Abstract
In November 2014, OPEC announced a new strategy geared towards improving its market share. Oil-market analysts interpreted this as an attempt to squeeze higher-cost producers including US shale oil out of the market. Over the next year, crude oil prices crashed, with large repercussions for the global economy. We present a simple equilibrium model that explains the fundamental market factors that can rationalize such a "regime switch" by OPEC. These include: (i) the growth of US shale oil production; (ii) the slowdown of global oil demand; (iii) reduced cohesiveness of the OPEC cartel; (iv) production ramp-ups in other non-OPEC countries. We show that these qualitative predictions are broadly consistent with oil market developments during 2014-15. The model is calibrated to oil market data; it predicts accommodation up to 2014 and a market-share strategy thereafter, and explains large oil-price swings as well as realistically high levels of OPEC output.
I. Introduction
In 2014, global oil supply overtook demand and the oil price started to decline. In its November 2014 meeting, OPEC1 decided not to reduce supply and prices fell further. Many oil-market analysts interpreted this as the formal decision to squeeze higher-cost competitors, including US shale oil extracted using hydraulic fracturing (“fracking”), out of the market. It also stood in contrast with OPEC’s coordinated cut during the Global Financial Crisis. A former adviser to Saudi Arabia’s Oil Minister summarized: “His biggest move was the latest one of defending … market share”.2
OPEC’s actions occurred against the backdrop of weakening global demand for crude and several years of steadily rising capacity from non-OPEC sources—most notably from unconventional sources in the US. Since mid-2014, the oil price fell from above $100 to an average of $50 during 2015. In its December 2015 meeting, OPEC reiterated its commitment to a “market-share” strategy. Many have opined on whether OPEC is taking a sensible perspective by driving competitors out of business or whether it is a misguided move tantamount to “hara-kiri”.3
Our goal in this paper is to understand the fundamental market factors that induced the shift in OPEC’s strategy. We present a simple economic model of the oil market: OPEC has a degree of market power and competes against a set of non-OPEC producers who act as a price-taking competitive fringe.4 OPEC has a choice between two strategies. The first strategy, which we call “accommodate”, is to maximize profits via a “high” oil price which allows high-cost non-OPEC producers to remain profitable. The second strategy, referred to as “squeeze”, is to drive up production—and hence drive down price—and thereby induce high-cost producers to exit the market. We show that either of these two strategies can be optimal for OPEC depending on market fundamentals in demand and supply.
Our theory shows that the market-share strategy becomes relatively more attractive for OPEC in the face of: (i) slower global oil demand; (ii) greater high-cost oil production (for example shale); (iii) reduced cohesiveness within OPEC; and (iv) higher output in other non-OPEC countries. We show that a regime switch from accommodate to squeeze becomes optimal when high-cost oil production grows beyond a specific point. The model can rationalize OPEC’s decision to raise output in the face of weaker demand, and explain a large drop in the oil price.
In the empirical part of the paper, we begin with a description of oil-market developments which highlight how the model’s comparative-statics are pertinent. We give an account of OPEC’s strategy shift and the market responses of non-OPEC players. We then calibrate the model to oil5 market data across a range of scenarios. First, we show how the model rationalizes the oil market in the period preceding the price collapse as a high-price accommodate scenario where OPEC chooses not to squeeze US shale oil—despite already substantial market-share erosion and sufficient spare capacity for a squeeze. Second, we show how some parameter changes can prompt a rational decision by OPEC to squeeze US shale out of the market. Third, we show that the model generates squeeze equilibria when calibrated to forecasts of future data that yield higher OPEC output and lower prices.
Our model exposes the fallacy of interpreting a decline in OPEC’s revenues or profit as evidence that a market-strategy is necessarily misguided. The simple point is that the relevant comparison is not how profits compare to an earlier period, but rather how they would compare to pursuing a different strategy today—for which profits could be even lower. By showing how a market-share strategy can be optimal for OPEC in a formal framework, we offer the model as a potential rational economic explanation for its 2014 strategy switch and the subsequent oil price crash. However, we do not wish to claim that it is the most likely of a range of possible economic or political motivators.6
Our theory makes a number of simplifying assumptions. The model is static and partial-equilibrium; it does not explicitly incorporate dynamics such as a producer’s intertemporal decision to sell today or leave the oil in the ground.7 Relatedly, the model does not feature inventory behaviour—although we do account for this in the empirical part of the paper. We also do not address the potential roles of uncertainty and asymmetric information. Finally, the production of non-OPEC players is modelled as a binary decision: they produce up to capacity if price exceeds their cost, and otherwise shut down.
OPEC’s market share strategy is broadly aimed at its high-cost competitors.8 The organization has disputed a common perception in the industry by stating that it is not targeting specific countries or production technologies.9 A number of conventional producers have sustained current production but reduced investment in future capacity, which suggests they will also be squeezed over a longer time horizon (IEA, 2016; Toews and Naumov, 2016). Furthermore, OPEC’s market-share strategy could deter the adoption of fracking in other countries that would otherwise have been profitable in a high-price environment.
Nonetheless, the market-share strategy illustrated in this paper emphasizes US shale, which is the main focus of oil-market analysts. It is premised on OPEC having “low” costs and US shale having “high” costs. US shale’s costs are almost surely higher than OPEC’s, yet are the world’s highest only over a given time frame. Conventional oil extraction entails large upfront sunk costs but low subsequent marginal variable costs. As a result, it would take extremely low prices to induce exit from “high long-run cost” conventional resources such as the Canadian oil sands.10 The US shale life-cycle is much shorter, which makes the US supply response to prices quicker (Bjornland, Nordvik and Rohrer, 2016). Our static model’s marginal costs include upfront expenses for US shale but exclude initial investments for other producers. In our empirical work, we consider a wide range of cost parameters for US shale in light of the considerable variation in existing cost estimates.
Related literature. Although there has been a lot of policy-related discussion since November 2014, we believe ours is at the forefront of papers beginning to offer a formal economic model of OPEC’s strategy shift and its repercussions. Fattouh, Poudineh and Sen (2015) analyze the trade-offs between a strategy of market share and one of curtailing output to generate near-term revenue. Introducing uncertainty about the nature of US shale tends to favor accommodation but, as further information reduces this uncertainty, a switch in strategy becomes more likely.11
There are a number of analyses of the oil price crash; for discussions of its causes and implications see, e.g., Arezki and Blanchard (2014), Hussain et al. (2015), Baumeister and Killian (2015), Hamilton (2015), and Mohaddes and Raissi (2016). Many suggest that supply has been more important in explaining the oil price crash than demand.12 Smith (2009) demonstrates how the combination of low demand and supply elasticities in the oil market can account for historical levels of oil price volatility—without any role for any volatility-enhancing financial speculation. Our results show how an oil price decline induced by weaker demand or an exogenous rise in supply can be magnified because it induces a regime switch in OPEC behaviour and a further endogenous increase in supply. In a similar vein, Verleger (2016) emphasizes the vital role that market structure plays for oil prices.
There remains considerable debate on the extent to which OPEC members cooperate when setting output or prices (Smith, 2005; Bremond, Hache, and Mignon, 2012; Nakov and Nuno, 2013; Huppmann, 2013). Almoguera, Douglas and Herrera (2011) suggest that OPEC’s behaviour is a mix of near-collusive episodes and subsequent non-cooperative breakdowns. Pierru, Smith and Zamrik (2016) analyze how OPEC (or a subset of members) stabilizes prices through optimization of spare capacity. Huppmann and Holz (2012) find that OPEC’s degree of market power has declined, and Fattouh and Mahadeva (2013) attribute fluctuations in this power to market conditions.
Our approach is flexible in that we calibrate OPEC’s market power to fit the data across each of our scenarios. We obtain parameters that describe the level of competition in the market and are broadly in line with those from the empirical literature. Pricing regimes fall short of a perfect cartel but still allow low-cost producers (OPEC and non-OPEC) to earn rents. Our accommodate strategy also has OPEC offset other producers’ production changes, and our squeeze strategy has some similarity with Stackelberg behaviour (Huppmann, 2013). OPEC’s decision between these strategies is influenced by its time-varying ability to coordinate and its market-dependent choice means that its market power is endogenous. Complementing the longer-term views in the existing literature, we focus on market developments since 2014.
The strategy pursued by OPEC in our model is a form of “limit pricing”. One advantage of our approach over classic industrial-organization theory is that it does not rely on the dynamic of a later period with again-higher prices in which OPEC can recoup “lost” profits.13 Our model shows that OPEC’s profits under a low-price squeeze can be permanently higher than under accommodate. In related work, Andrade de Sá and Daubanes (2014) suggest that OPEC prices out of the market any “backstop technology” which has large potential to erode oil demand. Their main focus is on how this behaviour differs from a Hotelling rule and the implications for carbon-tax design.
Plan for the paper. Section 2 sets up our model of the oil market, and analyses the equilibrium outcomes under “accommodate” and “squeeze”. Section 3 presents the comparative statics that favour a regime switch, and a testable condition on when it occurs. Section 4 argues that the comparative-statics predictions from the model are consistent with market experience. Section 5 presents our quantitative calibration of the model to oil market data over a range of scenarios. Section 6 concludes.
II. A simple equilibrium model of the oil market
A. Setup of the model
We assume that the global demand curve for oil takes the linear form D(P) = (α − P)/β, with parameters α, β > 0. This is a common assumption in the literature, and will facilitate empirical calibration of the model later on. On the supply side, there are N + 1 oil producers, namely OPEC, denoted as i, plus N other non-OPEC players. OPEC has production capacity Ki with a marginal cost of production of Ci. Of the other producers, player n ∈ N has capacity Kn and unit cost Cn; it is a price-taker which sells up to capacity if P > Cn and zero otherwise. Let Cj ≡ maxn∈N{Cn} > Ci denote the player j with the highest unit cost, and capacity Kj. In the present analysis, we take this to be US shale oil because it is the highest-cost producer in our chosen period of analysis, but this could generalize to any highest-cost producer. Let K𝓁 ≡ Σ n∈N\{j} Kn denote the combined production capacity of all other non-OPEC players. Note that the setup implies that all non-OPEC players produce up to capacity whenever US shale oil does so (but not necessarily vice versa).
OPEC has market power and can choose between two strategies:
“Accommodate”: Maximizing its profits taking as given that player j produces up to its capacity level Kj;
“Squeeze”: Lowering the market price to Cj, thus squeezing player j out of the market.
The first of these corresponds to what is often called a “price” strategy whilst the second is about “market share”. Our main question is, which of these two strategies is more profitable for OPEC?
In practice, OPEC is not an efficient cartel: its internal ability to restrict output has fallen short of what monopoly pricing would require. To capture this, we use a parameter λ ∈ (0,1] as a reduced form of OPEC’s pricing power under the accommodation strategy. The case with λ = 1 corresponds to a fully-efficient cartel (facing a competitive fringe); lower values of λ represent weaker pricing power.14 As will become clear, our theory does not hinge on the precise value of λ, but this parameter plays an important role in the calibration exercise later on.
B. Analysis of the strategies
We begin by deriving OPEC’s profits under each of the two strategies. Two assumptions on parameter values are made:
A1. (Cj − Ci) < λ[(α − Cj) − β(Kj + K𝓁)]
A2. (α − Cj) ≤ β(Ki + K𝓁)
The first assumption ensures that player j (US shale oil in this paper) is viable under the “accommodation” strategy. It implies that all other non-OPEC producers are also viable, and that OPEC is too (since they all have lower costs); in particular, note that λ cannot be too small. The second assumption ensures that OPEC has sufficient capacity to be able to carry out the “squeeze” strategy. A1 and A2 together imply (Cj − Ci) < λ [(α − Cj) − β(Kj + Kl)] ≤ λβ(Ki − Kj), so that OPEC has significantly higher production capacity than US shale, specifically Ki > Kj + (Cj – Ci)/λβ (where Cj > Ci). We verify that these parameter assumptions are satisfied in the empirical calibration of the model.
B.1 Strategy 1: Accommodate
So OPEC optimally absorbs higher production capacity of non-OPEC players, Kj + Kℓ, at a rate of [100/(1 + λ)]%, that is,
The profits of non-OPEC player n ∈ N, which produces Kn by construction, are simply equal to Kn(P* − Cn), and are positive by A1.
B.2 Strategy 2: Squeeze
Thus OPEC’s profits under the squeeze do not depend on the λ parameter which captures its pricing power under the previous accommodate strategy. The profits of non-OPEC player n ∈ N\{j} are Kn(P** − Cn), and are positive since Cj ≡ maxn∈N{Cn} = P** > Cn for all n ∈ N\{j}.
III. Model results
We now turn to our main results on the different market factors which can lead to a “regime switch” under which OPEC finds it optimal to squeeze US shale.
The preceding analysis already pins down the difference in profits between the two strategies,
Proposition 1 The “squeeze” strategy becomes relatively more attractive compared to the “accommodate” strategy, in that it offers relatively higher profits (that is, higher ΔΠi), for OPEC under the following conditions:
(i) the production capacity of high-cost player j (Kj) is larger;
(ii) the internal cohesiveness of OPEC λ is lower;
(iii) the global demand for crude oil α is lower;
(iv) the marginal cost of high-cost player j (Cj) is higher;
(v) the production capacity of other non-OPEC players Kl is larger.
The comparative statics from Proposition 1 are intuitive. First, larger production by player j depresses price under the accommodation strategy but its production is zero by construction under the squeeze strategy, regardless of capacity. This makes squeezing player j out of the market look relatively more attractive to OPEC.
Similarly, if OPEC is less internally cohesive, then it cannot raise price as strongly and extract as much profit under accommodation. Under the squeeze, the degree of price coordination is not a factor so this again favours the squeeze strategy.
Third, weaker global demand for crude depresses profits under both the accommodate and the squeeze strategies. The difference is that, under accommodation, lower demand reduces both OPEC’s sales and its profit margin. By contrast, under the squeeze, lower demand only reduces sales—since the price is pinned down by the marginal cost of the squeezed-out player. Thus lower demand relatively favours the squeeze strategy.16
Fourth, higher costs of player j have no impact on the accommodate equilibrium from OPEC’s viewpoint: since player j remains viable by A1, and produces up to capacity, higher costs simply mean less profits for player j but no change in the market equilibrium. However, the squeeze strategy becomes more attractive as less of a price decline is needed to squeeze player j out of the market.
Finally, higher production by other non-OPEC players also makes the squeeze relatively more attractive. Similar to the demand effect, this reduces both price and OPEC sales under accommodate but solely its sales under the squeeze strategy.
Proposition 1 delivers a clean set of qualitative “all-else-equal” results which can be taken to the data. In practice, many of these market factors—global demand patterns, oil production capacities and costs, OPEC’s internal dynamics—change simultaneously. Our empirical analysis in Sections 4 and 5 therefore considers the evolution of all of these market factors together.
The comparison of profits between the two strategies leads to the following quantitative prediction:
Put simply, it is a profitable strategy for OPEC to squeeze out a rival selling Kj units at cost Cj whenever “the prize” is sufficiently large in that
Proposition 2 thus delivers a critical value
We stress that the optimality of the market-share strategy does not rely on a subsequent “harvesting” period with again-higher prices after the high-cost players have been squeezed out of the market.
We thus obtain a further result on OPEC supply following a regime switch:
Proposition 3 (i) Suppose that an increase in the capacity of player j, from
(ii) Suppose that a decline in global oil demand, from α′ to α″, induces a regime switch from accommodate to squeeze, that is,
Proposition 3 shows how OPEC’s optimal supply responses can take an unexpected form. Standard intuition from economic theory, as well as the logic of a “swing producer”, suggest that higher rival output and lower demand should prompt a “soft” response in form of lower OPEC supply. While this is true within an accommodate strategy, the situation is different if these market factors induce a regime switch. Then higher player j production can induce a “fighting response” from OPEC, and lower demand can make it optimal to produce more.
Numerical example. A simple example illustrates the workings of the model. Let α = 250, β = 1, Ci = 0, Cj = 50 and λ = 1; all players except i,j are inactive, Kℓ ≡ 0. A1 and A2 boil down to Kj < 150 and Ki ≥ 200. OPEC’s profits under accommodation
Impact of US shale capacity on OPEC profits and optimal production
Citation: IMF Working Papers 2016, 131; 10.5089/9781498351638.001.A001
IV. Qualitative Empirical Discussion
This section begins with a discussion of how oil market developments in the run up to late 2014 would have driven a regime switch in light of our comparative-statics results from Proposition 1. We then give an account of OPEC’s decision in its November 2014 meeting to adopt a “market-share strategy” and its actions since. Finally, we explain the subsequent responses of other oil-market players.
A. Drivers of regime switch
This section describes the five developments that favoured OPEC’s decision to squeeze US shale17, namely: (i) weakening demand; strengthening supply from (ii) US shale, (iii) non-OPEC non-shale sources, and (iv) OPEC, as well as (v) coordination difficulties among OPEC members. One factor acting against these is (vi) falling US shale oil costs.
1. Weakening global demand (lower α). Having grown weakly in recent years, demand growth slowed further from 1.2 million barrels per day (mbd) in 2013 to only 0.9 mbd in 2014, a growth rate of less than 1 percent (Figures 2 and 3). The slowdown was largely unanticipated. In particular, Q3 2014 actual demand levels were 0.5 mbd lower than forecast in the International Energy Agency’s (IEA) June Monthly Oil Market Report (MOMR) and Q4 demand levels were almost 0.4 mbd lower than forecast in the September report. According to Proposition 1, weakening demand makes a decision to squeeze more likely.
Oil demand, supply and price
Citation: IMF Working Papers 2016, 131; 10.5089/9781498351638.001.A001
Oil supply and demand growth, mbd
Citation: IMF Working Papers 2016, 131; 10.5089/9781498351638.001.A001
Demand for oil is structurally restrained by disappointing economic growth after the Global Financial Crisis. Global GDP grew on average by 3⅓ percent in 2013-4, which is slower than in previous years and less than had been forecast (IMF, 2012; 2014). In addition, the sources of GDP growth are the less energy-intensive sectors. Further constraints to oil demand include efficiency improvements, fuel switching to natural gas and biofuels, and environmental restrictions (IEA, 2014; Verleger, 2016).
2. Higher US shale output (higher Kj). Reversing a decline since the early 1980s, US crude oil output rose from about 5 mbd in 2008 to 6½ mbd in 2012. Accelerating output reached about 8½: mbd in 2014 and an estimated 9¼ mbd in 2015 (Energy Information Administration, 2013, 2015). (Using the slightly broader definition of oil reported by the IEA, US output reached an estimated 12.8 mbd in 2015.) The increase is attributable to growth in oil extracted from unconventional sources. Production of light tight oil (LTO), which is one measure of shale production, almost doubled from 2¼ mbd in 2012 to 4¼ mbd in 2014.18 Over the two years, this was the primary source of incremental global supply and almost matched growth in global demand.
Realized values repeatedly exceeded forecasts by agencies, indicating a surprise element. For example, US output in 2014 was ¾ mbd higher than anticipated by the Energy Information Administration (EIA) early in its January 2013 Short-term Energy Outlook, and output for the third quarter of 2014 alone exceeded IEA forecasts for that quarter made in June 2014 by the same amount. Moreover, forecasts for future output also rose due to base effects and revised expectations about the pace of technical progress. For example, EIA estimates for 2019 LTO output were revised upwards by about ¾ mbd between the 2014 and 2015 editions of their Annual Energy Outlook (2014, 2015) despite a decline in prices that had already begun. In terms of our framework, actual and anticipated US shale production volumes were becoming too large for OPEC to accommodate.
3. Higher non-OPEC non-shale output (higher Kℓ). After accounting for the rise in US shale, non-OPEC output from other sources also rose. The contribution to global supply growth was small in 2013, but output rose by 1.4 mbd in 2014. Much of the increase came from Brazil and Canada. Russia’s oil output was until recently higher than for the United States, holding steady at 10.9 mbd in 2014. There was also a surprise element; output for Q4 of 2014 was some 0.3 mbd higher than anticipated by the IEA in September of that year. The rise in non-OPEC output made a decision by OPEC to squeeze US shale more likely.
4. Higher OPEC spare capacity (higher Ki). The “call on OPEC crude” is the difference between global oil demand and non-OPEC supply (and OPEC NGLs).19 In 2014, the call declined by 1.8 mbd to 30 mbd, leaving it 1 mbd short of crude output. The implied 5½ mbd of spare crude capacity compares with only about 3 mbd in 2011. Over the same period, OPEC’S NGL capacity increased by ½ mbd.20
In 2011, Libya’s conflict saw its oil output collapse by 1 mbd. Production was restored in 2012, but renewed political and security disruptions once again cut output in 2013-14. Saudi Arabia increased output to offset Libya’s disruptions, while other countries including the UAE and Kuwait also decided to raise output. When Libya’s output began to recover, there was no corresponding net decrease by other members. In fact, Saudi Arabia and other countries increased output further in 2012 and sustained high oil output in subsequent years.
Trends in Iran and Iraq broadly offset one another between 2011 and 2014. Iraq continued to increase its capacity. 2014 was no exception, which surprised many given Islamic State’s territory gains in that country. Although Iran’s technical capacity may have remained intact, the US oil embargo imposed binding constraints on Iran’s ability to sell oil. However, the interim deal signed with the P5+1 in August 2013 helped Iran’s output stabilize in 2014.21
5. OPEC coordination difficulties (lower λ). Increased coordination difficulties would make OPEC producers less likely to cooperate to accommodate non-OPEC producers in the face of weakening demand. Although OPEC is literally the textbook model of cartels, there is an extensive literature debating this. OPEC has at times been characterized as being closer to a fringe of non-cooperative (OPEC and non-OPEC) producers that is led by Saudi Arabia (Huppmann and Holz, 2012; Huppmann, 2013; Nakov and Nuno, 2013) or a small subset of OPEC members (Bremond, Hache and Mignon, 2012). Smith (2005) argues that OPEC members are more cooperative as a cartel which is possibly led by a core group. Almoguera et al. (2011) conclude OPEC behaves more like (uncooperative) Cournot competitors with a non-OPEC fringe.22
Structural factors that could contribute to this lack of coordination include differences in characteristics across members—with those in worse fiscal situations feeling less able to cut output and those with more reserves having a longer-term perspective; the absence of internal compensation or an effective enforcement mechanism; and monitoring costs. Iraq’s formal exemption from the quota following its history of sanctions and OPEC’s relatively low global market share by historical standards may have acted to reduce scope for coordination (Fattouh and Mahadeva, 2013; Huppmann and Holz, 2015).
Huppmann and Holz (2012) find that OPEC’s degree of market power declined significantly in the aftermath of the 2008 financial crisis, which in our context corresponds to a drop in λ. The media has recently reported widening rifts among members, including increasingly unproductive OPEC meetings. Long accustomed to arriving early at OPEC’s two meetings per year to build consensus among members, Saudi Arabia’s oil minister reportedly arrived at the last minute to the mid-2014 event, stayed only for a few hours, and suggested a reduction in meeting frequency to just once a year as he believed there was little point in talking.23
6. Lower marginal costs for US shale (lower Cj). Cost estimates for US shale vary considerably due to uncertainties as well as inconsistencies in cost definition.24 Arezki and Blanchard (2014), citing Rystad Energy, indicate an average breakeven for North American shale of $62, but have a range of ±$20 to reflect variation across different US shale plays. They interpret breakeven as the price at which it becomes profitable to extract. Ebinger (2014) indicates a similar range but also distinguishes between costs that include drilling and wells that have already been completed. Consistent with this, Citi estimates that half-cycle costs (around $40) could be half as low as full cycle costs.25 Some proprietary estimates include only the costs of finding and extracting the oil, while others add overheads, transportation, or a hurdle rate for the cost of capital. Sigonney (2015) presents long term marginal costs including a 10 percent profit hurdle rate ranging from $40 to $100 as at 2014.
It has been widely reported that these costs have been falling, which further complicates comparability across references. Rostand (2015) calculates that breakevens including finding, development and extraction but excluding overheads, transport, or the weighted average cost of capital have declined from $93 in 2009 to $58 in 2013. The main drivers include technology improvements such as shorter well completion times;26 superior seismic data thanks to software, sensors and lasers; the use of sand, better liquids, or even microbes for fracking; refracking of wells; and stripping idle rigs for parts (The Economist, 2015; Brousseau, 2016). These improvements would have acted to discourage or postpone OPEC’s decision to try to curtail shale production.
B. OPEC’s actions
As the oil price decline continued in the second half of 2014, many OPEC members repeatedly signaled a regime switch, indicating they opposed cutting output and intended to defend market share (Middle East Economic Survey, 2014). Saudi officials have indicated their belief that shale producers’ costs are high (approaching $100), that Saudi costs are less than $10 (Middle East Economic Survey, 2014), and that market equilibrium should be restored by reductions in supply from high cost producers.27
Nonetheless, the OPEC meeting in November 2014 surprised many by the seemingly collective decision not to reduce its quota to match the demand for its crude, or at least to reduce actual output to meet the quota. This is consistent with the formal announcement by OPEC to squeeze high-cost production (US shale in our framework) rather than accommodate it. In its December 2015 meeting, OPEC reiterated its commitment to market-share.
Consistent with this, in 2015, the call on OPEC remained at 30 mbd, yet OPEC production increased by 1.2 mbd. The biggest contributors were Saudi Arabia (0.4) mbd and Iraq (0.7) mbd, while no other major OPEC members scaled back output. OPEC capacity increased by ¼ mbd in 2015 alongside upward revisions of future capacity growth acted to re-enforce the decision to squeeze. Confidence in Iraq’s ability to continue capacity growth was restored and, unlike before, this would potentially coincide with growth from Iran. In particular, the final nuclear deal signed in July 2015 and subsequent actions taken by Iran brought with it the prospect of rising Iranian capacity in 2016 and beyond. Finally, Indonesia rejoined OPEC in late 2015, making more capacity available for an OPEC squeeze.28 Consequently, despite some scaling back of investments in response to lower oil prices, the IEA (2015, 2016) increased its estimates of OPEC capacity in 2016 by ¾ mbd to 42.6 mbd between the 2015 and 2016 editions of its Medium Term Oil Market Report.
Because of an increase in the number of OPEC members and because much of the capacity growth is accounted for by traditional political rivals, discord among OPEC intensified and arguably acted to make a coordinated cut less feasible.
C. Market responses
The November 2014 OPEC decision accelerated the oil price decline to about $50 in the first quarter of 2015. A subsequent recovery during 2015 proved short-lived, as the excess supply pressures that had built up in 2014 did not unwind. As a result, oil was cheaper at the end of 2015 than at the start, and averaged $50 for the year as a whole. Since that decision, other structural factors have continued to favor pursing market share. In particular, US and other non-OPEC capacity has continued rising, and global demand has continued to disappoint.
US shale supply started showing signs of scaling back
Following the decline in oil prices, debate shifted to the speed of the US shale supply response. As of early 2015, the response of shale was hard to determine; some commentators emphasized slowing growth in output as weakness while others pointed to ongoing rises in levels as strength. There is empirical evidence that lower oil prices reduce drilling for new wells (Toews and Naumov, 2016). Rig counts initially gave mixed signals but ended the year 62 percent lower than at the end of 2014 and at their lowest level since 1999 (Williams, 2016). Yet rig counts can be an imperfect leading indicator of output or output growth. The number of existing wells being fracked, arguably a better predictor, was still rising (The Economist, 2016a).
Those expecting resilient production to continue refer to efficiency gains from learning-by-doing and cost cutting. Rystad Energy data cited in The Economist (2016b) for selected US shale plays suggests breakeven oil prices declined by about 40 percent between 2013 and 2015,29 and recent corporate filings report cost savings of 25-30 percent per well (IEA, 2016). Others cite oil-price hedging by producers and their ability to secure ongoing financing to sustain operations in hope of a price recovery as merely temporary factors that were delaying the inevitable.30
Nonetheless, there is econometric evidence that US shale oil is more price-responsive than conventional oil (Bjornland et al, 2016). In the latter parts of 2015, there were indications that US LTO levels had peaked in the middle of that year as well as clearer signs of declining output levels reported in the September 2015 edition of the OMR. This is consistent with US shale production starting to be squeezed. 2016 LTO output was revised down to about 4 mbd (EIA, 2015b). At face value, output in 2016 would be substantially lower than its peak and than in 2014. An alternative measure of the squeeze is a comparison between the latest available projections and earlier ones before low oil prices had been factored into projections. 2016 shale output was forecast to be about 5 mbd in early 2015, some 1 mbd higher than the latest available forecasts.
Non-OPEC non-shale capacity investment was cut drastically.31
Multinationals like BP, Chevron, ExxonMobil, Shell and Total have responded to the weaker oil price by laying off workers, cutting investment, and in some cases postponing and canceling some of their exploration projects (The Economist, 2016a). According to Wood MacKenzie, close to $400 billion worth of large upstream oil & gas projects have been put on hold (as of January 2016), and Rystad Energy estimate capital spending declined by $215 billion between 2014 and 2015 alone.32 Non-OPEC non-shale supply is also expected to be negatively affected by output decreases in Russia due to the sanctions as well as the lower oil price outlook (IEA, 2015).
However, 2015 saw net growth of an estimated 1.2 mbd. Shell’s Chief Financial Officer has stated that, having already incurred investment costs, the incentive is to produce “as flat out as you can” (The Economist, 2016a) and that true marginal variable costs are much lower after factoring in mothballing expenses. Russia’s production increased marginally in 2015, but the sources of growth were again Canadian sands and Brazilian waters. Both are high cost oil sources which by many measures exceed US shale costs. However, the price responses are much slower than for US shale as the projects entail high upfront capital costs, which have already been incurred, and long project lifecycles. In other words, the coming months are the long run for many shale plays and only the short run for other oil resources.33 As a result of this non-OPEC non-shale growth, OPEC’s market share increased only slightly from 40 percent in 2014 to 41 percent in 2015.
Demand growth rose as result of the lower oil price
Lower prices contributed to demand acceleration of 1:6 mbd in 2015 (IEA, 2016). However, this rise is relatively small considering the oil price decline, suggesting renewed weakness that has acted to re-enforce the market share strategy. 2015 GDP growth expectations were revised down to 3¼ percent, which is lower than every year since 2009. In particular, growth in relatively energy-hungry Emerging Market and Developing Economies including China likely declined for the fifth consecutive year in 2015 (IMF, 2015). Moreover, demand growth is expected to slow again to 1.2 mbd in 2016 (IEA, 2016) and structural pressures on demand could also intensify after the December 2015 Paris Climate Change conference.
V. Quantitative empirical calibration
This section matches the events described above to the model by combining observed data and empirically supported parameter values. We start with two snapshots from before the oil price crash (in 2012 and 2014) that confirm that the model predicts the high oil prices and relatively restrained OPEC production consistent with an “accommodate”equilibrium. We proceed to a set of three illustrative scenarios in which to demonstrate a squeeze. They show in a stylized way how market developments or a revised calculation by OPEC could induce a change of strategy. Finally, we have two instances where we apply the model to predicted data for the future to show it generates a squeeze equilibrium, which in turn predicts higher OPEC supply and low prices in line with forecasts.
A. Calibration approach and data
Actual prices and forecasts (based on futures markets) are the Average Petroleum Spot Price (APSP) taken from the IMF’s World Economic Outlook database, specifically those used for the January 2016 World Economic Outlook Update.
On the demand side of the model, actual historical or future forecast demand quantities in millions of barrels per day (mbd) are sourced from various issues of the MOMR and IEA (2016). A key parameter is β, which is chosen so as to ensure demand elasticities that are consistent with estimates in the literature for a relevant range of observed prices and quantities. Setting β = 8 implies an elasticity of demand of almost -0.15 when oil prices are $100 and around -0.07 when oil prices are $50. This range falls comfortably within the confines of empirical work.34 We solve for the demand shift parameter (a) using actual demand, actual prices, and β (recall that our demand curve is D(P) = (α – P)/β).
Actual historical global supply and inventory changes, which account for discrepancies with respect to global demand, are also sourced from MOMR issues, as are OPEC and non-OPEC supply. However, to distinguish US shale production from more conventional US output, we refer to the Energy Information Administration (EIA, 2015).35 For non-OPEC supply, capacity is assumed to equal actual output. For OPEC, sustainable capacity estimates are taken from the IEA (2013, 2015, 2016). As mentioned earlier, non-OPEC statistics do not distinguish between crude and NGLs, but OPEC statistics do. We add NGLs to OPEC crude output/capacity, resulting in volumes that are higher than more widely reported crude-only volumes. For supply forecasts, non-OPEC capacity/output is derived from IEA (2016) and shale capacity is taken from EIA (2015). The IEA does not produce OPEC supply forecasts but OPEC capacity is taken from IEA (2016).
We set marginal cost for US shale based on presentations of proprietary information and the references given in Section 4. We include “full-cycle” marginal costs for shale and short-run marginal costs for conventional producers including OPEC. Numerical values will be indicated in the subsections that follow. OPEC’s pricing power λ is solved for the value that makes calculated prices and quantities consistent with the data and other parameters as per equation (2) which determines OPEC’s supply behaviour.
B. Accommodate examples
We present results for the second quarter of 2014 because it preceded the decline in oil prices as well as 2012 for robustness; these are represented as examples 1A and 1B in Table 1. Our main finding is that it was then still optimal for OPEC to follow an accommodate strategy.
In both years, oil prices (P) were close to $105. Actual demand (D) was 90.7 mbd in 2012 and 92 mbd in 2014. Setting β = 8 implies a price elasticity of demand of about -0.15 in both years. Then P, D, and β can be substituted into the demand function to solve for α for each year. Global supply exceeded demand by 0.2 mbd in 2012 and by 3.4 mbd in the second quarter of 2014, implying large inventory builds. As discussed earlier, shale capacity (Kj) was 2 mbd in 2012 and 4 mbd in 2014, while OPEC capacity (Ki) remained constant and other non-OPEC capacity (Kℓ) rose.
Marginal costs are set at Ci = $10 for OPEC in both years and and Cj = $90 in 2012 and Cj = $85 for US shale in 2012 and 2014, respectively. As discussed in subsections 4A and 5A, this variable is difficult to pin down, but we choose values towards the top of the range to represent full cycle costs and allow for a modest cost reduction between 2012 and 2014. We calculate that λ ≈ ⅓ for both 2012 and 2014. This is broadly consistent with the OPEC literature discussed earlier, including numerical model simulations and econometric estimates (Huppmann and Holz, 2012; Almoguera et al., 2011) which imply λ < ½.
The fitted data confirm that our theory assumptions A1 and A2 hold in both scenarios 1A and 1B. Consistent with A1, US shale oil is viable given that price exceeds its cost. A2 also holds in both 2012 and 2014, which means that OPEC had sufficient spare capacity to carry out the squeeze strategy.
The data are consistent with an accommodate equilibrium as per Proposition 2, so OPEC optimally chose not to pursue the squeeze. In particular, the parameters and data imply
The calculated quantity supplied by OPEC under such an equilibrium (denoted in Table 1 by
C. Illustrative squeeze scenarios
Taking 2014 as a starting point, this subsection presents three constructed scenarios where a squeeze is triggered and US shale output is zero. The first two separately show how higher US shale capacity and lower OPEC coordination individually trigger the switch. The third illustrative scenario combines the first two of these with lower marginal costs for US shale and lower global demand, thus capturing four of the five drivers discussed above, to generate a squeeze.36
Although stylized, these scenarios show our key point that the regime switch may have been optimal for OPEC from an ex ante viewpoint, given the information they may have incorporated in deciding how to react to the initial price decline in the 2nd half of 2014.
We in scenario 2A illustrate a case in which all demand and cost parameters (as well as λ) are held constant at 2014 levels but allow Kj = 5.5. Although illustrative, we chose this value because shale output was forecast to reach 5.5 mbd in 2018-2024 (EIA, 2015).37 These forecasts would already imply a capacity in excess of the values of
The model assumptions A1 and A2 again hold: shale output would have been positive under the counterfactual of an accommodation strategy, and OPEC indeed has the capacity required for a squeeze.
Another important development discussed in Section 4 is a decline in λ, representing OPEC’s lower ability to push up prices. In scenario 2B, we again hold all the 2014 parameters constant, including Kj = 4, but now use Proposition 2 to solve for the critical value of λ such that
The illustrative scenarios so far imply prices well above those observed in late 2014 and early 2015. Our scenario 2C generates a lower price by allowing multiple parameters to shift in a manner that is qualitatively consistent with Section 4. As discussed earlier, a number of commentators have pointed to the declining marginal costs of US shale, especially since oil prices began to fall, so we set Cj = 55 = P** .38 Given this lower price, setting demand to that observed for 2015 implies a sizeable decline in the solved value of α (relative to 2014), which implies a weakening in global demand. Thus, although lower US costs discourage the squeeze, the negative demand shift encourages it. Letting US shale capacity Kj = 5.5, we again use Proposition 2 to find the value of λ for which
In summary, scenario (2C) generates a squeeze equilibrium with a more realistic oil price through higher US shale capacity, lower OPEC pricing power, weaker demand, and falling production costs. OPEC’s market share is 42 percent of demand compared to 35 percent under the accommodate counterfactual. A1 continues to hold, which implies that shale would have been viable (aided by lower costs but harmed by inter alia weaker demand) had it been accommodated. A2 also still holds. In terms of our qualitative discussion from Section 4, this shows that the various factors favoring a squeeze can quantitatively outweigh lower US shale costs (which point to accommodation).
D. Future squeeze equilibria
This subsection recalibrates the model using forecasts of 2020 oil market data. The first retains the notion of all US shale being squeezed out of the market by then. The second allows for some US shale to remain active. These squeeze equilibria imply that the market-share strategy can be rationalized economically as a “less-bad” option for OPEC in the future, and also yield more plausible forecasts for OPEC output than would be the case in an accommodate equilibrium.
In equilibrium 3A, the oil price for 2020 of $58 is used to pin down marginal cost for US shale oil of Cj = $58. By assumption, β is unchanged. The demand shift parameter (α) is solved as before, except this time using third-party forecasts of P and D rather than historical values, and increases by a plausibly moderate amount between 2014 and 2020. As per Proposition 2, Kj = 5.6 based on EIA (2015) and so
Hence, OPEC supply is
Calibrating the model
Calibrating the model
| Scenario: | Accomodate examples | Squeeze scenarios [using 2014Q2] | Squeeze equilibria | |||||
|---|---|---|---|---|---|---|---|---|
| 1A | 1B | 2A | 2B | 2C | 3A | 3B | ||
| 2012 | 2014Q2 | High kj | Low λ | Multiple* | 2020 All shale | 2020 Some shale | ||
| p | Price[$/barrel) | 105 | 106 | 85 | 85 | 55 | 58 | 58 |
| D | Demand [mbd] | 90.7 | 92.0 | 94.7 | 94.7 | 94.4 | 100.5 | 100.5 |
| β | Inverse demand slope parameter | 8.0 | 8.0 | 8 | 8 | 8 | 8.0 | 8.0 |
| P(Dβ)-1 | Demand elasticity | −0.14 | −0.14 | −0.11 | −0.11 | −0.07 | −0.07 | −0.07 |
| α | Demand shift parameter (mbd) | 831 | 843 | 843 | 843 | 810 | 862 | 862 |
| α(β)-1 | Demand shift parameter (mbd) | 104 | 105 | 105 | 105 | 101 | 108 | 108 |
| S | Global supply actual | 90.9 | 95.4 | 100.5 | 100.5 | |||
| Si | Leader supply actual | 37.6 | 36.4 | |||||
| Leader supply (accommodate) | 37.4 | 33.1 | 32.0 | 34.2 | 32.8 | 34.8 | 38.0 | |
| Leader supply (squeeze) | 41.2 | 39.7 | 39.7 | 39.7 | 39.4 | 41.6 | 38.6 | |
| Ki | Leader capacity | 41.3 | 41.4 | 41.4 | 41.4 | 41.4 | 43.5 | 43.5 |
| Kj+Kl | Follower supply (mbd) | 53.3 | 59.0 | 60.5 | 59.0 | 60.5 | 58.9 | 61.9 |
| Kj | Stale capacity (mbd) | 2.0 | 4.0 | 5.5 | 4.00 | 5.5 | 5.6 | 3.0 |
| Kl | ROW capacity (mbd) | 51.3 | 55.0 | 54.96 | 54.96 | 54.96 | 58.9 | 58.9 |
| Ki+Kl | Nor-shale capacity (mbd) | 92.7 | 96.3 | 96.35 | 96.35 | 96.35 | 102.4 | 102.4 |
| Ci | Marginal cost (OPEC) | 10 | 10 | 10 | 10 | 10 | 10 | 10 |
| Cj | Marginal cost(non-OPEC) | 90 | 85 | 85 | 85 | 55 | 58 | 58 |
| λ | OPEC co-ordination power | 0.32 | 0.36 | 0.36 | 0.32 | 0.21 | 0.21 | 0.17 |
| Conditions checks | ||||||||
| A1 | holds | holds | holds | holds | holds | holds | holds | |
| left | 80 | 75 | 75 | 75 | 45 | 48 | 48 | |
| right | 100 | 104 | 100 | 91 | 56 | 69 | 53 | |
| A2 | holds | holds | holds | holds | holds | holds | holds | |
| left | 741 | 758 | 758 | 758 | 755 | 804 | 804 | |
| right | 741 | 771 | 771 | 771 | 771 | 819 | 819 | |
| Regime conditions | ||||||||
| Proposition 1 | Accomm. | Accomm. | Squeeze | Squeeze | Squeeze | Squeeze | Squeeze | |
| −1.8 | −1.5 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ||
| 3.8 | 5.5 | 5.5 | 4.0 | 5.5 | 5.6 | 3.0 | ||
Calibrating the model
| Scenario: | Accomodate examples | Squeeze scenarios [using 2014Q2] | Squeeze equilibria | |||||
|---|---|---|---|---|---|---|---|---|
| 1A | 1B | 2A | 2B | 2C | 3A | 3B | ||
| 2012 | 2014Q2 | High kj | Low λ | Multiple* | 2020 All shale | 2020 Some shale | ||
| p | Price[$/barrel) | 105 | 106 | 85 | 85 | 55 | 58 | 58 |
| D | Demand [mbd] | 90.7 | 92.0 | 94.7 | 94.7 | 94.4 | 100.5 | 100.5 |
| β | Inverse demand slope parameter | 8.0 | 8.0 | 8 | 8 | 8 | 8.0 | 8.0 |
| P(Dβ)-1 | Demand elasticity | −0.14 | −0.14 | −0.11 | −0.11 | −0.07 | −0.07 | −0.07 |
| α | Demand shift parameter (mbd) | 831 | 843 | 843 | 843 | 810 | 862 | 862 |
| α(β)-1 | Demand shift parameter (mbd) | 104 | 105 | 105 | 105 | 101 | 108 | 108 |
| S | Global supply actual | 90.9 | 95.4 | 100.5 | 100.5 | |||
| Si | Leader supply actual | 37.6 | 36.4 | |||||
| Leader supply (accommodate) | 37.4 | 33.1 | 32.0 | 34.2 | 32.8 | 34.8 | 38.0 | |
| Leader supply (squeeze) | 41.2 | 39.7 | 39.7 | 39.7 | 39.4 | 41.6 | 38.6 | |
| Ki | Leader capacity | 41.3 | 41.4 | 41.4 | 41.4 | 41.4 | 43.5 | 43.5 |
| Kj+Kl | Follower supply (mbd) | 53.3 | 59.0 | 60.5 | 59.0 | 60.5 | 58.9 | 61.9 |
| Kj | Stale capacity (mbd) | 2.0 | 4.0 | 5.5 | 4.00 | 5.5 | 5.6 | 3.0 |
| Kl | ROW capacity (mbd) | 51.3 | 55.0 | 54.96 | 54.96 | 54.96 | 58.9 | 58.9 |
| Ki+Kl | Nor-shale capacity (mbd) | 92.7 | 96.3 | 96.35 | 96.35 | 96.35 | 102.4 | 102.4 |
| Ci | Marginal cost (OPEC) | 10 | 10 | 10 | 10 | 10 | 10 | 10 |
| Cj | Marginal cost(non-OPEC) | 90 | 85 | 85 | 85 | 55 | 58 | 58 |
| λ | OPEC co-ordination power | 0.32 | 0.36 | 0.36 | 0.32 | 0.21 | 0.21 | 0.17 |
| Conditions checks | ||||||||
| A1 | holds | holds | holds | holds | holds | holds | holds | |
| left | 80 | 75 | 75 | 75 | 45 | 48 | 48 | |
| right | 100 | 104 | 100 | 91 | 56 | 69 | 53 | |
| A2 | holds | holds | holds | holds | holds | holds | holds | |
| left | 741 | 758 | 758 | 758 | 755 | 804 | 804 | |
| right | 741 | 771 | 771 | 771 | 771 | 819 | 819 | |
| Regime conditions | ||||||||
| Proposition 1 | Accomm. | Accomm. | Squeeze | Squeeze | Squeeze | Squeeze | Squeeze | |
| −1.8 | −1.5 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | ||
| 3.8 | 5.5 | 5.5 | 4.0 | 5.5 | 5.6 | 3.0 | ||
A less stylized 2020 equilibrium includes non-zero US shale output in a way that reduces OPEC supply while leaving global supply, prices, and demand unaltered. In particular, equilibrium 3B relaxes the assumption that US shale is a homogenous group and instead allows for varying costs such that the futures price of $58 would only squeeze out those with higher costs. In terms of the model setup, this is equivalent to n being the subset of US shale capacity consisting of those shale plays with costs above $58.
In particular, setting Kj = 2.8 to represent the more expensive half of US shale and following the same procedure as in equilibrium 3A,39 a squeeze equilibrium would result in OPEC producing 38.8 mbd and US shale producing 2.8 mbd from the cheaper oil fields instead of the full 5.6 mbd. We find that
An interesting implication of this low value of λ is that the counterfactual price under the accommodate equilibrium is now only $5 higher than the squeeze price. In this sense, US shale oil effectively becomes the price-setter in this future scenario regardless of which equilibrium is played.
VI. Conclusions
The debate on the rationale for and the repercussions of OPEC’s November 2014 switch to a “market-share” strategy has drawn considerable attention. Many oil-market analysts view the decision as a battle of “OPEC vs shale” aimed at squeezing higher-cost US players out of the market. We have contributed to this debate with an equilibrium model that helps understand how fundamental market developments can rationalize OPEC’s regime switch as a profit-maximizing strategy. Such a shift can explain why OPEC supply can optimally rise in response to high-cost supply growth (such as from US shale) or weaker global demand and induce an oil price collapse.
Our calibration of the model shows it was better for OPEC to accommodate expanding US shale production up to 2014 despite having the spare capacity to squeeze them out of the market. Stylized comparative statics show how plausible updates to OPEC’s information set prompted a switch to a market-share strategy in late 2014. Calibration to forecasts of future market data shows how evolving developments can sustain a regime switch to a squeeze. Through the lens of the model, the market-share strategy can be the better of the two options—given US shale capacity, OPEC coordination prospects, weak global oil demand, and other market factors.
It remains to be seen whether the initial logic of the squeeze will play out and vindicate the strategy in the coming years. As of early 2016, the squeeze appears to have been less successful than OPEC might have calculated: a substantial decline in US shale output does not (yet) appear imminent, and the squeeze has perhaps been more costly than anticipated given the continued depression of oil prices (IEA, 2016). One potential reason is that the costs of US shale have fallen more than might have been anticipated. In terms of our framework, this could prompt a further OPEC regime switch back to accommodate. It is also possible that the attempted squeeze and the re-entry of Iran have made coordinated accommodation so problematic that OPEC reluctantly yet rationally persists with the squeeze. Moreover, high-cost producers beyond US shale may be squeezed out by lower oil prices. This paper has not pretended to forecast the future of the industry but rather to provide a coherent economic framework to think about the key drivers of such regime switches, including the one that took place at the end of 2014.
Finally, while we have focused on oil, our approach can be applied to other energy sectors. For example, natural gas is also characterized by significant supply-side concentration. In the EU, Gazprom plays a dominant role in the sense that it accounts for around 30% of gas imports. It competes against domestic supplies in some EU countries, other pipeline exporters, and liquefied natural gas (LNG) – which likely all have higher production costs. Recent gas policy discussions suggest that the demand slowdown and likely future competition from US shale gas arriving in Europe as LNG mean that Gazprom should begin a “price war” to regain market share and squeeze higher-cost LNG players (and possibly coal production) out of the European market (Henderson 2016). This regime choice has some close parallels with the oil-market setting, and our model could similarly be used to quantify the conditions under which a market-share strategy becomes optimal for Gazprom.
References
Almoguera, Pedro A. Christopher C. Douglas, and Ana Maria Herrera (2011). “Testing for the cartel in OPEC: Non-cooperative collusion or just non-cooperative?” Oxford Review of Economic Policy, 27(1):144–168
Andrade de Sa, Saraly and Julien Daubanes (2014). “Limit Pricing and the (In)Effectiveness of a the Carbon Tax.” Working Paper at ETH Zurich, November 2014.
Arezki, Rabah and Olivier Blanchard (2014), “Seven Questions About The Recent Oil Price Slump.” iMFdirect blog. Posted 22 December 2014.
Atkins, Frank and Tayyebi Jazayeri (2004), “A Literature Review of Demand Studies in World Oil Markets”. University of Calgary Department of Economics Discussion Paper, 2004−07.
Baffes, John, M. Ayhan Kose, Franziska Ohnsorge, and Marc Stocker. “The Great Plunge in Oil Prices: Causes, Consequences, and Policy Responses.” Development Economics Group Research Note, PRN/15/01. World Bank.
Baumeister, Christiane and Lutz Kilian (2015). “Understanding the Decline in the Price of Oil Since June 2014”. CEPR Discussion Paper, 10404
Baumeister, Christiane and Gert Peersman (2013). “The Role of Time-varying Price Elasticities in Accounting for Volatility Changes in the Crude Oil Market”. Journal of Applied Econometrics, 28:1087−1109
Behar, Alberto and Malika Pant (2015). “Outlook for the Global Oil Market and Implications for Saudi Arabia,” In Ahmed Al-Darwish, Naif Alghaith, Alberto Behar, Tim Callen, Pragyan Deb, Amgad Hegazy, Padamja Khandelwal, Malika Pant, and Haonan Qu, Saudi Arabia: Tackling Emerging Economic Challenges to Sustain Growth. International Monetary Fund Middle East and Central Asia Department. Washington, DC.
Beidas-Strom, Samya and Carolina Osorio-Buitron (2015). “The Oil Price Collapse: Demand or Supply?”. Uneven Growth: Short- and Long-Term Factors, World Economic Outlook, April 2015. International Monetary Fund. Washington, DC.
Bremond, Vincent, Emmanuel Hache and Valerie Mignon (2012). “Does OPEC still exist as a cartel? An empirical investigation”. Energy Economics, 34:125−131.
Bjørnland, Hilde C., Frode Nordvik, and Maximilian Rohrer (2016), “How Flexible is US Shale Oil Production? Evidence from North Dakota”, Presentation at conference on ’Oil, Middle East, and the Global Economy’, University of Southern California, 1 April 2016.
Brusseau, Alain (2016). Oil Market Developments. Mimeograph. International Monetary Fund. 3 February.
Cairns, Robert and Enrique Calfucura (2012). “OPEC: Market failure or power failure?”. Energy Policy, 50: 570−580.
Ebinger, Charles (2014), “World Oil Demand: And Then There Was None”. Brookings PlanetPolicy. Posted 17 October 2014.
Energy Information Administration (2014). Annual Energy Outlook 2014. Washington, DC.
Energy Information Administration (2015). Annual Energy Outlook 2015. Washington, DC.
Energy Information Administration (2015b). Short-term Energy Outlook, September 2015. Washington, DC.
Energy Information Administration (2016). “Hydraulic fracturing accounts for about half of current U.S. crude oil production.” Today in Energy, March 15, 2016.
Fattouh, Bassam and Lavan Mahadeva (2013). “OPEC: What Difference Has It Made?”. Annual Review of Resource Economics, 5:427−443
Fattouh, Bassam, Rahmatallah Poudineh and Anupama Sen (2015). “The Dynamics of the Revenue Maximization - Market Share Trade-off: Saudi Arabia’s Oil Policy in the 2014-2015 Price Fall”. Oxford Institute for Energy Studies Working Paper, WPM 61.
Hamilton, James D. (2009). “Understanding crude oil prices.” The Energy Journal, 30(2): 179–206
Hamilton, James D. (2015). “Demand factors in the collapse of oil prices.” Econbrowser blog, posted 11 January 2015.
Henderson, James (2016). “Gazprom – Is 2016 the Year for a Change of Pricing Strategy in Europe?” Working Paper at Oxford Institute for Energy Studies, January 2016.
Huppmann, Daniel (2013). “Endogenous Shifts in OPEC Market Power: A Stackelberg Oligopoly with Fringe.” Working Paper at DIW Berlin, July 2013.
Huppmann, Daniel and Franziska Holz (2012). “Crude oil market power: A shift in recent years?” The Energy Journal, 33(4):1–2
Huppmann, Daniel and Franziska Holz (2015). “What about the OPEC Cartel?” DIW Roundup: Politik im Fokus, 58.
Hussain, Aasim, Rabah Arezki, Peter Breuer, Vikram Hassar, Thomas Helbling, Paulo Medas, Martin Sommer, and an IMF Staff Team (2015). “Global Implications of Lower Oil Prices”. International Monetary Fund Staff Discussion Note, SDN/15/15.
International Energy Agency (2013). Medium-term Oil Market Report, 2013. Paris.
International Energy Agency (2014). Medium-term Oil Market Report, 2014. Paris.
International Energy Agency (2015). Medium-term Oil Market Report, 2015. Paris.
International Energy Agency (2016). Medium-term Oil Market Report, 2016. Paris.
International Monetary Fund (2012). Coping with High Debt and Sluggish Growth, World Economic Outlook, October 2012. Washington, DC
International Monetary Fund (2014). Legacies, Clouds, Uncertainties, World Economic Outlook, October 2014. Washington, DC
International Monetary Fund (2015). Adjusting to Lower Commodity Prices, World Economic Outlook, October 2015. Washington, DC
Killian, Lutz and Daniel Murphy (2014). “The Role of Inventories and Speculative Trading in the Global Market for Crude Oil”. Journal of Applied Econometrics, 29:454−478.
Middle East Economic Survey (2014). Saudi Sees $90/B Crude As Price Floor. 3 October 2014.
Mabro, Robert (1998). “The Oil Price Crisis of 1998”. Oxford Institute for Energy Studies, SP 10.
Mohaddes, Kamiar and M. Hashem Pesaran (2015). “Country-Specific Oil Supply Shocks and the Global Economy: A Counterfactual Analysis.” Cambridge EPRG Working Paper, 1512.
Mohaddes, Kamiar and Mehdi Raissi (2016). “The U.S. Oil Supply Revolution and the Global Economy.” Cambridge EPRG Working Paper, 1604.
Nakov, Anton and Galo Nuno (2013). “Saudi Arabia and the Oil Market.” Economic Journal 123, 1333−1362.
Pierru, Axel, James Smith and Tamim Zamrik (2016). On the Size and Significance of OPEC’s Spare Production Capacity. Working Paper at Southern Methodist University, 21 March 2016.
Rostand, Antoine (2015). What Future Evolution for US Light Tight Oil?. Schlumberger Presentation, World Policy Conference, 21 November 2015.
Sigonney, Pierre (2015). International Oil Companies in the Evolving Energy Landscape. Total Presentation, World Policy Conference, 21 November 2015.
Smith, James (2005) “Inscrutable OPEC: Behavioral Tests of the Cartel Hypothesis,” The Energy Journal 26(1), 51–82.
Smith, James (2009). “World Oil: Market or Mayhem?”. Journal of Economic Perspectives (vol. 23, no. 3, 2009).
The Economist (2015). The oil industry: After OPEC. 16 May 2015.
The Economist (2016a). Oil Conundrum. 23 January 2016.
The Economist (2016b). DUC and cover. 12 March 2016.
Tirole, Jean (1988). The Theory of Industrial Organization. MIT Press.
Toews, Gerhard and Alexander Naumov (2016). The Relationship Between Oil Price and Costs in the Oil Industry. The Energy Journal 36, 237–254.
Verleger, Philip K. Jr. (2016). “Structure Matters: Oil Markets Enter the Adelman Era.” The Energy Journal 36, 129–157.
Westelius, Niklas (2013). “ Assessing Saudi Arabia’s Systemic Role in the Oil Market and Global Economy.” International Monetary Fund Selected Issues Paper SM/13/173.
Williams, James (2016). North America Rotary Rig Counts. WTRG Economics. 5 January 2016.
Yergin, Daniel (2008). The Prize: The Epic Quest for Oil, Money & Power. Free Press. New York.
Appendix A: Proofs
Appendix B: Data Sources
Oil prices (historical and assumed): IMF World Economic Outlook database (January 2016 World Economic Outlook Update vintage).
Demand volumes (historical and forecast): International Energy Agency Medium Term Oil Market Report (2015, 2016) and Monthly Oil Market Report (numerous issues).
Demand parameters: β = 8, in line with existing empirical work; α solved using P, D, and β.
Global supply volumes; inventory changes (realized): International Energy Agency Medium Term Oil Market Report (2015, 2016) and Monthly Oil Market Report (numerous issues).
OPEC supply volumes (historical): International Energy Agency Medium Term Oil Market Report (2015, 2016) and Monthly Oil Market Report (numerous issues).
OPEC supply volumes (forecast): Solved endogenously.
OPEC capacity (historical and forecast): International Energy Agency Medium Term Oil Market Report (2015, 2016).
US shale capacity/supply (realized and forecast): Energy Information Administration (2015).
Non-OPEC non-shale capacity/supply (realized and forecast): International Energy Agency Medium Term Oil Market Report (2015, 2016) and Monthly Oil Market Report (numerous issues).
US shale marginal cost: In line with industry reports or equal to oil price forecasts (squeeze).
OPEC marginal cost: As per industry reports.
OPEC coordination power: Determined endogenously.
Behar is in the Middle East and Central Asia Department, International Monetary Fund, and would like to thank staff of the International Energy Agency, colleagues in the IMF Commodities Unit, and participants in the Middle East and Central Asia Department Seminar Series. Ritz is in the Energy Policy Research Group, University of Cambridge, and would like to thank the IMF Commodities Unit, especially Rabah Arezki, as well as the Center for Energy and Environmental Policy Research (CEEPR) at MIT for their hospitability during the early stages of this project.
As at the end of 2015, the members of The Organization of the Petroleum Exporting Countries (OPEC) are (in descending order of crude oil capacity for 2015): Saudi Arabia, Iraq, Iran, United Arab Emirates, Kuwait, Venezuela, Nigeria, Angola, Algeria, Indonesia, Qatar, Ecuador, and Libya, although Libya’s capacity is at present highly constrained by its security situation. This amounts to cumulative production capacity of 35½ mbd. OPEC’s actual crude (31½ mbd) and NGL (6½ mbd) output exceeded 40% of global demand in 2015.
Quoted in Wall Street Journal (4 June 2015) “Saudi Arabia’s Celebrity Oil Minister Ali al-Naimi Prepares for Potential OPEC Swan Song”.
Ise (1926) quoted in Yergin (2008).
Although Saudi Arabia is the dominant player in OPEC, we refer to the broader group as a collective. Saudi Arabia has accounted for the bulk of OPEC adjustment when responding to moderate changes in the oil market, but large adjustments in OPEC output have included participation from multiple parties, including collective cuts during the Global Financial Crisis and some increases in output during the recovery and in response to supply outages during the Arab Spring. In addition, a lot of recent growth in OPEC capacity and output has come from Iraq, representing the choice of Iraq to produce more and of other members not to keep collective OPEC output constant.
Unless crude is specifically mentioned, oil refers to liquids, namely crude oil and natural gas liquids (NGLs) as these are very close substitutes. The IEA does not distinguish between the two when reporting demand or non-OPEC supply. For OPEC, these are separated out by the IEA in part because NGLs are not formally part of OPEC’s quota. Gas, whether natural gas or associated gas generated from the production of liquids, is excluded.
As argued by Fattouh, Poudine and Sen (2015) for Saudi Arabia, many OPEC countries remain undiversified and hence reliant on oil for meeting domestic spending pressures, which makes revenue the prime consideration.
The Hotelling rule is well-known to have little empirical explanatory power. Cairns and Calfucura (2012) argue it is only relevant for producers with a limited resource horizon, which is not the case for the large oil producers.
“[The policy to defend market share] is also a defense of high efficiency producing countries, not only of market share. We want to tell the world that high efficiency producing countries are the ones that deserve market share. That is the operative principle in all capitalist countries.” Oil Minister at the time, Mr Al-Naimi, Middle East Economic Survey Interview, 21 December 2014.
We have not declared war on shale or on production from any given country or company. - Mr Al-Naimi, cited in CBS News Moneywatch, "Saudi Arabia: We have not declared war on shale," 23 February 2016.
Mabro (1998) suggests a market-share strategy is not sensible: since conventional oil producers traditionally have operating costs that are well below prevailing prices, it would take too large price decline to induce their exit. Our analysis revisits this issue with a more formal economic framework geared towards the distinction between conventional and unconventional oil production. Earlier, having incurred substantial losses in the early 1980s following accommodative production cuts (Westelius, 2013), the subsequent rise in output was arguably a shift to a market-share strategy.
They also note that OPEC allowing for more price volatility introduces uncertainty for prospective entrants and can discourage entry as a result.
Although the relative importance of each factor is difficult to pin down, OPEC’s renouncement of price support and rapid expansion of oil supply from unconventional sources appear to have played a crucial role since mid-2014. Empirical estimates also indicate that supply has accounted for the lion’s share of the plunge in oil prices (Baffes et al, 2015; Beidas-Strom and Osorio-Buitron, 2015).
Classic limit-pricing theory relies on the “incumbent” player raising price again following the exit of the weaker “entrant” (Tirole, 1988: Chapter 9). Under perfect information, this leads to a credibility problem: the entrant realizes that price will go back up (making re-entry profitable), so cannot be induced to exit in the first place. Thus limit pricing does not work without the addition of another market imperfection such as asymmetric information (which allows the incumbent to build a “tough” reputation by pricing low).
Lower pricing may also be the result of dynamic considerations which we do not model explicitly here, or because some domestic OPEC stakeholders wish to maximize revenue rather than profits.
It is easy to check that the second-order condition is satisfied for any λ > 0.
The industrial-organization literature on collusion comes to conflicting views on how the cycle affects the stability of price coordination (Tirole, 1988: Chapter 6). On the one hand, there is a greater short-term temptation to cheat when demand is high; equilibrium prices are thus lower in booms in order to limit this incentive to cheat. On the other hand, with imperfect observability of actions, firms cannot perfectly distinguish between rivals cheating and low demand; thus price wars are more likely during busts. Similarly, the incentive to deviate is typically stronger when future demand is falling. Our model results are consistent with the latter perspective.
As mentioned earlier, OPEC has stated it is not targeting specific countries or production methods. Nonetheless, the shorter production lifecycle of US shale oil has directed market and media attention to this production source.
Alternative measures yield similar results. Production in the Eagle Ford and Bakken formations alone doubled to about 2½ mbd, while proxies reported by the World Bank (Baffes et al, 2015) indicate a doubling from 2 mbd to 4 mbd. US oil extracted by fracking rose by a similar magnitude to account for about half of US crude production in 2014, while conventional output declined slightly (EIA, 2016).
As mentioned earlier, NGLs are not part of OPEC’s quota of 30 mbd.
Further discussion is available in Behar and Pant (2015).
Libya and Iran were not the only countries to experience supply disruptions. Verleger (2016) notes that unanticipated global supply outages rose from 1 mbd to 3 mbd after 2011.
Others have emphasized the dominant role of Saudi Arabia as a swing producer that has targeted a specific price that balances the trade-off between short-term government funding needs and discouraging long-term incentives to substitute away from oil before reserves are exhausted (Behar and Pant, 2015; Cairns and Calfucura, 2012). In a 1998 interview, Mr Al-Naimi stated that Saudi Arabia had formally abandoned the role of swing producer in the 1980s (Westelius, 2013).
Reported by The Wall Street Journal, 5 October 2014 “OPEC Members’ Discord Adds to Slide in Oil Prices”.
Ebinger (2014), notes "While various pundits have opined on this question, the truth of the matter is that no analyst really knows the full range of production costs across the unconventional crude oil production continuum since this information is highly proprietary."
Reported by FTAlphaville 8 October 2014 “It’s a super market price war! (in oil)“.
For example, the time between permit applications and production declined by about 10 percent between the start of 2012 and 2014 (Currie, 2016).
“Saudi Arabia … enjoys very low production costs. And we are more efficient than other producers. It is an advantage we will use, as any producer would…”- Saudi Arabia’s Oil Minister at the time, Mr Al-Naimi (2015: www.saudiembassy.net/announcement/announcement03041501.aspx).
Indonesia’s crude output amounts to about 0.7 mbd. To facilitate comparison in the figures and charts presented in this section, Indonesia is classified as being part of OPEC in all years. In the calibrations to be preseted in the next section, Indonesia is only part of OPEC in the predicted data for future years.
The breakevens refer to rig and drilling costs reaching $50, which likely exclude transportation as well as capital and other "fixed" costs. For wells that have already been completed The Economist (2016a) reports a decline in cash costs to below $20.
Verleger (2016) argues that financial market innovation has allowed allow disruptive smaller producers to withstand low prices.
Recall that, in our setup, this group of producers facilitates or impedes the squeezing of the high-cost producer. With small modifications to the time horizon, some of these producers could be cast as the target of a potential squeeze.
See, respectively, “Delayed oil projects total nears $400bn” Financial Times, 14 January 2016 and “Oil price: ’Shot in the arm’ misses economic target”, Financial Times, 5 April 2016.
Non-OPEC capacity forecasts for the next 5 years have been reduced between IEA (2015) and IEA (2016), reflecting some scaling back of investment as well as the exclusion of Indonesia after it rejoined OPEC.
Surveys by Atkins and Jayazeri (2004) and Smith (2009) indicate a range of 0 to -0.11. Hamilton (2009) finds elasticities that are very close to zero, but some more recent studies have found higher demand responses. Kilian and Murphy (2014) have a preferred estimate of -0.27, and Mohaddes and Pesaran (2015) offer -0.21. Both of these are similar to the median among a time-varying range of elasticities in Baumeister and Peersman (2013), who themselves find elasticities have declined over time.
Specifically, we use their data for tight oil in the lower 48 US states. Similar levels or growth rates are attained using proxies based on individual states or for the main shale oil fields (Baffes et al, 2015).
Changes in OPEC capacity are only indirectly important for ensuring A1 and A2 hold.
The rise in (forecast) shale oil capacity can be seen as the latest in a sequence of persistent positive surprises and upward revisions to forecasts by the EIA. It can also be seen as OPEC having some lag in incorporating these revisions in its internal calculation of the tradeoffs.
This value is only illustrative and was chosen to bring price close to the average observed in 2015. Nonetheless, it is close to the mid-point of more recent cost estimates and would also imply a decline broadly in line with some claimed cost reductions since the start of the squeeze.
This choice of Kj = 2.8 is in line with proprietary estimates of values by shale oil field and with the median and range published by Arezki and Blanchard (2014).
