Closer to One Great Pool? Evidence from Structural Breaks in Oil Price Differentials.

• Author: Plante, Michael

1. INTRODUCTION

   Morris Adelman (1984) famously wrote "The world oil market, like the world ocean, is one great pool." If this were literally true, it would imply that all crude streams would be perfectly sub-stitutable for one another in the refining process. We would then expect to see generally small price differentials between different crude streams over the long-run, reflecting primarily transportation costs. In reality, crude oil can have a wide range of physical properties that play an important role in how one crude stream is priced relative to another, and one often observes large price differentials between crude streams of different qualities.

   Price differentials between different grades of crude are important for many oil market participants. For refiners, they can affect profitability and influence investment decisions about specific equipment, such as cokers, that could improve the profitability of processing lower grades of crude. (1) Oil producers and fiscal authorities are concerned about these differentials because of the impacts they can have on revenues earned from producing or taxing certain types of oil. (2) They can also affect a government's choice of the benchmark used to set official selling prices. (3) Finally, for academics, analysts and others interested in understanding the upstream and downstream oil markets, these differentials provide important signals about how supply and demand conditions change over time for one type of crude relative to others.

   In a certain sense, these price differentials reflect limits on the global refining sector's ability to treat various crude streams as substitutes for one another when it comes to transforming them into the valuable petroleum products that consumers desire. In this paper, our question of interest is whether the average values of price differentials between different quality crude oils have declined over time. That is, can we find evidence that crude oils of different types may have become more substitutable for one another and that the oil market has become closer to "one great pool"?

   To answer this question, we construct price differentials between numerous crude oils of various types and then test whether these differentials have experienced shifts in their means using a structural breakpoint test. While it is well known in the industry and literature that changing market conditions can cause short-run variations in these price differentials, little has been said about whether they have been affected by structural breaks that have more permanently changed their average levels.

   To provide some motivation for our interest in structural breaks, we plot in Figure 1 examples of price differentials between higher and lower grade crudes for four parts of the world: Midland, TX; the U.S. Gulf Coast; Europe; and Asia. These are log-differentials using monthly data from 1997 to 2018 that consider the price of a high-quality crude relative to a lower-quality one. Visually, there is strong evidence of at least one break in the means of these differentials, occurring sometime around 2007 or 2008. The vertical lines denote the breakpoints as determined by a more formal statistical test. Visual inspection of other differentials, not shown here, strongly suggests the existence of structural breaks in many of their means, as well.

   Our first contribution is to document more rigorously and systematically the extent to which differentials between crude oils of different types have experienced structural breaks in their means. Using the sequential breakpoint test of Bai (1997), we find that almost all of the differentials we look at have experienced at least one break in their mean. In particular, we find that most price differentials between crude oils of different types--25 out of 27 cases, to be exact--experienced a significant break around 2008. We then investigate how the means have shifted over time and find that most of those differentials have narrowed. A reduction in volatility has accompanied those changes. After the break, the means and volatilities are often half of their pre-break levels. Related to these findings, we show that differentials between higher-valued petroleum products, such as gasoline and diesel, and residual fuel oil, a low-quality fuel produced in greater abundance in lowgrade crude oil, have also narrowed significantly and become less volatile following breaks that also occurred around 2008.

   We then investigate what has changed in the oil market that would be consistent with these differentials having become smaller. We document that the global refinery sector has added a significant amount of upgrading capacity, thereby increasing its ability to process low-grade crude oils into high-value petroleum products. At the same time, we show that the U.S. shale oil boom has boosted the relative supply of high-quality, light crude oil, which does not require highly sophisticated refiners to fully process. Both changes are consistent with the findings of smaller price differentials. We also show that changes in environmental regulations on sulfur content in petroleum products and changes in the relative demand for residual fuel oil are not consistent with smaller price differentials between high and lower-quality crude oil.

   Regarding the timing of the cluster of breaks, we consider the potential roles of the Great Recession, supply changes, and speculation. Data show the Great Recession significantly reduced demand for petroleum products, especially lighter petroleum products such as gasoline and diesel. At the same time, due to long lead times associated with refinery investment decisions, significant upgrading capacity additions continued unabated during the downturn. We use simple supply and demand curves to highlight how these two occurrences would help generate smaller differentials. We find no notable changes in the supply of different grades of crude oil around that time. The limited available data also does not point to any particular role for speculation with regard to the changes in the price differentials.

   Finally, we also investigated whether oil price differentials between crudes of the same type, for example, two light, sweet crude oils, experienced a similar set of breaks, particularly around 2008. If that were true, it would suggest a broader change in the oil market not necessarily connected to crude quality. Overall, we do not find any evidence for this. We do, however, find that differentials between similar-type crude oils have experienced their own set of breaks. Many appear connected to changing market conditions in the United States, occurring either in the mid-2000s or after 2010, and affecting numerous differentials related to crude oils in the U.S. Gulf Coast, particularly light, sweet crude oil. A modest contribution on our part is to show that these breaks are more prevalent than previously documented in the literature.

   We note that our work is connected with previous research papers, such as Weiner (1991), Sauer (1994), Ripple and Wilamoski (1995), Gulen (1997), Gulen (1999), and Bachmeier and Griffin (2006) that have considered to what extent Adelman's statement holds true. Those works have mainly looked at the degree to which oil prices move together across space and time, often using cointegration models. Or, to elaborate on Adelman's metaphor, these works ask if there is a disturbance in one part of the pool that generates waves (price movements), do the waves spread out and affect other parts of the pool?

   In our work, we consider the idea that long-run price differences also tell us something about how close the oil market is to being "one great pool" but from the quality perspective. Or, elaborating on the metaphor again, to what extent can the global refining sector literally dip into any part of the pool to get the crude oil it needs? Because of the different perspective, we focus on structural breaks in the long-run average size of the price differentials between high and low-quality crude, rather than modeling dynamics using cointegration models. (4)

   Prior work in the literature has also discussed the occurrence and importance of structural breaks affecting price differentials related to key benchmarks for light, sweet crude, such as West Texas Intermediate (WTI) and Brent. See, for example, Buyuksahin et al. (2013), Borenstein and Kellogg (2014), Scheitrum et al. (2018), and Agerton and Upton (2019). These primarily focused on the implications of the shale oil boom and the ensuing logistical bottlenecks. Buyuksahin et al. (2013) also discussed Canadian production growth and issues related to storage. Our work differs from the prior literature in two main regards: (1) our main focus is on price differentials between crude oils of differing qualities; (2) we offer additional insight into the role of the downstream (refining) sector as an important market factor.

   The rest of the paper is organized as follows. Section 2 offers a brief introduction to crude quality and oil price differentials. In Section 3, we discuss our data and econometric methodology. Section 4 presents evidence regarding the presence of structural breaks and documents how they have affected the differentials. Section 5 discusses what changes in the oil market are consistent with our econometric findings. We then conclude.

2. CRUDE OIL PROPERTIES AND PRICE DIFFERENTIALS

   While the previous literature has found that oil prices tend to move together over time, i.e. they are cointegrated, crudes usually do not sell for the same price because of differences in their physical characteristics. Two properties of particular importance are a crude oil's American Petroleum Institute gravity, hereafter API gravity, and sulfur content. (5) The industry has found it convenient to lump crude oils into several major groups based on these properties. It is common to label oils as light, medium or heavy depending upon their API...