Modelling the Global Price of Oil: Is there any Role for the Oil Futures-spot Spread?

AuthorVolenti, Daniele

    This paper investigates the main economic and financial drivers of the real price of oil and it relates to the strand of the literature explaining oil prices by supply and demand shocks. The empirical approach is based on a revised version of the Structural Vector Autoregressive (SVAR) model developed by Kilian and Murphy (2014).

    Our main idea is to retrieve the forward-looking expectations of oil traders by replacing a physical proxy for global above-ground crude oil inventories with the oil futures-spot spread (henceforth, spread). In this work, the spread is defined as the ratio of oil futures prices overthe relative oil spot prices minus one and the free-risk interest rate, after accounting for the time to maturity of the futures contract. According to the theory of competitive storage, the spread is a proxy for the net-convenience yield of oil stocks, although expressed with an opposite sign (1) Therefore the spread accounts for a stream of implicit benefits to the holder of the commodity inventory during periods of oil market stress and it is highly informative about the slope of the term structure of the oil futures curve. The latter provides intertemporal price signals for all traders participating to the financial and the physical markets for crude oil.

    Relative to the existent literature on modelling the real price of oil, our analysis provides three main contributions.

    First, our study proposes a spread-based SVAR model of the international market for crude oil. The spread is derived by crude oil Brent futures prices with maturity 3-months, since about two-thirds of oil purchases at world level use Brent as a reference price. This suggests that the Brent market is exposed to worldwide oil price shocks and it represents a natural choice for examining the dynamics of the convenience yield at the global level. Moreover, the time to maturity of the spread plays a crucial role in the economic properties of the SVAR model since it contains information about short-term and long-term convenience yields, respectively. Thus, the short-maturity spread reflects the perceived relative importance of the amount of inventory that is available in the near future (see Alquist et al., 2014). In contrast, the long-maturity spread is less sensitive to oil price shocks, consistent with the view that consumers and producers have more time to make consumption decisions and adjust production in the long period (see Lee and Zeng, 2011).

    In our analysis, the benefits of using the spread as a measure of market expectations can be motivated on the basis of several economic considerations. First, the spread exploits the price discovery role in the crude oil futures markets. Second, the spread allows for a feedback effect from futures to spot markets and it accounts for the presence of informational frictions faced by the market participants (see Singleton, 2014; van Huellen, 2020). Third, the proxy for global above-ground crude oil inventories is affected by measurement error (see Baumeister and Hamilton, 2019). Our model provides empirical evidence that the spread responds to oil price shocks differently, depending on the economic motivations behind each shock. On average, oil supply disruptions and positive shocks to global business cycle cause a large and persistent drop in the spread, consistent with the fact that inventories are used for consumption and production smoothing, respectively. Conversely, shocks to the demand for storage driven by fears of production shortage cause a small decline in the spread. Therefore, we document a negative relationship between the impact responses of the price of oil and the spread to global oil market-driven shocks.

    Second, our work provides fresh evidence on how the spread-based model helps to identify the speculative component of the real price of oil triggered by the oil futures markets. Therefore, the empirical approach used in our paper allows us to provide an economic interpretation of the residual structural shock, namely the financial market shock. The latter implies that an unanticipated rise in the spread might be interpreted as a market signal of higher future oil spot prices. In this context, oil producers have the incentive to defer production, causing the spot price of oil to rise. This last type of shock induces an increase in the demand for below-ground crude oil inventories because the spot price of oil is expected to rise.

    It is also important to highlight that, in presence of asymmetric information, the speculative activities in the futures markets can drive up the spot price of oil without necessarily reducing the aggregate consumption and boosting inventories, as discussed by Sockin and Xiong (2015). The temporary distortion between spread and inventories is a function of the elasticity of arbitrage, which in turn depends on physical and financial constraints faced by the arbitrageurs (see Ederington et al., 2020; Acharya et al., 2013; Etula, 2013).

    Third, our study provides a clear picture of the historical dynamic of the real price of oil and the spread. To illustrate this point, we focus on four exogenous events in global crude oil markets: the 1990-1991 Persian Gulf War, the 2003-2008 oil price surge, the 2008-2009 global financial crisis and the 2014-2016 oil price slump.

    The rest of the paper is organized as follows. The next section provides a review of the relevant literature. Section 3 discusses the economic motivations, which support the use of the spread in the SVAR models. Section 4 describes the data. Section 5 illustrates the econometric approach. Empirical results are presented in Sections 6. Finally, Section 7 concludes.


    The traditional literature explaining oil prices by supply and demand shocks is vast, see for example Kilian (2009), Kilian and Murphy (2012), Baumeister and Peersman (2013) and Lutkepohl and Netsunajev (2014). Previous studies rely on a different set of identifying assumptions of SVAR models of the global market for crude oil. In these works, the oil demand shocks play the most important role in accounting for the historical oil price movements. However, given the lack of a forward-looking measure in the set of endogenous variables, these models cannot identify the speculative component of the real price of oil. Kilian and Murphy (2014) contribute to the issue of a measure of traders' expectations by proposing a proxy for global crude oil inventories above the ground. (2) The latter is designed to capture the expected demand and supply conditions that are not contained in the past data available to the econometrician. Therefore, the crude oil inventory plays a crucial role in the identification of the speculative component of the real price of oil.

    The SVAR model proposed by Kilian and Murphy (2014) shows that shocks to the aggregate demand (likely driven by a strong growth in the economy) were the main factors in driving up the real price of oil, from early 2003 until mid-2008. These results are robust to changes in the proxy for global above-ground crude oil inventories, as discussed by Kilian and Lee (2014). Moreover, Herrera and Rangaraju (2020) show that the dynamic effect of oil supply shocks on the real price of oil is mainly related to the methodology for the identification of the structural shocks and the bounds of the implied price elasticities of oil demand and oil supply. Finally, Zhou (2019) proposes a refined version of the inventory-based detection strategy, developed by Kilian and Murphy (2014). The identification of the structural shocks is obtained by means of the narrative sign restrictions, as discussed by Antolin-Diaz and Rubio-Ramirez (2018).

    In contrast to previous studies, Juvenal and Petrella (2015) use a Factor Augmented VAR model with a different set of structural assumptions. The authors find that speculative shocks were responsible of a large increase in the price of oil between 2004 and 2008. However, their results suggest that oil consumption demand shocks were the most important factors in explaining the fluctuations in the real price of oil, during the period of interest. Finally, a recent work by Baumeister and Hamilton (2019) provides some relevant contributions in this literature, which are summarised as follows. First, oil supply shocks appear to be more important in explaining the path of the real price of oil compared to earlier studies. Second, oil supply disruptions cause a reduction in the economic activity after significant lags, while a rise in the real price of oil triggered by oil consumption demand shocks are not responsible of a large drop in the global economic activity. Finally, the traditional proxy for the global above-ground crude oil inventories is considerably affected by measurement error.


    Understanding the speculative component of the real price of oil is not a simple endeavour. Most of the oil market VAR models use an inventory-based detection strategy to identify the speculative demand for crude oil. (3) In contrast, our study provides several economic reasons to consider the spread as a reliable measure of oil market expectations.

    First of all, the spread accounts for the price discovery role in the futures market (see Garbade and Silber, 1983; Gospodinov and Ng, 2013). For example, a work by Alquist et al. (2014) shows that the first-two principal components extracted from a panel of oil futures-spot spreads (with different maturities) have predictive power for the future path of the real price of oil. Moreover, according to the theory of competitive storage, the spread plays an important role in explaining the value of holding crude oil stocks, as conveyed by the futures markets. For example, a post-shock increase in the spot price of oil causes a reduction in the level of inventories and a rise in the convenience yield. The crude oil stocks are drawn down in an effort to smooth...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT