We study the probabilities of predicting fluctuations in oil price returns, based on the information in the expected skewness derived from West Texas Intermediate (WTI) oil price returns for the period from 1859:11 – 2023:04. The data set we have analyzed encompasses the entire period of modernization in the petroleum industry. It began with the dredging into the world’s first petroleum well United States (US) in 1859 in Titusville, Pennsylvania. Expected Skewedness is an indicator of the risk in the oil market since it measures the anticipated asymmetry in the distribution of future returns.

Intuitively, the variation in the expected skewness of the market is likely to be a result of extreme, directed changes in supply and demand (Yu and al., 2022; Salisu et al., 2022a), geopolitical actions and threats (Salisu and al., 2021), geopolitical threats (Salisu and al. 2021) as well as rare disasters and pandemics (Demirer et al. (2018); Bouri et al. 2020; Qin et al. 2020; Salisu et al. 2021) and, in some cases, spillovers from financial markets (Salisu et al., 2022b). 2022b). These changes could alter the risk of returns on oil by triggering a “leverage effect” in the oil market, as first investigated in the work of Geman as well as Shih (2009) as well in more recent studies and by Asai and colleagues. (2019, 2020). The idea behind the leverage effect, developed for the analysis of fluctuations in the stock market through Black (1976), suggests that positive (positive) returns are typically followed by up (downward) revisions to volatility.

Aboura as well as Chevallier (2013) Aboura and Chevallier (2013), however, they argue that volatility could be increased following a rise in the price of oil because a surge in price could be a sign that oil consumers are concerned about the rising cost of oil. In this respect, Demirer et al. (2020) claim that the effect of volatility and an increase in price actually depends upon the type of shocks. While oil supply shocks largely reflect geopolitical developments and country/region-specific surprises, demand shocks reflect unexpected changes in the aggregate, preventive, or speculative demand for oil, which, in turn, are driven by market participants’ expectations of future economic conditions and concerns regarding future supply shortfalls. For instance, positive oil price returns that are brought about by a supply shock would be associated with a subsequent slowdown of economic activity, and the resulting recessionary momentum would likely reduce demand and trading activity in the oil market, thereby leading to a leverage-effect-consistent lowering of volatility. If there is an increase in demand, contrary to this positive oil price returns are associated with expansion of the economy, which could increase the volume of trading and volatility or lower the volatility of markets by reducing macroeconomic uncertainty. If it is the case that a demand shock is triggered by an unexpected increase in demand for protection because market participants are worried about the possibility of supply shortages in the future, or because of speculation, it may have negative consequences, i.e., raising the volatility.

In conclusion, negative (negative) anticipated skewness of oil returns, purely theoretic grounds, may be correlated with a decreasing or growing (increasing or reducing) future oil price returns volatility. Thus, the indication of the relationship between skewness expectations and volatility is only discovered by examining the data. In this way by encompassing the longest possible period of data that can be used to construct an estimate of expected skewness across the time period 1859 to 2023, it is possible to identify a variety of negative and positive oil shocks that are associated with like for instance, the U.S. Civil War, the two World Wars, the West coast gas famine, the Great Depression, the Korean conflict as well as The Suez crisis, Suez Crisis, the OPEC oil embargo as well as the Iranian revolution as well as The Iran-Iraq War, the Gulf War and worldwide financial crises, emergence of the Coronavirus pandemic in 2020 and of course in the last few months, this continuing Russia-Ukraine War. 1 Additionally, by studying the predictability of volatility in oil prices due to the risk of oil that is correlated with expected skewness, we can avoid the problem of bias in the selection of samples and provide a complete solution to the query regarding the nature of the relationship between expected skewness and market volatility.

To accomplish our goal, we forecast both out-of-sample and in-sample the real-time the volatility ( RV) of the oil price return using expected skewness using a quantiles-based predictive regression model. In this context it is worthwhile to emphasize that we use an RV, that in our instance is measured in the form of the square root total of the daily price returns over the course of a month (following Andersen and Bollerslev, 1998) This provides us with an inexplicably and unambiguous measurement of the volatility (unlike previously (see, Chan and Grant (2016)) which is derived via generalized autoregressive conditionsal heteroscedasticity (GARCH) and stochastic variation (SV) model) that is, in reality, an inactive process. In addition, using quantiles is more reliable in comparison to the linear model (also being considered) because it focuses on the capability of expected skewness to forecast the whole conditional distribution that is RV and is not limited to only the conditional mean. This is crucial because focusing only on the conditional mean alone that of RV can “hide” interesting characteristics (Meligkotsidou and co. 2014) since it can result in low predictive accuracy, whereas it can be useful for forecasting certain aspects in the conditional spectrum that is the RV. The quantile-regression model is unlike the Markov-switching or smooth-transition threshold model there is no need to determine an ad-hoc the different scenarios of the RV. Additionally, the model for quantile regression maintains the basic structure of a linear predictive regression model for a given quantity of RV. However, it allows us to include the non-linearity element in our research methodology because it is possible to alter the coefficients that make up the model’s predictive coefficients can vary over the various levels of the distribution conditionally of RV. Based on the skewness in the distribution of RV as well as the evidence from statistics of nonlinearity and structural cracks that we observe in the relationship between it and expected skewness, a derivation of our empirical findings using a model of quantile-regression consequently, is highly recommended.

Recent research has revealed that the skewness of daily data has been used to determine the daily RV in the last few decades, it has not been used to predict daily RV for the last couple of decades. (see for instance, Gkillas et al. (2020), Demirer et al. (2022), Luo et al. (2022) and the references in those papers) We have the distinction of being the only ones to provide insight into the significance of expected skewness when the driving of RV throughout the time period of WTI oil market that spans over one hundred and 165 years. In this way we contribute to the existing vast literature on the ability to predict oil market volatility, based on a diverse array of models including macroeconomic and financial behavioral and climate patterns-related prediction factors (see, Bouri et al. (2021) and Salisu and et al. (2022c) for more in-depth reviews) and focusing on the significance of expected Skewedness. The only paper that is related to ours is one written by Salisu et al. (2022a) which claims negative tail risks which are derived from the Conditional Autoregressive Value at risk (CAViaR) model, could be used to forecast the actual variances in oil returns during the period from 1859:10 until 2020:10. The ability to use skewness to identify both extreme positive and negative shocks, and employing an equation of quantiles to determine the complete variation in the distribution of oil market volatility, is a technological enhancement over the existing conditional means-based, negative shocks-only model studied by Salisu and colleagues. (2022a). In the same way our findings will be beneficial for researchers, investors, and policy makers, as our analysis combines Salisu and co. (2022a) with respect to an overall analysis of the shocks that we have studied and also provides information regarding predictability that covers the whole contingent variation in the RV.

Apart from being an important research question, Our empirical research is crucial for policymakers and investors as our findings could aid in making “optimal” decisions. This is due to the increasing financialization of the market for oil in the past two decades or so, in particular (Bampinas as well as Panagiotidis 17th, 2017) has resulted in an increase in the presence of important investors like the hedge fund industry, pension funds and insurance firms in the market, making oil a viable alternative investment option in the asset allocation decision-making for financial institutions (Degiannakis and Filis 2017). Furthermore, considering that volatility, when seen to mean uncertainty is a key factor in investment decisions and the selection of portfolios (Poon and Granger 2003) The precise forecasts of the volatility of oil price returns are crucial for those who participate in the market for oil. Furthermore, considering that moment-to-moment fluctuations in crude prices for oil have historically been linked to negative effects on economic activities (van Eyden and co. 2019, 2019) forecasting the likely direction in the direction of volatility in RV is of utmost importance for policymakers.

The remainder of the document is structured in the following manner Section 2 provides an overview of the methodology and data Section 3 discusses the forecasting results Section 4 wraps up.