Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Oil Prices and the Stock Market

Oil Prices and the Stock Market Abstract This paper develops a novel method for classifying oil price changes as supply or demand driven using information in asset prices. Motivated by a simple model, demand shocks are identified as returns to an index of oil producing firms which are orthogonal to unexpected changes in the VIX index, with supply shocks capturing the remaining variation in oil prices. Demand shocks are strongly positively correlated with market returns and economic output, whereas supply shocks have a strong negative correlation. The negative correlation of supply shocks and returns is strongest in industries that produce consumer goods, while the positive correlation of demand shocks is stronger for industries which use relatively large amounts of oil as an input. 1. Introduction Oil prices have long been cited as a leading economic indicator, with Hamilton (2003) and others documenting a strong negative relation between increases in oil prices and future economic growth. Given these findings, it is natural to examine the relations between oil prices and other traded assets, such as equities, to help better understand the link between oil prices and the economy. However, in doing this a puzzling fact emerges; oil price changes and stock market returns have very low correlation. For instance, from 1986 to 2011, a simple regression of monthly US market returns on contemporaneous changes in oil prices suggests essentially zero relation between the two variables. More simply put: Where Is Oil Price Beta?1 This paper attempts to address this puzzle by introducing a novel method of classifying changes in oil prices as demand driven or supply driven. The simple intuition behind the strategy is that oil producing firms are likely to benefit from increases in oil demand, but may have a natural hedge against shocks to oil supply. If oil prices increase due to higher demand, producers are able to sell more oil at the higher price and are likely to experience positive returns. In contrast, if prices rise because oil becomes more difficult to produce, the impact on the value of oil producing firms is less clear. They will sell less, but they will benefit from the higher prices. If producers’ equity values are relatively unaffected by these shocks, then producer stock returns can be used as a control variable to identify price changes coming from the two different sources. Following this intuition, demand shocks are defined as the portion of contemporaneous returns of a global index of oil producing firms which is orthogonal to unexpected changes in the log of the VIX index, which are included as a proxy for aggregate changes in market discount rates, potentially driven by changing attitudes toward risk. Supply shocks are constructed as the portion of contemporaneous oil price changes which is orthogonal to demand shocks as well as to innovations in the VIX. By construction, innovations to the VIX (risk shocks), demand shocks, and supply shocks, are orthogonal and account for all of the variation in oil prices. Since the VIX has very low correlation with oil prices over the sample period, nearly all of the variance is captured by the demand and supply shocks, with supply shocks accounting for 78% of the total variation and demand shocks another 21%. When the supply and demand shocks are examined separately, it appears that the lack of relation between oil prices and stock returns is an artifact of the conflicting effects of the two types of shocks. Instead of no relation, both supply and demand shocks are strongly correlated with aggregate stock returns over the sample period. Supply shocks have a strongly significant negative relation with stock returns and with future economic output, while demand shocks have a strong positive relation. The identified shocks also have differing effects across industries, shedding light on how oil shocks impact the economy. The negative relation of supply shocks is strongest for firms which produce consumer goods, suggesting that the main impact of an oil shock may be a reduction in consumer spending, consistent with the hypothesis of Hamilton (2003). In contrast, positive oil demand shocks coincide with high returns to manufacturing industries which use large amounts of oil as an input to production. The insight that price shocks from different sources will have different effects on the economy is not a new one, and has lead to several methods of classifying shocks as supply or demand driven. One common approach is to find instruments for oil price changes that are exogenous with respect to the rest of the economy, such as a time series of events affecting oil production. This strategy is pursued by several authors, including Hamilton (1983), Hamilton (2003), Kilian (2008), and Cavallo and Wu (2006). One shortcoming of this technique is that it requires a clear event to identify a supply shock, such as a war or hurricane which disrupts oil production. In contrast, how would one classify a month where oil prices rise slowly day by day, as the amount of oil produced fails to meet expectations? The resulting change in prices is as much a supply shock as a one time major disruption in production from a natural or political event, but is much more difficult to identify from news reports. Another literature identifies various shocks using data on oil production. For instance, Kilian (2009) identifies demand and supply shocks using a structural vector autoregression (SVAR) with data on oil production and shipping prices as proxies for supply and demand, and Kilian and Park (2009) extend this methodology to examining different shocks’ impact on the US stock market. However, they find very little contemporaneous explanatory power for stock returns (less than 2% combined), and this correlation is entirely driven by the residual changes in oil prices related to neither supply nor aggregate demand.2 One weakness of this framework is that the data included in the SVAR need to correlate with contemporaneous or future changes in oil prices to effectively identify shocks. For instance, the identified supply and demand shocks in Kilian (2009) explain only 4% of the contemporaneous variation in oil prices from 1986 to 2011. Of the remaining variation in oil price changes, 19% is classified as predictable by the SVAR, and the residual 77% is classified as “precautionary demand shocks”.3 Unfortunately, there is no way to ascertain if these changes in precautionary demand are driven by concerns over supply or expectations of changes in demand. For example, an increase in oil prices driven by an increase in the probability of a supply constraint which never materializes will not be identified in the VAR. Likewise, an increase in prices due to increases in demand which are not simultaneously reflected in increased shipping prices will also not be identified.4 Both of these changes will be classified as precautionary demand shocks, though presumably the two events would have different implications for aggregate stock returns and economic output. An identification technique which relies upon prices of traded assets can make use of the forward looking nature of prices to avoid these issues. To provide motivation for the classification strategy used here, a model of a competitive commodity producing sector is introduced, similar to the models of Carlson, Khokher, and Titman (2007) and Casassus, Collin-Dufresne, and Routledge (2005). The model illustrates that certain characteristics of oil production, namely the depletable nature of oil resources, the highly inelastic demand, and the significant difficulties in developing new reserves, give producers a natural hedge against shocks to the aggregate productivity of the sector. This in turn will yield producer stock returns that are unresponsive to changes in this productivity, and therefore can provide an effective control to identify supply shocks. The model is highly stylized to provide clear intuition for the identification strategy. Empirical evidence is therefore provided that the constructed variables are indeed effective proxies for supply and demand shocks. Regressions of changes in economic aggregates and variables related to oil consumption and production show that the identified shocks impact these variables in ways consistent with supply and demand shocks. Price increases from supply shocks are strongly negatively correlated with oil production and consumption, and have negative impacts on aggregate economic variables. In contrast, oil price increases from demand shocks are associated with increased economic output and steady or slightly increasing levels of oil production and consumption. The rest of the paper is organized as follows: Section 2 introduces a model of competitive oil producers and discusses the shock identification strategy in the context of the model. Section 3 empirically implements the identification strategy, presents the relations between the shocks and the aggregate stock market, and presents empirical support for the validity of the identification technique. Section 4 presents detailed results on the relations between the two constructed shocks and industry stock returns. Section 5 concludes. 2. A Simple Model of Oil Production The model introduced here is a model of atomistic competitive firms which take the price as given and choose both investment in oil reserves (with very high adjustment costs) and the level of a flow input (with low adjustment costs).5 The model is used to examine the potential efficacy of using stock returns to oil producing firms to identify different sources of oil price variation, with the caveat that the effectiveness of the strategy will be somewhat dependent on the parameters. The model is meant as a demonstration of why such a strategy might work, with the primary burden falling on the empirical analysis to validate the technique. The model views oil as a depletable resource as in Carlson, Khokher, and Titman (2007) and Ghoddusi (2010). The first important extension here is the inclusion of exogenous supply shocks, in addition to the standard demand shocks, so that the relative impacts on prices and producer stock returns can be examined. The supply shocks are modeled as increases to the flow of oil produced by a given level of flow input and oil wells. These shocks both increase current production as well as the speed with which oil wells are depleted. The model also deviates from existing work in allowing for exogenous shocks to the expected rate of return on oil producing firms, which in the model are generated by exogenous shocks to the price of risk associated with aggregate demand shocks. These shocks impact oil producer returns without creating a large impact on oil prices, and therefore need to be controlled for in order to effectively identify demand and supply shocks. 2.1 Consumers Following Kogan, Livdan, and Yaron (2009), consumers of oil are represented by an inverse demand curve, so that spot prices Pt are given by   Pt=At(Ot)−1α. (1) The price is dependent upon At, representing the aggregate level of oil demand in the economy, Ot, the total production of oil, and α, the elasticity of demand. 2.2 Firms The model consists of a continuum of competitive oil producing firms with Cobb–Douglas production technology.   Ot=ZtFtνWt1−ν. (2) Oil wells Wt, and a flow input Ft, are used to produce oil output Ot. The level of productivity is also affected by an oil industry production shock, Zt. For simplicity there are no firm-specific shocks, as well as no entry and exit, as in Kogan, Livdan, and Yaron (2009). In the context of the model, Wt represents developed oil reserves in the ground, and allows the model to capture the storable nature of oil. The costs of increasing production in a given period are not only the direct costs to the producer of a higher level of the input Ft, but also the reduction of oil reserves available to produce in future periods. This cost is reflected in the evolution of oil reserves   Wt+1=Wt−Ot+It. (3) The producer chooses investment in new reserves (It), which are depleted as oil is produced. Given an oil price, Pt, firms sell their output earning a profit Πt  Πt=PtOt−cFFt−It−Φ(Ft,Ft−1,It,Wt), (4) where Φ is a function representing costs to adjusting both the level of the flow input and the levels of investment in oil wells, and takes a quadratic form   Φ(Ft,Ft−1,It,Kt)=aF2(Ft−Ft−1Ft−1)2Ft−1+aW2(ItWt−I¯W¯)2Wt, (5) where I¯ and W¯ are the deterministic steady-state values of investment and oil well stock, and aF and aW govern the level of adjustment costs for the flow input and oil reserves, respectively. Note that Equations (3) and (4) imply that, at the deterministic steady-state investment amount, an additional oil well costs one unit of profit. The competitive producers take the price as given and choose It and Ft in each period to maximize firm value, calculated as the discounted expectation of future profits.   Vt=Πt+Et[∑s=1∞Mt+sMtΠt+s], (6) where Mt+s is the stochastic discount factor and is defined below. It is also assumed for simplicity that firms are purely equity financed, so that returns to an index of oil producing firms are given by   (1+RtProd)=Vt+ΠtVt−1. (7) 2.3 Dynamics From Equation (1), it is clear that there are two possible channels in the model for generating a change in the oil price. The first is a rise in the level of demand At, and the second is a reduction in the level of supply Ot. Although producers of final goods and household consumers are omitted for parsimony, simple intuition suggest that rises in the oil price from increases in At reflect positive economic news, while rises in price from a reduction in Ot generated by a decrease in productivity, Zt, would represent negative news for the aggregate stock market. Both aggregate oil demand and oil productivity are stochastic and their logs (indicated by lower case) evolve according to   at+1=a0+ρa(at−a0)+σaea,t+1, (8)  zt+1=z0+ρz(zt−z0)+σzez,t+1, (9) where ea,t+1 and ez,t+1 are independent normally distributed shocks with mean zero and a variance of one. High realizations of either ea,t+1 or ez,t+1 correspond to “good” times, and therefore both command positive prices of risk. To capture this the stochastic discount factor is given by   Mt+1Mt=β  exp(−λa,tea,t+1−λzez,t+1). (10) In order to capture market wide shocks to expected returns, λa,t is time varying and evolves according to   λa,t+1=λ¯a+ρλ(λa,t−λ¯a)+σλeλ,t+1. (11) 2.4 Model Results A competitive equilibrium is defined as a sequence of choices of It and Ft such that the firms are maximizing firm value while taking Pt as given, and the market clearing condition Pt=AtOt−1α is met. The producers’ first-order equation for the choice of the flow input, Ft is   cF=νOtFt(Pt−qt). (12) Here, qt is the Lagrangian multiplier associated with investment constraint on oil well accumulation, and represents the marginal value of an extra oil well in time t + 1. As this equation implies, producers will take into account that selling oil incurs a cost not only through the direct cost of the flow input, but also through the depletion of reserves. This effect is a standard feature of exhaustible resource models, dating back to Hotelling (1931). In contrast to models where reserves are fully exhaustible, here it is possible to invest in additional reserves, so prices in this model are stationary. Figure 1 shows impulse response functions to the three different types of shocks in the model. Given a negative shock to oil well productivity, the oil price rises as production falls, and producers respond by increasing the flow input to offset some of the fall in production. The increase in the flow variable is enough to prevent a rise in profitability, and hence value, for the producing firms, so this increase in oil prices is not accompanied by an increase in oil producer value. Figure 1 View largeDownload slide Model impulse response functions. Plots for model impulse responses to positive demand shocks (positive realization of ea,t), and negative productivity shocks (negative realization of ez,t), and increases in the price of risk associated with demand shocks (positive realization of eλ,t). Figure 1 View largeDownload slide Model impulse response functions. Plots for model impulse responses to positive demand shocks (positive realization of ea,t), and negative productivity shocks (negative realization of ez,t), and increases in the price of risk associated with demand shocks (positive realization of eλ,t). An increase in oil demand generates an increase in oil prices, an increase in oil production, and increased use of the oil flow input by producers. Since both oil prices and production increase, this leads to an increase in the value of oil producing firms. Finally, a shock to the price of risk associated with demand shocks makes oil firms riskier, and therefore increases the discount rate applied to the profit of oil producing firms, and leads to a negative stock return for oil producers. In contrast, this shock has a relatively small impact on the production decision of the firms, and therefore a small impact on oil prices. This differential impact makes discount rate shocks an important control in the context of the identification strategy. 2.5 Simulations The model is solved using parameters given in Table I. One important parameter for the identification technique is the elasticity of oil demand, which is significantly less than one in the data. The calibration uses a value of α=0.5, consistent with estimates in the literature (See Kogan, Livdan, and Yaron, 2009). The remaining parameters are calibrated to match volatilities and correlations of prices and returns. Table I Model parameters Parameter  Description  Value  ν  Share of flow input in production  0.6  α  Elasticity of demand  0.5  σa  Volatility of demand shock  0.15  σz  Volatility of productivity shock  0.065  σλ  Volatility of risk shock  0.25  ρa  Persistence of demand shock  0.95  ρz  Persistence of productivity shock  0.95  ρλ  Persistence of demand risk price  0.95  a0  Mean of demand shock  0  z0  Mean of productivity shock  0  cF  Cost of flow input  2.5  aF  Flow input adjustment cost  3  aW  Oil well adjustment cost  15  d  Depletion rate  1  δ  Deprecation of oil wells  0.01  λ¯a  Mean price of demand risk  2.0  λz  Price of supply risk  0.3  β  Discount rate  0.995  Parameter  Description  Value  ν  Share of flow input in production  0.6  α  Elasticity of demand  0.5  σa  Volatility of demand shock  0.15  σz  Volatility of productivity shock  0.065  σλ  Volatility of risk shock  0.25  ρa  Persistence of demand shock  0.95  ρz  Persistence of productivity shock  0.95  ρλ  Persistence of demand risk price  0.95  a0  Mean of demand shock  0  z0  Mean of productivity shock  0  cF  Cost of flow input  2.5  aF  Flow input adjustment cost  3  aW  Oil well adjustment cost  15  d  Depletion rate  1  δ  Deprecation of oil wells  0.01  λ¯a  Mean price of demand risk  2.0  λz  Price of supply risk  0.3  β  Discount rate  0.995  Table II reports volatilities and the correlation of oil producers’ stock returns and changes in oil prices for the simulated model. The table also reports the correlations between the simulated price innovations and returns and the unobservable supply and demand shocks. Oil producer stock returns are highly correlated with demand shocks while being nearly uncorrelated with supply shocks. If the shocks to risk-aversion can be controlled for, this makes producer returns an effective control for identifying the sources of price variation. Table II Model simulated prices and returns This table shows monthly correlations of simulated model shocks, oil prices, oil producer returns. The table also shows as the volatilities of oil prices and producer returns, along with the percentage of variance of these variables explained by the structural shocks in the model. Δpt are simulated changes in the log spot price of oil, Rtprod are simulated oil producer returns, ea,t are simulated oil demand shocks, ez,t are simulated producer productivity (supply) shocks, and, eλ,t are simulated shocks to the price of risk associated with demand shocks. The model is simulated monthly for 10,000 periods.   Observable variables     Correlation matrix    RtProd  Δpt  RtProd  1.00          Volatility  20.68  34.16  Δpt  0.28  1.00        % of var. from:      ez,t  0.01  −0.83  1.00      ez,t  0.0%  85.5%  ea,t  0.96  0.32  0.00  1.00    ea,t  92.4%  13.0%  eλ,t  −0.28  0.11  0.00  0.00  1.00  eλ,t  7.6%  1.5%    Observable variables     Correlation matrix    RtProd  Δpt  RtProd  1.00          Volatility  20.68  34.16  Δpt  0.28  1.00        % of var. from:      ez,t  0.01  −0.83  1.00      ez,t  0.0%  85.5%  ea,t  0.96  0.32  0.00  1.00    ea,t  92.4%  13.0%  eλ,t  −0.28  0.11  0.00  0.00  1.00  eλ,t  7.6%  1.5%  Although the model is meant to be stylized, it does provide some insight into how an empirical identification technique would work. A supply shock should be reflected higher oil prices, and lower oil output and consumption, while a positive demand shock should be reflected in higher prices, and potentially a rise in oil production and consumption, but not one as marked as the observed decrease from the supply shock. The differential exposure of oil producer returns to the two types of shocks also allows for identifying the sources of changes in price. The next section constructs these two shocks using changes in the VIX index as a proxy for shocks to risk aversion, and attempts to provide empirical evidence that the classified shocks are indeed related to oil supply and demand. 3. Classification of Oil Shocks 3.1 Data and Shock Construction The three variables necessary to construct the series of supply and demand shocks are an index of oil producing firms, a measure of oil price changes, and a proxy for changes in expected returns. To cover as much of the global oil industry as possible, the index used is the World Integrated Oil and Gas Producer Index available from Datastream. This index covers the large publicly traded oil producing firms (Exxon, Chevron, BP, Petrobras, etc.), but does not include nationalized oil producers such as Saudi Aramco. For changes in the oil price, the 1-month returns on the second nearest maturity NYMEX Crude—Light Sweet Oil contract are used for the bulk of the analysis, in order to focus on unexpected changes to oil prices. Aggregate US stock market data are from the universe of CRSP stocks.6 To proxy for changes in the discount rate, innovations to the VIX index are used. The VIX is calculated from options data, and therefore provides a measure of the risk-neutral expectation of volatility. As shown in Bollerslev, Tauchen, and Zhou (2009), the variance risk premium captured in the VIX is both strongly negatively correlated with stock returns and positively predict stock returns in the time series, suggesting that it may be a reasonable proxy for changes in risk. The VIX is published by the CBOE back to 1986, and this limits the data for the main portion of the analysis. In order to isolate unexpected changes in the VIX, an ARMA(1,1) process is estimated and the residuals from this process are used as innovations. These innovations are denoted ξVIX. Panel A of Table III provides summary statistics of standard deviations and correlations of the four variables for the sample period. This correlation matrix is at the heart of the strategy pursued in this paper. The high correlation between oil producer returns and both oil prices and aggregate market returns, along with a very low correlation between oil prices and aggregate market returns, suggests that there may be some source of variation which loads negatively on aggregate stock returns but positively on oil prices and is uncorrelated with producers. Additionally, changes in the VIX have a correlation with both aggregate and oil producer stock returns, but a minimal correlation with oil prices. Table III Summary of data and identified shocks Panel A shows annualized means, standard deviations, and monthly correlations. Oil price changes are the returns to the NYMEX second nearest maturity future. Oil producer returns are from the Datastream World Integrated Oil and Gas Producer Index. US Market return is the aggregate CRSP return. ξVIX is the residual value from an estimated ARMA(1,1) for the log of the VIX index. Data are monthly and returns are calculated in logs. Panel B shows decomposition of oil price changes into component shocks using the method described in Section 3.1. By construction the shocks account for all variation in oil prices. Data are monthly with annualized standard deviations reported. Sample is January 1986 to December 2011. Panel A. Summary of data  Variable  Description  Mean  Standard deviation  Correlation matrix    RtUSA  US market return  0.04  0.16  1        RtProd  Oil Producer Index Returns  0.05  0.18  0.62  1      ξVIX,t  Innovations to VIX  0.00  0.60  −0.67  −0.45  1    Δpt  Changes in oil prices  0.06  0.35  0.07  0.45  −0.11  1    Panel B. Identified oil shocks  Variable  Description  Standard deviation  % of Var.  Correlation matrix    Δpt  Changes in oil prices  0.34    1        st  Supply shocks  0.30  0.78  0.88  1      dt  Demand shocks  0.15  0.21  0.45  0  1    vt  Risk shocks  0.04  0.01  0.11  0  0  1  Panel A. Summary of data  Variable  Description  Mean  Standard deviation  Correlation matrix    RtUSA  US market return  0.04  0.16  1        RtProd  Oil Producer Index Returns  0.05  0.18  0.62  1      ξVIX,t  Innovations to VIX  0.00  0.60  −0.67  −0.45  1    Δpt  Changes in oil prices  0.06  0.35  0.07  0.45  −0.11  1    Panel B. Identified oil shocks  Variable  Description  Standard deviation  % of Var.  Correlation matrix    Δpt  Changes in oil prices  0.34    1        st  Supply shocks  0.30  0.78  0.88  1      dt  Demand shocks  0.15  0.21  0.45  0  1    vt  Risk shocks  0.04  0.01  0.11  0  0  1  For the primary analysis the supply shocks st, demand shocks dt, and risk shocks vt are orthogonal and defined in the following manner. Define   Xt≡[ΔptRtProdξVIX,t],Zt≡[stdtvt],A≡[1110a22a2300a33]. The matrix A maps the identified shocks into the observable variables so that   Xt=AZt. (13) To impose orthogonality a22,a23,a33 and σs, σd, σv satisfy   A−1ΣX(A−1)T=[σs2000σd2000σv2], (14) where ΣX is the covariance matrix of the observable Xt, and σs, σd, and σv are the volatilities of the identified shocks. Note that this is simply a renormalization of the standard orthogonalization used to define the structural shocks in an SVAR setting. Here, rather than normalize the volatility of the shocks to one, the shocks are constrained to sum up to the total change in the oil price. Panel B of Table III shows the summary statistics for the constructed shocks. The annualized volatility of changes in oil prices over the sample period is roughly 33.8% while the annualized volatility of the supply and demand shocks is 30.5% and 15.6%, respectively. Equivalently, roughly 78% of the variance in oil prices is classified as supply shocks and 21% as demand shocks, with the shocks to the VIX explaining only 1% of the total variance in oil price shocks. 3.2 Oil Shocks and Aggregate Stock Returns Once the shocks are constructed, it is a simple matter to use them in a basic regression of aggregate stock market returns. Since the risk shocks vt are simply a constant multiple of ξVIX,t, ξVIX,t is used in the regression in order to provide a more straightforward interpretation of the economic magnitude of the regression coefficients. Table IV reports regressions of the form   RtMkt=α+βddt+βsst+βVIXξVIX,t (15) as well as the R2 from univariate regressions of the market return on each shock, for both the aggregate CRSP stock index and a world wide stock index from Global Financial Data. On their own, oil prices have little or no relation to aggregate market returns. However, when they are decomposed into two series, which together explain nearly all of the variation in oil prices, oil prices have a very clear and intuitive link to the stock market. High oil prices from supply shocks are bad news for aggregate stock returns, and can explain 3.6% of the monthly variation in the aggregate US market return, while rises in prices from demand shocks are associated with positive excess returns and can explain an additional 12.4% of the variation. Table IV Aggregate stock market returns and oil price shocks Panel A shows regressions of Aggregate US Stock Market Returns (return to the index of all CRSP stocks) on changes in oil prices and on constructed demand and supply shocks. Shocks are defined as in Section 3.1. Panel B shows regressions of world aggregate stock market index on constructed shocks. World index is from Global Financial Data. White (1980) standard errors in parentheses. Data are monthly from 1986 to 2011 and market returns are in logs. Panel A. US stock market returns and oil shocks  Description  Variable  US market ret. ( RtUSA)        Oil price changes  Δpt  0.031      (0.027)  Univariate R2  Demand shock  dt    0.370**  0.124  (0.046)    Supply shock  st    −0.102**  0.036  (0.021)    Innovation in VIX  ξVIX,t    −0.184**  0.444  (0.012)    Constant    0.005  0.003    (0.003)  (0.002)  Observations    315  315    R-squared    0.004  0.604      Panel B. World stock market returns and oil shocks  Description  Variable  World Market Ret. ( RtWorld)        Oil Price Changes  Δpt  0.046      (0.026)  Univariate R2  Demand shock  dt    0.493**  0.231  (0.040)    Supply shock  st    −0.109**  0.043  (0.023)    Innovation in VIX  ξVIX,t    −0.158**  0.345  (0.011)    Constant    0.005  0.002    (0.003)  (0.002)  Observations    315  315    R-squared    0.010  0.619    Panel A. US stock market returns and oil shocks  Description  Variable  US market ret. ( RtUSA)        Oil price changes  Δpt  0.031      (0.027)  Univariate R2  Demand shock  dt    0.370**  0.124  (0.046)    Supply shock  st    −0.102**  0.036  (0.021)    Innovation in VIX  ξVIX,t    −0.184**  0.444  (0.012)    Constant    0.005  0.003    (0.003)  (0.002)  Observations    315  315    R-squared    0.004  0.604      Panel B. World stock market returns and oil shocks  Description  Variable  World Market Ret. ( RtWorld)        Oil Price Changes  Δpt  0.046      (0.026)  Univariate R2  Demand shock  dt    0.493**  0.231  (0.040)    Supply shock  st    −0.109**  0.043  (0.023)    Innovation in VIX  ξVIX,t    −0.158**  0.345  (0.011)    Constant    0.005  0.002    (0.003)  (0.002)  Observations    315  315    R-squared    0.010  0.619    These results are robust to the exclusion of outliers, the largest of which occur during the 2008 Financial Crisis. Removing these observations considerably strengthens the impacts of supply shocks on aggregate returns. Moreover, when the same strategy is performed with copper and aluminum producers and prices as a placebo test, the supply shocks identified using these two, less essential, commodities have no impact on aggregate stock returns, suggesting that the documented effects reflect the impacts of oil supply shocks and not a mechanical correlation.7 The next section focuses on presenting additional evidence in support of the efficacy of the identification strategy. 3.3 Evidence for the Validity of the Supply and Demand Classification It is clear that decomposing changes in oil prices in the manner described in the previous section leads to strong correlations with aggregate stock market returns. In order to provide support for the proposed interpretation that this decomposition allows for effective identification of supply and demand shocks, Section 3.3.a examines the behavior of producer returns and prices around a known supply shock, the first Gulf War, and finds them to be consistent with the story. Section 3.3.b presents univariate vector autoregressions showing that the two shocks relate to variables reflecting macroeconomic output and the oil supply in ways which are consistent with the supply and demand shock interpretation. 3.3.a. Oil producer returns and oil prices in the First Gulf War. This section examines the behavior of stock returns and prices around the First Gulf War. This period is chosen because it is the most severe oil shock in the sample. Although the war interrupted production of the major nationalized oil producers in the Middle East, it also impacted many of the large western oil firms in the world wide producer index which were active in the region at the time. Figure 2 shows the patterns of returns and prices over this time period. Figure 2 View largeDownload slide Oil producers stock prices, oil prices, and the Gulf War. Plots of cumulative returns for oil producers, the global oil producer index, the aggregate stock market, as well as cumulative changes in oil prices from June 1990 to July of 1991. Oil producer index is the Datastream World Integrated Oil and Gas Producer Index. US Stock Return is the value weighted CRSP index. Stock market returns are from CRSP. Figure 2 View largeDownload slide Oil producers stock prices, oil prices, and the Gulf War. Plots of cumulative returns for oil producers, the global oil producer index, the aggregate stock market, as well as cumulative changes in oil prices from June 1990 to July of 1991. Oil producer index is the Datastream World Integrated Oil and Gas Producer Index. US Stock Return is the value weighted CRSP index. Stock market returns are from CRSP. In the first panel, the stock prices of several multinational oil producing corporations are shown along with the spot price of oil during the first Gulf War. Companies active in the Middle East, such as Ashland, Inc. and Occidental Petroleum, suffered drops in value due to concerns about their ability to produce during the conflict. In contrast companies, such as British Petroleum, which were not operating in the region, saw their valuations rise with the higher price of crude. The second panel shows the behavior of the Datastream World Integrated Oil and Gas Producers index over this period, as well as the performance of the US stock market. The initial invasion of Kuwait saw a spike in oil prices, and a negative return for the aggregate stock market, but very little response in the returns of oil producers. Although some individual producers saw large returns, in aggregate it appears that oil producers as an industry enjoyed a natural hedge against the potential supply shock. At the same time, aggregate stock values fell, suggesting a potential relation between changes in oil prices and stock returns. 3.3.b. Macroeconomic variables and oil shocks. To provide further evidence that the identified shocks represent distinct sources of changes in oil prices, univariate vector autoregressions are estimated for variables related to macroeconomic output and oil markets, with the constructed supply and demand shocks entering as exogenous variables. Following Hamilton (2008), these VARs have the following form   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (16) where yt is the log of an economic variable of interest, and the supply and demand shocks st and dt, which are included contemporaneously as well as with N lags. For this portion of the analysis, a longer sample is used to provide more power with tests of quarterly data. The shocks are constructed in the same manner as described in Section 3.1, but the log change of the WTI is used in place of futures returns, and residuals from an ARMA(1,1) of aggregate US stock market volatility are used in place of ξVIX,t. The resulting supply and demand shocks have a correlation of 99% and 93%, respectively, with the shocks constructed using futures and the VIX. Recent work by Bekaert and Hoerova (2014) shows that the volatility portion of the VIX correlates most strongly with macroeconomic aggregates, suggesting that this control is appropriate for this analysis. Since US data for both aggregate output and oil use are readily available for the sample, five of the six variables chosen for the analysis are related to US economic output and oil consumption. US GDP and Total US Oil consumption are included to measure aggregate economic output and oil use. Total US Oil Consumption is US Oil Production plus Imports less Exports less Changes in Inventories, with all data taken from the EIA. Additionally, Ready (2016) shows that relative levels of household aggregate consumption and household gasoline consumption are very tightly related to the oil price. This motivates the choice of two household consumption variables, the household consumption of gasoline, given by “Gasoline and Other Energy Goods” from the BEA’s NIPA survey, and a Cobb–Douglas aggregate of durable and nondurable household consumption (net of household gasoline consumption), similar to the construction in Yogo (2005). Although not explicitly included in the model, inventories are also central to discussions of the oil price, so US petroleum stocks (net of the Strategic Petroleum Reserve) provided by the EIA are also included. The last variable is the level of international oil production again from the EIA.8 All data are aggregated to the quarterly level, and regressions are done using four lags. Figure 3 reports the impulse response functions of six variables to an oil supply shock and Figure 4 reports the same for a demand shock. For an increase in oil prices from a supply shock, there are significant decreases in all six variables, consistent with the hypothesis that this shock measures a reduction in available oil, leading to lower oil consumption, which in turn lowers aggregate economic activity. Conversely, for a demand shock, there are significant positive increases in economic output and total consumption, and an increase in aggregate economic oil use. There is no significant change in household oil consumption, and a delayed increase in world oil production, consistent with the model. Also, supply shocks result in a sharp decrease in inventories while demand shocks do not. The different responses of oil production and consumption are notable, since both shocks are associated with an increase in the oil price. Figure 3 View largeDownload slide Supply shock impulse response functions. Figure plots cumulative impulse response functions to a one standard deviation oil supply shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil supply shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (17) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 3 View largeDownload slide Supply shock impulse response functions. Figure plots cumulative impulse response functions to a one standard deviation oil supply shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil supply shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (17) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 4 View largeDownload slide Demand shock impulse response functions. Figure plots VAR cumulative impulse response functions to a one standard deviation oil demand shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil demand shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsΔst−n+βndΔdt−n), (18) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 4 View largeDownload slide Demand shock impulse response functions. Figure plots VAR cumulative impulse response functions to a one standard deviation oil demand shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil demand shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsΔst−n+βndΔdt−n), (18) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. 4. The Impact of Oil Shocks on Industry Returns In order to further illustrate the relation between oil prices and stock returns, and to gain more insight into the mechanisms by which oil price shocks impact the economy, this section explores the relation between the oil shocks and industry returns. The main takeaways are that consumer goods industries have the strong negative loadings on both oil supply shocks and demand shocks, while the industries with both high levels of oil use and high market betas have the strongest positive loadings on demand shocks. This provides support to the hypothesis of Hamilton (2003), who suggests that oil prices shocks act primarily on consumer expenditure rather than a direct effect from increased cost of inputs. Additionally, the fact that supply shock exposure is unrelated to market beta suggests that these shocks are a distinct source of return variation from aggregate market shocks. Table V shows the results for regressions of industry returns on supply and demand shocks over the pre-crisis sample for a partition of ten industry portfolios constructed by Fama and French (1997). Returns for these industries are available on Ken French’s website. The industries are sorted by their loading on the oil supply shock. What is immediately apparent is that nearly all of the industries do load negatively on the shock, but that three of the top four industries, Consumer Durables, Consumer Nondurables, and Retail, are highly dependent on consumer spending. Table V Fama-French ten industry portfolios and oil shocks Table reports univariate regressions of industry returns on supply shocks and univariate regressions of industry returns on demand shocks. Constants are suppressed. Industry portfolios are from Fama and French (1997) and are taken from Ken French’s website. White (1980) standard errors in parentheses. Data are monthly from 1986 to 2011. Description  Univariate regressions   Supply shock   Demand shock   Beta  R2  Beta  R2  Consumer durables  −0.235**  0.117  0.256**  0.036  (0.040)    (0.081)    Retail  −0.232**  0.154  0.123  0.011  (0.033)    (0.071)    Other  −0.220**  0.157  0.190**  0.031  (0.031)    (0.066)    Consumer nondurables  −0.205**  0.174  0.215**  0.050  (0.027)    (0.058)    Manufacturing  −0.188**  0.121  0.315**  0.089  (0.031)    (0.062)    Healthcare  −0.171**  0.096  0.161*  0.022  (0.032)    (0.065)    High Tech  −0.139**  0.026  0.199  0.014  (0.052)    (0.103)    Telecommunications  −0.134**  0.051  0.191**  0.027  (0.035)    (0.070)    Utility  −0.117**  0.066  0.337**  0.143  (0.027)    (0.051)    Energy  −0.017  0.001  0.862**  0.601  (0.035)    (0.043)    Description  Univariate regressions   Supply shock   Demand shock   Beta  R2  Beta  R2  Consumer durables  −0.235**  0.117  0.256**  0.036  (0.040)    (0.081)    Retail  −0.232**  0.154  0.123  0.011  (0.033)    (0.071)    Other  −0.220**  0.157  0.190**  0.031  (0.031)    (0.066)    Consumer nondurables  −0.205**  0.174  0.215**  0.050  (0.027)    (0.058)    Manufacturing  −0.188**  0.121  0.315**  0.089  (0.031)    (0.062)    Healthcare  −0.171**  0.096  0.161*  0.022  (0.032)    (0.065)    High Tech  −0.139**  0.026  0.199  0.014  (0.052)    (0.103)    Telecommunications  −0.134**  0.051  0.191**  0.027  (0.035)    (0.070)    Utility  −0.117**  0.066  0.337**  0.143  (0.027)    (0.051)    Energy  −0.017  0.001  0.862**  0.601  (0.035)    (0.043)    To further illustrate this, Figure 5 plots supply shock betas, this time for each of forty-nine Industry Portfolios again available from Ken French’s website. The industry “Other” in this partition is replaced by an Airline industry, constructed as the value-weighted return to SIC 4512 in the CRSP database, and the “Oil” industry is excluded.9 The top plot of the figure shows industry supply shock betas plotted against the industries’ market beta measured over the sample. The lack of relation here suggests that the identified shock, which negatively impacts nearly all industries, is a distinct source of variation from other aggregate market shocks. The second plot shows supply betas plotted against industry oil use, which is defined as the dollar of oil necessary to produce a dollar of industry output, constructed using input–output tables from the BEA. Again there is no significant pattern, suggesting that high oil intensity does not correlate with strong exposure to supply shocks. The last plot shows industry supply betas plotted against the percentage of the industry output used for personal consumption and residential investment, constructed using industry use tables again from the BEA. Here, we see a strong relation, with industries selling directly to consumers most impacted by supply shocks. Figure 5 View largeDownload slide Industry oil supply shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 5 View largeDownload slide Industry oil supply shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 6 repeats these plots but this time for oil demand betas. The figure shows that all industries load positively on the oil demand shock, but, in contrast with supply shocks, there is a positive significant (at the 10% level) relation between these loadings and industry market beta. There is also a significant relation between oil use and oil demand beta. This suggests that when manufacturing and oil intensive areas of the economy pick up in activity, the high demand translates to high oil prices. In the last plot, we also see a strong negative relation between oil demand beta and the percent of industry output used by households, suggesting that the high oil prices caused by increase demand are a drag on consumer industries, and that shocks to these industries do not seem to be driving oil demand.10 Figure 6 View largeDownload slide Industry oil demand shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 6 View largeDownload slide Industry oil demand shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. 5. Conclusion This paper presents a new, simple, method for identifying sources of oil supply shocks using changes to the VIX and oil producer stock returns. Using this method, both oil supply and demand shocks are shown to have a highly significant impact on US and world stock prices, in contrast to the very small correlations observed when using aggregate changes in oil prices. The impact of supply shocks is more significant for firms that depend on consumer expenditure rather than those which rely on oil as an input, while demand shocks are associated with high returns to oil using firms. These findings provide insight into the way oil prices affect the economy, and suggest that the important effect of oil price shocks may be pain at the pump for consumers rather than higher prices for oil using firms. The classification of the shocks provides a new technique for researchers studying the effects of changes in oil prices, particularly those interested in studying impacts at frequencies higher than quarterly data, or over short sample periods. The results may also be of interest to practitioners, since they directly imply that holding a portfolio of oil producing firms is not a hedge for harmful oil shocks. Supplementary Material Supplementary data are available at Review of Finance online. Footnotes 1 Other authors have examined this question and found similarly low contemporaneous relations. See for instance: Jones and Kaul (1996); Kilian and Park (2009); Chen, Roll, and Ross (1986); Huang, Masulis, and Stoll (1996); and Sadorsky (1999). In related work, Chiang et al. (2015) provide complementary evidence on the importance of various types of oil shocks for the cross-section of equity returns, and Gilje, Ready, and Roussanov (2015) provide evidence on the stock market impacts of increases in US shale oil production. 2 First line of Table I in Kilian and Park (2009). 3 See Internet Appendix for these results and the relation of the shocks identified in Kilian (2009) to those identified here. 4 See Kolodzeij and Kaufmann (2014) for a detailed discussion of the issues with the SVAR methodology in this context. 5 The Internet Appendix extends the model to include a price-leading monopolist along side a competitive fringe, and discusses the implications for the classification strategy in that setting. 6 See Internet Appendix for results using other sources of oil producer indices, oil prices, and risk levels, and the effects of controlling for aggregate levels of prices and exchange rates. There is no significant impact on the results. 7 See Internet Appendix. 8 International oil production is lagged 1 month as lagged production seems to correlate most strongly with prices and US quantities. This is possibly due to a delay in the availability of information. The US data are released weekly in the “Weekly Petroleum Status Report”, while international data are published with a 3-month lag in the “Monthly Energy Report”. 9 See Internet Appendix for regressions of individual industry portfolio returns on oil shocks. 10 Similar results hold for demand and supply shock impacts on international returns of various countries. See Internet Appendix. References Bekaert G., Hoerova M. ( 2014) The vix, the variance premium and stock market volatility, Journal of Econometrics  183, 181– 192. Google Scholar CrossRef Search ADS   Bollerslev T., Tauchen G., Zhou H. ( 2009) Expected stock returns and variance risk premia, Review of Financial Studies  22, 4463– 4492. Google Scholar CrossRef Search ADS   Carlson M., Khokher Z., Titman S. ( 2007) Equilibrium exhaustible resource price dynamics, Journal of Finance  62, 1663– 1703. Google Scholar CrossRef Search ADS   Casassus J., Collin-Dufresne P., Routledge B. R. ( 2005) Equilibrium commodity prices with irreversible investment and non-linear technology. Working paper series. Cavallo M., Wu T. ( 2006) Measuring oil–price shocks using market-based information. Technical report. Chen N.-F., Roll R., Ross S. A. ( 1986) Economic forces and the stock market, Journal of Business  59, 383– 403. Google Scholar CrossRef Search ADS   Chiang I., Ethan H., Keener Hughen W., Sagi J. S. ( 2015) Estimating oil risk factors using information from equity and derivatives markets, The Journal of Finance  70( 2), 769– 804. Google Scholar CrossRef Search ADS   Fama E. F., French K. R. ( 1997) Industry costs of equity, Journal of Financial Economics  43, 153– 193. Google Scholar CrossRef Search ADS   Ghoddusi H. ( 2010) Dynamic investment in extraction capacity of exhaustible resources, Scottish Journal of Political Economy  57, 359– 373. Google Scholar CrossRef Search ADS   Gilje E., Ready R., Roussanov N. ( 2015) Fracking, drilling, and asset pricing: estimating the economic benefits of the shale revolution, Working Paper. Hamilton J. D. ( 1983) Oil and the macroeconomy since World War II, Journal of Political Economy  91, 228– 248. Google Scholar CrossRef Search ADS   Hamilton J. D. ( 2003) What is an oil shock?, Journal of Econometrics  113, 363– 398. Google Scholar CrossRef Search ADS   Hamilton J. D. ( 2008) Understanding crude oil prices. NBER Working papers 14492, National Bureau of Economic Research, Inc. Hotelling H. ( 1931) The economics of exhaustible resources, The Journal of Political Economy  39, 137– 175. Google Scholar CrossRef Search ADS   Huang R., Masulis R. W., Stoll H. R. ( 1996) Energy shocks and financial markets, Journal of Futures Markets  16, 1– 27. Google Scholar CrossRef Search ADS   Jones C., Kaul G. ( 1996) Oil and the stock markets, Journal of Finance  51, 463– 491. Google Scholar CrossRef Search ADS   Kilian L. ( 2008) Exogenous oil supply shocks: how big are they and how much do they matter for the us economy?, The Review of Economics and Statistics  90, 216– 240. Google Scholar CrossRef Search ADS   Kilian L. ( 2009) Not all oil price shocks are alike: disentangling demand and supply shocks in the crude oil market, The American Economic Review  99, 1053– 1069. Google Scholar CrossRef Search ADS   Kilian L., Park C. ( 2009) The impact of oil price shocks on the U.S. stock market*, International Economic Review  50, 1267– 1287. Google Scholar CrossRef Search ADS   Kogan L., Livdan D., Yaron A. ( 2009) Oil futures prices in a production economy with investment constraints, The Journal of Finance  64, 1345– 1375. Google Scholar CrossRef Search ADS   Kolodzeij M., Kaufmann R. K. ( 2014) Oil demand shocks reconsidered: a cointegrated vector autoregression, Energy Economics  41, 33– 40. Google Scholar CrossRef Search ADS   Ready R. C. ( 2016) Oil consumption, economic growth, and oil futures: the impact of long-run oil supply uncertainty on asset prices. Working paper. Sadorsky P. ( 1999) Oil price shocks and stock market activity, Energy Economics  21, 449– 469. Google Scholar CrossRef Search ADS   White H. ( 1980) A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity, Econometrica  48, 817– 838. Google Scholar CrossRef Search ADS   Yogo M. ( 2005) A consumption-based explanation of expected stock returns, Journal of Finance  61, 539– 580. Google Scholar CrossRef Search ADS   © The Authors 2017. Published by Oxford University Press on behalf of the European Finance Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Review of Finance Oxford University Press

Oil Prices and the Stock Market

Review of Finance , Volume 22 (1) – Feb 1, 2018

Loading next page...
 
/lp/ou_press/oil-prices-and-the-stock-market-IvpmeRGDUH

References (3)

Publisher
Oxford University Press
Copyright
© The Authors 2017. Published by Oxford University Press on behalf of the European Finance Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
1572-3097
eISSN
1573-692X
DOI
10.1093/rof/rfw071
Publisher site
See Article on Publisher Site

Abstract

Abstract This paper develops a novel method for classifying oil price changes as supply or demand driven using information in asset prices. Motivated by a simple model, demand shocks are identified as returns to an index of oil producing firms which are orthogonal to unexpected changes in the VIX index, with supply shocks capturing the remaining variation in oil prices. Demand shocks are strongly positively correlated with market returns and economic output, whereas supply shocks have a strong negative correlation. The negative correlation of supply shocks and returns is strongest in industries that produce consumer goods, while the positive correlation of demand shocks is stronger for industries which use relatively large amounts of oil as an input. 1. Introduction Oil prices have long been cited as a leading economic indicator, with Hamilton (2003) and others documenting a strong negative relation between increases in oil prices and future economic growth. Given these findings, it is natural to examine the relations between oil prices and other traded assets, such as equities, to help better understand the link between oil prices and the economy. However, in doing this a puzzling fact emerges; oil price changes and stock market returns have very low correlation. For instance, from 1986 to 2011, a simple regression of monthly US market returns on contemporaneous changes in oil prices suggests essentially zero relation between the two variables. More simply put: Where Is Oil Price Beta?1 This paper attempts to address this puzzle by introducing a novel method of classifying changes in oil prices as demand driven or supply driven. The simple intuition behind the strategy is that oil producing firms are likely to benefit from increases in oil demand, but may have a natural hedge against shocks to oil supply. If oil prices increase due to higher demand, producers are able to sell more oil at the higher price and are likely to experience positive returns. In contrast, if prices rise because oil becomes more difficult to produce, the impact on the value of oil producing firms is less clear. They will sell less, but they will benefit from the higher prices. If producers’ equity values are relatively unaffected by these shocks, then producer stock returns can be used as a control variable to identify price changes coming from the two different sources. Following this intuition, demand shocks are defined as the portion of contemporaneous returns of a global index of oil producing firms which is orthogonal to unexpected changes in the log of the VIX index, which are included as a proxy for aggregate changes in market discount rates, potentially driven by changing attitudes toward risk. Supply shocks are constructed as the portion of contemporaneous oil price changes which is orthogonal to demand shocks as well as to innovations in the VIX. By construction, innovations to the VIX (risk shocks), demand shocks, and supply shocks, are orthogonal and account for all of the variation in oil prices. Since the VIX has very low correlation with oil prices over the sample period, nearly all of the variance is captured by the demand and supply shocks, with supply shocks accounting for 78% of the total variation and demand shocks another 21%. When the supply and demand shocks are examined separately, it appears that the lack of relation between oil prices and stock returns is an artifact of the conflicting effects of the two types of shocks. Instead of no relation, both supply and demand shocks are strongly correlated with aggregate stock returns over the sample period. Supply shocks have a strongly significant negative relation with stock returns and with future economic output, while demand shocks have a strong positive relation. The identified shocks also have differing effects across industries, shedding light on how oil shocks impact the economy. The negative relation of supply shocks is strongest for firms which produce consumer goods, suggesting that the main impact of an oil shock may be a reduction in consumer spending, consistent with the hypothesis of Hamilton (2003). In contrast, positive oil demand shocks coincide with high returns to manufacturing industries which use large amounts of oil as an input to production. The insight that price shocks from different sources will have different effects on the economy is not a new one, and has lead to several methods of classifying shocks as supply or demand driven. One common approach is to find instruments for oil price changes that are exogenous with respect to the rest of the economy, such as a time series of events affecting oil production. This strategy is pursued by several authors, including Hamilton (1983), Hamilton (2003), Kilian (2008), and Cavallo and Wu (2006). One shortcoming of this technique is that it requires a clear event to identify a supply shock, such as a war or hurricane which disrupts oil production. In contrast, how would one classify a month where oil prices rise slowly day by day, as the amount of oil produced fails to meet expectations? The resulting change in prices is as much a supply shock as a one time major disruption in production from a natural or political event, but is much more difficult to identify from news reports. Another literature identifies various shocks using data on oil production. For instance, Kilian (2009) identifies demand and supply shocks using a structural vector autoregression (SVAR) with data on oil production and shipping prices as proxies for supply and demand, and Kilian and Park (2009) extend this methodology to examining different shocks’ impact on the US stock market. However, they find very little contemporaneous explanatory power for stock returns (less than 2% combined), and this correlation is entirely driven by the residual changes in oil prices related to neither supply nor aggregate demand.2 One weakness of this framework is that the data included in the SVAR need to correlate with contemporaneous or future changes in oil prices to effectively identify shocks. For instance, the identified supply and demand shocks in Kilian (2009) explain only 4% of the contemporaneous variation in oil prices from 1986 to 2011. Of the remaining variation in oil price changes, 19% is classified as predictable by the SVAR, and the residual 77% is classified as “precautionary demand shocks”.3 Unfortunately, there is no way to ascertain if these changes in precautionary demand are driven by concerns over supply or expectations of changes in demand. For example, an increase in oil prices driven by an increase in the probability of a supply constraint which never materializes will not be identified in the VAR. Likewise, an increase in prices due to increases in demand which are not simultaneously reflected in increased shipping prices will also not be identified.4 Both of these changes will be classified as precautionary demand shocks, though presumably the two events would have different implications for aggregate stock returns and economic output. An identification technique which relies upon prices of traded assets can make use of the forward looking nature of prices to avoid these issues. To provide motivation for the classification strategy used here, a model of a competitive commodity producing sector is introduced, similar to the models of Carlson, Khokher, and Titman (2007) and Casassus, Collin-Dufresne, and Routledge (2005). The model illustrates that certain characteristics of oil production, namely the depletable nature of oil resources, the highly inelastic demand, and the significant difficulties in developing new reserves, give producers a natural hedge against shocks to the aggregate productivity of the sector. This in turn will yield producer stock returns that are unresponsive to changes in this productivity, and therefore can provide an effective control to identify supply shocks. The model is highly stylized to provide clear intuition for the identification strategy. Empirical evidence is therefore provided that the constructed variables are indeed effective proxies for supply and demand shocks. Regressions of changes in economic aggregates and variables related to oil consumption and production show that the identified shocks impact these variables in ways consistent with supply and demand shocks. Price increases from supply shocks are strongly negatively correlated with oil production and consumption, and have negative impacts on aggregate economic variables. In contrast, oil price increases from demand shocks are associated with increased economic output and steady or slightly increasing levels of oil production and consumption. The rest of the paper is organized as follows: Section 2 introduces a model of competitive oil producers and discusses the shock identification strategy in the context of the model. Section 3 empirically implements the identification strategy, presents the relations between the shocks and the aggregate stock market, and presents empirical support for the validity of the identification technique. Section 4 presents detailed results on the relations between the two constructed shocks and industry stock returns. Section 5 concludes. 2. A Simple Model of Oil Production The model introduced here is a model of atomistic competitive firms which take the price as given and choose both investment in oil reserves (with very high adjustment costs) and the level of a flow input (with low adjustment costs).5 The model is used to examine the potential efficacy of using stock returns to oil producing firms to identify different sources of oil price variation, with the caveat that the effectiveness of the strategy will be somewhat dependent on the parameters. The model is meant as a demonstration of why such a strategy might work, with the primary burden falling on the empirical analysis to validate the technique. The model views oil as a depletable resource as in Carlson, Khokher, and Titman (2007) and Ghoddusi (2010). The first important extension here is the inclusion of exogenous supply shocks, in addition to the standard demand shocks, so that the relative impacts on prices and producer stock returns can be examined. The supply shocks are modeled as increases to the flow of oil produced by a given level of flow input and oil wells. These shocks both increase current production as well as the speed with which oil wells are depleted. The model also deviates from existing work in allowing for exogenous shocks to the expected rate of return on oil producing firms, which in the model are generated by exogenous shocks to the price of risk associated with aggregate demand shocks. These shocks impact oil producer returns without creating a large impact on oil prices, and therefore need to be controlled for in order to effectively identify demand and supply shocks. 2.1 Consumers Following Kogan, Livdan, and Yaron (2009), consumers of oil are represented by an inverse demand curve, so that spot prices Pt are given by   Pt=At(Ot)−1α. (1) The price is dependent upon At, representing the aggregate level of oil demand in the economy, Ot, the total production of oil, and α, the elasticity of demand. 2.2 Firms The model consists of a continuum of competitive oil producing firms with Cobb–Douglas production technology.   Ot=ZtFtνWt1−ν. (2) Oil wells Wt, and a flow input Ft, are used to produce oil output Ot. The level of productivity is also affected by an oil industry production shock, Zt. For simplicity there are no firm-specific shocks, as well as no entry and exit, as in Kogan, Livdan, and Yaron (2009). In the context of the model, Wt represents developed oil reserves in the ground, and allows the model to capture the storable nature of oil. The costs of increasing production in a given period are not only the direct costs to the producer of a higher level of the input Ft, but also the reduction of oil reserves available to produce in future periods. This cost is reflected in the evolution of oil reserves   Wt+1=Wt−Ot+It. (3) The producer chooses investment in new reserves (It), which are depleted as oil is produced. Given an oil price, Pt, firms sell their output earning a profit Πt  Πt=PtOt−cFFt−It−Φ(Ft,Ft−1,It,Wt), (4) where Φ is a function representing costs to adjusting both the level of the flow input and the levels of investment in oil wells, and takes a quadratic form   Φ(Ft,Ft−1,It,Kt)=aF2(Ft−Ft−1Ft−1)2Ft−1+aW2(ItWt−I¯W¯)2Wt, (5) where I¯ and W¯ are the deterministic steady-state values of investment and oil well stock, and aF and aW govern the level of adjustment costs for the flow input and oil reserves, respectively. Note that Equations (3) and (4) imply that, at the deterministic steady-state investment amount, an additional oil well costs one unit of profit. The competitive producers take the price as given and choose It and Ft in each period to maximize firm value, calculated as the discounted expectation of future profits.   Vt=Πt+Et[∑s=1∞Mt+sMtΠt+s], (6) where Mt+s is the stochastic discount factor and is defined below. It is also assumed for simplicity that firms are purely equity financed, so that returns to an index of oil producing firms are given by   (1+RtProd)=Vt+ΠtVt−1. (7) 2.3 Dynamics From Equation (1), it is clear that there are two possible channels in the model for generating a change in the oil price. The first is a rise in the level of demand At, and the second is a reduction in the level of supply Ot. Although producers of final goods and household consumers are omitted for parsimony, simple intuition suggest that rises in the oil price from increases in At reflect positive economic news, while rises in price from a reduction in Ot generated by a decrease in productivity, Zt, would represent negative news for the aggregate stock market. Both aggregate oil demand and oil productivity are stochastic and their logs (indicated by lower case) evolve according to   at+1=a0+ρa(at−a0)+σaea,t+1, (8)  zt+1=z0+ρz(zt−z0)+σzez,t+1, (9) where ea,t+1 and ez,t+1 are independent normally distributed shocks with mean zero and a variance of one. High realizations of either ea,t+1 or ez,t+1 correspond to “good” times, and therefore both command positive prices of risk. To capture this the stochastic discount factor is given by   Mt+1Mt=β  exp(−λa,tea,t+1−λzez,t+1). (10) In order to capture market wide shocks to expected returns, λa,t is time varying and evolves according to   λa,t+1=λ¯a+ρλ(λa,t−λ¯a)+σλeλ,t+1. (11) 2.4 Model Results A competitive equilibrium is defined as a sequence of choices of It and Ft such that the firms are maximizing firm value while taking Pt as given, and the market clearing condition Pt=AtOt−1α is met. The producers’ first-order equation for the choice of the flow input, Ft is   cF=νOtFt(Pt−qt). (12) Here, qt is the Lagrangian multiplier associated with investment constraint on oil well accumulation, and represents the marginal value of an extra oil well in time t + 1. As this equation implies, producers will take into account that selling oil incurs a cost not only through the direct cost of the flow input, but also through the depletion of reserves. This effect is a standard feature of exhaustible resource models, dating back to Hotelling (1931). In contrast to models where reserves are fully exhaustible, here it is possible to invest in additional reserves, so prices in this model are stationary. Figure 1 shows impulse response functions to the three different types of shocks in the model. Given a negative shock to oil well productivity, the oil price rises as production falls, and producers respond by increasing the flow input to offset some of the fall in production. The increase in the flow variable is enough to prevent a rise in profitability, and hence value, for the producing firms, so this increase in oil prices is not accompanied by an increase in oil producer value. Figure 1 View largeDownload slide Model impulse response functions. Plots for model impulse responses to positive demand shocks (positive realization of ea,t), and negative productivity shocks (negative realization of ez,t), and increases in the price of risk associated with demand shocks (positive realization of eλ,t). Figure 1 View largeDownload slide Model impulse response functions. Plots for model impulse responses to positive demand shocks (positive realization of ea,t), and negative productivity shocks (negative realization of ez,t), and increases in the price of risk associated with demand shocks (positive realization of eλ,t). An increase in oil demand generates an increase in oil prices, an increase in oil production, and increased use of the oil flow input by producers. Since both oil prices and production increase, this leads to an increase in the value of oil producing firms. Finally, a shock to the price of risk associated with demand shocks makes oil firms riskier, and therefore increases the discount rate applied to the profit of oil producing firms, and leads to a negative stock return for oil producers. In contrast, this shock has a relatively small impact on the production decision of the firms, and therefore a small impact on oil prices. This differential impact makes discount rate shocks an important control in the context of the identification strategy. 2.5 Simulations The model is solved using parameters given in Table I. One important parameter for the identification technique is the elasticity of oil demand, which is significantly less than one in the data. The calibration uses a value of α=0.5, consistent with estimates in the literature (See Kogan, Livdan, and Yaron, 2009). The remaining parameters are calibrated to match volatilities and correlations of prices and returns. Table I Model parameters Parameter  Description  Value  ν  Share of flow input in production  0.6  α  Elasticity of demand  0.5  σa  Volatility of demand shock  0.15  σz  Volatility of productivity shock  0.065  σλ  Volatility of risk shock  0.25  ρa  Persistence of demand shock  0.95  ρz  Persistence of productivity shock  0.95  ρλ  Persistence of demand risk price  0.95  a0  Mean of demand shock  0  z0  Mean of productivity shock  0  cF  Cost of flow input  2.5  aF  Flow input adjustment cost  3  aW  Oil well adjustment cost  15  d  Depletion rate  1  δ  Deprecation of oil wells  0.01  λ¯a  Mean price of demand risk  2.0  λz  Price of supply risk  0.3  β  Discount rate  0.995  Parameter  Description  Value  ν  Share of flow input in production  0.6  α  Elasticity of demand  0.5  σa  Volatility of demand shock  0.15  σz  Volatility of productivity shock  0.065  σλ  Volatility of risk shock  0.25  ρa  Persistence of demand shock  0.95  ρz  Persistence of productivity shock  0.95  ρλ  Persistence of demand risk price  0.95  a0  Mean of demand shock  0  z0  Mean of productivity shock  0  cF  Cost of flow input  2.5  aF  Flow input adjustment cost  3  aW  Oil well adjustment cost  15  d  Depletion rate  1  δ  Deprecation of oil wells  0.01  λ¯a  Mean price of demand risk  2.0  λz  Price of supply risk  0.3  β  Discount rate  0.995  Table II reports volatilities and the correlation of oil producers’ stock returns and changes in oil prices for the simulated model. The table also reports the correlations between the simulated price innovations and returns and the unobservable supply and demand shocks. Oil producer stock returns are highly correlated with demand shocks while being nearly uncorrelated with supply shocks. If the shocks to risk-aversion can be controlled for, this makes producer returns an effective control for identifying the sources of price variation. Table II Model simulated prices and returns This table shows monthly correlations of simulated model shocks, oil prices, oil producer returns. The table also shows as the volatilities of oil prices and producer returns, along with the percentage of variance of these variables explained by the structural shocks in the model. Δpt are simulated changes in the log spot price of oil, Rtprod are simulated oil producer returns, ea,t are simulated oil demand shocks, ez,t are simulated producer productivity (supply) shocks, and, eλ,t are simulated shocks to the price of risk associated with demand shocks. The model is simulated monthly for 10,000 periods.   Observable variables     Correlation matrix    RtProd  Δpt  RtProd  1.00          Volatility  20.68  34.16  Δpt  0.28  1.00        % of var. from:      ez,t  0.01  −0.83  1.00      ez,t  0.0%  85.5%  ea,t  0.96  0.32  0.00  1.00    ea,t  92.4%  13.0%  eλ,t  −0.28  0.11  0.00  0.00  1.00  eλ,t  7.6%  1.5%    Observable variables     Correlation matrix    RtProd  Δpt  RtProd  1.00          Volatility  20.68  34.16  Δpt  0.28  1.00        % of var. from:      ez,t  0.01  −0.83  1.00      ez,t  0.0%  85.5%  ea,t  0.96  0.32  0.00  1.00    ea,t  92.4%  13.0%  eλ,t  −0.28  0.11  0.00  0.00  1.00  eλ,t  7.6%  1.5%  Although the model is meant to be stylized, it does provide some insight into how an empirical identification technique would work. A supply shock should be reflected higher oil prices, and lower oil output and consumption, while a positive demand shock should be reflected in higher prices, and potentially a rise in oil production and consumption, but not one as marked as the observed decrease from the supply shock. The differential exposure of oil producer returns to the two types of shocks also allows for identifying the sources of changes in price. The next section constructs these two shocks using changes in the VIX index as a proxy for shocks to risk aversion, and attempts to provide empirical evidence that the classified shocks are indeed related to oil supply and demand. 3. Classification of Oil Shocks 3.1 Data and Shock Construction The three variables necessary to construct the series of supply and demand shocks are an index of oil producing firms, a measure of oil price changes, and a proxy for changes in expected returns. To cover as much of the global oil industry as possible, the index used is the World Integrated Oil and Gas Producer Index available from Datastream. This index covers the large publicly traded oil producing firms (Exxon, Chevron, BP, Petrobras, etc.), but does not include nationalized oil producers such as Saudi Aramco. For changes in the oil price, the 1-month returns on the second nearest maturity NYMEX Crude—Light Sweet Oil contract are used for the bulk of the analysis, in order to focus on unexpected changes to oil prices. Aggregate US stock market data are from the universe of CRSP stocks.6 To proxy for changes in the discount rate, innovations to the VIX index are used. The VIX is calculated from options data, and therefore provides a measure of the risk-neutral expectation of volatility. As shown in Bollerslev, Tauchen, and Zhou (2009), the variance risk premium captured in the VIX is both strongly negatively correlated with stock returns and positively predict stock returns in the time series, suggesting that it may be a reasonable proxy for changes in risk. The VIX is published by the CBOE back to 1986, and this limits the data for the main portion of the analysis. In order to isolate unexpected changes in the VIX, an ARMA(1,1) process is estimated and the residuals from this process are used as innovations. These innovations are denoted ξVIX. Panel A of Table III provides summary statistics of standard deviations and correlations of the four variables for the sample period. This correlation matrix is at the heart of the strategy pursued in this paper. The high correlation between oil producer returns and both oil prices and aggregate market returns, along with a very low correlation between oil prices and aggregate market returns, suggests that there may be some source of variation which loads negatively on aggregate stock returns but positively on oil prices and is uncorrelated with producers. Additionally, changes in the VIX have a correlation with both aggregate and oil producer stock returns, but a minimal correlation with oil prices. Table III Summary of data and identified shocks Panel A shows annualized means, standard deviations, and monthly correlations. Oil price changes are the returns to the NYMEX second nearest maturity future. Oil producer returns are from the Datastream World Integrated Oil and Gas Producer Index. US Market return is the aggregate CRSP return. ξVIX is the residual value from an estimated ARMA(1,1) for the log of the VIX index. Data are monthly and returns are calculated in logs. Panel B shows decomposition of oil price changes into component shocks using the method described in Section 3.1. By construction the shocks account for all variation in oil prices. Data are monthly with annualized standard deviations reported. Sample is January 1986 to December 2011. Panel A. Summary of data  Variable  Description  Mean  Standard deviation  Correlation matrix    RtUSA  US market return  0.04  0.16  1        RtProd  Oil Producer Index Returns  0.05  0.18  0.62  1      ξVIX,t  Innovations to VIX  0.00  0.60  −0.67  −0.45  1    Δpt  Changes in oil prices  0.06  0.35  0.07  0.45  −0.11  1    Panel B. Identified oil shocks  Variable  Description  Standard deviation  % of Var.  Correlation matrix    Δpt  Changes in oil prices  0.34    1        st  Supply shocks  0.30  0.78  0.88  1      dt  Demand shocks  0.15  0.21  0.45  0  1    vt  Risk shocks  0.04  0.01  0.11  0  0  1  Panel A. Summary of data  Variable  Description  Mean  Standard deviation  Correlation matrix    RtUSA  US market return  0.04  0.16  1        RtProd  Oil Producer Index Returns  0.05  0.18  0.62  1      ξVIX,t  Innovations to VIX  0.00  0.60  −0.67  −0.45  1    Δpt  Changes in oil prices  0.06  0.35  0.07  0.45  −0.11  1    Panel B. Identified oil shocks  Variable  Description  Standard deviation  % of Var.  Correlation matrix    Δpt  Changes in oil prices  0.34    1        st  Supply shocks  0.30  0.78  0.88  1      dt  Demand shocks  0.15  0.21  0.45  0  1    vt  Risk shocks  0.04  0.01  0.11  0  0  1  For the primary analysis the supply shocks st, demand shocks dt, and risk shocks vt are orthogonal and defined in the following manner. Define   Xt≡[ΔptRtProdξVIX,t],Zt≡[stdtvt],A≡[1110a22a2300a33]. The matrix A maps the identified shocks into the observable variables so that   Xt=AZt. (13) To impose orthogonality a22,a23,a33 and σs, σd, σv satisfy   A−1ΣX(A−1)T=[σs2000σd2000σv2], (14) where ΣX is the covariance matrix of the observable Xt, and σs, σd, and σv are the volatilities of the identified shocks. Note that this is simply a renormalization of the standard orthogonalization used to define the structural shocks in an SVAR setting. Here, rather than normalize the volatility of the shocks to one, the shocks are constrained to sum up to the total change in the oil price. Panel B of Table III shows the summary statistics for the constructed shocks. The annualized volatility of changes in oil prices over the sample period is roughly 33.8% while the annualized volatility of the supply and demand shocks is 30.5% and 15.6%, respectively. Equivalently, roughly 78% of the variance in oil prices is classified as supply shocks and 21% as demand shocks, with the shocks to the VIX explaining only 1% of the total variance in oil price shocks. 3.2 Oil Shocks and Aggregate Stock Returns Once the shocks are constructed, it is a simple matter to use them in a basic regression of aggregate stock market returns. Since the risk shocks vt are simply a constant multiple of ξVIX,t, ξVIX,t is used in the regression in order to provide a more straightforward interpretation of the economic magnitude of the regression coefficients. Table IV reports regressions of the form   RtMkt=α+βddt+βsst+βVIXξVIX,t (15) as well as the R2 from univariate regressions of the market return on each shock, for both the aggregate CRSP stock index and a world wide stock index from Global Financial Data. On their own, oil prices have little or no relation to aggregate market returns. However, when they are decomposed into two series, which together explain nearly all of the variation in oil prices, oil prices have a very clear and intuitive link to the stock market. High oil prices from supply shocks are bad news for aggregate stock returns, and can explain 3.6% of the monthly variation in the aggregate US market return, while rises in prices from demand shocks are associated with positive excess returns and can explain an additional 12.4% of the variation. Table IV Aggregate stock market returns and oil price shocks Panel A shows regressions of Aggregate US Stock Market Returns (return to the index of all CRSP stocks) on changes in oil prices and on constructed demand and supply shocks. Shocks are defined as in Section 3.1. Panel B shows regressions of world aggregate stock market index on constructed shocks. World index is from Global Financial Data. White (1980) standard errors in parentheses. Data are monthly from 1986 to 2011 and market returns are in logs. Panel A. US stock market returns and oil shocks  Description  Variable  US market ret. ( RtUSA)        Oil price changes  Δpt  0.031      (0.027)  Univariate R2  Demand shock  dt    0.370**  0.124  (0.046)    Supply shock  st    −0.102**  0.036  (0.021)    Innovation in VIX  ξVIX,t    −0.184**  0.444  (0.012)    Constant    0.005  0.003    (0.003)  (0.002)  Observations    315  315    R-squared    0.004  0.604      Panel B. World stock market returns and oil shocks  Description  Variable  World Market Ret. ( RtWorld)        Oil Price Changes  Δpt  0.046      (0.026)  Univariate R2  Demand shock  dt    0.493**  0.231  (0.040)    Supply shock  st    −0.109**  0.043  (0.023)    Innovation in VIX  ξVIX,t    −0.158**  0.345  (0.011)    Constant    0.005  0.002    (0.003)  (0.002)  Observations    315  315    R-squared    0.010  0.619    Panel A. US stock market returns and oil shocks  Description  Variable  US market ret. ( RtUSA)        Oil price changes  Δpt  0.031      (0.027)  Univariate R2  Demand shock  dt    0.370**  0.124  (0.046)    Supply shock  st    −0.102**  0.036  (0.021)    Innovation in VIX  ξVIX,t    −0.184**  0.444  (0.012)    Constant    0.005  0.003    (0.003)  (0.002)  Observations    315  315    R-squared    0.004  0.604      Panel B. World stock market returns and oil shocks  Description  Variable  World Market Ret. ( RtWorld)        Oil Price Changes  Δpt  0.046      (0.026)  Univariate R2  Demand shock  dt    0.493**  0.231  (0.040)    Supply shock  st    −0.109**  0.043  (0.023)    Innovation in VIX  ξVIX,t    −0.158**  0.345  (0.011)    Constant    0.005  0.002    (0.003)  (0.002)  Observations    315  315    R-squared    0.010  0.619    These results are robust to the exclusion of outliers, the largest of which occur during the 2008 Financial Crisis. Removing these observations considerably strengthens the impacts of supply shocks on aggregate returns. Moreover, when the same strategy is performed with copper and aluminum producers and prices as a placebo test, the supply shocks identified using these two, less essential, commodities have no impact on aggregate stock returns, suggesting that the documented effects reflect the impacts of oil supply shocks and not a mechanical correlation.7 The next section focuses on presenting additional evidence in support of the efficacy of the identification strategy. 3.3 Evidence for the Validity of the Supply and Demand Classification It is clear that decomposing changes in oil prices in the manner described in the previous section leads to strong correlations with aggregate stock market returns. In order to provide support for the proposed interpretation that this decomposition allows for effective identification of supply and demand shocks, Section 3.3.a examines the behavior of producer returns and prices around a known supply shock, the first Gulf War, and finds them to be consistent with the story. Section 3.3.b presents univariate vector autoregressions showing that the two shocks relate to variables reflecting macroeconomic output and the oil supply in ways which are consistent with the supply and demand shock interpretation. 3.3.a. Oil producer returns and oil prices in the First Gulf War. This section examines the behavior of stock returns and prices around the First Gulf War. This period is chosen because it is the most severe oil shock in the sample. Although the war interrupted production of the major nationalized oil producers in the Middle East, it also impacted many of the large western oil firms in the world wide producer index which were active in the region at the time. Figure 2 shows the patterns of returns and prices over this time period. Figure 2 View largeDownload slide Oil producers stock prices, oil prices, and the Gulf War. Plots of cumulative returns for oil producers, the global oil producer index, the aggregate stock market, as well as cumulative changes in oil prices from June 1990 to July of 1991. Oil producer index is the Datastream World Integrated Oil and Gas Producer Index. US Stock Return is the value weighted CRSP index. Stock market returns are from CRSP. Figure 2 View largeDownload slide Oil producers stock prices, oil prices, and the Gulf War. Plots of cumulative returns for oil producers, the global oil producer index, the aggregate stock market, as well as cumulative changes in oil prices from June 1990 to July of 1991. Oil producer index is the Datastream World Integrated Oil and Gas Producer Index. US Stock Return is the value weighted CRSP index. Stock market returns are from CRSP. In the first panel, the stock prices of several multinational oil producing corporations are shown along with the spot price of oil during the first Gulf War. Companies active in the Middle East, such as Ashland, Inc. and Occidental Petroleum, suffered drops in value due to concerns about their ability to produce during the conflict. In contrast companies, such as British Petroleum, which were not operating in the region, saw their valuations rise with the higher price of crude. The second panel shows the behavior of the Datastream World Integrated Oil and Gas Producers index over this period, as well as the performance of the US stock market. The initial invasion of Kuwait saw a spike in oil prices, and a negative return for the aggregate stock market, but very little response in the returns of oil producers. Although some individual producers saw large returns, in aggregate it appears that oil producers as an industry enjoyed a natural hedge against the potential supply shock. At the same time, aggregate stock values fell, suggesting a potential relation between changes in oil prices and stock returns. 3.3.b. Macroeconomic variables and oil shocks. To provide further evidence that the identified shocks represent distinct sources of changes in oil prices, univariate vector autoregressions are estimated for variables related to macroeconomic output and oil markets, with the constructed supply and demand shocks entering as exogenous variables. Following Hamilton (2008), these VARs have the following form   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (16) where yt is the log of an economic variable of interest, and the supply and demand shocks st and dt, which are included contemporaneously as well as with N lags. For this portion of the analysis, a longer sample is used to provide more power with tests of quarterly data. The shocks are constructed in the same manner as described in Section 3.1, but the log change of the WTI is used in place of futures returns, and residuals from an ARMA(1,1) of aggregate US stock market volatility are used in place of ξVIX,t. The resulting supply and demand shocks have a correlation of 99% and 93%, respectively, with the shocks constructed using futures and the VIX. Recent work by Bekaert and Hoerova (2014) shows that the volatility portion of the VIX correlates most strongly with macroeconomic aggregates, suggesting that this control is appropriate for this analysis. Since US data for both aggregate output and oil use are readily available for the sample, five of the six variables chosen for the analysis are related to US economic output and oil consumption. US GDP and Total US Oil consumption are included to measure aggregate economic output and oil use. Total US Oil Consumption is US Oil Production plus Imports less Exports less Changes in Inventories, with all data taken from the EIA. Additionally, Ready (2016) shows that relative levels of household aggregate consumption and household gasoline consumption are very tightly related to the oil price. This motivates the choice of two household consumption variables, the household consumption of gasoline, given by “Gasoline and Other Energy Goods” from the BEA’s NIPA survey, and a Cobb–Douglas aggregate of durable and nondurable household consumption (net of household gasoline consumption), similar to the construction in Yogo (2005). Although not explicitly included in the model, inventories are also central to discussions of the oil price, so US petroleum stocks (net of the Strategic Petroleum Reserve) provided by the EIA are also included. The last variable is the level of international oil production again from the EIA.8 All data are aggregated to the quarterly level, and regressions are done using four lags. Figure 3 reports the impulse response functions of six variables to an oil supply shock and Figure 4 reports the same for a demand shock. For an increase in oil prices from a supply shock, there are significant decreases in all six variables, consistent with the hypothesis that this shock measures a reduction in available oil, leading to lower oil consumption, which in turn lowers aggregate economic activity. Conversely, for a demand shock, there are significant positive increases in economic output and total consumption, and an increase in aggregate economic oil use. There is no significant change in household oil consumption, and a delayed increase in world oil production, consistent with the model. Also, supply shocks result in a sharp decrease in inventories while demand shocks do not. The different responses of oil production and consumption are notable, since both shocks are associated with an increase in the oil price. Figure 3 View largeDownload slide Supply shock impulse response functions. Figure plots cumulative impulse response functions to a one standard deviation oil supply shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil supply shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (17) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 3 View largeDownload slide Supply shock impulse response functions. Figure plots cumulative impulse response functions to a one standard deviation oil supply shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil supply shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsst−n+βnddt−n), (17) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 4 View largeDownload slide Demand shock impulse response functions. Figure plots VAR cumulative impulse response functions to a one standard deviation oil demand shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil demand shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsΔst−n+βndΔdt−n), (18) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. Figure 4 View largeDownload slide Demand shock impulse response functions. Figure plots VAR cumulative impulse response functions to a one standard deviation oil demand shock for the logs of US real GDP, US total oil use, aggregate household consumption (net of gasoline), household gasoline consumption, US inventories, and world oil production. Data are from the BEA and EIA. Oil demand shocks are constructed monthly using changes to the log of the WTI, the Datastream Integrated World Oil and Gas Producers index, and unexpected changes in aggregate US stock market volatility and then aggregated up to a quarterly frequency. Each panel shows the impulse response estimated from a univariate autoregressive process in which lags of the supply and demand shocks are included as exogenous variables:   Δyt=∑n=1NβnyΔyt−n+∑n=0N(βnsΔst−n+βndΔdt−n), (18) where Δyt is the log innovation of the variable at issue. Regressions are at a quarterly frequency from 1979 to 2011 with four lags. Dashed lines represent 90% confidence intervals. 4. The Impact of Oil Shocks on Industry Returns In order to further illustrate the relation between oil prices and stock returns, and to gain more insight into the mechanisms by which oil price shocks impact the economy, this section explores the relation between the oil shocks and industry returns. The main takeaways are that consumer goods industries have the strong negative loadings on both oil supply shocks and demand shocks, while the industries with both high levels of oil use and high market betas have the strongest positive loadings on demand shocks. This provides support to the hypothesis of Hamilton (2003), who suggests that oil prices shocks act primarily on consumer expenditure rather than a direct effect from increased cost of inputs. Additionally, the fact that supply shock exposure is unrelated to market beta suggests that these shocks are a distinct source of return variation from aggregate market shocks. Table V shows the results for regressions of industry returns on supply and demand shocks over the pre-crisis sample for a partition of ten industry portfolios constructed by Fama and French (1997). Returns for these industries are available on Ken French’s website. The industries are sorted by their loading on the oil supply shock. What is immediately apparent is that nearly all of the industries do load negatively on the shock, but that three of the top four industries, Consumer Durables, Consumer Nondurables, and Retail, are highly dependent on consumer spending. Table V Fama-French ten industry portfolios and oil shocks Table reports univariate regressions of industry returns on supply shocks and univariate regressions of industry returns on demand shocks. Constants are suppressed. Industry portfolios are from Fama and French (1997) and are taken from Ken French’s website. White (1980) standard errors in parentheses. Data are monthly from 1986 to 2011. Description  Univariate regressions   Supply shock   Demand shock   Beta  R2  Beta  R2  Consumer durables  −0.235**  0.117  0.256**  0.036  (0.040)    (0.081)    Retail  −0.232**  0.154  0.123  0.011  (0.033)    (0.071)    Other  −0.220**  0.157  0.190**  0.031  (0.031)    (0.066)    Consumer nondurables  −0.205**  0.174  0.215**  0.050  (0.027)    (0.058)    Manufacturing  −0.188**  0.121  0.315**  0.089  (0.031)    (0.062)    Healthcare  −0.171**  0.096  0.161*  0.022  (0.032)    (0.065)    High Tech  −0.139**  0.026  0.199  0.014  (0.052)    (0.103)    Telecommunications  −0.134**  0.051  0.191**  0.027  (0.035)    (0.070)    Utility  −0.117**  0.066  0.337**  0.143  (0.027)    (0.051)    Energy  −0.017  0.001  0.862**  0.601  (0.035)    (0.043)    Description  Univariate regressions   Supply shock   Demand shock   Beta  R2  Beta  R2  Consumer durables  −0.235**  0.117  0.256**  0.036  (0.040)    (0.081)    Retail  −0.232**  0.154  0.123  0.011  (0.033)    (0.071)    Other  −0.220**  0.157  0.190**  0.031  (0.031)    (0.066)    Consumer nondurables  −0.205**  0.174  0.215**  0.050  (0.027)    (0.058)    Manufacturing  −0.188**  0.121  0.315**  0.089  (0.031)    (0.062)    Healthcare  −0.171**  0.096  0.161*  0.022  (0.032)    (0.065)    High Tech  −0.139**  0.026  0.199  0.014  (0.052)    (0.103)    Telecommunications  −0.134**  0.051  0.191**  0.027  (0.035)    (0.070)    Utility  −0.117**  0.066  0.337**  0.143  (0.027)    (0.051)    Energy  −0.017  0.001  0.862**  0.601  (0.035)    (0.043)    To further illustrate this, Figure 5 plots supply shock betas, this time for each of forty-nine Industry Portfolios again available from Ken French’s website. The industry “Other” in this partition is replaced by an Airline industry, constructed as the value-weighted return to SIC 4512 in the CRSP database, and the “Oil” industry is excluded.9 The top plot of the figure shows industry supply shock betas plotted against the industries’ market beta measured over the sample. The lack of relation here suggests that the identified shock, which negatively impacts nearly all industries, is a distinct source of variation from other aggregate market shocks. The second plot shows supply betas plotted against industry oil use, which is defined as the dollar of oil necessary to produce a dollar of industry output, constructed using input–output tables from the BEA. Again there is no significant pattern, suggesting that high oil intensity does not correlate with strong exposure to supply shocks. The last plot shows industry supply betas plotted against the percentage of the industry output used for personal consumption and residential investment, constructed using industry use tables again from the BEA. Here, we see a strong relation, with industries selling directly to consumers most impacted by supply shocks. Figure 5 View largeDownload slide Industry oil supply shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 5 View largeDownload slide Industry oil supply shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 6 repeats these plots but this time for oil demand betas. The figure shows that all industries load positively on the oil demand shock, but, in contrast with supply shocks, there is a positive significant (at the 10% level) relation between these loadings and industry market beta. There is also a significant relation between oil use and oil demand beta. This suggests that when manufacturing and oil intensive areas of the economy pick up in activity, the high demand translates to high oil prices. In the last plot, we also see a strong negative relation between oil demand beta and the percent of industry output used by households, suggesting that the high oil prices caused by increase demand are a drag on consumer industries, and that shocks to these industries do not seem to be driving oil demand.10 Figure 6 View largeDownload slide Industry oil demand shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. Figure 6 View largeDownload slide Industry oil demand shock betas. This figure plots betas for the forty-nine Fama–French industries (excluding the Oil industry) with respect to oil demand shocks against three explanatory variables. The “Other” industry is replaced by a value weighted index of Airlines (SIC 4512). In the top panel, X-axis represents the industry beta with respect to the aggregate market beta over the sample. In the middle panel, the X-axis is the amount of oil in dollars necessary to produce a dollar of industry output. This ratio is calculated as the equal weighted average of BEA industries, which are hand mapped to the Fama–French industry classifications. The X-axis in the bottom panel is a measure of the percentage of oil used by households, which is calculated as industry output used for personal consumption and residential investment divided by the total industry output from the “Industry Use” table published by the BEA. Data are from the BEA input–output tables, CRSP, Ken French’s website. T-statistics for the trend line are calculated using panel regressions including an interaction of the explanatory variable and the oil shock, with standard errors clustered by industry. Data are from 1986 to 2011. 5. Conclusion This paper presents a new, simple, method for identifying sources of oil supply shocks using changes to the VIX and oil producer stock returns. Using this method, both oil supply and demand shocks are shown to have a highly significant impact on US and world stock prices, in contrast to the very small correlations observed when using aggregate changes in oil prices. The impact of supply shocks is more significant for firms that depend on consumer expenditure rather than those which rely on oil as an input, while demand shocks are associated with high returns to oil using firms. These findings provide insight into the way oil prices affect the economy, and suggest that the important effect of oil price shocks may be pain at the pump for consumers rather than higher prices for oil using firms. The classification of the shocks provides a new technique for researchers studying the effects of changes in oil prices, particularly those interested in studying impacts at frequencies higher than quarterly data, or over short sample periods. The results may also be of interest to practitioners, since they directly imply that holding a portfolio of oil producing firms is not a hedge for harmful oil shocks. Supplementary Material Supplementary data are available at Review of Finance online. Footnotes 1 Other authors have examined this question and found similarly low contemporaneous relations. See for instance: Jones and Kaul (1996); Kilian and Park (2009); Chen, Roll, and Ross (1986); Huang, Masulis, and Stoll (1996); and Sadorsky (1999). In related work, Chiang et al. (2015) provide complementary evidence on the importance of various types of oil shocks for the cross-section of equity returns, and Gilje, Ready, and Roussanov (2015) provide evidence on the stock market impacts of increases in US shale oil production. 2 First line of Table I in Kilian and Park (2009). 3 See Internet Appendix for these results and the relation of the shocks identified in Kilian (2009) to those identified here. 4 See Kolodzeij and Kaufmann (2014) for a detailed discussion of the issues with the SVAR methodology in this context. 5 The Internet Appendix extends the model to include a price-leading monopolist along side a competitive fringe, and discusses the implications for the classification strategy in that setting. 6 See Internet Appendix for results using other sources of oil producer indices, oil prices, and risk levels, and the effects of controlling for aggregate levels of prices and exchange rates. There is no significant impact on the results. 7 See Internet Appendix. 8 International oil production is lagged 1 month as lagged production seems to correlate most strongly with prices and US quantities. This is possibly due to a delay in the availability of information. The US data are released weekly in the “Weekly Petroleum Status Report”, while international data are published with a 3-month lag in the “Monthly Energy Report”. 9 See Internet Appendix for regressions of individual industry portfolio returns on oil shocks. 10 Similar results hold for demand and supply shock impacts on international returns of various countries. See Internet Appendix. References Bekaert G., Hoerova M. ( 2014) The vix, the variance premium and stock market volatility, Journal of Econometrics  183, 181– 192. Google Scholar CrossRef Search ADS   Bollerslev T., Tauchen G., Zhou H. ( 2009) Expected stock returns and variance risk premia, Review of Financial Studies  22, 4463– 4492. Google Scholar CrossRef Search ADS   Carlson M., Khokher Z., Titman S. ( 2007) Equilibrium exhaustible resource price dynamics, Journal of Finance  62, 1663– 1703. Google Scholar CrossRef Search ADS   Casassus J., Collin-Dufresne P., Routledge B. R. ( 2005) Equilibrium commodity prices with irreversible investment and non-linear technology. Working paper series. Cavallo M., Wu T. ( 2006) Measuring oil–price shocks using market-based information. Technical report. Chen N.-F., Roll R., Ross S. A. ( 1986) Economic forces and the stock market, Journal of Business  59, 383– 403. Google Scholar CrossRef Search ADS   Chiang I., Ethan H., Keener Hughen W., Sagi J. S. ( 2015) Estimating oil risk factors using information from equity and derivatives markets, The Journal of Finance  70( 2), 769– 804. Google Scholar CrossRef Search ADS   Fama E. F., French K. R. ( 1997) Industry costs of equity, Journal of Financial Economics  43, 153– 193. Google Scholar CrossRef Search ADS   Ghoddusi H. ( 2010) Dynamic investment in extraction capacity of exhaustible resources, Scottish Journal of Political Economy  57, 359– 373. Google Scholar CrossRef Search ADS   Gilje E., Ready R., Roussanov N. ( 2015) Fracking, drilling, and asset pricing: estimating the economic benefits of the shale revolution, Working Paper. Hamilton J. D. ( 1983) Oil and the macroeconomy since World War II, Journal of Political Economy  91, 228– 248. Google Scholar CrossRef Search ADS   Hamilton J. D. ( 2003) What is an oil shock?, Journal of Econometrics  113, 363– 398. Google Scholar CrossRef Search ADS   Hamilton J. D. ( 2008) Understanding crude oil prices. NBER Working papers 14492, National Bureau of Economic Research, Inc. Hotelling H. ( 1931) The economics of exhaustible resources, The Journal of Political Economy  39, 137– 175. Google Scholar CrossRef Search ADS   Huang R., Masulis R. W., Stoll H. R. ( 1996) Energy shocks and financial markets, Journal of Futures Markets  16, 1– 27. Google Scholar CrossRef Search ADS   Jones C., Kaul G. ( 1996) Oil and the stock markets, Journal of Finance  51, 463– 491. Google Scholar CrossRef Search ADS   Kilian L. ( 2008) Exogenous oil supply shocks: how big are they and how much do they matter for the us economy?, The Review of Economics and Statistics  90, 216– 240. Google Scholar CrossRef Search ADS   Kilian L. ( 2009) Not all oil price shocks are alike: disentangling demand and supply shocks in the crude oil market, The American Economic Review  99, 1053– 1069. Google Scholar CrossRef Search ADS   Kilian L., Park C. ( 2009) The impact of oil price shocks on the U.S. stock market*, International Economic Review  50, 1267– 1287. Google Scholar CrossRef Search ADS   Kogan L., Livdan D., Yaron A. ( 2009) Oil futures prices in a production economy with investment constraints, The Journal of Finance  64, 1345– 1375. Google Scholar CrossRef Search ADS   Kolodzeij M., Kaufmann R. K. ( 2014) Oil demand shocks reconsidered: a cointegrated vector autoregression, Energy Economics  41, 33– 40. Google Scholar CrossRef Search ADS   Ready R. C. ( 2016) Oil consumption, economic growth, and oil futures: the impact of long-run oil supply uncertainty on asset prices. Working paper. Sadorsky P. ( 1999) Oil price shocks and stock market activity, Energy Economics  21, 449– 469. Google Scholar CrossRef Search ADS   White H. ( 1980) A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity, Econometrica  48, 817– 838. Google Scholar CrossRef Search ADS   Yogo M. ( 2005) A consumption-based explanation of expected stock returns, Journal of Finance  61, 539– 580. Google Scholar CrossRef Search ADS   © The Authors 2017. Published by Oxford University Press on behalf of the European Finance Association. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Journal

Review of FinanceOxford University Press

Published: Feb 1, 2018

There are no references for this article.