تخفیف قیمت، موجودی و انحراف از معیار WTI

Applying a rolling estimation to the Hasbrouck information share model, this study investigates the changing status of WTI benchmark over time. Results show the ability of WTI in reflecting market conditions decreases sharply, and WTI's efficiency in processing information has been surpassed by Brent's since the second half of 2004. In the short run, the WTI distortion is related to its price discount problem, but the distortion cannot be indicated by contangos. In the long run, WTI's price discount problem coexists with a positive forward curve and both have harmed the price discovery role of WTI. The rising inventories in Cushing significantly deteriorate WTI's ability in serving as a world benchmark.

The usefulness of the West Texas Intermediate (WTI) benchmark in pricing the crude oils has been increasingly debated in the world oil market because of the frequent dislocations of WTI from other major benchmarks. From early 2006, WTI has been occasionally priced at large discounts to Brent and other grades of oils (including Dubai). These large discounts violate the traditional pricing principles that command a premium on WTI to reflect the better quality of oil delivered, and to account for the cost of shipping crude oil across the Atlantic.1 Doubts on the benchmark status of WTI are expressed by Bentzen, 2007, Hammoudeh et al., 2008 and Kaufmann and Ullman, 2009 who argue for the inability of WTI to reflect market conditions. But these arguments contradict the results put forth by Gülen, 1999, Brunetti and Gilbert, 2000, Lin and Tamvakis, 2001 and Lin and Tamvakis, 2004. Market participants often attribute WTI dislocations to the logistic constraints at Cushing, Oklahoma — a delivery and settlement point of WTI futures contract (Fattouh, 2007, Fattouh, 2009 and Fattouh, 2010). The bottleneck in Cushing's ability to shift oils out of the region has sometimes caused a larger-than-expected build-up of crudes in Cushing. Rising crude oil stocks at Cushing depress WTI prices in both the physical and paper markets. In addition, a self-feeding ‘reinforcing feedback’ of local storage built-up is inflamed by exploiting the contango price arbitrage with a ‘cash and carry’ strategy. The reinforcing contango leads to a continuous decoupling of WTI from other U.S. and international crude grades, making WTI ineffective in hedging crude oil futures. Fattouh (2007) argues as a consequence that WTI is a ‘broken benchmark’. Although discussions regarding the appropriateness of using WTI as a pricing marker are intensively carried on the media,2 little attention has been paid to the evidence relating the glut of Cushing's inventory to the inefficiency of WTI benchmark. This paper is going to prove the relation of inventory problem and dislocation of WTI, and to examine how the reference place of WTI changes over time. We apply the Hasbrouck (1995) information share model to a rolling estimation process in order to analyze the relations between the benchmark status of WTI, the inventory accumulations in Cushing and WTI price discounts to Brent. Quantitative analyses are used to test the following hypotheses: (1) The WTI benchmark has been ‘broken’. (2) The price disparity between WTI and Brent (i.e. the WTI discount problem) is a warning against using WTI benchmark. (3) The crude oil stocks in Cushing prevent WTI from serving as a benchmark for the world oil market. Using Hasbrouck information share model to investigate the benchmark role of WTI is supported by Kaufmann and Ullman (2009) in which a benchmark is defined as the market from which the price changes first appear, and toward which the prices of other crude oils equilibrate. Since the information share given by the Hasbrouck model denotes one market's efficiency in reflecting relevant news for markets in a cointegrated system, the Hasbrouck (1995) information share model has been widely used in the literature to discuss market efficiency.3 If one market is more information-efficient, it will get higher information share for reflecting a better ability in indicating future price changes. In a cointegrated system, the market with the highest information share acts as the price discovery center. Empirically, evidence on the price discovery center (or equivalently, the benchmark) for world oil markets has not been drawn, yet. Earlier studies that dispute over the globalization/regionalization hypotheses tend to acknowledge the legitimacy of WTI rather than Brent in determining the world oil prices. Gülen, 1997 and Gülen, 1999 indicates the benchmark role of WTI and Brent comes from the large progress in the futures markets respectively launched by the New York Mercantile Exchange (NYMEX) and the London International Petroleum Exchange (IPE, renamed as the IntercontinentalExchange/or the ICE later) in the 1980s. The largest trading volume on the NYMEX's crude oil contracts has in particular contributed WTI to be more relevant in pricing the crudes than Brent. Other researches that use GARCH-type models give similar conclusions. For example, Brunetti and Gilbert, 2000 and Lin and Tamvakis, 2001 demonstrate the WTI contracts can efficiently incorporate London's information into its price dynamics, and therefore suggest WTI is superior than Brent in leading the oil price. However, recent studies that extend the investigation period beyond the year of 2000 express controversial opinions about the world oil pricing center, most of which overrule the position of WTI as a world benchmark. For example, Hammoudeh et al. (2008) indicate Dubai is more influential in directing the oil price dynamics than WTI and Brent. Bentzen (2007) highlights the increasing influence of OPEC on the prices of light crude oils. Kaufmann and Ullman (2009) find the world oil market is bi-centric. The first center is the spot market of Dubai-Fateh, and the other is the far month trading of WTI. Market innovations first appear in the two centers, and subsequently spread to other regions. In this paper, the changes in the reference place of WTI are investigated by a rolling estimation over the period from October, 1991 to February, 2009. Our results show the ability of WTI in processing and reflecting market conditions decreases dramatically over time. Additionally, since the second half of 2004, the information share of WTI has been surpassed by that of Brent. The WTI distortion is highly related with the appearance of its price discount problems. The information share of WTI is significantly and negatively influenced by the increasing frequencies of WTI discount. The shape of a forward curve cannot explain WTI distortion in the short run. But in the medium and long terms, contango is entangled with discount problem and both indicate the failure of WTI benchmark. Since contango serves as a proxy of inventories, this last result is consistent with the conjectures of market participants that the increasing stocks in Cushing causes continuing discounts and contangos in WTI price, and explains the breakdowns in WTI benchmark (Fattouh, 2007 and Fattouh, 2009). The following section, Methodology, presents the Hasbrouck (1995) information share model and describes the empirical models used in this paper. Section 3 describes data used in this paper and offers preliminary analyses. Section 4 reports empirical results. Section 5 gives concluding remarks and provides a brief discussion of possible future investigations.

This paper applies the Hasbrouck (1995) information share model to a rolling estimation process to trace the changing status of WTI benchmark. The dynamics of WTI information share demonstrate the fact that WTI has lost its place as a pricing center for years. This result supports the opinion prevailing in markets that WTI benchmark has broken down. Signals indicating WTI dislocations are tied up with some eccentric price structures. The first one is the unusual discount on WTI price. The appearance of WTI discounts may significantly indicate the breakdowns in WTI benchmark in the short run. Another clue alerting to the malfunction in WTI benchmark is the shape of a forward curve. The frequency of contango, along with the forward spread, may signal the distortion problem in the medium/and long terms. But the shape of a forward curve is not a useful sign for WTI dislocation in the short run. Moreover, WTI discount problem is highly collinear with an upward-sloped forward curve in the medium-to-long terms, illustrating there are common factors contributing to the appearance of discount and contango problems in the long run. Since the slope of a forward curve is a proxy for inventories, the rising stocks at Cushing generated by the logistic bottlenecks may explain the declining efficiency of WTI benchmark. Results from regressing on Cushing's actual inventory data strongly indicate the inventory-accumulation has negative effect on the benchmark role of WTI. The WTI-Brent decoupling, starting at the beginning of 2006, has existed for years. Up to May, 2010 we could still see the WTI contract was sometimes sold at a large discount to Brent. To replace WTI, a new benchmark — the Argus Sour Crude Index (ASCI) was introduced in May 2009. As of 2010, Saudi Arabia abandoned WTI and priced its oil exporting to the US on ASCI. The ASCI index tracks the price in the physical market of US Gulf Coast crude oils, and is published as a differential to WTI crude oil and as a flat price. The ICE and the NYMEX respectively announced to launch futures contracts on ASCI index. One of the contracts is an ASCI outright contract. The other is a differential contract priced at the differential between the ASCI and WTI prices. From the perspective of NYMEX, launching the ASCI contract is to strengthen the benchmark status of the existing WTI contract, helping their customers to hedge price exposure to the global crude market with greater precision. Yet from the viewpoint of ICE, ICE hopes the outright ASCI index price may become more independent of its WTI base, especially when the differentials to WTI continue to be less stable than the differentials to Brent. With the growing global production and trading of the sour grades, quite a few of the traders support the launch of ASCI contracts. Since many refineries located in the US Gulf Coast region have upgraded their processing capacity for refining sour crude oils, traders believe it is better to use a sour crude grade as the benchmark for pricing the crude oils into the US. The future development of the ASCI contract could be one of the issues deserving concerns. Will the ASCI contract successfully replace the WTI benchmark? Without the supports of trading volumes, it is immature at present to discuss this question. However, it is worth to keep an eye on the developments of the ASCI contract and study the price discovery function of ASCI among several alternative markers, such as the WTI, Brent, and Dubai. Another possible direction of future study is re-examining the price discovery role of WTI with a structural model where the sources of shocks are indentified by Yan and Zivot, 2007, Yan and Zivot, 2010 and Wang and Yang, 2011. As commented in Lehmann (2002), the information share in the Hasbrouck model is based on the residuals of a vector error correction model in the reduced form, which does not make the interpretations of information share measure always clear. Yan and Zivot, 2007 and Yan and Zivot, 2010 propose to use the cumulative impulse response from the structural conitegrated VAR to gain new insights to the price discovery dynamics of information share measure. In this regard, a structural VAR model developed by Wang and Yang (2011) for examining price discovery in sequential markets can be used in further studies.