Scott Sumner has a new job where he will be free to focus on monetary policy and NGDP futures. Good for him.
NGDP isn’t a terrible way to help think about monetary policy. However, NGDP futures are a dead end. They cannot work.
I feel a bit bad for the person who is funding Scott S over the next few years, because they will spend a decent amount of money finding out something I tried to give away for free.
There are only a few possible contract designs for NGDP futures. All of them have deadly flaws. The flaws are not solvable with technology or new rules, because they happen at the underlying level. The problems happen outside of the realm of futures contracts.
NGDP Futures are *Naturally* Illiquid
One huge issue which is not easy to overcome for any futures contract is the use of the contract as a hedging vehicle. ALL successful futures market rely on a large pool of market participants using the futures contract to hedge real world money losses. What does this mean?
Hedging is the exchange of unknown prices in the future for a known price today, all because there is a firm relationship between real world money losses/gains and the hedge contract.
This is where NGDP futures have the most problem – there is essentially no company in the world who has a firm enough relationship between NGDP of the economy and some aspect of their business.
As a result ALL NGDP market designs start with a severe handicap of an illiquid market. Every design has to somehow overcome this problem. There is no technology which makes this problem go away. The relationship in question is not at a technology level – it’s at the legally defined contract level. There is no way to legally define a contract which has a high enough money relationship for hedgers because the real world relationship between NGDP and their profits and losses is not high enough AND not solid enough.
It’s important to remember right here there are two giant parts of statistical relationships. First, is how much do they co-move? That is when one moves, how much does the other part of the relationship move. Second, how much of the movement in the second is explained by the movement in the first?
For hedges to be effective, both of these relationships have to be relatively strong. Unfortunately for NGDP futures, both of these relationships are weak for essentially all businesses. NGDP movements could be related to profits and losses, but the relationship is so vague in a statistical sense, no trader in their right mind would recommend using it to hedge.
For example, S&P 500 futures are a success story, because mutual fund managers can easily figure out how many futures they need to buy for fund inflows and outflows, AND as those futures move, the amount of money the futures make or lose will closely approximate the gains/losses of the hedged portfolio over minutes, days, weeks, and months.
This cannot exist for NGDP futures, because the relationship between NGDP and, well, anything else companies want to hedge is so mushy.
Companies would not be able to figure out how many futures contracts to use. Then, once they had that number of contracts, the money link between profits and losses in NGDP futures and their real world numbers would be weak. Essentially, this would be speculation because the links are so vague.
You might say “how about corporate profits”? I think it’s important to keep in mind the purpose of a firm at this point. Firms exist to take idiosyncratic risk in specific industries. If you are hedging the profits you make in the business- are you still in that business? Hmmm.
So it is extremely likely NGDP futures markets will be handicapped with illiquid markets. This is an enormous problem, because illiquid markets are prone to being manipulated!
Problems because they are Illiquid
George Soros and Charles/David Koch have some of the largest and most sophisticated trading operations on the planet. Additionally, there are thousands of other people with highly sophisticated trading operations, who may have political interest in monetary policy.
Remember, the point of NGDP futures is to set monetary policy. Monetary policy is a powerful lever for the political landscape, and also helps set the interest rates on the rest of the government bond yield curve.
All of these people many want to influence NGDP futures in a way which has absolutely nothing to do with the actual level of NGDP at contract expiration. They may be willing to take rather large losses to express this view in the NGDP futures contracts.
You can easily make up scenarios where people choose to lose $10 million to influence an election in an illiquid market.
Problems with Bizarre Incentives to participants
Just asking, how would Pimco play the NGDP futures markets? And why in the world would we want Pimco to play in NGDP futures? My strong guess is bond guys *might* sometimes play in NGDP futures markets to influence the short end of the curve in ways which benefits them, and not ways which accurately predicts NGDP! They would be willing to take losses on NGDP futures to influence their far larger bond portfolio.
This is not hedging. They won’t be hedging money risk because there is no strong relationship between NGDP and interest rate levels, or even NGDP and changes in interest rate levels. Of course, they would try to strip out the inflation component of NGDP, which is really the part of NGDP which matters to them. They would do this with TIPS or regular old Treasuries (and futures on those).
This problem exists because monetary policy influences NGDP via interest rates. Unfortunately, the relationship between monetary policy and interest rates is far, far stronger than the relationship between monetary policy and realized NGDP.
This is due to the weak links between NGDP and everything worth hedging, and the weak link between monetary policy and NGDP.
Too liquid with people who do not care about NGDP
The next, obvious question: “Hey Mike, you said the markets were illiquid, now you seem to be worried too many people will participate! You’re contradicting yourself!”
Well, no. I predict either: nobody will participate, or lots of people who do not care about NGDP will participate.
What I am claiming is that any liquidity in the NGDP Futures market will come from firms who have close to zero concerns with the settlement price of the NGDP futures contract and therefore almost no concern with actual, realized NGDP.
For a futures market to work, someone, somewhere has to care about where the real world market settles in the first place. Ok, fine.
Then, someone has to force these two markets to the “same” price as settlement gets closer – they must converge because someone who cares is going to take delivery at expiration price.
I am worried about a market where zero of the participants care about the settlement price of either the futures contract or the underlying NGDP.
We care about underlying NGDP. But the participants in NGDP futures market probably will not care about where NGDP settles. This is a problem. This is a huge problem.
This could easily happen because the impact of NGDP futures on short term interest rates and/or the political process will be far larger than the impact on realized NGDP itself!
The impact NGDP futures markets (if they are linked to actual fed monetary policy) for interest rates is short and measurable, while the impact of monetary policy on actual NGDP is “long and variable”. This is due to the statistical relationships we talked about a few paragraphs ago. The relationship between fed policy and short term interest rates is far stronger than the relationship between fed policy and NGDP.
It is a fundamental flaw for NGDP futures if we want NGDP futures to impact NGDP. NGDP futures will impact short term interest rates more then NGDP. It’s a huge flaw.
This is not solvable by technology or contract design.
It *might* be solvable by some sort of improved monetary policy lever which instantaneously impacted economic activity. The fed could link this imaginary policy lever to NGDP futures which would then react more quickly than interest rates.
But if we had that policy lever, NGDP futures would not be necessary in the first place, right? We could just increase our frequency of measurement of NGDP and use that policy lever. No need for speculating on the future of NGDP – the link between this policy lever and realized NGDP is fast. So we just measure NGDP, use the lever, and presto, NGDP is fixed.
What we should do instead of NGDP futures (an aside)
We already measure payroll tax receipts in real time. These are put into tax accounts every day, and are never revised. It’s a near real time measure of how much wages are being paid across the U.S. This tax data, combined with Billion Prices Project daily information, is almost certainly enough real time data to set policy.
In fact, these two data points are unlikely to be improved upon, ever. Additional real time granularity – say hours, minutes, or seconds – won’t give us a better picture of how inflation is unfolding than current BPP. “How much taxes paid in a single day” is not improved by having hourly information on tax payments. The data is already never revised.
If we have to make decisions faster than this data requires, the economy has worse problems than adjusting monetary policy could solve.
So we could just use realized NGDP to change monetary policy, and let the lags take care of themselves. Better yet, we could directly monitor the two numbers we care about, inflation and people making money, and adjust policy based on these numbers.
We could do this today, with the policy levers we have at hand. When above the zero bound, use monetary policy. When at the zero bound, send out checks to everyone monthly.
This is real time data, not based on speculation. Given the long and variable lags of policy, why risk speculators being crazy wrong about the most important part of our government? We already measure what we need to know in real time. We could just deal with it as it comes our way, using information we already collect.
We could improve this tax information simply by requiring companies to report the number of employees this particular payroll tax payment covers. That small additional information would give us powerful real time data on the economy, and probably could not be improved, ever.
Using crazy contract designs to overcome the basic problem
The big problems with NGDP futures are due to the lack of a useful relationship with anything anyone wants to hedge, and the stronger relationship monetary policy has with interest rates.
To overcome these dual problems, Scott has proposed a few market/contract designs. He’s not hugely consistent on this and shifts between several designs. Something to keep in mind is “you can only choose one.” Some of these designs have benefits, but have other flaws. In the end, you can only trade one contract, so you are stuck with the “worst” flaw of each design.
This is the old Theory of Constraints, applied to futures contract specifications.
Here are some of the designs he proposed.
1. Price never changes of the NGDP futures, but the Open interest does. Fed acts as market maker, willing to buy/sell at the target NGDP price.
Problem: Speculation! You can’t lose money here! Why not try to influence monetary policy for profit in interest rate markets. Why not influence presidential elections – you can’t lose!
2. Price changes, Fed is market maker with specific amounts at each price.
Problem: Fed policy has stronger impact on short term interest rates than it does on NGDP. People will bet their short term interest rate book. Political speculation still possible and even likely
3. Price changes, fed not market maker.
Problem: Illiquid markets. Nobody has incentive to trade. if they do trade, Interest rate speculators and political speculators want to influence other market besides realized NGDP
4. Any contract structure, Fed incentivizes liquidity by paying interest rate.
This is the one which bothers me the very most. It is infuriating. Yes, infuriating.
Scott would have people deposit money at the fed and get an interest rate on that money. This should sound familiar to you, because this is almost exactly what T-Bills do.
T-Bills can be considered tradable, term, interest bearing, risk free savings accounts. Scott proposes setting up term, interest bearing, risk free savings accounts – and then have the owners of the accounts tell the fed what monetary policy should be.
As soon as the interest rate on these accounts was higher than the rate on T-Bills of similar tenor, people would sell T-bills and flood into the accounts, then tell the fed what they want short term interest rates to be. Of course they would! They have a term, interest bearing, risk free account with a higher interest rate available to them!
This is a caviler and thoughtless proposal for anyone to make. Disrupting the U.S. T-Bill market is not something undertaken lightly and would impact fed monetary policy in unknown ways.
The short end of the nominal curve would be directly set by fed interest policy on NGDP futures – and everyone participating in the short end would be compelled to give an opinion on NGDP. The T-bill market – one of the main ways the federal government finances spending – would disappear.
Paying above market rates on futures margin for NGDP would undoubtedly be the largest change to U.S. fiscal and monetary policy, ever. It would be an incredible change in how monetary policy manifests, and have unknown and gigantic consequences for government financing.
Scott makes this proposal without discussing the fairly obvious and certainly important possible consequences of the proposal, hence my fury for this proposal. How can you be concerned about NGDP if you make a proposal like this?
5. Price does not change until settlement, Fed is market maker. NGDP futures settle and give/take cash from participants
Problem: Staggeringly huge Windfall profits for last minute traders
Keeping the price at one level until settlement almost would give sophisticated traders massive windfall profits. It would be very profitable to “bet with the crowd” and reverse direction last minute.
Bet with the crowd, force interest rates to some crazy level. Talk about your plans openly on CNBC. You are not colluding with other traders, just letting the public know your plans.
Of course people would do this – it’s free money for them.
Feel free to make your own flow chart of possible contract and market structures. You won’t find some magic combination of specifications which makes all of these go away.
No Thread thin enough for NGDP futures
There isn’t some way to solve these problems with NGDP futures. Launching actual NGDP futures does not make these problems disappear, nor does changing contract design. The low viability of hedging with NGDP combined with high impact on short term interest rates means disaster.
Any possible contract design must deal with some form of fatal problem. Most of the problem is due to the weak and variable link between NGDP and anything people and companies might want to hedge, and the far stronger link between monetary policy and short term interest rates. This weak relationship cannot be fixed by playing with the contract/market design.
Either nobody will participate in NGDP futures – which opens up for speculation. Or, the participants will not care about where NGDP actually settles which is the entire reason to have an NGDP futures market.
There is no thread thin enough to get through a hole which does not exist. NGDP futures have flaws which cannot be fixed by any contract design.
We already have better information than NGDP futures would provide
Additionally, we have real time data for taxes and inflation. We can use this real time information to use traditional policy levers to impact NGDP. There is no reason to rely on speculative markets to tell us anything – we already have very close to the best real time information possible.
This real time information could be improved to the best possible with a single additional amount of information being collected.