8. Gravatar of ssumner ssumner
    14. May 2009 at 16:51

    Nick, I completely agree, and in an earlier paper I discussed the costs and benefits this way:

    1. Short time horizon: More stable price level, but more relative price distortions as a relatively small number of flexible prices must adjust to offset shocks.

    2. Long time horizon: Less stable price level, but less distortion of relative prices due to price stickiness.

    I have a new post criticizing the “long and variable lags” view, but I do agree that sticky prices and output respond with somewhat of a lag.

    123, I don’t have any big problem with 18 or 24 months, indeed is might be better, as you say. I am focused right now on explaining the idea, and I suppose 12 months just seemed an easy way to explain it.

  9. Gravatar of Bill Woolsey Bill Woolsey
    16. May 2009 at 05:24


    Your explanation about short vs. long sounded to me like it was based upon price level stability.

    I realize that nominal income is made up up the price level and real income. (I prefer using final sales. While that leaves out all inventory investment, the _benefit_ is that it leaves off involuntary inventory investment. I think this is relevant to the question at hand. Goods that are produced but not sold to anyone… where does that fit into M*V?)

    To me, a long time horizon means than if MV is high now, sales are strong now, and that will stay high for a while, though in a year (or 18 months or 2 years) it will be back where it should be. It seems like 2 years is plenty of time for firms to start responding to this by changing their price and production, based on this gradually dissapating disturbance. That isn’t a good thing.

    With a short horizon, if it is high now, sales are strong now, but it will get back to target soon. Seems to me that the response should be to use up inventories. And then not bother to adust prices or production for a disturbance that won’t last very long. That seems good to me.

    I think you are thinking about Py. Some price rises, and this raises the price level, and given production decisions, Py is too high. And so dropping M is going to push down flexible prices (or ouput, which you forgot to mention above.)

    Anyway, I think that the increase in MV is going to be cleared by an increase in flexible prices and output. And if M drops fast, the disequilibrium increase in flexible prices will be reversed faster and the disquilibrium increase in output will reversed faster. What is the problem?

    Now, suppose there are sectorial shifts. So demand rises for something in particular, so output and prices start to rise. The “problem” then is that base money falls “too rapidly” so that other sorts of output fall too much and flexibly priced items fall too much.

    I know that you don’t like to think about interest rates, but I am pretty sure that particular goods that are especially sensitive to changes in short term interest rates have falling demand. And this lowers the prices of those with flexible prices (and production) and lowers the production of things with sticky prices. This happens enough to offset the increase in prices and production of the items in demand.

    Oh.. need to translate all this into slower growth in dmeand… right?

    Now, my question is…. why is the greater variance of prices and output in short term interest sensitive industries a bad thing? Changes in velocity are changes in the demand for money. IF this is a shift in asset preference, then failure of base money to accomoidate this results in excessively low short term interest rates. Why is rerversing this quickly a bad thing?

    IF it is a change in saving that happens to manifest it in a change in the demnad for money, why shouldn’t industries whose demand is sensitve to changes in short term interest rates contract rapidly to offset the effects of an expansion to increases in the demand for consumer goods.

    Well, maybe I don’t understand. But I am pretty sure that thinking about a price level targeting and the impact of supply side shock (and really, a price level shock) and rapid reversal, doesn’t apply very well to nominal income index futures convertibility.

    And, by the way, (for Nick,) my point is that the period at which the futures are targetted is important. In my earlier post, I said why not one quarter ahead or 18 months. (Shorter or longer) As Scott said, he is just explaining the idea and using one year for convenience. For years, I have been very concerned with this timing issue. (I think I understood the idea 15 years ago.)

  10. Gravatar of ssumner ssumner
    18. May 2009 at 04:09

    Bill, You raise some good points, and I am not sure whether I have a good answer. You are right that I was thinking about prices, not nominal output. It seems to me that a key question is once you get of the target path, how soon should you try to return? The farther out in the future you target, the more gradual the adjustment to any temporary deviation. Any quick adjustment will distort relative prices and output, because it is easy to change prices and output in some industries as compared to others. I just don’t know how severe this micro inefficiency would be in practice. I find it very plausible that the optimal targeting period might be very short.

    I still have an open mind about the output/final sales distinction. I think the reason I initially picked output to stabilize was that I saw employment fluctuations as the biggest cost of business cycles.

    Another answer is that this question is simply too complicated to answer without experimentation. We could start with a one year target, and then adjust as we were able to see how well it worked in practice. I think one reason why I choose one year is that I can see good arguments for both short (6 months) and long (24 months) targets. So one year seemed like a number that would be fairly safe, not likely to cause huge problems. As we know, the harm for macro policy mistakes is highly skewed toward the big screw-ups, so taking a “moderate” position initially has some merit.

  11. Gravatar of Is the Bank of England targeting nominal GDP? | Quotar Blog Is the Bank of England targeting nominal GDP? | Quotar Blog
    31. July 2009 at 21:06

    […] Sumner will surely be pleased to hear it: he has been advocating for a while that central banks should target nominal GDP. What’s more, they are looking not at the rate of change, but […]

  12. Gravatar of FT Alphaville » Fantasy Fed options FT Alphaville » Fantasy Fed options
    24. August 2011 at 07:08

    […] More on how that works here. […]

  13. Gravatar of rtah100 rtah100
    24. August 2011 at 09:55

    I don’t get this. Manipulating the level of base money in the financial system will improve the real economy *how* exactly?

    This proposal strikes me as financialisation in pursuit of “fiscalisation” of central banking, when the correct response to a debt deflation is for the government, the proper fiscal authority, to create non-credit money and place it directly in the hands of the non-bank private sector.

  14. Gravatar of ssumner ssumner
    10. September 2011 at 08:08

    rtah100, Don’t take this the wrong way, but more money means more RGDP (in the short run) is the standard model in economics. If you don’t agree with the model–fine, but tell me why. I can’t teach macro 101 to everyone who appears here, I don’t have time.

Leave a Reply