The Negative Pricing of the May 2020 WTI Contract.

• Author: Fernandez-Perez, Adrian

1. INTRODUCTION

   This article attempts to shed light on the anomalous pricing of the May 2020 NYMEX West Texas Intermediate (WTI) crude oil futures contract, denoted as CLK20. The price of CLK20 fell dramatically from $18.27 (April 17, 2020) to -$37.63 (April 20, 2020), effectively meaning that sellers paid buyers to take crude oil barrels off their hands. (1) The price then climbed back to $10.01 at maturity (April 21, 2020). It was the first time that a WTI futures contract had experienced a negative price since trading began on March 30, 1983. This article tries to uncover what happened not only in the few weeks preceeding the negative pricing but also on the day that the price turned negative.

   The crude oil glut inherited from the 2010s was exacerbated in the early months of 2020 by demand shattered by COVID-19 lockdowns and an oversupply exaccerbated by geopolitical tensions between Russia and Saudi Arabia. Against this background, the WTI futures market steered into a super-contango state; namely, the futures-spot spread became very positive, exceeding its 95th percentile on March 23, 2020. The super-contango that then prevailed incentivized cash and carry (C&C) traders to open long positions on CLK20 and short positions in more distant contracts, while simultaneously booking storage at Cushing, Oklahoma, the delivery hub of WTI futures contracts. As corroborating evidence for the presence of C&C arbitrageurs, we note that from as early as March 5, 2020, the futures-spot spread exceeded the cost of financing and carrying the spot asset, and the profit and loss (P&L) of the C&C arbitrage strategy thus turned positive. To substantiate further the presence of C&C arbitrage, we borrow and extend the research framework of Ederington et al. (2021) and test whether crude oil inventories rose at Cushing in March and April 2020 in response to the widening of the futures-spot spread. We note an increased participation of C&C arbitrageurs around that time: The response of inventories to changes in the futures-spot spread was 4.3 times stronger in March-April 2020 than it typically would be, suggesting that C&C arbitrageurs were then particularly present. We also bring forward evidence that exceptional C&C arbitrage took place outside Cushing, suggesting that, as spare storage capacity started to diminish at Cushing, C&C arbitrageurs considered other hubs as depositories for the crude oil that they had "collected" through their long positions.

   Even within one day of the negative price event, some energy market commentators (2) had blamed index traders for distorting the price of CLK20. The line of reasoning advocated by these market pundits was simply that by rolling their long CLK20 positions to more distant contracts, index traders had triggered the negative pricing of CLK20. We substantiate the veracity of this claim--or lack thereof--in two directions. First, as a direct test of the price impact of long index trackers, and in reference to the United States Oil fund (USO), the largest WTI crude oil exchange-traded fund, we investigate whether its flows influenced either the return or the change in the volatility of CLK20. Second, as an indirect test of the price impact of index traders, we examine whether the prespecified rolling of long index positions ahead of maturity impacted futures-spot spreads in March and April 2020, thereby triggering further C&C trades and indirectly contributing to the observed negative pricing. Both tests conclude that there was an absence of price impact and index traders cannot thus be made responsible for disrupting the price of CLK20.

   On April 20, 2020, or one day before maturity, a large number of open positions combined with the lack of storage at Cushing created an unprecedented problem of illiquidity. Long CLK20 traders who had not secured storage at Cushing either had to pay an exorbitant cost for storage, if any spare capacity was still available, or close their positions at any price. In the end, most of them chose to close their positions at negative prices. As Nagy and Merton (2020) put it, WTI crude oil became like "toxic waste or even garbage" to long CLK20 traders without prebooked storage on April 20, 2020. In effect, long traders cannot just dump crude in a lake or ocean to dispose of it; for that reason, they deemed paying up to $37.63 per WTI crude oil barrel sold to be a lesser loss than taking physical delivery. Aggravating factors included i) the staggering margin calls that long traders inexorably had to pay as the price of CLK20 fell and ii) the likely price distortion that occurred as a consequence of the trading-at-settlement (TAS) mechanism. Under that mechanism, it became profitable to buy CLK20 at TAS price during the April 20, 2020 trading day and simultaneouly sell it outright so as to push the TAS price down, offsetting both positions at settlement. This only served to aggravate the price fall.

   Our findings speak to the empirical literature on the theory of storage (Kaldor, 1939; Working, 1949; Brennan, 1958) (3) by bringing indirect evidence that the law of one price implied by the cost-of-carry model does not hold in the presence of storage constraints. In so doing, it complements the analysis of Ederington et al. (2021) by focusing on the anomalous negative pricing of CLK20, and by showing that limits in the availability of storage (i.e., reaching the nightmare scenario of exhausted storage) can drive futures prices into negative territory. Practical implications from this research include lessons to traders with long front-end positions (irrespective of whether they engage in C&C trades) right before maturity, calling them to exert caution in super-contangoed futures markets, since at maturity, the long position can suddenly become unfeasible if the asset cannot be physically stored. Likewise, traders not seeking to take physical delivery (e.g., long index trackers) need to exert caution in rolling their long positions sufficiently ahead of maturity to avoid being caught in such liquidity freeze-outs in the future. Our findings thus call for regulators to monitor the long positions of traders close to delivery so that they do not dislocate the natural convergence of futures and spot prices at maturity. To prevent price distortion and market abuse through the TAS mechanism, it might also be of interest for regulators to limit the netting of speculative TAS positions with speculative outright positions during the contract delivery month. This would ensure the integrity of the TAS pricing mechanism.

   Our findings also speak to the literature on the financialization of energy futures markets by showing that index traders, such as USO, did not drive the price of CLK20 away from its fundamental value nor distort the futures-spot spread while rolling their positions ahead of maturity (see, e.g., Till, 2009; Buyuksahin and Harris, 2011; Irwin and Sanders, 2012; Fattouh et al., 2013; Hamilton and Wu, 2015; Brunetti et al., 2016). This suggests that calls for the regulation of long index trackers might be, at this stage, premature. Further regulation could, in fact, be detrimental, as it may discourage index trackers, who are important providers of risk-absorption and liquidity, from trading crude oil futures.

   The article unfolds as follows. Section 2 describes the data. Section 3 briefly presents the macroeconomic and political environment of early 2020 that led to a super contango. Sections 4 through 6 analyze the positions of C&C arbitrageurs (Section 4), index trackers (Section 5), and speculators and hedgers (Section 6) and test whether the long CLK20 positions of these traders influenced inventory levels and price formation in the days preceeding the crash. Section 7 explains the dramatic price swing observed in April 20, 2020, and Section 8 concludes.

2. DATA

   2.1 Futures Prices

   Our analysis relies on the daily settlement prices of all 446 NYMEX WTI crude oil futures that were traded over the period from March 30, 1983 to June 29, 2020, as provided by Refinitiv Datastream. These prices are used to measure the returns of various contracts using the formula ([F.sub.t,T] - [F.sub.t-1,T])/|[F.sub.t-1,T]|, where [F.sub.t,T] is the time t settlement price of a contract that matures at time T. The use of absolute prices in the return definition is dictated by the negative price observed on April 20, 2020. For comparison with the price evolution of the WTI contracts, we also obtained the daily settlement prices of front- and second-nearest maturity futures contracts on Brent crude oil over the available period from December 12, 1988 to June 29, 2020.

   To demonstrate the lack of liquidity of CLK20 in its last two days of trading, we downloaded from Refinitiv Tick History intraday best bid, best ask, and transaction prices at a one-minute frequency for CLK20 and its two adjacent WTI contracts (with April 2020 and June 2020 maturities) over their last two days of trading. (4) As another proxy for storage costs, we adopt the prices of ICE Permian WTI storage futures contracts which give the holder the right to store 1,000 barrels of Permian WTI crude oil at Magellan's terminal in East Houston, Texas. The data are daily front-end settlement prices from the inception of this futures market on March 4, 2019 until June 29, 2020 that we obtain from Refinitiv Tick History.

   2.2 Control Variables

   Throughout our analysis, we investigate whether a certain group of traders impacted inventory levels or the price and volatility of CLK20. The analysis is implemented after accounting for various oil market-specific and macroeconomic factors that have been shown in the literature to explain inventory levels and/or price commodity futures contracts (e.g., Singleton, 2014; Algieri and Leccadito, 2019; Wimmer et al., 2021). The factors that we consider as control variables are: i) the futures-spot spread of Erb and Harvey (2006) and Gorton and Rouwenhorst (2006) defined...