کمی کردن مولفه سوداگرانه در قیمت واقعی نفت : نقش موجودی جهانی نفت

One of the central questions of policy interest in recent years has been how many dollars of the inflation-adjusted price of oil must be attributed to speculative demand for oil stocks at each point in time. We develop statistical tools that allow us to address this question, and we use these tools to explore how the use of two alternative proxies for global crude oil inventories affects the empirical evidence for speculation. Notwithstanding some differences, overall these inventory proxies yield similar results. While there is evidence of speculative demand raising the price in mid-2008 by between 5 and 14 dollars, depending on the inventory specification, there is no evidence of speculative demand pressures between early 2003 and early 2008. As a result, current policy efforts aimed at tightening the regulation of oil derivatives markets cannot be expected to lower the real price of oil in the physical market. We also provide evidence that the Libyan crisis in 2011 shifted expectations in oil markets, resulting in a price increase of between 3 and 13 dollars, depending on the inventory specification. With regard to tensions with Iran in 2012, the implied price premium ranges from 0 to 9 dollars.

The real price of crude oil depends on shocks to the flow supply of oil (defined as the amount of oil being pumped out of the ground), on shocks to the flow demand for crude oil that reflect the state of the global business cycle, on shocks to the speculative demand for oil stocks above the ground, and on other more idiosyncratic oil demand shocks. Especially, the quantification of speculative oil demand shocks has long eluded researchers because it raises difficult problems of identification. A speculator is someone who buys crude oil with the intent of storing it for future use in anticipation of rising oil prices. Such forward-looking behavior invalidates standard econometric oil market models if speculators respond to information not available to the econometrician attempting to disentangle demand and supply shocks based on historical data. Recent theoretical and empirical work by Alquist and Kilian, 2010 and Kilian and Murphy, 2013, and Baumeister and Kilian (2012a) made considerable strides in addressing these problems within a framework that is theoretically sound and empirically tractable.1 These studies generalized the structural oil markets models pioneered by Kilian, 2009 and Baumeister and Peersman, 2013, and Kilian and Murphy (2012) to examine the role of speculation and forward-looking behavior with careful attention to the role of spot and futures prices. The key insight on which the Kilian and Murphy (2013) model builds is that otherwise unobservable shifts in expectations about future oil demand and supply conditions must be reflected in shifts in the demand for above-ground crude oil inventories. Shocks to this expectations-driven or speculative component of inventory demand may be identified and estimated jointly with all other shocks within the context of a fully specified structural vector autoregressive model. This fact allows one to assess how quantitatively important the speculative component in the real price of oil has been at each point in time from the late 1970s until today. The latter question has been of central policy interest since 2003 when oil prices began to surge to unprecedented levels, raising the question of how policy makers should respond to rising oil prices (see, e.g., Fattouh et al., 2012). Models aimed at quantifying the speculative component in the real price of oil depend crucially on the quality of the oil inventory data. There are no readily available data for global crude oil inventories. Kilian and Murphy (2013) instead relied on a proxy constructed from publicly available U.S. Energy Information Administration (EIA) data. The objective of this paper is to explore how sensitive the conclusions reached by Kilian and Murphy are to the use of an alternative proxy compiled by the Energy Intelligence Group (EIG), a private sector company which provides detailed accounts of crude oil inventory stocks by region as well as oil at sea and oil in transit. We examine how the use of this alternative proxy affects our assessment of the causes of the oil price surge from 2003 to mid-2008 and of the subsequent collapse and partial recovery of the real price of oil. We also examine for the first time the role of speculative demand during the Libyan Revolution, the Arab Spring, and recent tensions with Iran ranging from the Iranian nuclear threat to the EU's decision in early 2012 to impose an oil import embargo on Iran. These recent episodes are of particular interest both because they provide additional evidence about the role of expectations shifts and because many pundits have conjectured that rising oil prices in recent years may be attributed to these events. Our focus throughout the paper is on providing results in a format that is immediately useful for policy makers. For this purpose, we design two new presentation tools that summarize – at each point in time – how many dollars of the inflation-adjusted price of oil must be attributed to which demand or supply shock in the global market for crude oil. The remainder of the paper is organized as follows. Section 2 reviews the structure and identifying assumptions of the vector autoregressive model to be used throughout this paper. Section 3 compares the two alternative proxies for changes in global above-ground crude oil inventories. In Section 4, we re-estimate the Kilian-Murphy model using these alternative proxies on data extending to 2012.5. We quantify the effects of speculative demand using measures of their cumulative effects as well as counterfactuals for the real price of oil. The conclusion in Section 5 links our discussion of speculation in the physical market for crude oil to recent debates about the role of speculation in the paper market for crude oil.

Global commodity markets play an increasingly important role in the world economy, yet economists are only beginning to study these markets. In this paper, we focused on the role of inventories or stocks of crude oil for the determination of the real price of oil. The fact that crude oil is storable allows market participants to speculate in oil by storing purchases of oil for future use in anticipation of rising prices. Shifts in expectations about future oil prices may greatly and immediately influence the real price of oil by shifting the speculative demand for oil. Indeed, such speculative demand shifts have been held responsible for the remarkable surge in oil and other industrial commodity prices that took place between 2003 and mid-2008. Compared with markets for other storable commodities, the market for crude oil lends itself to a formal econometric analysis of this question not only because of the importance of crude oil for the global economy, but because of the availability of monthly global data on oil production and aboveground oil inventories dating back many years. Even for crude oil, however, the quality of the inventory data is less than perfect. This paper explored in detail how the use of alternative proxies for global oil inventories affects the empirical results of the structural oil market model of Kilian and Murphy (2013), suitably updated to 2012.5. We concluded that, despite some differences in emphasis, both inventory proxies yield very similar results in general. We found evidence of speculation driving up the real price of oil in the physical market for crude oil in 1979 after the Iranian Revolution, in 1990 near the time of the invasion of Kuwait, in 2002 in the months leading up to the 2003 IraqWar, in early 2011 during the Libyan crisis and in early 2012 during the Iranian crisis. A common feature of all these episodes of speculative pressures is that they reflect concerns about the stability of oil supplies from the Middle East. We also found evidence that speculation may lower the real price of oil. We identified several episodes in which a reduction in speculative demand contributed to lower oil prices. One example is in 1986 after the collapse of OPEC; another example of speculative downward pressures on the price is late 2008 and early 2009. The latter episode presumably was associated with expectations of a prolonged global downturn rather than improved oil supplies. Episodes of increased speculative demand in the physical market for crude oil do not line up at all with increases in measures of the participation of financial investors in oil futures markets. Indeed, the view that an exogenous shift in the participation of financial investors in oil futures markets explains the surge in the real price of oil during 2003–08 can be ruled out on the basis of our results. By standard arbitrage arguments, speculation in financial markets for oil cannot affect the real price of oil in physical markets unless there is a shift in inventory demand. Our analysis found no evidence of such a shift, consistent with a general lack of evidence for the hypothesis that the financialization of oil markets caused oil price increases (see, e.g., Büyüks¸ ahin and Harris (2011), Irwin and Sanders (2012), Fattouh and Mahadeva (2012), Hamilton and Wu (2013b)). This does not necessarily mean that the financialization of oil futures markets did not matter, but that it should be modeled as part of the endogenous propagation of shocks to economic fundamentals rather than as an exogenous intervention. This interpretation is consistent with the view that index funds simply followed market trends set in motion by earlier shocks to economic fundamentals rather than creating market trends of their own for reasons not related to economic fundamentals. Despite evidence that speculation raised the real price of oil by between 5 and 14 dollars from March of 2008 until July of 2008, the bulk of the cumulative increase of 95 dollars (measured in 2012.5 dollars) from early 2003 until mid-2008 (and much of the evolution of the real price of oil since then) must be attributed to shifts in flow demand, associated with shifts in the global demand for oil from emerging Asia and from the OECD. Flow demand shocks account for as much as 61 dollars of that increase with flow supply and idiosyncratic demand shocks adding between 17 and 30 dollars, depending on the specification. In short, the surge in the price of oil and other industrial commodities appears to be driven primarily by economic fundamentals. This fact has important implications for policy makers. For example, current policy efforts aimed at tightening the regulation of oil derivatives markets cannot be expected to lower the real price of oil, given that excessive speculation in these markets was not the cause of earlier increase in the price of oil in the physical oil market. To the extent that higher demand for oil from emerging Asia caused that surge, as has been suggested by Kilian (2009) and Kilian and Hicks (2013) among others, one would not expect higher oil prices to disappear, unless global growth slows down further. Finally,we examined for the first time the evolution of the real price of oil since 2010.We confirmed that for this period as well, flow demand shocks have been the primary driver of the real price of oil. We also examined the role of speculative shocks. It has been conjectured that the Libyan Revolution in early 2011 affected the real price of oil by shifting speculative demand (see Baumeister and Kilian, 2012a). Ours is the first study to examine this question formally. We provided evidence that the Libyan crisis indeed shifted expectations in oil markets, resulting in a price increase of between 3 and 13 dollars (in 2012.5 consumer prices), depending on the specification of oil inventories. This increase is short-lived and not related to the Arab Spring more generally. In fact, there is no evidence that the Arab Spring caused an increase in speculative demand in 2011. With regard to tensions with Iran in early 2012 (ranging from the decision to impose an EU oil import embargo to the Iranian nuclear threat), the evidence is more mixed. The implied price premium ranges from 0 to 9 dollars, depending on the specification. Finally, we found no indication that higher demand for strategic oil inventories from emerging Asia (or for that matter Iranian storage of oil on tankers in recent years) played an important role determining global oil inventories or the real price of oil after 2009. Regarding the flow supply of oil, we showed that to the extent that flow supply shocks mattered for the real price of oil after 2010, they tended to increase the real price of oil with estimates ranging from 7 to 19 dollars. There is no indication that the supply side of the oil market has been a key determinant of the real price of oil, however. For example, between 20 and 23 dollars of the 29 dollar decline in the real price of oil since its peak in mid-2008 is accounted for by flow demand shocks compared with between
2 and þ3 dollars explained by flow supply shocks.