Recent Posts

Wednesday, July 27, 2016

No Need For NGDP Futures, We Have Market-Based Monetary Policy Already

The "nominal GDP futures targeting regime" proposal has once again popped up in internet discussion. I have previously avoided the topic, as nominal GDP futures ("NGDP futures") targeting is an inherently silly idea. The only usefulness of the topic is that it allows us to examine the chain of errors that leads to the recommendation. The justification for nominal GDP futures is that the markets help set monetary policy -- which is exactly the situation right now. Which means that the reason for creating these futures in the first place is flawed; the fact that they cannot be implemented is just icing on the cake. This article discusses the issue of the interaction of markets with central bank policy, and does not attempt to delve into the explanation why nominal GDP futures markets would be useless for policymakers.

(The discussion in this article may raise eyebrows amongst my post-Keynesian readers, and it may appear to contradict what I have written elsewhere. The explanation for this is found later in the article; whether interest rate policy is truly market-based depends upon your stance regarding the debate about the effectiveness of interest rates in steering the economy. I am officially neutral on that debate.)


For readers who are interested in nominal GDP futures, I would suggest the following articles.
My view is straightforward: even if we wave a magic wand to make all of the technical issues highlighted by Michael Sankowski go away (spoiler: we can't), NGDP futures would still not work. No market structure which is tied to policy setting is going to generate useful information for policymakers, and the means to link the futures to monetary policy appears questionable.

Unless my readers want the painful details of why this is the case, I am going to skip to what I view as the interesting bits.

Market-Based Monetary Policy?

The entire premise of NGDP futures is that we need to get markets to set policy. A 2013 paper by Scott Sumner (which he cites in his article above) starts with:
In recent decades, there has been a worldwide shift toward market-driven economic policies, including privatization, deregulation of market access, bandwidth auctioning, congestion pricing, and tradable pollution permits. Yet monetary policy has been relatively unaffected by the “neoliberal revolution.” Governments have retained a monopoly in the production of fiat money, the setting of policy targets, and the implementation of monetary policy.
Unfortunately, that is a rather poor characterisation of how monetary policy works. Under normal circumstances, monetary policy is effected by interest rates, which drive the economy purely in a market-based fashion.

The only exception to this was the failed attempt by Monetarists to centrally control the level of money in the economy. Luckily, the free markets prevailed, and red-blooded capitalist financial institutions adapted their behaviour to frustrate that attempt to Sovietise credit creation.

Whenever someone walks into a bank to get a mortgage or business loan, they will be faced with a banker with quote sheets for interest rates on a wide variety of loan structures. Those bank interest rates are entirely priced off of market-based interest rates. The attractiveness of those interest rates will influence decisions that drive the economy. (Under the assumption that interest rates have a measurable effect on economic activity, of course.)

Err, What About The Central Bank?

The fact that a committee of government bureaucrats sets a policy interest rate appears to contradict this rosy pure free market Eden. In my view, this is not really an issue. Just because interest rates are market-based does not imply that they will trade where any particular investor wants them to be (such as self-described loudmouth "bond vigilantes" on financial television).

I will address this in two parts: firstly the question of open market operations, and secondly, the setting of the policy rate.

Open Market Operations: Should Be Market-Based

The central bank transacts in the fixed income markets as part of its normal operations (except for the case of an overdraft economy). These open market operations change the size of the central bank's balance sheet. These open market operations attract a lot of mysticism, because the central bank's balance sheet size is roughly the monetary base (ignoring some items, such as government deposits at the central bank). Since this is one of the "money supply" numbers, clear thinking goes out the window. (This inability to discuss "money" clearly within economics is the central theme of my next book: Abolish Money (From Economics)!".)

If we have a sensible financial system in which there are no bank reserve requirements, the monetary base is equal to currency in circulation. This is the basis of what I refer to as the "Simplified Framework of Government Finance" (link to article which was incorporated into Understanding Government Finance). The Simplified Framework is a good approximation of the normal operation of the modern Canadian financial system.

The amount of currency (notes and coins) in circulation is purely a decision of households and businesses. There is very little government policy can do to affect it (other than cracking down on the underground economy), although there is long-term correlation with nominal incomes. (This is chalked up to either "money neutrality" or a stock-flow norm; a debate I will discuss in the upcoming book.)

Within this system, the only open market operations that the central bank must do is to buy government bonds in exchange for demands for currency from private banks (who then distribute it to the rest of the private sector). Some operations may be needed to keep short-term rates near a policy rate, but the assumption is that these operations do not affect the size of the central bank's balance sheet. (I return to the policy rate below.)

Unlike the central planners of Monetarism, or the brain trust that is behind Quantitative Easing, a central bank within the Simplified Framework leaves its balance sheet size to be determined by the needs of the private sector for currency.

If we mistakenly insist on banks holding reserves, the picture is only slightly more complicated. The central bank needs to engage in extra operations to create new required reserves. However, the magnitude of these operations should be entirely guided by market signals. (For example, is the traded rate for fed funds greater than target? If so, add reserves.) It is only when policymakers do something that is blatantly stupid (for example, quantitative easing), that they have to think about the magnitude of open market operations. (In any event, the complete lack of impact of Quantitative Easing on the economy tells us that these operations do not matter a great deal.)

The Policy Rate -- Planned, Or Not?

 Obviously, a committee of government bureaucrats sets the overnight rate. (This is normally unchanged until the next policy meeting, so you might think of it as a six-week rate, so long as you embed an emergency rate cut/hike option into that six-week rate.)

The question is: how do they set it?

If we take the pre-Financial Crisis mainstream economic consensus at face value, the central bank is following an inflation target. The policy rate needs to follow the dictates of a "reaction function" which keeps the inflation rate near target. The central bank has some leeway in the parameters of its reaction function, but once it is picked, rates are determined by the conditions of the markets within the economy. That is, it is market determined, as the markets have set whether the economy is above or below potential, and the current set of inflation expectations, etc.

An alternative (mainstream) way of looking at the above argument is that the market determines the real interest rate; the nominal interest rate is indeterminate. The government sets the inflation rate by policy, and nominal rates have to end up where they are by the assumed action of the Fisher effect. Forcing the government to keep the price level (or gold price level) stable is exactly what free market zealots have been screaming about for decades.

We can easily argue that real world is complicated; there are various qualitative factors that affect the reaction function. This is not particularly novel; qualitative factors always influence behaviour. The desire pursue virtual monsters on cell phones is probably going to damage productivity growth over the next few months; it would be hard to quantitatively model a "chasing virtual monsters" factor into a DSGE model.

It is no secret that I do not take the pre-Financial mainstream economic consensus at face value. For skeptics like myself, there are two options to pick from.
  1. Monetary policy roughly acts in the same way that the economic consensus believes. (Higher rates slow the economy; lower rates boost economic growth.) If the policy committee sets the overnight rate at the "wrong" level, interest rate market participants can signal the error by moving the yield curve. For example, if the policy rate is "too low," the curve would steepen, which raises term rates for borrowers in the real economy. The market action tightens financial conditions, which means that the interest rates that matter are not fully affected by the policy rate mistake. If the central bank is indeed wrong, it will later miss its inflation objective, and will have to follow the path of pricing indicated by the forward curve.
  2. Monetary policy is ineffective in controlling the economy. In this case, who cares where they set the rate of interest?
In other words, if you think interest rates matter, they will have to end up close to where the market is supposed to have put them in the first place. Given that the credit and equity markets can misprice assets for years, industrial capitalism can survive a mispricing of the overnight interbank rate for six weeks.

If you are skeptical about the importance of interest rates (such as many MMT'ers), yes, the central bank has considerable latitude to determine where interest rates are set.

Throwing Out Market Information

My discussion here is somewhat idealistic; I assume that central bankers pay attention to market signals about interest rates. This ignores the reality that a spectacular amount of mathematical work has been undertaken to obliterate the information that is easily extracted from market prices.

On the analytical side, half-baked affine term structure models are used to explain that most yield movements are the result of term or inflation-risk premia. "Sure, breakeven inflation rates have collapsed, but that does not mean that we are making a policy error -- that is just inflation risk premia moving!"

Meanwhile, we have been treated to the spectacle of hawkish central bankers showing off their inability to forecast inflation by lecturing bond market participants that they are setting forward rates too low (for about six years in a row).

The obvious solution is to force a return to a culture of secrecy and opacity at the central bank; all the Federal Reserve System should have is a two sentence policy statement (one for Fed Funds, one for the discount rate).

Advantages Versus NGDP Futures

Circling back to NGDP futures, the current framework has two huge advantages versus them.
  1. There are extremely deep and liquid yield curves that provide strong signals to policymakers where they should set the policy rate. (When is the last time that the Fed set a rate that was wildly different from where Fed Funds futures were trading, not counting emergency rate cuts? All the argle-bargle by policymakers could just be a sham; they could just look at the Fed Funds futures strip and then get some interns to make up stuff for the policy statement and minutes.)
  2. The forward rate curve provides guidance in exactly the form needed to set the policy rate. There is no need to translate the market signal into new units. (If the NGDP futures market transacts 100 contracts at an implied growth rate 0.5% below target, what does that imply about changes to the stance of monetary policy?)
The fact that there would be no useful market signals provided by NGDP futures was implied by Michael Sankowski's critique, but it could be extended to any conceivable market framework. There is no way the Fed would get useful information from any market based on predicting nominal GDP, it it were tied to monetary policy in the manner that Sumner suggests.

The second issue -- that the units of NGDP futures are mismatched versus policy instruments -- is what I find most interesting about this topic. Scott Sumner's discussion of how monetary policy would be affected by NGDP futures is yet another example of why money needs to be abolished from economic theory. (I hope to discuss that topic within my book.)

See Also:

(c) Brian Romanchuk 2016


  1. I don't understand how or if the NGDP futures market would work. But would there be a large downside to experimenting with it? If not, then why not try and if it didn't work, then at least we learned something. And if it did work then that's just much better. The Fed can afford to lose a few hundred million, it costs them nothing to create dollars. Of course, I would be happy to make something up to explain it to them for much less. Say 20 million. I would need help presenting something that sounded reasonable though... what would your consulting fee look like? (Aim high, I hear the Fed has more money than God)

    1. I would note that this article discusses "convertibility", which as I show, does not work if
      NGDP futures are structured like other futures. This matters, as this is apparently the preferred structure of Scott Sumner, so I have to assume that he envisions the structure as being different than a market participant would understand what a "NGDP futures" would be.

      NGDP futures without Fed involvement could be tried, and they would end up in the dustbin along with CPI futures. (There was an attempt to build up a over-the-counter economic derivatives market before the Financial Crisis; that push died when all the derivatives traders were busy scrambling to stay alive.)

      Central banks target inflation now, and they have a relatively deep market in inflation-linked bonds now. This was supposed to give the Fed information to allow them to hit their target. Guess what? Chair Yellen has downgraded inflation breakevens to "inflation compensation," and resolutely denies that the market is saying that she is doing her job wrong. At the same time, wonks at the Fed are inventing complex models to obliterate any information embedded in the yield and breakeven curve - which coincidentally also makes it look like they are doing their job better. In summary, the existence of a purely private NGDP future market would prove nothing, since the brain trust at the Fed would just invent excuses to ignore the pricing.

      As for tying it to policy, I was going to cover that in a later post. Without a fixed contract and intervention structure, it is impossible to judge whether it would work. I have not seen such a fixed structure from Scott Sumner (not that I looked very hard), and so it is difficult to write about. I do not see any structure as being feasible, but once again, that would be a big article.

    2. With regards to the Fed losing money, it's very easy to turn small losses into very large losses using derivatives. Although the Fed cannot go bust, they do have many Republicans breathing down their neck. Handing a couple hundred billion over to trading desks on Wall Street would not look particularly good in the current political environment...

    3. You are of course right that handing over MORE money to Wall Street would not look good politically. Nor would it be good as far as I am concerned. But estimates of what the Fed did during the great recession range from "well, they made money" to Randy Wray's 29 Trillion worth of intervention commitments. Now I realize that those commitments were not all cash handed out, but they for sure were worth a whole lot. I think that even I could make a pile of money (or avoid losing it) if the Fed was backstopping me.

      But anyways, having read Sumner for years now, I feel that he uses the idea of a NGDP futures market primarily as a means to push the Fed towards NGDP level targeting, rather than strict inflation targeting. I can't see how a test of that idea would end up handing a couple hundred billion to Wall Street. But if that is a reasonable outcome, I would agree that that constituted a large downside to an experiment.

      An NGDP futures market appeals to those who hold two premises. One is that the Market will almost always come to a better solution than any individual or committee will. The second is that Monetary Policy can Always be very effective if it is just used in the right way at the right time. Although I have many, many doubts about these premises, supporting an NGDP futures market of some type seems logical if they were in fact true. Oh, and of course one more premise is that NGDP level targeting is better than current central bank policy.

      I am looking forward to your next post on the subject. And also as to your opinions about the worthiness of NGDPLT as a (new?) goal for the central bank. With or without a futures market.

    4. I will have to get back to this later; some of this was going to be a section of my book.

      NGDP targeting (without a level target) is probably what the Fed should be doing. There are political issues with it, and one could question whether they can get it to work. If inflation is supposed to be 2%, that gives me useful information about structuring my affairs. Nominal GDP growth of 5%? Since most people have no idea what household growth is, nor can they forecast business revenue, what information does that give me? The idea only really works in neo-classical models, where the model structure is simple, and known with absolute certainty.

      The level target raises problems. During the aftermath of the financial crisis, it sounded like one way to get the economy moving (if you believed neoclassical models...). However, it would involve some serious acceleration/deceleration during normal operation of the economy. The central bank would be treated with serious disdain if it tried it.

      Meanwhile, when people did the numbers, the pre-2008 inflation targeting regime had almost exactly the outcome that NGDPLT was supposed to produce, without the downside.

  2. Brian I'm intrigued by your proposition of "abolishing money (from economics)". I'm not sure what do you mean by that, isn't neoclassical position basically this one, although disguised?

    Most economics (mainstream, and not so mainstream, as heterodox is arguing that mainstream is wrong specially about this) are a pretending exercise that money does not matter, so it's abstracted away ('abolished') while they talk about only money, in reality. They do this assuming that Say's law is true (although we know is false, even Say's himself said so).

    But if you agree that Say's law is false, you cannot really abolish money from economics, you cannot ignore it, as it's an active object of economic exchange itself. Add to this the capital controversies.

    I think economics should deal exactly what in theory it says it does, instead of pretending it does deal with real economic output and distribution, but instead is just a collection of assumptions and underlying it all is money. This would pull economics closer to both social and natural sciences (paradoxically), dealing both with material and inmaterial relations and distribution issues, so I think the importance of money should be greatly diminished and mostly ignored, but it cannot be completely abolished (because it forms part of the social domain, as we know precisely because the denial of Say's law).

    Sorry for the tangential topic, I guess I'll have to wait for your book to find out what do you exactly mean with this proposition.

    1. My idea is that money should have no special status within economics, it is just another liability.

      This is not news to people in the financial markets; they have been ignoring money supply numbers since Monetarism jumped the shark. (it also matches what you wrote.)

      But economists? Look at QE, helicopter money, gold standard adviocates, bitcoin, 100% reserve banking, and of course, all the varieties of Monetarism. The magical power of money still acts upon economists (and techno-libertarians).

  3. This idea cannot be from Scott Sumner. You must be wrong. He is AGAINST voodoo economics.


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.