Working Paper Series: Much Ado About Nothing? The Shale Oil Revolution and The Global Supply Curve
Working Paper Series: Much Ado About Nothing? The Shale Oil Revolution and The Global Supply Curve
Working Paper Series: Much Ado About Nothing? The Shale Oil Revolution and The Global Supply Curve
Disclaimer: This paper should not be reported as representing the views of the European Central Bank
(ECB). The views expressed are those of the authors and do not necessarily reflect those of the ECB.
Abstract
We focus on the implications of the shale oil boom for the global supply of oil.
We begin with a stylized model with two producers, one facing low production costs
and one higher production costs but potentially lower adjustment costs, competing à
la Stackelberg. We find that the supply function is flatter for the high cost producer,
and that the supply function for shale oil producers becomes more responsive to
demand shocks when adjustment costs decline. On the empirical side, we apply an
instrumental variable approach using estimates of demand-driven oil price changes
derived from a standard structural VAR of the oil market. A main finding is that
global oil supply is rather vertical, practically all the time. Moreover, for the global
oil market as a whole, we do not find evidence of a major shift to a more price elastic
supply as a result of the shale oil boom.
Keywords: Oil supply, shale oil, oil shocks, structural VAR, instrumental vari-
ables, sign restrictions.
This paper focuses on the question of whether the world oil supply function has changed
in a fundamental way with the advent of the shale oil on the market. A change in the
elasticity of the oil supply function can have, in fact, potential consequences for oil prices
and broader macroeconomic consequences given oil’s still central role in the world economy.
Our contribution is twofold. On the conceptual side, we propose a very stylized model
for the oil market with two producers, one representing conventional oil producers and one
shale oil producers, we find that the supply function is flatter for the high cost producer
(shale). Further, we find that the supply function for shale oil producers becomes more
responsive to demand shocks when adjustment costs decline. On the other hand, more
efficient production by the shale oil producer actually reduces its production elasticity.
We find that the oil supply curve is typically relatively vertical. At the same time,
we find a flattening of the supply curve for the U.S., in particular for oil demand shocks
around 2010. Taking our model as the guide to analyze the empirical evidence, we find
that this is consistent with the idea that a fall in adjustment costs has been a major factor
for shale oil production, rather than an outright reduction in marginal production costs.
For the global oil market as a whole, however, we do not find evidence of a major shift
to a flatter supply curve around 2010. This may be explained by the fact, its impressive
growth notwithstanding, US shale oil production only represents a relatively small fraction
of world oil production.
Since the summer of 2014, the oil industry has been on the roller-coaster, with the oil
price falling to 30 dollars per barrel from above 100, before recovering to the level of 50
dollars per barrel since mid 2016, and increasing again to 80 dollars in mid 2018. In the
meanwhile, we have seen the OPEC first implicitly trying to fight the shale oil by dragging
the shale oil companies into unprofitable territories, then agreeing in cutting the production
of oil to sustain the price.
While the financial press and policy institutions are evoking the shale oil as a major
change in the oil market, and the reason behind the plunge in the oil price, there is surpris-
ingly still few studies which look at the shale phenomenon in the economic literature. Kilian
(2017b) aims at quantifying the impact of the U.S. shale oil boom on U.S. imports and ex-
ports of crude and refined oil. Further, he aims at estimating the effects of a change in the
balance trade of the U.S. on the Arab countries, and in particular on Saudi Arabia. Based
on a structural VAR model, he shows that the U.S. shale oil revolution is not the cause of
the 2014/15 decline in the Brent oil price. He furthermore quantifies the cumulative effects
of the decline in the Brent price caused by the shale oil revolution on the foreign exchange
revenue of Saudi Arabia. Kilian (2016) examines how the increased availability of shale oil
has shaped the U.S. oil and gasoline prices. A broader review on how the shale oil affected
the U.S. economy is provided by Kilian (2017a). Mohaddes and Pesaran (2016) take a
different methodological perspective and look at the effects of lower oil prices to the global
economy, with the use a multi-country GVAR model. With a similar GVAR approach,
Mohaddes and Raissi (2015) investigate more specifically the macroeconomic consequences
of the U.S. oil revolution for the global economy in general and the Middle East and North
Africa in particular. Kilian (2009) was the first paper to attribute the sluggishness of the
supply response to the costs of adjusting oil production and the uncertainty about the
state of the crude oil market (p. 1059).This observation was formalized by Anderson et al.
(2018) who showed that in equilibrium the price elasticity of oil supply should be zero,
given the high costs of shutting down and reopening conventional oil wells. In this case,
the optimal response of oil producers to an oil price change induced by oil demand shifts is
to adjust investment in future oil production rather than the level of oil production from
existing wells. This theoretical prediction is supported by extraneous micro evidence. Al-
though there are no microeconomic estimates of the global one-month price elasticity of oil
supply, recent microeconomic estimates based on regional data from the United States are
The goal of our paper is to understand whether there has been a change in the oil
market, induced by the shale technology. In particular, we want to check whether with
the advent of the shale oil, the oil supply has become more responsive1 . In other words,
while a few studies in the literature typically found oil production to be inelastic in the
short run, we wonder whether this is still true, especially after the shale oil boom. If shale
oil production is more elastic than conventional oil production, this can have potential
consequences for the evolution of the price of oil.
To conduct our analysis, we start with a stylized model for the oil market, in which we
have two producers, one facing low costs but with very high adjustment costs (represent-
ing conventional oil producers), and one facing a higher cost, but lower adjustment costs
(representing shale oil producers). The two players compete à la Stackelberg: they produce
a homogeneous good, they do not collude, but only Saudi Arabia has market power and
internalises the demand constraint in its choice. They compete in quantities and choose
them simultaneously. Further, they act rationally and strategically, aiming at maximizing
their own profits. This is in keeping with most of the literature, which considers typically a
dominant firm with a competitive fringe model (see, for example, Nakov and Nuño (2013)
and Golombek et al. (2018)).
Based on simple but realistic calibrations of our model, we find that the supply function
is flatter for the high cost producer. This reflects the fact that a positive demand shock
pushes production in the area that makes it profitable for the high cost producer, whereas
the low cost producer finds it profitable even when demand is scarce. Further, we find that
the supply function for shale oil producers becomes more responsive to demand shocks
when adjustment costs decline, while we find the opposite for the case of a lower marginal
production cost. The intuition for the latter result is that the decline in production cost
1
For simplicity, we will talk throughout the paper about oil supply curve. More precisely, we refer to
the short-run oil supply curve conditional on past data. The estimated model is shown in equation (7).
We introduce a model of the oil market which features two producers, one facing lower
production costs (henceforth Saudi Arabia, or SA), and one facing a higher cost, but
potentially lower adjustment costs (henceforth Shale, or SH). The model is in real terms,
i.e. all the variables are real.
Our intent is to construct a small model to highlight mainly the role of marginal and
adjustment shocks in the optimal supply of oil and guide the empirical analysis. It is
therefore far less rich than the model introduced by Nakov and Nuño (2013) to describe
the behaviour of Saudi Arabia as a dominant producer with competitive fringe. Further,
that model is constructed into a dynamic stochastic general equilibrium setting, while ours
is a simplified, partial equilibrium one. Similar to Nakov and Nuño (2013) and other
papers in the literature we choose to model the market with only one strategic player,
Saudi Arabia.2
In addition, note that we do not consider the fact that oil is an exhaustible resource
(Hotelling rule) as most of the existing literature considers this dimension not to be relevant
for producers decisions (see, e.g., (Anderson et al. (2018) and Cairns and Calfucura (2012)).
Likewise, we do not consider the question of how OPEC coordinates its actions and how
it behaves like a cartel on not; on that question see for example Almoguera et al. (2011)
and Alkhathlan et al. (2014). We consider one block of low cost producers as acting as one
in the context of Cournot competition. Unlike Behar and Ritz (2016) we do not focus on
the OPEC or Saudi decision to accommodate or squeeze out shale oil out of the market.
Moreover, our modelling strategy is different, and notably the marginal cost of production
is an endogenous variable in our setting, while it is a parameter in Behar and Ritz (2016).
2
A previous version of this paper has a Cournot model and results are similar. For a comparison between
the Cournot and Stackelberg models, see Anderson and Engers (1992). While in a Stackelberg model firms
choose output sequentially, in the Cournot model they choose it simultaneously, without knowing the
competitors’ choice.
Pt = a − dQt + t , (1)
where Qt is the total quantity of oil produced in the market, and Pt is its price.
The two producers produce a homogeneous good, do not collude, but only Saudi Arabia
has market power (i.e., internalises the demand constraint in its choices and does not
take the oil price as given). They compete in quantities and choose them simultaneously.
Further, they act rationally and strategically, namely they want to maximize their profits.
ηi γi
Ci,t = αi Qδi,ti Pi,t + (Qi,t − Qi,t−1 )2 (2)
2
where i = SA, SH. Costs follow a power function and adjustment costs to production
are assumed to be quadratic. The term Ptη i captures the fact that production costs may
be a function of the oil price, which is particularly plausible for Shale producers due to the
cost of rigs.3 In the calibration, we set ηi > 0 for Shale but ηi = 0 for Saudi Arabia.
Further, αSA < αSH and δSA < δSH but γSA = γSH . Hence, we assume that Saudi
Arabia has lower variable and fixed production costs and, in the baseline, the same adjust-
ment costs as Shale. In the calibration, we also consider a version where Shale has lower
adjustment costs than Saudi Arabia, and also where it has lower marginal production costs
than in the baseline, while still higher than Saudi Arabia. In both cases, oil production hits
diminishing marginal returns (that is, there are increasing production costs); this implies
that δi > 1.4
3
We thank Axel Pierru for this suggestion.
4
Note that we are assuming that the adjustment costs are symmetric, which may not be the case in
reality (in particular, reducing production may be easier than increasing it). We are working at an extension
which allows for asymmetry.
Each producer chooses the quantity to produce Qi to maximize its profits. That is, each
of the two producers solve the following maximization problem:
∞
X ηi γi
max β t {Pt Qi,t − αi Qδi,ti Pi,t − (Qi,t − Qi,t−1 )2 } (3)
t=0
2
with i = SA, SH, subject to the demand function in equation (1) and the market
clearing condition:
Qt = QSA,t + QSH,t . (4)
In the calibration of the slope of the demand function we follow Behar and Ritz (2016)
and impose a slope parameter of -8.5 Production is expressed in million barrels per day.
Our cost calibration, when setting adjustment costs at zero, ensures that the low cost
player (Saudi Arabia) has a marginal production cost of 15 in USD per barrel for a produc-
tion of 10 million barrels per day, while the high cost producer (Shale) has a marginal cost
of 50 (again USD per barrel) for a production of 5 million barrels per day. Further, ad-
justment costs are set at 15 for both Saudi Arabia and Shale; this value gives a reasonable
degree of persistence in the impulse responses.
Given the parameters chosen for the calibration, we obtain a steady state of the model
with Saudi Arabia of 7.7 million barrels per day, a Shale production of 3.86 million, and
an oil price of 76.9 dollars per barrel. While these do not literally correspond to real world
5
Observe that they, unlike us, define the demand function as Q = f (P ).
values (although, at the time of writing, the oil price is around 78 dollars per barrel), they
are at least in the ballpark of plausible values.
We also propose an alternative calibration, in which we assume that the marginal cost
of Shale declines to 40 USD per barrel for the same production of 5 million barrels a day.
All the other parameters remain as in the baseline. In a second alternative, we consider
a change in the adjustment costs, which decrease for the shale oil producers from 15 to 5.
The calibration of our baseline model and of our two alternatives is summarized in Table
1.
In Figure 1 we plot the supply reaction of Saudi Arabia (left panel) and of Shale (right
panel) to a demand shock in our baseline calibration (red solid line), when the adjustment
cost for shale oil decreases (turquoise dotted line), and when marginal costs for Shale
decrease (blue dashed line).
In general, we find that the supply function is (as expected) positively sloped for both
producers and flatter for the high cost producer. This reflects the fact that a positive
demand shock pushes production in the area that makes it profitable for the high cost
producer, whereas the low cost producer finds it profitable to produce even when demand
is scarce. With this general principle in mind, we find that the supply function for the high
cost producer (Shale) becomes flatter (more responsive to demand shocks) when adjustment
costs decline. For the case of a lower marginal production cost we find the opposite to be
the case, namely that the high cost producer becomes less responsive (the supply curve
steepens); see Figure 1. The intuition for the latter result is that the decline in production
Note: We report impulse responses after an oil demand shock increasing the oil price by 1%. The red solid
line refers to the baseline calibration, the blue dashed line to a calibration with lower marginal costs for
Shale (α2 = 40, δ2 = 1.2) and the green dotted line to an alternative calibration with lower adjustment
costs for the shale oil producers (γ2 = 5). Data are in percentage points. The calibrated parameters are
summarized in Table 1.
cost makes the high cost producer more similar to the low cost producer (Shale closer to
Saudi), and hence steepens the supply curve.
In Table 2 we report the impact elasticity from the Cournot competition model, for
Saudi, Shale and total (world) production. We find, for example, that a demand shocks
leading to an increase in the oil price by 1% leads to a rise in world oil production by
approximately 0.02%, i.e. the supply elasticity is 0.02. The elasticity is larger for Shale
and lower for Saudi and is of course different in the alternative calibrations.
All in all, our simple model suggests two testable implications for the Shale (high cost)
producer:
1. More flexible production, as captured by lower adjustment costs in the model, should
lead to a steeper supply curve, i.e. to a larger reaction of supply to demand-induced
price increases;
2. More efficient production, as measured by less decreasing returns to scale in oil pro-
duction, leads to a less steep supply curve.
For the Saudi (low cost) producer, the implications are the opposite to those of Shale,
but more muted, at least in the calibration that we use.
In this section we introduce our data, which are monthly and cover the period from
1985 to 2018. Figure 2 offers an overview of all the variables that we will be using in this
paper. A summary on the data sources are provided in Table 3.
Note: Data for the Brent oil price are in USD, oil production data are in million barrels per day. See Table
3 on the sources of the data.
A standard model in the literature to analyze the global oil market is the one proposed
by Kilian and Murphy (2014), which in a sign-restricted VAR framework, allows to identify
three types of shocks in the global oil market: shocks to the storage demand for oil, to flow
demand and to supply.
Therefore, following Kilian and Murphy (2014) we estimate a monthly model, and derive
estimates of oil demand shocks. The model is a VAR estimated on monthly data, including
a vector z with the percent change in crude oil production, the Kilian index of real economic
activity (see Kilian (2009) for a description of the index), the real price of oil in log-levels
(Brent price deflated with the US CPI) and the change of global crude oil inventories above
the ground. The representation of the structural VAR model is the following:
24
X
A0 zt = α + Ai zt−i + t . (6)
i=1
where t is a white noise process with mean zero and covariance matrix Σt . The identifica-
tion of the oil shocks is achieved by imposing the sign restrictions as in Kilian and Murphy
(2014) (see Table 4).
Further, as in Kilian and Murphy (2014) we impose a bound on the impact price elastic-
ity of oil demand. In particular, we impose that the impact elasticity of oil demand must be
weakly negative on average over the sample, included between -0.8 and 0.6 We also impose
the additional restriction that the response of the real price of oil to a negative flow supply
shock must be positive not only on impact, but on the first 12 periods (to be consistent
with the conventional views on unanticipated oil supply disruptions). However, given that
the goal of our paper is exactly to look at potential changes in the supply elasticity, we
impose only a loose bound on the impact price elasticity of oil supply (corresponding to
the loosest one in the Kilian and Murphy (2014)), equal to 0.1.
Further, in order to take into account potential time-variation in the oil supply elasticity
around the time of the shale oil revolution, we cannot estimate our SVAR on the full
sample, which would imply no change in the structure of the economy. We therefore allow
for one-time break in the data in 2010, the year when US shale oil production started
to rise markedly. In practice, we estimate our model first on the sample 1985-2009, and
then on the sample 2010-2018. Given that the second sample is short, we deviate from
the estimation method implemented by Kilian and Murphy (2014) and we use a Bayesian
method. A relatively tight Minnesota prior allows us in this case to maintain unchanged
the number of lags in the estimation (equal to 24 months). In fact, this type of prior implies
that the further the lag, the more confident we should be that coefficients linked to this
lag have a value of zero. This means that the variance of the lagged parameters should be
smaller on further lags. The same is true for coefficients relating variables to past values of
other variables. The variance (i.e. the tightness) of these parameters is driven by the choice
of only 4 hyper-parameters. We optimize the hyper-parameters of the prior by using a grid
search, and retain the combination of hyper-parameters that maximizes the value of the
marginal likelihood for the model. In practice, we estimate the sign-restricted VAR with
standard Bayesian algorithms (implemented in the ECB BEAR toolbox), by constructing a
6
These bounds are standard in the literature, see the discussion in Kilian and Murphy (2014) and their
references to Sweeney (1984) and Hausman and Newey (1995).
We estimate the oil supply curve using local projections as in Jorda (2005). The em-
pirical model is the following:
where ∆P rodi,t+h is the change in oil production for country i between t and t + h, h =
1, ...12, Oilp is the log oil price in USD deflated with the US CPI and the parameter βh
indicates the slope of the curve. We consider h > 0 in order to allow for lags in producers’
reaction to new information (consistent with, e.g., Anderson et al. (2018). One would
normally expect this parameter to be positive in a supply function, as it is more convenient
to produce more when the price is high. In order to study whether the boom in the shale
oil production has influenced oil supply by other producers and globally, we estimate the
supply equation in the 1985-2009 and 2010-2018 separately, and compare them.
The estimation of this equation by OLS may lead to inconsistent results, because supply
shocks to oil production (say, a disruption in supply due to a natural disaster or political
event) may well influence the oil price, creating a simultaneity bias. For this reason, we
instrument ∆Oilpt with the two demand shocks, [b StorageDemand
Ft lowDemand , bt ].
5 Results
Before describing the results in detail, it is useful to provide an overview of the main
findings. First, we look at the first stage regressions, and conclude that our choice of
instruments appears appropriate as our estimated demand shocks are strong instruments
and affect oil prices with the expected sign. Second, we focus on the full sample and
confirm previous results in the literature that the supply curve in the short run is relatively
vertical, with short term elasticity between 0.02 and 0.04. Finally, has the oil supply curve
has changed over time? We find that the shale oil boom around 2010 does not appear to
have fundamentally affected the slope of the global oil supply curve, which remains rather
vertical. Moreover, the boom appears to have increased the medium term responsiveness
of US production to demand-driven changes in oil prices, which in our theoretical model is
consistent with lower adjustment costs but not with lower marginal costs of production.
We begin by reporting the results of the first stage regressions in Table 5, where we
regress the growth rate of the oil price in USD on the estimated aggregate demand and
precautionary demand shocks (estimates from the Bayesian version), in the 1985-2009 and
2010-2018 sample periods. In both samples the instruments are correctly signed and strong,
as suggested by the F statistic much above the typical benchmark of 10.
Armed with the instruments, we estimate equation (7) and derive impulse responses
from the βh coefficients. In Figure 3 we compare the estimates based on instruments from
7
Of course, the strategy of using extraneous information for estimating supply or demand elasticity is
not new in the literature; see, e.g., Kilian and Lee (2014).
Table 5: OLS regression. Dependent variable: monthly change in the oil price in USD.
the original Kilian-Murphy VAR model (black lines) and from the Bayesian version (blue
lines) on the pre-shale sample, 1985-2009. Two results are noteworthy.
First, the impact elasticity is positive but small, between .02 and 0.4, implying a rela-
tively vertical curve. This implies that a 10% increase in the oil price leads to a rise in oil
production by 0.2-0.4%. The effect is relatively persistent but loses statistical significance
after one month. Observe that this number is well within the range of the previous findings
in the literature. For example, it is well in line with the 0.025 estimated in Kilian and
Murphy (2012). Bjornland et al. (2017) find low and insignificant elasticity for conven-
tional production, although they find a high elasticity (between 0.6 and 1) for shale oil
wells. Anderson et al. (2018) find that production from existing wells in Texas does not
respond to price signals (implying a vertical short term supply curve), although drilling
activity does. Caldara et al. (2016) report larger elasticities between 0.05 and 0.1, as do
Bornstein et al. (2017), who however use annual data.8 Finally, it accords surprisingly well
with our stylised theoretical model, where the supply elasticity turned out to be in the
neighborhood of 0.02.
Second, it is noteworthy that the two lines are practically overlapping, signalling that,
at least for the purpose of this exercise, the full Kilian-Murphy and our more parsimonious
Bayesian version are largely interchangeable at least for the purpose of this analysis.
Overall, based on these two results, we then turn to compare the pre- and post-shale
supply curve estimates.
5.3 Did the shale oil boom bring a structural break in oil supply?
Figure 4 compare estimates of the oil supply for the world, Saudi Arabia, OPEC, Russia
and the United States in the pre-shale sample 1985-2009 (black lines) and in the post shale
8
They also find that OPEC production is less price responsive, which is consistent with our results.
The figure reports impulse responses derived from local projections estimates as in equation (7) es-
timated with instrumental variables on the sample 1985-2009, where ∆Oilpt is instrumented with
Demand
[bt Oildemand
,bt ] The black lines refers to the Kilian-Murphy original VAR model, the blue lines to the
Bayesian version with x lags. Dashed lines indicate confidence bands at the 90% level.
Seeing these results through the lenses of our simple theoretical model, we would reach
two main conclusions: (i) the shale oil boom does not appear to have fundamentally changed
the contours of global oil production, where the supply function remains relatively vertical;
(ii) tentative evidence, although not statistically significant, points to the oil supply curve
becoming more vertical in Saudi Arabia, and more price responsive in the US. This in turn
in consistent with the change being brought about by a decline in production adjustment
costs in Shale, and not by a fall in marginal production costs compared with Saudi.
6 Conclusions
With this paper, we contribute to the understanding of the oil supply function, and in
particular to whether it has changed with the advent of the shale oil on the market. A
change in the elasticity of the oil supply function can have, in fact, potential consequences
for oil prices and broader macroeconomic consequences given oil’s still central role in the
world economy.
With a very stylized model for the oil market with two producers, one representing
conventional oil producers and one shale oil producers, we find that the supply function
is steeper for the high cost producer (Shale). Further, we find that the supply function
for Shale producers becomes more responsive to demand shocks when adjustment costs
decline. These results find confirmation in our empirical analysis. Conducting a structural
VAR exercise, we find that the oil supply curve is difficult to estimate, generally stable
and not very responsive to fluctuations in oil demand - i.e., the curve is typically vertical.
9
Observe that our data on US production cover both conventional and shale oil.
The figure reports impulse responses derived from local projections estimates as in equation (7) es-
timated with instrumental variables on the sample 1985-2009, where ∆Oilpt is instrumented with
Demand
[bt Oildemand
,bt ] The black lines refers to the VAR model (Bayesian version) estimated on 1985-2009,
the blue lines to 2010-2018. Dashed lines indicate confidence bands at the 90% level.
The rise in shale oil production is a relatively recent phenomenon and it is still early to
make predictions about its effect on the global oil market. This paper should be seen as a
first step, but its conclusions should be revisited at a later point in time as technological
advances in fracking technology continue and further evidence accumulates.
Almoguera, P. A., Douglas, C. C., and Herrera, A. M. (2011). Testing for the cartel in
opec: non-cooperative collusion or just non-cooperative? Oxford Review of Economic
Policy, 27(1):144–168.
Anderson, S. T., Kellogg, R., and Salant, S. W. (2018). Hotelling under pressure. Journal
of Political Economy, 126(3):984–1026.
Bjornland, H. C., Nordvik, F. M., and Rohrer, M. (2017). Supply Flexibility in the Shale
Patch: Evidence from North Dakota. Working Papers No 2/2017, Centre for Applied
Macro- and Petroleum economics (CAMP), BI Norwegian Business School.
Bornstein, G., Krusell, P., and Rebelo, S. (2017). Lags, Costs, and Shocks: An Equilibrium
Model of the Oil Industry. NBER Working Papers 23423, National Bureau of Economic
Research, Inc.
Cairns, R. D. and Calfucura, E. (2012). OPEC: Market failure or power failure? Energy
Policy, 50(C):570–580.
Caldara, D., Cavallo, M., and Iacoviello, M. (2016). Oil Price Elasticities and Oil Price
Fluctuations. International Finance Discussion Papers 1173, Board of Governors of the
Federal Reserve System (U.S.).
Golombek, R., Irarrazabal, A. A., and Ma, L. (2018). Opec’s market power: An empirical
dominant firm model for the oil market. Energy Economics, 70:98 – 115.
Kilian, L. (2008). Exogenous Oil Supply Shocks: How Big Are They and How Much Do
They Matter for the U.S. Economy? The Review of Economics and Statistics, 90(2):216–
240.
Kilian, L. (2009). Not All Oil Price Shocks Are Alike: Disentangling Demand and Supply
Shocks in the Crude Oil Market. American Economic Review, 99(3):1053–69.
Kilian, L. (2016). The Impact of the Shale Oil Revolution on U.S. Oil and Gasoline Prices.
Review of Environmental Economics and Policy, 10(2):185–205.
Kilian, L. (2017a). How the Tight Oil Boom Has Changed Oil and Gasoline Markets.
Papeles de Energia, 3:79–102.
Kilian, L. (2017b). The Impact of the Fracking Boom on Arab Oil Producers.
Kilian, L. and Lee, T. K. (2014). Quantifying the speculative component in the real price
of oil: The role of global oil inventories. Journal of International Money and Finance,
42(C):71–87.
Kilian, L. and Murphy, D. (2012). Why agnostic sign restrictions are not enough: Un-
derstanding the dynamics of oil market var models. Journal of the European Economic
Association, 10(5):1166–1188.
Kilian, L. and Murphy, D. P. (2014). The Role Of Inventories And Speculative Trading In
The Global Market For Crude Oil. Journal of Applied Econometrics, 29(3):454–478.
Mohaddes, K. and Pesaran, M. H. (2016). Oil Prices and the Global Economy; Is It
Different This Time Around? IMF Working Papers 16/210, International Monetary
Fund.
Mohaddes, K. and Raissi, M. (2015). The U.S. Oil Supply Revolution and the Global
Economy. IMF Working Papers 15/259, International Monetary Fund.
Sweeney, J. L. (1984). The response of energy demand to higher prices: What have we
learned? The American Economic Review, 74(2):31–37.
Claudia Foroni
European Central Bank, Frankfurt am Main, Germany; email: claudia.foroni@ecb.europa.eu
Livio Stracca
European Central Bank, Frankfurt am Main, Germany; email: livio.stracca@ecb.europa.eu