Early Warning System for Futures Market Nonconvergence
Buy low. Sell high. It's a foolproof way to make money. Unfortunately, in the real world, such opportunities are often illusory.
On September 2, 2008, traders could have bought wheat in Toledo OH for $5.49 per bushel. At the same time, futures traders were willing to buy the same kind of wheat in the same location for $7.44. It appears like there was an easy $1.95 per bushel of profit to be made by buying low and selling high.
This event was not fleeting, which is our first indication that it didn't represent a true profit opportunity. The plot below shows the futures price for immediate delivery minus cash price (known as the basis) for soft red winter wheat. The apparent profit opportunity persisted throughout 2008 and 2009.
Why didn't traders buy large amounts of cash wheat and sell it by delivering on futures? In theory, this would have pushed up the price of cash wheat and pushed down the futures price until they converged. But, that didn't happen for over two years.
In this article, I describe what caused this nonconvergence episode (and others), and why it didn't signal a true profit opportunity. In so doing, I'll introduce our new data app, which provides an early warning system for potential future nonconvergence events. I link to the research behind this work at the end of the article.
It is important to understand how futures market delivery works for grains. A futures contract obligates the trader to buy or sell a commodity at some point in the future and an agreed price. At the time of delivery, the seller does not deliver actual grain. Instead, she delivers the promise that when the buyer is ready to pick up the grain it will be available.
The buyer can hold this promise, known as a shipping certificate, indefinitely as long she reimburses the seller for the cost of storing the commodity. But, here's the key: the maximum storage fee the seller can charge is set by the futures exchange.
Sometimes the actual cost of storing the commodity exceeds this maximum allowable fee. In such cases, the only recourse is for the seller to charge more for the shipping certificate than the physical commodity is worth. Thus, when the storage fee is too low, futures prices for delivery of shipping certificates exceed the cash price of the physical commodity and we have non-convergence.
The Chicago Mercantile Exchange recognized this phenomenon and implemented a system in 2010 for soft red winter wheat that allows the maximum storage fee to adjust. The plot below shows that the price of storing wheat increased in 2008, which caused the non-convergence episode. When the futures storage rate increased in 2010, it solved the non-convergence problem. The price of physical storage also increased in 2016, but the futures storage rate adjusted before non-convergence became an issue.
Holbrook Working first showed in 1933 that the market price of storage is highest when inventories are plentiful because space is scarce, which makes it expensive to find space to store the next bushel. The plot below shows that non-convergence arises when prices are low or decreasing, which tends to be when inventories are accumulating.
The variable storage rate system is based on the futures spread, which is the difference between the futures price for delivery two or three months hence and the price for immediate delivery. The spread measures the price of storing the commodity over the two or three month period. If the price is $7.44 for delivery in September and $7.67 for delivery in December, the market value of storage is $7.67 - $7.44 = 23c.
The plot below shows the spread as a percentage of the maximum allowable storage fee. The non-convergence episodes occurred when this percentage was high.
Unlike wheat, corn and soybeans do not have a variable storage rate system; they still operate with a fixed storage rate. Early in the trade war in 2018-19, soybean inventories accumulated because of reduced demand from China. This plot shows significant non-convergence in this period. The basis reached 72c per bushel (about 10% of the value of the commodity)
The plot below shows that an increase in the physical price of storage provided a warning in 2018 that non-convergence in soybeans was coming. This year, the CME increased that rate from 5c to 8c per bushel per month, which will alleviate the potential for non-convergence, although 8c may not be high enough.
The plots of the market value of physical storage, along with the percent of full carry, are an early warning system for potential nonconvergence. We will keep them up to date.
These plots are available in our new Convergence data app for corn, soybeans, SRW wheat and HRW wheat. Huge kudos to UC Davis ARE PhD student Shanchao Wang for putting this together.
Non-Convergence in Domestic Commodity Futures Markets: Causes, Consequences, and Remedies. Adjemian, M.; Garcia, P.; Irwin, S. H; and Smith, A. USDA-ERS Economic Information Bulletin, (115). 2013. This paper provides an accessible discussion of the convergence problem and its causes.
Futures Market Failure?. Garcia, P.; Irwin, S. H; and Smith, A. American Journal of Agricultural Economics, 97(1): 40–64. 2015. This paper develops the theory behind and empirical evidence of nonconvergence.
Trilogy for troubleshooting convergence: Manipulation, structural imbalance, and storage rates. Irwin, S.H. Journal of Commodity Markets, 17: 100083. This paper updates the empirical evidence in the Futures Market Failure paper and discusses policy options.