Skip to main content

How do supply or demand shocks affect the US oil market?


The study of the relationship between crude oil and its refined products prices may be perceived as an important tool for testing how are the dynamics and the type of integration of the petro-derivatives market in the United States. In this sense, we have applied a set of causality tests to study the possible presence of asymmetries in the relationship between WTI crude oil and each refined product price and to explore the type of market integration. Furthermore, the application of these causality tests lets us explore the validation of different hypotheses in the literature, such as the Rocket and Feathers hypothesis and the Verleger hypothesis. Our findings reveal that Reformulated Gasoline Blendstock for Oxygen Blending (RBOB), heating oil, diesel and kerosene are supply-driven integrated and conventional gasoline and kerosene are demand-driven integrated when linear effects are assessed. This behaviour changes deeply when the existence of asymmetries is tested, noticing that the Rocket and Feathers hypothesis is not fulfilled when a negative shock appears. Conversely, the Verleger hypothesis is supported when a negative shock appears for conventional gasoline and kerosene. These results provide important policy implications for investors, energy policymakers and refiners.


The study of the shocks in the relationship between crude oil prices and refined product prices bears significant information to refiners, investors, policymakers or practitioners, who would monitor the dynamics to establish different strategies or measures (Fousekis and Grigoriadis 2017). The abovementioned shocks are often produced by supply and demand forces, and the market participants would be on the lookout to obtain the maximum possible benefit and the reduction of the risk through portfolio diversification and hedging strategies (Hamilton and Wu 2014; Kyritsis and Serletis 2018) or taking into account sectoral financial innovations as elastic alternative investment options in order to improve the risk-adjusted returns for oil investors during a crisis (Salisu and Obiora 2021).

The COVID-19 outbreak triggered in Wuhan (China) and its rapid spread around the world caused a collapse in West Texas Intermediate (WTI, hereafter) prices in April 2020, for reasons such as the weakness of the demand and ongoing production versus storage limitations (Jefferson 2020). Another factor that could play an important role is the different lockdown measures implementedFootnote 1 by the countries to prevent and impede contagions. These lockdown measures have affected businesses, job securities and services and have brutally impacted all types of transport, either by road, rail, sea or/and air. As previously mentioned, the government's response faces the COVID-19 pandemic. People's behaviour has dramatically changed by the imposition of restrictions on mobility and economic activity, letting to work from home those who could and domestic and international travel severely restricted. People stopped driving, flying, or travelling on public transport (see Billio and Varotto 2020) so, the flows of people and goods have been reduced. This triggered a decline in demand for energy sources (transport fuels, for instance), leading to a drop in oil prices (Ozili and Arun 2020) or being a source of systematic risk, provoking a decline in global financial indices (see Ahundjanov et al. 2020; or Sharif et al. 2020). Given this, it is not unreasonable to think that, especially these factors have played an essential role in the shock influences the crude oil market so, as Bakas and Triantafyllou (2020) or Zhang et al. (2021) affirmed, in times of higher uncertainty (as the case of a pandemic like COVID-19), supply and demand drop dramatically and gradually over time because of the rise in the price elasticity.

Likewise, in 2021 the prices become to rise and recover their original path due to the reactivation of the economy and mobility. On the other hand, there is a combination of the deliberate production cuts agreed by OPEC and other non-OPEC producers, such as Russia; and a drop in production in the United States and elsewhere due to the low prices registered per barrel the previous year. This was translated into the explosiveness of crude oil and its derivatives prices that society is experimenting with nowadays. However, and more importantly, by the beginning of 2022 the ‘Ukraine–Russian war’ pushed the refined product prices up to high levels, around 5 dollars per gallon in the United States, provoking that inflation to reach high levels. In this sense, Li et al. (2020) suggest actual extreme political events will drive crude oil prices and provide that geopolitical factors affect crude oil prices.

In this sense, the study of energy prices linkages lets researchers evaluate the degree and type of integration, which market (or product) may be the price setter, the asymmetries that could incur in the information transmission, the co-movements between prices, or to value the price risk (see Girma and Paulson 1999; Grasso and Manera 2007; Douglas 2010; Bremmer and Kesselring 2016; or Caro et al. 2020; for instance). In this respect, as it is well-known, crude oil is the raw ingredient with which refined products are made so, it stands to reason that a change in the price of oil could affect the price of refined products, being this an example of supply-driven market integration (Asche et al. 2003). Conversely, according to Kilian (2010, 2014), when refined product prices condition crude oil prices due to crude oil is necessary to generate these products, i.e., gasoline, diesel or heating oil, that is consumed straightaway so, the flow demand for oil increases, being this an example of demand-driven market integration, i.e., the demand from crude oil is given from the demand for refined products (see Kilian 2010). Thus, accounting the nature of the shocks would provide us information regarding the type of market integration.

In this context, the study of the causality link regarding the crude oil and its refined products has not been discussed in great depth but has been more related to other commodities (see Villar and Joutz 2006; Brown and Yucel 2008; Zhang and Wei 2010; Wolfe and Rosenman 2014; Yu et al. 2015; or Rezitis 2015) or economicFootnote 2 and financial variables (see Kilian 2008; Kilian and Vigfusson 2011; Brahmasrene et al. 2014; Bal and Rath 2015; Ding et al. 2016).

Thus, the contribution of this work is a further step in this area of research so, as far as we know, this is the first time that this methodology (the methodology developed by Hatemi-J (2012), which is an expansion of the model developed by Toda and Yamamoto (1995)) has been used to capture how the different nature of the shocks (positive and negative) can affect the relationships between crude oil and its derivatives and the type of market integration. Based on Hatemi-J (2012) test, which is an extension of the Toda and Yamamoto (1995) test, is a causality approach that allows for asymmetry. This means that we can assess causality by considering the cumulative sum of positive and negative shocks separately. The impact of these shocks on causal relationships can differ since shocks can be either positive or negative, a characteristic not considered in other causality tests (Mishra et al. 2022). Following Hatemi-J (2012), this method has the advantage of causality can be tested among variables of a different order of integration and even in the absence of cointegration (Masih and Masih 1998).

In fact, the objective of the present study is to investigate whether crude oil and its refined products may represent a demand-driven or supply-driven market integration by analysing the asymmetric causal links between crude oil and its refined products, studying the behaviour when a negative or positive shock affects the variables. To this end, in this paper, we use the asymmetric causality analysis that allows us to explore the type of market integration. Afterwards, when we seek for the different type of market integration, i.e., demand-driven or supply-driven market integration, we could subsequently test the existence of the Rockets and Feathers hypothesis (if the market follows an asymmetric supply-driven market integration) or the Verleger hypothesis (if the market follows an asymmetric demand-driven market integration).

In this respect, our findings indicate the heterogeneity in the behaviour of the refined products attending to their causal relation to WTI, exposing the need to study the asymmetries regarding the causal relationships given. In this regard, the dependent role to crude oil of kerosene is represented in each type of integration when assessing the effects of linear relation or positive and negative shocks, which makes sense. This aspect would convey us to explore the treatment of this refined product, which is essential to the airlines and the travel industry. In this sense, the hedging strategies followed by airlines are well-known to protect themselves from volatility and explosiveness in fuel prices. Furthermore, our results indicate the importance of attending when negative shocks appear, where all refined products would be located under the supply-driven market integration. On the other hand, conventional gasoline joints to kerosene are also placed on the side of demand-driven market integration. In this respect, it catches the eye that, in contrast to linear effects where conventional gasoline is only situated under the demand-driven market integration when the negative effects make their appearance conventional gasoline prices become fully integrated.

The rest of the paper is organized as follows. The following “Literature review” section reviews the existing literature on this topic. “Data and methodology” section describes the methodology applied to this topic. “Results” section details and displays the results, and “Conclusions” section identifies the conclusions and policy implications.

Literature review

Nowadays, due to globalization process and the free market economy, the prices of refined products are projected to adjust with the change in the crude oil price. Nonetheless, some studies evidence that prices of refined products may react faster and on a greater scale to the increase of crude oil price, while the price of the former plunges at a slower rate and slighter size in reaction to drop in crude oil prices. Therefore, the relationship and the integration between crude oil price and energy (or its refined products) price is essential to disentangle the comportment of the price formation.Footnote 3

Indeed, following Ederington et al. (2019), the literature has dealt with this topic in two ways. The first strand argues that the causal relation flows from crude oil prices to refined product prices (see Frey and Manera 2007) for a survey). Furthermore, Bacon (1991) explain an influential paper where the Rockets and Feathers Hypothesis is demonstrated. He showed that when prices of crude oil rise, this causes an increment of gasoline prices (or other refined products) in a faster way. Conversely, as previously mentioned, the drop in crude oil prices does not provoke a faster response to gasoline prices. Indeed, the influential paper of Borenstein et al. (1997) shows that when crude oil prices fluctuate, i.e., increase or decrease, the response of the gasoline prices is asymmetric. Gjolberg and Johnsen (1999) used the Error Correction Model (ECM hereafter) to examine the relationship between crude oil monthly prices and refined product prices. They confirmed that any crude oil price deviation from equilibrium could give information on how the refined products behaviour could be in the short-run. Adrangi et al. (2001) employ VAR methodology and a bivariate GARCH model to prove that movements in crude oil prices cause shifts in the price of the refined product used, i.e., the Los Angeles diesel. This can also happen in the opposite direction due to demand inferences and Venditti (2013) evidence proofs on the rockets and feathers issue by using weekly data on gasoline and gasoil prices for the U.S., the euro area, by using nonlinear impulse response functions and forecast accuracy tests and displaying evidence of asymmetries in the U.S.

Another influential paper, as Bachmeier and Griffin (2003) apply the basic and asymmetric ECM, could not evidence a pass-through response from crude oil prices to gasoline prices. Hammoudeh et al. (2003) apply a cointegration, an error-correction representation, and GARCH models to crude oil and gasoline daily prices, suggesting a bidirectional causal relationship between daily crude oil and gasoline prices. They also evidence a unidirectional causality from crude oil price to heating oil. Additionally, Yang and Ye (2008) developed a simple search model to explain the performance of refined products prices when a positive (or negative) shock affects crude oil price. Honarvar (2009) explains the behaviour of gasoline and crude oil prices for the USA by using a Crouching ECM, which describes the hidden cointegration. He demonstrates that the behaviour of the forms depends on the fluctuations of the crude oil price, being conditioned by the origin of the shocks, i.e., if there is a shock in the crude oil price, it could be due to market conditions. In contrast, if the shock occurs in the refined product price, the origin may be induced by “efficiency or consumers' incentives”, i.e., the demand side. Later, Kaufmann and Laskowski (2005) use monthly data and ECM to study the possible existence of asymmetries between the crude oil and refined petroleum products prices in the United States, evidencing that the asymmetric relation between crude oil and motor gasoline is explained by refinery utilization rates and inventory behavior. Additionally, the asymmetric relation between crude oil and heating oil maybe is produced by contractual arrangements between retailers and consumers. Indeed, these results would suggest that price asymmetries might be caused by efficient markets.

For its part, Karali and Ramirez (2014) examine the volatility and spillover effects in some energy markets such as crude oil, heating oil, and natural gas, by using daily prices in a GARCH model and incorporating changes in important macroeconomic variables and major political and weather-related events. They obtain the presence of asymmetric effects in crude oil following major political, financial, and natural events. Furthermore, seasonality and day-of-the-week effects in the crude oil and heating oil markets are encountered. Atil et al. (2014) assessed the pass-through from crude oil prices into gasoline and natural gas prices when the non-linear autoregressive distributed lags (NARDL) model is applied. So, if a shock in the crude oil price transfers in a non-linear manner to gasoline and natural gas prices and shows that energy products prices dynamics possess a complex relationship. For its part, Han et al. (2015) uses a battery of techniques such as VAR and ECM, Granger causality and impulse-response functions, finding bidirectional causality between crude oil and gasoline prices when facing a supply shock. They show that the interaction between crude oil price and gasoline price possess long-live. Kpodar and Abdallah (2017) also apply the VAR model and impulse-response function to study the price pass-through between retail gasoline prices and crude oil prices for a large dataset that encompasses 162 countries. Their results point out that positive shocks on crude oil prices possess a larger impact than negative shocks on gasoline prices response, and Martinez et al. (2018) apply a novel approach, the wavelet local multiple correlation (WLMC), to evaluate the relationship between oil time series in the time-scale domain. In fact, they use the prices of seven commodities: crude oil (WTI) and six refined products (conventional gasoline, RBOB regular gasoline, heating oil, Ultra-Low-Sulphur diesel fuel, kerosene and propane). They find that the wavelet correlations are strong throughout the period studied and show a strong decline in correlation values from 2013 to 2015 (due to the overproduction of tight oil in the U.S. and a slowdown in global demand for oil).

More recently, Martin-Moreno et al. (2019a, b) applied a Markov-switching model and used it to contrast the robustness of the results obtained with the TAR-ECM methodology to Spanish and German, French, Italian and UK data, respectively, getting significative evidence of asymmetric adjustments of gasoline and diesel prices to variations in crude oil price. Indeed, if changes in retail prices are translated by increases or decreases in input prices thus, the standard economic theory states that the market is perfectly competitive (Martin-Moreno et al. 2019a). Furthermore, Kyritsis and Andersson (2019) explored the causal relationship between crude oil and its refined products. They compared different causality tests and indicated a unidirectional causality relation from crude oil price returns to refined products price returns. Remarkably, they show a bidirectional causal relationship between heating and crude oil price returns and between natural gas and crude oil price returns, respectively. It is worth noting that they interpret causality as a predictability way instead of studying structural economic relations. Finally, Gosinska et al. (2020) examined the European wholesale market, analyzing the relationship between Brent crude oil and unleaded 95 petrol and diesel using a threshold cointegration, concluding that the pass-through of crude oil price to the retail prices is asymmetric. Olayungbo and Ojeyinka (2021) show that an increase in crude oil prices turns out a stronger effect on refined product prices than the effect of a decrease in crude oil prices. This could be indicator of asymmetries in Nigerian crude oil market and further convincing evidence in favour of the Rocket and Feather hypothesis in the pass-through between crude oil and refined product prices. More recently, Bakhat et al. (2022) provide new evidence on asymmetric price transmissions, by using a NARDL model. They pay attention to the asymmetric response of both diesel and gasoline weekly prices to oil price shocks, particularly not just to the magnitude but also to the duration of the effect, offering an understanding of how diesel and gasoline markets operate and how these markets are efficient.

The second strand explores the relationship between refined product prices to crude oil prices. It is worth noting that this link has received limited attention but increasing lately. In this respect, it is worth noting Verleger (1982), who states that refined product prices are the primary determinant of crude oil prices, i.e., this is known as the Verleger hypothesis. Indeed, Verleger (2011) exposes that refiners could reduce their production if they do not find a price for crude oil which matches with refined product prices, triggering a decline in the crude oil demand, reducing the price of crude oil and the subsequent fall in the supply of refined products, which would increase their prices, that is, shifts in demand for a determined refined product could have predictive power for the crude oil price (Baumeister et al. 2018). For its part, Kaufmann et al. (2009) apply an ECM to show that shifts in prices of refined products do not affect crude oil prices. Hence, changes in the demand for a given refined product have effects on the product mix of refined products and thus, for the demand for crude oil, arising a situation in which excess demand may result in rising prices (Alquist et al. 2013). Furthermore, Bilgin and Ellwander (2019) evidenced that changes in global demand for fuel explained most of the fluctuations in crude oil prices when a structural VAR model is applied. Finally, Vides et al. (2021) employ the fractionally cointegrated VAR model, showing that the Verleger hypothesis is rejected for all refined products and Ederington et al. (2021) finds that spot crude oil prices Granger-cause spot gasoline and actual spot heating oil prices.

Data and methodology

Our econometric strategy implicates obtaining and analysing the model estimation at a weekly frequency, assessing the long-run relationship. Hence, we perform statistical tests for causality and the possibility of asymmetries in those possible causal relations based on the fundamental equation for the bivariate relationship between crude oil price and each of its refined products in an econometric context.


In this article, we focus on the US oil market, considering the relationship between the spot price of West Texas Intermediate (WTI) crude oil and its refined products,Footnote 4 named conventional gasoline NY, Reformulated Gasoline Blendstock for Oxygen Blending (RBOBFootnote 5 hereafter), Ultra-Low-Sulphur Diesel NY (diesel hereafter) and kerosene. According to the US Energy Information Administration (EIA), the refinement process converts approximately 49% of crude oil barrels into gasoline, 25% into distillate fuel oil, i.e., diesel and heating oil, 9% in kerosene and the rest into other products. These proportions may be used as a preliminary notion of affecting the refining process to the price formation. All data were obtained from the US Energy Information Administration (EIA),Footnote 6 and each series covers a period spanning from June 16th 2006 to August 12th 2022, collecting 844 observations weekly.Footnote 7 Regarding the dynamics of the data, Fig. 1 plots the behaviour of the prices over time. It is apparent that crude oil and its refined products prices display a similar pattern, explained by these products directly extracted from crude oil. As we can see, those prices (both crude oil prices and refined products) follow a similar trend, which could suggest the existence of a long-run relationship. As previously mentioned in the literature review section, this view has been generally treated. It exposes that a variation in refined products prices may be driven by a change in the crude oil price (Ederington et al. 2019).

Fig. 1
figure 1

Crude oil and its refined products’ price dynamics. Notes Left-axis refers to dollars per Gallon for refined products and the right-axis refers to dollars per barrel for WTI crude oil


Traditionally, this topic has been studied by using the linear Granger causality (Granger 1969). The purpose of this work is to check the possible existence (or not) of Granger causality relationships between the WTI crude oil and its refined products by using a set of econometric procedures to achieve more robust and comparable results, from asymmetric point of view. Firstly, we apply the methodology suggested by Hatemi-J (2012), which expands the model developed by Toda and Yamamoto (1995) to capture the behaviour of the whole US fuel market. The Toda and Yamamoto (1995) method and the Hatemi-J (2012) method focus on performing an asymmetric analysis that lets us examine the origin of the price formation, i.e., if the price is given by the demand or, otherwise, the supply. Those latter techniques are an extension of Toda and Yamamoto (1995).

Granger causality, by product: Toda–Yamamoto test

In the literature, one of the most common methods for testing the causality effects between different variables is the Granger causality method based on the estimation of VAR models. The method developed by Toda and Yamamoto (1995) measures causality by solving problems derived from cointegrating relationships and non-stationary series. Delving into the relationship suggested, we follow the Toda–Yamamoto causality approach as an enlarged form of the Granger causality test based on augmented-VAR models in levels and extra lags, providing more efficient and robust results than the standard VAR model because it may provide biased results with finite samples (see Johansen and Juselius 1990; Zapata and Rambaldi 1997; Maddala and Kim 1998; Pesaran et al. 2001; Clarke and Mirza 2006). The core advantage of this test is the possibility of being applied regardless of whether the series are or are not cointegrated. In the case of cointegration, the order of integration is not crucial. In this exercise, a bivariate model including the West Texas Intermediate crude oil and each refined product price are examined. Thus, we can describe the benchmark model for this test:

$$WTI_{t} = \alpha_{1} + \mathop \sum \limits_{i = 1}^{{l + d_{max} }} \beta_{1i} WTI_{t - i} + \mathop \sum \limits_{j = 1}^{{l + d_{max} }} \gamma_{1j} refined\;product_{t - j} + \varepsilon_{1t}$$
$$refined\;products_{t} = \alpha_{2} + \mathop \sum \limits_{i = 1}^{{l + d_{max} }} \beta_{2i} \;refined\;product_{t - i} + \mathop \sum \limits_{j = 1}^{{l + d_{max} }} \gamma_{2j} WTI_{t - j} + \varepsilon_{2t}$$

where \(l\) is the optimal lag structure for the VAR model according to the Schwarz Information Criterion (SIC); \(d_{max}\), extra lagged explanatory variables, corresponds to the maximum order of integration for the variables considered in the model; and the error terms \(\varepsilon_{1t}\) and \(\varepsilon_{2t}\) follow a Gaussian Distribution and are white noise processes. Therefore, this test estimates a \(VAR (l + d_{max}\)) model employing a Modified Wald test (MWALD), which is statistically asymptotically distributed as a \(\chi^{2}\) with p degrees of freedom. Furthermore, \(\alpha_{i}\) is the intercept and \(\beta_{ji}\) and \(\gamma_{ij}\) are the parameters of the model which denote the relationship among the different variables.

To test the Granger causality between the two selected variables, attending to the first equation, if \(\sum\nolimits_{j = 1}^{l} {\gamma_{1j} } \ne 0\), suggests a given \(refined product_{t}\) Granger causes \(WTI_{t}\). Similarly, in the second equation, if \(\sum\nolimits_{j = 1}^{l} {\gamma_{2j} } \ne 0\), \(WTI_{t}\) Granger causes a given \(refined\;product_{t}\). Subsequently, if both hypotheses are rejected, this implies that there may exist a bi-directional causality in the examined relationship.

Looking for asymmetric causality relationships

Following the empirical literature, causality is often rejected due non-linear relationships are not contemplated. For this, we apply a non-linear test developed by Hatemi-J (2012) based on the indications of Granger and Yoon (2002), enabling us to determine if cumulative positive and negative shocks can have different impacts on the causal relationship between WTI and the refined products. Ensuing, we begin denoting our two variables through a random walk model:

$$WTI_{t} = WTI_{t - 1} + \varepsilon_{1t} = WTI_{0} + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{1i}$$


$$refined\;product_{t} = refined\;product_{t - 1} + \varepsilon_{2t} = refined\;product_{0} + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{2i}$$

where \(t = 1, 2, \ldots T\), the constants \(WTI_{0}\) and each \(refined\;product_{0}\) are the primary constant values, and the variables \(\varepsilon_{1i}\) and \(\varepsilon_{2i}\) are white noise error terms. The maximum and minimum initial constant values of both the positive and negative shocks are identified as:

\(\varepsilon_{1i}^{ + } = \max (\varepsilon_{1i} , 0) {\text{and}} \varepsilon_{2i}^{ + } = \max (\varepsilon_{2i} , 0) {\text{and}} \varepsilon_{1i}^{ - } = \min (\varepsilon_{1i} , 0) {\text{and}} \varepsilon_{2i}^{ - } = \min (\varepsilon_{2i} , 0),\) respectively. Clustering these terms as follow, \(\varepsilon_{1i} = \varepsilon_{1i}^{ + }\) + \(\varepsilon_{1i}^{ - }\) and \(\varepsilon_{2i} = \varepsilon_{2i}^{ + }\) + \(\varepsilon_{2i}^{ - }\), we reach:

$$WTI_{t} = WTI_{t - 1} + \varepsilon_{1t} = WTI_{0} + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{1i}^{ + } + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{1i}^{ - }$$
$$refined\;product_{t} = refined\;product_{t - 1} + \varepsilon_{2t} = refined\;product_{0} + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{2i}^{ + } + \mathop \sum \limits_{i = 1}^{t} \varepsilon_{2i}^{ - } .$$

Therefore, positive and negative shocks can be written as follows:

$$WTI_{t}^{ + } = \mathop \sum \limits_{i = 1}^{t} \varepsilon_{1i}^{ + } ; WTI_{t}^{ - } = \mathop \sum \limits_{i = 1}^{t} \varepsilon_{1i}^{ - } ;\;refined\;product_{t}^{ + } = \mathop \sum \limits_{i = 1}^{t} \varepsilon_{2i}^{ + } ;\;refined\;product_{t}^{ - } = \mathop \sum \limits_{i = 1}^{t} \varepsilon_{2i}^{ - } .$$

Assuming that:

\(y_{t}^{ + } = \left( {WTI_{t}^{ + } ,refined\;product_{t}^{ + } } \right),\;y_{t}^{ - } = \left( {WTI_{t}^{ - } ,\;refined\;product_{t}^{ - } } \right),\;y_{t}^{ \pm } = \left( {WTI_{t}^{ + } ,\;refined\;product_{1t}^{ - } } \right),\) and

$$y_{t}^{ \mp } = \left( {WTI_{t}^{ - } , refined product_{1}^{ + } } \right),$$

the causal relationship between the variables could be examined using a VAR model of order \(p\)¸ for lag order \(r = \left( {1, \ldots , p} \right)\). To perform a Wald test, the VAR (p) model can be expressed in a compressed form (for the first combination, \(y_{t}^{ + }\), for instance),

$$\begin{aligned} & Y = DZ + \delta ,\;{\text{where}} \\ & Y: = \left( {y_{1}^{ + } , \ldots , y_{T}^{ + } } \right) \left( {n\;x\;T} \right) matrix, \\ & D: = \left( { \upsilon , A_{1} \ldots , A_{p} } \right) \left( {n\;x\;\left( {1 + np} \right)} \right) matrix, \\ & Z_{t} : = \left( {\begin{array}{*{20}c} 1 \\ {y_{t}^{ + } } \\ {y_{{\begin{array}{*{20}c} {t - 1} \\ \vdots \\ {y_{t - p + 1}^{ + } } \\ \end{array} }}^{ + } } \\ \end{array} } \right) \left( {\left( {1 + np} \right)\;x\;1} \right)matrix,\;{\text{for}}\;t = 1, \ldots ,T, \\ & Z: = \left( {Z_{0} , \ldots , Z_{T - 1} } \right) \left( {\left( {1 + np} \right)\;x\;T} \right)matrix,\;{\text{and}} \\ & \delta : = \left( {u_{1}^{ + } , \ldots ,u_{T}^{ + } } \right)\left( {n\;x\;T} \right)\;matrix. \\ \end{aligned}$$

The Wald statistic is \(\left( {C\beta } \right)^{\prime } \left[ {C\left( {\left( {Z^{\prime } Z} \right)^{ - 1} \otimes S_{U} } \right)C^{\prime } } \right]^{ - 1} \left( {C\beta } \right),\) where \(\beta = vec\left( D \right),\) being vec(·) the column-stacking operator; \(\otimes\) is the Kronecker product; C is a \(p x n\left( {1 + np} \right)\) indicator matrix with elements for restricted parameters and zeros for the rest of the parameters; and \(S_{U} = \frac{{\hat{\delta }_{U}^{\prime } \hat{\delta }_{U} }}{T - q}\), q is the number of parameters in each equation of the VAR model. Under the assumption of normality, the Wald statistic follows an asymptotic \(\chi^{2}\) distribution with equal degrees of freedom of the number of constraints to be tested (in this case, equal to p). The null hypothesis of non-Granger causality, \(H_{0} :C\beta = 0\), is rejected at the \(\alpha\) level of significance (1%, 5% or 10%) according to the bootstrap critical values generated by GAUSS software.


According to the econometric strategy above-described, in this section, we show the results to investigate the Granger causality relationships not only product-by-product but also by allowing asymmetries between the variables WTI and the different refined products and the inverse. In this sense, we apply the method of Hatemi-J, which is an extension of Toda and Yamamoto’s work, for a product versus product study that also permits the study of asymmetries. Depending on the hypotheses to be tested, the results of these methods are described and differentiated by the direction of the causality. In this sense, Table 1 shows the empirical hypotheses coined for our goal and additionally, the estimated results are presented in Tables 2, 3 and 4, while Table 5 shows a summary of the main results. Prior to conducting our analysis, we perform four different tests: The augmented Dickey–Fuller (ADF) test offered by Dickey and Fuller (1979), the Phillips–Perron (PP) test offered by Phillips and Perron (1988), the Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test offered by Kwiatkowski et al. (1992) and the Ng-Perron test developed by Ng and Perron (2001). We test for the stationarity of the series by means of the conventional ADF, PP, KPSS and Ng-Perron unit root tests. The unit root test results are presented in Table 7.

Table 1 Empirical hypotheses
Table 2 Linear effects
Table 3 Positive shocks on market integration
Table 4 Negative shocks on market integration
Table 5 Summary of results

Table 7 shows mixed results on whether the petro-derivatives prices follow a stationary process or not. Evidence shows that at a 1% significance level, the null hypothesis of a unit root cannot be rejected at levels in almost all the variables, while it is soundly rejected at first-differences.

Our first approach is shown in Table 2, which reveals that causality runs from WTI crude oil prices to all refined products prices except for conventional gasoline. In contrast, the inverse relationship shows causality from conventional gasoline, RBOB and kerosene to WTI. In general, as an upstream commodity, the price of crude oil is a major driving force for refined product prices. Thus, a significant movement in the price of crude oil is usually accompanied by a substantial movement in the prices of refined products. Conversely, extreme increases in refined product prices due to a demand shock can ultimately push crude oil prices higher (Tong et al. 2013). These results evidence the divergence between supply-driven and demand-driven market integration (Asche et al. 2003). In our case, our results suggest that all refined products except conventional gasoline could follow a supply-driven market integration, and for its part, conventional gasoline would follow a demand-driven market integration. Furthermore, we note that, at first sight, RBOB and kerosene would be included in both types of market integration, being interpreted as a feedback mechanism or a full integration, i.e., both prices would be influenced by supply and demand. These results are in line with those obtained by Martinez et al. (2018).

However, the nature of the previous results shown in Table 2 may shed more light when assessing the existence of asymmetries between crude oil and each refined product prices. In other words, we will analyze how positive or negative shocks could influence the causal relationship between our variables, being this idea the main contribution of this paper. In this sense, when the analysis is executed by contemplating the asymmetries, these results vary. In Table 3, the results of the positive effects show that a positive WTI shock causes positive shocks in kerosene. In this regard, these cases show that the price formation of that refined products is affected or caused by changes in crude oil prices. This corresponds to the so-called supply-driven market integration. At the same time, a positive shock on RBOB causes a positive shock on crude oil, denoting a demand-driven market integration. When the shock is positive, the significance of conventional gasoline, diesel or kerosene has disappeared, suggesting the importance of the study of the asymmetries existence. These results imply that external aspects such as the role of the valued chain in the price formation process or the information content of futures markets for these products.

Conversely, Table 4 shows the negative effects of these relationships. These results argue that a dropping price of WTI would cause a fall in price to each refined product. This result, in contrast to those obtained in Table 3, evidence a fall in WTI crude oil prices may cause the prices of refined products to drop, being strongly integrated when a negative shock appears. Thus, a negative shock on conventional gasoline and kerosene prices would cause a drop in WTI crude oil prices, being evidence for the fulfilment of Verleger hypothesis when a negative shock is given. Following Baumeister et al. (2018), this negative shock on conventional gasoline and kerosene prices will negatively affect the price of crude oil, suggesting that the difference between, in this case, both refined product prices and the purchase price of crude oil should have predictive power for the price of crude oil. The case of conventional gasoline must be highlighted. As we could see in Table 2, conventional gasoline appeared on the demand-driven side, but when asymmetries are considered, the conventional gasoline behaviour has changed, becoming a fully integrated refined product when the shock is negative.

The following Table 5 summarizes the results previously shown,Footnote 8.Footnote 9 First, the causalities of the overall effects are also found when asymmetries are allowed. In addition, these results show that the causalities differ depending on the nature of the shock in which the relationship is observed, demonstrating that an analysis that only contemplates the overall effects could be biased.

In this regard, we can see a mix of results, following a heterogeneous pattern and revealing the need to study the asymmetries regarding the causal relationships given. In this sense, our results suggest the dependent role of crude oil, represented in each type of integration, i.e., supply-driven and demand-driven market integration, when assessing the effects of linear relation or positive and negative shocks. Furthermore, our results indicate that all refined products would be under supply-driven market integration when negative shocks occur. On the other, conventional gasoline joints to kerosene are also placed on the side of demand-driven market integration. In this respect, it catches the eye that, in contrast to linear effects where conventional gasoline is only situated under the demand-driven market integration when the negative effects make their appearance, conventional gasoline prices become fully integrated.

Finally, concerning the existence of asymmetries, our results align with those of Karali and Ramirez (2014). Thereby, the crude oil effect on refined products is more significant and more persistent during bearish markets than bullish ones. These findings provide some interesting information for policymakers and practitioners, being important to consider policies in favour of consuming or producing one product over another, depending on the market conditions at a given moment. Additionally, we know that effective competition in the market relies heavily on the information available to concerned parties. Consequently, government policies designed for informing about the different prices in real time may be a good measure to influence price asymmetries.


The paper addressed to analyze the causality relationship is a set of alternative tests that reveals Granger causality considering asymmetries in the relationships, following the approaches by Toda and Yamamoto (1995) and Hatemi-J (2012). This paper provides evidence to previous empirical studies by analyzing the asymmetric behaviour of the relationship between crude oil price and the price of its refined products, which play a pivotal role in the global economy and remains at the core of energy markets (Fazelabdolabadi 2019). Furthermore, as previously mentioned in the literature review section, the shocks in different economic sectors caused by ‘Ukraine–Russian war’ and, significantly, the COVID-19 pandemic and the measures applied by governments have impacted crude oil prices differently and refined products. Thus, we have to highlight that the period analyzed covers several financially risky economic events, such as the global financial crisis and rises in tight oil production, and the abovementioned recent geopolitical events.

As Karagiannis et al. (2015) stated, symmetry in crude oil and refined products prices might be viewed because of an effective regulatory policy of the energy sector in an economy. Indeed, the remarkable nature of our results let us identify heterogeneity in the pattern of refined products concerning the causality versus WTI, revealing the opportuneness of studying the asymmetries regarding the causal relationships given. In line with Tiwari et al. (2021), the presence of asymmetries could be explained by the different arrangements between every link in the chain in the energy markets, provoking an extra payment of costs. On this point, the dependent role to crude oil of kerosene is represented in each type of market integration, i.e., it appears as a result in the supply-driven and the demand-driven market integration, when measuring the effects of linear relation or positive and negative shocks, which makes sense. This aspect would convey us to explore the treatment of this refined product, which is essential to the airlines and the travel industry. In this sense, the hedging strategies followed by airlines are well-known to protect themselves from volatility and explosiveness in fuel prices. Contrary to the efficient market hypothesis, investors could protect themselves and their portfolios by forecasting the prices due to the asymmetries, providing a convenient way to reduce portfolio risk (see Tang and Xiong 2012).

Furthermore, how changes in the crude oil price affect refined product prices is known as the pass-through. People’s main sense is that the pass-through is not symmetric: The speed at which the price of any refined product changes is different depending on if the price of each refined product is relatively high or relatively low compared to the price of oil (Owyang and Vermann 2014). Thus, our results suggest the importance of attending when negative shocks appear, where all refined products would be located under the supply-driven market integration (this is evidence of rejection of the Rockets and Feathers hypothesis). On the other, conventional gasoline joints to kerosene are also placed on demand-driven market integration, implying that these products would be the driving factor in the petro-derivatives price-generating process and could be key for refiners, which may use this information for designing refining strategies and controlling product stocks. Additionally, in contrast to linear effects where conventional gasoline is only situated under the demand-driven market integration when the negative effects appear, conventional gasoline prices become fully integrated.

Thus, given the results obtained, it is well-known that there exist two price transmission mechanisms between crude oil and its refined product prices in the economic sense. An increase in crude oil price fosters refining costs, raising the refined product prices. Therefore, if the refined products prices rise, refiners would be able to offer a more considerable volume of products. Subsequently, the demand for crude oil would grow, triggering a higher price. Due to the presence of asymmetries and the impact of different shocks on crude oil price and the prices of the refined products, there is a high degree of uncertainty in prices. In this regard, the US regulatory authorities and policymakers should implement and carry out a national energy outlook in oil markets, by using effective monitoring tools to avoid adverse effects of energy price shocks on consumers and, consequently, for the overall welfare of the economy. Also, the use of those tools provide to policymakers the possibility of minimizing oil price risk in different oil-dependent sectors, such as transportation, heating, or agriculture, by diversification and elaboration energy mixes with a view to a clean energy system. Furthermore, it would be advisable the implementation of fiscal and monetary stimulus packages to alleviate the aftermath of COVID-19 on commodity markets (Polat 2020). Additionally, in these times of high levels of inflation, it seems necessary to monitor the sign of the shock that may be a valuable tool for economic authorities to prevent the inflationist tensions that might occur and affect the transmission channels to the rest of the economy, affecting the rest of the products and altering the whole economy (Barros et al. 2011).

Lastly, we have revealed that the existence of asymmetries could be translated into a more predictability and, subsequently, a weaker form of market efficiency. This provides valued information to market participants regarding the behaviour of the WTI and refined products prices. Moreover, the consideration of asymmetries would response how the oil market reacts when a movement of its inputs occurs, i.e., whether a given refined product price moves symmetrically o asymmetrically when crude oil moves (Perdiguero-García 2013).

Likewise, our research findings have practical applications for developing a more effective approach focused on time series, especially when dealing with market participants who have varying temporal horizons, such as investors, speculators, hedge funds, and refineries. The energy sector, economists, policymakers, and investors require a deeper understanding of the relationship between crude oil and its refined products, particularly in light of the global economic crisis and the surge in tight oil production. Therefore, our results contribute to improving the comprehension of energy market behavior. This latter idea can be complemented by the fact that the energy market is changing over time, requiring that the different actors establish new positions because governments and public opinion demand heavier regulations and greater environmental commitment (Kirikkaleli and Güngör 2021). Thus, from an environmental point of view, the energy markets (the petro-derivatives market in particular) have a growing impact on the carbon emission problem due to the use of fossil fuels in transport vehicles. In this sense, Kou et al. (2022) argue that one plausible solution to mitigate and fight this problem is the use of electric vehicles, especially in road transport, being these electric vehicles would be solar-energy-type. Furthermore, Anser et al. (2021) identified this issue, i.e., carbon emissions, as one of the main problems that must be reduced so that economies can improve in the long term.

Finally, future research could be oriented to the study of the relationship between WTI crude oil and the refined products, by using other methodologies which also could capture the nature of the relations in a different way (thresholds, spatial analysis, among others), or by using the Brent crude oil as a benchmark. Additionally, large-scale GDM and fuzzy cluster analysis (Li et al. 2021) or an MCDM-based approach (Kou et al. 2014) might be suitable when assessing refiner’s or investors’ preferences, attending to spot and future prices or the financial risk associated.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.


  1. According to Sovacool et al. (2020), by the end of April 2020, more than half of the entire world's population (54%) was under some form of COVID-19 lockdown.

  2. See Hamilton (2011) for a survey of the nonlinear relationship between macroeconomic effects and oil prices.

  3. See Table 8 for a selected review of empirical studies.

  4. Crude Oil are valued in Dollars per barrel and the refined products in Dollars per gallon, which are converted to dollars per barrel by using 1 barrel = 42 US gallons, considering the US production.

  5. RBOB is heavily dependent on crude oil, as it is a refined product of the commodity.


  7. A table with descriptive statistics is included in Table 6 in the “Appendix”.

  8. The same analysis has been carried out for the sample before COVID-19, and the results are robust with those obtained in the present analysis.

  9. We also have included the positive and negative cumulative sums (Figs. 2, 3) in the “Appendix” section.



Coronavirus disease 2019


Derivatives products


Error Correction Model


New York


Organization of the Petroleum Exporting Countries


Reformulated Gasoline Blendstock for Oxygen Blending


United States of America


Vector autoregression


West Texas Intermediate


  • Adrangi B, Chatrath A, Raffiee K, Ripple RD (2001) Alaska North Slope crude oil price and the behavior of diesel prices in California. Energy Economics 23(1):29–42

    Article  Google Scholar 

  • Ahundjanov BB, Akhundjanov SB, Okhunjanov BB (2020) Information search and financial markets under COVID-19. Entropy 22(7):791

    Article  Google Scholar 

  • Alquist R, Kilian L, Vigfusson RJ (2013) Forecasting the price of oil. In: Handbook of economic forecasting, vol 2, pp 427–507. Elsevier

  • Anser MK, Khan MA, Zaman K, Nassani AA, Askar SE, Abro MMQ, Kabbani A (2021) Financial development during COVID-19 pandemic: the role of coronavirus testing and functional labs. Financ Innov 7:1–13

    Article  Google Scholar 

  • Asche F, Gjølberg O, Völker T (2003) Price relationships in the petroleum market: an analysis of crude oil and refined product prices. Energy Econ 25(3):289–301

    Article  Google Scholar 

  • Atil A, Lahiani A, Nguyen DK (2014) Asymmetric and non-linear pass-through of crude oil prices to gasoline and natural gas prices. Energy Policy 65:567–573

    Article  Google Scholar 

  • Bachmeier LJ, Griffin JM (2003) New evidence on asymmetric gasoline price responses. Rev Econ Stat 85(3):772–776

    Article  Google Scholar 

  • Bacon RW (1991) Rockets and feathers: the asymmetric speed of adjustment of UK retail gasoline prices to cost changes. Energy Econ 13(3):211–218

    Article  Google Scholar 

  • Bakas D, Triantafyllou A (2020) Commodity price volatility and the economic uncertainty of pandemics. Econ Lett 193:109283

    Article  Google Scholar 

  • Bakhat M, Rosselló J, Sansó A (2022) Price transmission between oil and gasoline and diesel: a new measure for evaluating time asymmetries. Energy Econ 106:105766

    Article  Google Scholar 

  • Bal DP, Rath BN (2015) Non-linear causality between crude oil price and exchange rate: a comparative study of China and India. Energy Econ 51:149–156

    Article  Google Scholar 

  • Barros CP, Gil-Alana LA, Payne JE (2011) An analysis of oil production by OPEC countries: persistence, breaks, and outliers. Energy Policy 39(1):442–453

    Article  Google Scholar 

  • Baumeister C, Kilian L, Zhou X (2018) Are product spreads useful for forecasting oil prices? An empirical evaluation of the Verleger hypothesis. Macroecon Dyn 22(3):562–580

    Article  Google Scholar 

  • Bilgin D, Ellwanger R (2019) The simple economics of global fuel consumption (No. 2019-35). Bank of Canada Staff Working Paper

  • Billio M, Varotto S (2020) A new world post COVID-19: lessons for business, the finance industry and policy makers. Venezia Edizioni Ca’Foscari-Digital Publishing

    Book  Google Scholar 

  • Borenstein S, Cameron AC, Gilbert R (1997) Do gasoline prices respond asymmetrically to crude oil price changes? Q J Econ 112(1):305–339

    Article  Google Scholar 

  • Brahmasrene T, Huang JC, Sissoko Y (2014) Crude oil prices and exchange rates: causality, variance decomposition and impulse response. Energy Econ 44:407–412

    Article  Google Scholar 

  • Bremmer DS, Kesselring RG (2016) The relationship between US retail gasoline and crude oil prices during the Great Recession: “rockets and feathers” or “balloons and rocks” behavior? Energy Econ 55:200–210

    Article  Google Scholar 

  • Brown SP, Yucel MK (2008) What drives natural gas prices? Energy J 29(2)

  • Caro JMB, Golpe AA, Iglesias J, Vides JC (2020) A new way of measuring the WTI–Brent spread. Globalisation, shock persistence and common trends. Energy Econ 85:104546

    Article  Google Scholar 

  • Clarke JA, Mirza S (2006) A comparison of some common methods for detecting Granger noncausality. J Stat Comput Simul 76(3):207–231

    Article  Google Scholar 

  • Dickey D, Fuller WA (1979) Distribution of the estimators for autoregressive time series with a unit root. J Am Stat Assoc 74(366a):427–431

    Article  Google Scholar 

  • Ding H, Kim HG, Park SY (2016) Crude oil and stock markets: Causal relationships in tails? Energy Econ 59:58–69

    Article  Google Scholar 

  • Douglas CC (2010) Do gasoline prices exhibit asymmetry? Not usually! Energy Econ 32(4):918–925

    Article  Google Scholar 

  • Ederington LH, Fernando CS, Hoelscher SA, Lee TK, Linn SC (2019) A review of the evidence on the relation between crude oil prices and petroleum product prices. J Commod Mark 13:1–15

    Article  Google Scholar 

  • Ederington LH, Fernando CS, Lee TK, Linn SC, Zhang H (2021) The relation between petroleum product prices and crude oil prices. Energy Econ 94:105079

    Article  Google Scholar 

  • Fazelabdolabadi B (2019) A hybrid Bayesian-network proposition for forecasting the crude oil price. Financ Innov 5(1):1–21

    Article  Google Scholar 

  • Fousekis P, Grigoriadis V (2017) Price co-movement and the crack spread in the US futures markets. J Commod Mark 7:57–71

    Article  Google Scholar 

  • Frey G, Manera M (2007) Econometric models of asymmetric price transmission. J Econ Surv 21(2):349–415

    Article  Google Scholar 

  • Girma PB, Paulson AS (1999) Risk arbitrage opportunities in petroleum futures spreads. J Futures Mark 19(8):931–955

    Article  Google Scholar 

  • Gjolberg O, Johnsen T (1999) Risk management in the oil industry: Can information on long-run equilibrium prices be utilized? Energy Econ 21(6):517–527

    Article  Google Scholar 

  • Gosińska E, Leszkiewicz-Kędzior K, Welfe A (2020) Who is responsible for asymmetric fuel price adjustments? An application of the threshold cointegrated VAR model. Balt J Econ 20(1):59–73

    Article  Google Scholar 

  • Granger CW (1969) Investigating causal relations by econometric models and cross-spectral methods. Econom J Econom Soc 37:424–438

    Google Scholar 

  • Granger CW, Yoon G (2002) Hidden cointegration. U of California, Economics Working Paper (2002-02)

  • Grasso M, Manera M (2007) Asymmetric error correction models for the oil–gasoline price relationship. Energy Policy 35(1):156–177

    Article  Google Scholar 

  • Hamilton JD (2011) Nonlinearities and the macroeconomic effects of oil prices. Macroecon Dyn 15(S3):364–378

    Article  Google Scholar 

  • Hamilton JD, Wu JC (2014) Risk premia in crude oil futures prices. J Int Money Financ 42:9–37

    Article  Google Scholar 

  • Hammoudeh S, Li H, Jeon B (2003) Causality and volatility spillovers among petroleum prices of WTI, gasoline and heating oil in different locations. N Am J Econ Finance 14(1):89–114

    Article  Google Scholar 

  • Han ZH, Yang S, Chen ML, He LY (2015) Mean spillover effect between crude oil and gasoline markets: an empirical result. Int J Global Energy Issues 38(1–3):49–68

    Article  Google Scholar 

  • Hatemi-J A (2012) Asymmetric causality tests with an application. Empir Econ 43(1):447–456

    Article  Google Scholar 

  • Honarvar A (2009) Asymmetry in retail gasoline and crude oil price movements in the United States: an application of hidden cointegration technique. Energy Econ 31(3):395–402

    Article  Google Scholar 

  • Jefferson M (2020) A crude future? COVID-19s challenges for oil demand, supply and prices. Energy Res Soc Sci 68:101669

    Article  Google Scholar 

  • Johansen S, Juselius K (1990) Maximum likelihood estimation and inference on cointegration with applications to the demand for money. Oxf Bull Econ Stat 52(2):169–210

    Article  Google Scholar 

  • Karagiannis S, Panagopoulos Y, Vlamis P (2015) Are unleaded gasoline and diesel price adjustments symmetric? A comparison of the four largest EU retail fuel markets. Econ Model 48:281–291

    Article  Google Scholar 

  • Karali B, Ramirez OA (2014) Macro determinants of volatility and volatility spillover in energy markets. Energy Econ 46:413–421

    Article  Google Scholar 

  • Kaufmann RK, Laskowski C (2005) Causes for an asymmetric relation between the price of crude oil and refined petroleum products. Energy Policy 33(12):1587–1596

    Article  Google Scholar 

  • Kaufmann RK, Dees S, Mann M (2009) Horizontal and vertical transmissions in the US oil supply chain. Energy Policy 37(2):644–650

    Article  Google Scholar 

  • Kirikkaleli D, Güngör H (2021) Co-movement of commodity price indexes and energy price index: a wavelet coherence approach. Financ Innov 7(1):1–18

    Article  Google Scholar 

  • Kilian L (2008) Exogenous oil supply shocks: How big are they and how much do they matter for the US economy? Rev Econ Stat 90(2):216–240

    Article  Google Scholar 

  • Kilian L (2010) Explaining fluctuations in gasoline prices: a joint model of the global crude oil market and the US retail gasoline market. The Energy Journal 31(2):87–112

    Article  Google Scholar 

  • Kilian L (2014) Oil price shocks: causes and consequences. Annu Rev Resour Econ 6(1):133–154

    Article  Google Scholar 

  • Kilian L, Vigfusson RJ (2011) Are the responses of the US economy asymmetric in energy price increases and decreases? Quant Econ 2(3):419–453

    Article  Google Scholar 

  • Kou G, Peng Y, Wang G (2014) Evaluation of clustering algorithms for financial risk analysis using MCDM methods. Inf Sci 275:1–12

    Article  Google Scholar 

  • Kou G, Yüksel S, Dinçer H (2022) Inventive problem-solving map of innovative carbon emission strategies for solar energy-based transportation investment projects. Appl Energy 311:118680

    Article  Google Scholar 

  • Kpodar K, Abdallah C (2017) Dynamic fuel price pass-through: evidence from a new global retail fuel price database. Energy Econ 66:303–312

    Article  Google Scholar 

  • Kwiatkowski D, Phillips PC, Schmidt P, Shin Y (1992) Testing the null hypothesis of stationarity against the alternative of a unit root: How sure are we that economic time series have a unit root? J Econom 54(1–3):159–178

    Article  Google Scholar 

  • Kyritsis E, Andersson J (2019) Causality in quantiles and dynamic relations in energy markets:(De) tails matter. Energy Policy 133:110933

    Article  Google Scholar 

  • Kyritsis E, Serletis A (2018) The zero lower bound and market spillovers: evidence from the G7 and Norway. Res Int Bus Financ 44:100–123

    Article  Google Scholar 

  • Li F, Huang Z, Zhong J, Albitar K (2020) Do tense geopolitical factors drive crude oil prices? Energies 13(16):4277

    Article  Google Scholar 

  • Li G, Kou G, Peng Y (2021) Heterogeneous large-scale group decision making using fuzzy cluster analysis and its application to emergency response plan selection. IEEE Trans Syst Man Cybern Syst 52(6):3391–3403

    Article  Google Scholar 

  • Maddala GS, Kim I-M (1998) Unit roots, cointegration and structural change. Oxford University Press, Oxford

    Google Scholar 

  • Martín-Moreno JM, Pérez R, Ruiz J (2019a) Evidence about asymmetric price transmission in the main European fuel markets: from TAR-ECM to Markov-switching approach. Empir Econ 56(4):1383–1412

    Article  Google Scholar 

  • Martín-Moreno JM, Pérez R, Ruiz J (2019b) Rockets and feathers behaviour in the Spanish gasoline and diesel market: new evidence. Bull Econ Res 71(4):657–683

    Article  Google Scholar 

  • Martínez JMP, Abadie LM, Fernández-Macho J (2018) A multi-resolution and multivariate analysis of the dynamic relationships between crude oil and petroleum-product prices. Appl Energy 228:1550–1560

    Article  Google Scholar 

  • Masih AM, Masih R (1998) A multivariate cointegrated modelling approach in testing temporal causality between energy consumption, real income and prices with an application to two Asian LDCs. Appl Econ 30(10):1287–1298

    Article  Google Scholar 

  • Mishra AK, Ghate K, Renganathan J, Kennet JJ, Rajderkar NP (2022) Rolling, recursive evolving and asymmetric causality between crude oil and gold prices: evidence from an emerging market. Resour Policy 75:102474

    Article  Google Scholar 

  • Ng S, Perron P (2001) Lag length selection and the construction of unit root tests with good size and power. Econometrica 69(6):1519–1554

    Article  Google Scholar 

  • Olayungbo DO, Ojeyinka TA (2021) Crude oil prices pass-through to retail petroleum product prices in Nigeria: evidence from hidden cointegration approach. Econ Change Restruct 1–22

  • Owyang MT, Vermann EK (2014) Rockets and feathers: Why don’t gasoline prices always move in sync with oil prices? Reg Econ

  • Ozili PK, Arun T (2020) Spillover of COVID-19: impact on the global economy. Available at SSRN 3562570

  • Perdiguero-Garcia J (2013) Symmetric or asymmetric oil prices? A meta-analysis approach. Energy Policy 57:389–397

    Article  Google Scholar 

  • Pesaran MH, Shin Y, Smith RJ (2001) Bounds testing approaches to the analysis of level relationships. J Appl Econom 16:289–326

    Article  Google Scholar 

  • Phillips PC, Perron P (1988) Testing for a unit root in time series regression. Biometrika 75(2):335–346

    Article  Google Scholar 

  • Polat O (2020) Measuring dynamic connectedness networks in energy commodities: evidence from the D-Y and frequency connectedness approaches. OPEC Energy Rev 44(4):404–428

    Article  Google Scholar 

  • Rezitis AN (2015) The relationship between agricultural commodity prices, crude oil prices and US dollar exchange rates: a panel VAR approach and causality analysis. Int Rev Appl Econ 29(3):403–434

    Article  Google Scholar 

  • Salisu AA, Obiora K (2021) COVID-19 pandemic and the crude oil market risk: hedging options with non-energy financial innovations. Financ Innov 7(1):1–19

    Article  Google Scholar 

  • Sharif A, Aloui C, Yarovaya L (2020) COVID-19 pandemic, oil prices, stock market, geopolitical risk and policy uncertainty nexus in the US economy: fresh evidence from the wavelet-based approach. Int Rev Financ Anal 70:101496

    Article  Google Scholar 

  • Sovacool BK, Del Rio DF, Griffiths S (2020) Contextualizing the Covid-19 pandemic for a carbon-constrained world: Insights for sustainability transitions, energy justice, and research methodology. Energy Res Soc Sci 68:101701

    Article  Google Scholar 

  • Tang K, Xiong W (2012) Index investment and the financialization of commodities. Financ Anal J 68(6):54–74

    Article  Google Scholar 

  • Tiwari AK, Suleman MT, Ullah S, Shahbaz M (2021) Analyzing the connectedness between crude oil and petroleum products: evidence from USA. Int J Finance Econ

  • Toda HY, Yamamoto T (1995) Statistical inference in Vector Autoregressions with possibly integrated processes. J Econom 66:225–250

    Article  Google Scholar 

  • Tong B, Wu C, Zhou C (2013) Modeling the co-movements between crude oil and refined petroleum markets. Energy Econ 40:882–897

    Article  Google Scholar 

  • Venditti F (2013) From oil to consumer energy prices: How much asymmetry along the way? Energy Econ 40:468–473

    Article  Google Scholar 

  • Verleger PK (1982) The determinants of official OPEC crude prices. Rev Econ Stat 177–183

  • Verleger PK (2011) The margin, currency, and the price of oil. Bus Econ 46(2):71–82

    Article  Google Scholar 

  • Villar JA, Joutz FL (2006) The relationship between crude oil and natural gas prices. Energy Inf Adm Office Oil Gas 1–43

  • Vides JC, Carmona M, Feria J, Golpe AA (2021) Modeling the United States crack spread: market efficiency, persistence and the Verleger hypothesis. Energy Sources Part B Econ Plan Policy 1–20

  • Wolfe MH, Rosenman R (2014) Bidirectional causality in oil and gas markets. Energy Econ 42:325–331

    Article  Google Scholar 

  • Yang H, Ye L (2008) Search with learning: understanding asymmetric price adjustments. Rand J Econ 39(2):547–564

    Article  Google Scholar 

  • Yu L, Li J, Tang L, Wang S (2015) Linear and non-linear Granger causality investigation between carbon market and crude oil market: a multi-scale approach. Energy Econ 51:300–311

    Article  Google Scholar 

  • Zapata HO, Rambaldi AN (1997) Monte Cario evidence on cointegration and causation. Oxford Bull Econ Stat 59:285–298

    Article  Google Scholar 

  • Zhang YJ, Wei YM (2010) The crude oil market and the gold market: evidence for cointegration, causality and price discovery. Resour Policy 35(3):168–177

    Article  Google Scholar 

  • Zhang F, Narayan PK, Devpura N (2021) Has COVID-19 changed the stock return-oil price predictability pattern? Financ Innov 7(1):1–10

    Article  Google Scholar 

Download references


Not applicable.


Not applicable.

Author information

Authors and Affiliations



JV: Conceptualization, methodology, software, writing—original draft. AG: Investigation, data curation, software, validation, writing—review. JF: Data curation, methodology, writing—review and editing. JM: Conceptualization, visualization, methodology, investigation. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Antonio A. Golpe.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.



See Tables 6, 7, 8 and Figs. 2, 3.

Table 6 Descriptive statistics
Table 7 Unit root tests
Table 8 Summary of the literature review
Fig. 2
figure 2

Positive cumulative sums of the variables. Notes Own elaboration

Fig. 3
figure 3

Negative cumulative sums of the variables. Notes Own elaboration

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Vides, J.C., Feria, J., Golpe, A.A. et al. How do supply or demand shocks affect the US oil market?. Financ Innov 10, 16 (2024).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


JEL Classification