On the Oil Price Uncertainty.

AuthorFtiti, Zied

    Following Hamilton's (1983) seminal paper, it is established that oil shocks may have a significant negative effect on the real economy. Indeed, oil is considered a major input for several industries and manufactories. Moreover, oil still matters to entrepreneurs, investors, consumers, and policymakers (Elder and Serletis, 2010; Hamilton, 2011), even though several national ongoing programs have tried to permanently reduce the dependency of advanced countries on the oil sector and propose new alternative energy resources (e.g., COP21). (1) Furthermore, strong evidence of uncertainty/unpredictability is seen in the oil sector (supply, reserves, pricing, the US shale revolution, demand, market regulation). Considered together, two interesting questions can be raised in the context: (i) Do oil prices matter? and (ii) Does oil price volatility and uncertainty matter?

    In the aftermath of the recent global financial crisis (2007-2009), major changes were observed in oil prices. The West Texas Intermediate (WTI) oil increased from US$ 58/barrel in January 2007 to US$ 140/barrel in June 2008, but declined to US$ 41/barrel in January 2009 to reach US$ 133/barrel in April 2011 before hitting US$ 48/barrel in September 2016, indicating evidence of high oil price volatility. Thus, high oil price volatility and abrupt changes in oil prices may impact the overall economy and emerge as a source of economic uncertainty, (2) which requires more attention from consumers, investors, and policymakers. Indeed, oil effects occur through multiple channels: (i) supply-side shocks (fluctuations in oil prices are indicative of an increase in the marginal cost of production); (ii) wealth-effect transfer (transfer of trade surplus from oil consuming countries to oil producing countries); (iii) inflationary effects; and (iv) unexpected effects (i.e., uncertainty about oil prices). Hereafter, we focus on uncertainty in oil prices, which is a crucial issue because, to the best of our knowledge, relatively few previous studies have investigated the effects of oil uncertainty (Elder and Serletis, 2010).

    Oil prices exhibit uncertainty for many reasons. First, oil-producing countries do not reveal or communicate their data on oil reserves, real oil extraction costs, and oil reserve management. Saudi Arabia was earlier considered to have the largest proven oil reserves (267 billion barrels accounting for 24.4% of proven world oil reserves in 2000); however, since 2011, Venezuela has surpassed with higher oil reserves. However, these reserves may be subject to measurement errors. Accordingly, the announcement of new oil reserves could affect oil supply and reduce oil prices, while the inverse effect is expected for a pessimistic analysis of oil reserves. Second, oil supply is time-varying, as it depends on the geopolitical environment and agreement among oil producers. For example, Saudi Arabia is a major oil producer worldwide, but its production is variable, with a volume of 10.3 million barrels a day in 1980, 10.6 million barrels a day in 2006, 9.2 million barrels a day in 2008, and 9.76 million barrels a day in 2009. Furthermore, with the recent Saudi Arabia-led military conflict against Yemen and the US sanctions on Iran's oil could destabilize oil production. Third, oil demand is always cyclical and changes with the economic business-cycle fluctuations, increasing when the economy is booming and declining during economic downturns, thus affecting oil demand elasticity and price. For example, the slowdown in global economic growth and China's oil demand (i.e., China's economy grew 7.4% in 2014, its lowest since 1990, and decreased further to 6.9% in 2015) led to the sharp drop in oil prices. Finally, the US shale revolution, (Jawadi, 2018) could induce further uncertainty and volatility in oil prices in the future. Indeed, shale revolution has had an adverse impact, as it opposes technology and traditional oil production, thereby yielding oil production at a lower price and leading to a ceiling on oil prices.

    In the existing literature, a strand investigated the impact of oil price uncertainty (OPU) on the real economy. For example, Elder and Serletis (2010) measured OPU using the conditional standard deviation of the forecast error for the change in real oil prices. Using a structural vector autoregressive (VAR) generalized autoregressive conditional heteroskedasticity (GARCH) as the mean, the authors studied the effect of OPU on real economic activity and showed that OPU has a significant negative effect on gross domestic product (GDP), durable consumption, and investment. This effect occurs through two main channels: i) real balances and monetary policy, (3) and ii) income transfer. Rahman and Serletis (2012) also present evidence showing a significant relationship between OPU and Canada's real GDP growth rate. The authors suggested that oil price crashes witnessed in January 1986 and November 2008 involved similar levels of OPU, even if the causes were different. (4) Jo (2014) extended the study of Elder and Serletis (2010), measuring OPU through a stochastic volatility model and showing its negative effect on global industrial production and manufacturing activities. In fact, if oil prices double, quarterly activity growth decreases by 0.1%. Furthermore, Jo (2014) proxies OPU by a time-varying standard deviation as the one-quarter ahead oil price forecast error, for which the time-varying volatility evolves through a stochastic volatility process. Aye et al. (2014) also captured OPU by the conditional standard deviation of the one period-ahead forecast error for the change in oil prices and showed that OPU significantly and negatively affects South Africa's manufacturing production. Bashar, Wadud, and Ahmed (2013) found evidence indicating OPU's importance and showed its significant impact on the Canadian economy, which is in line with Hamilton's (1983, 1996, 2003, 2011) conclusion that oil price shocks are sources of uncertainty, as oil price increases are often followed by declines in the US GDP growth.

    Other studies investigated the investment channel hypothesis to characterize the impact of OPU on the real economy (Henriques and Sadorsky, 2011; Aye et al., 2014; Wang et al., 2017). This literature is disaggregated between studies analyzing the effect of OPU on economic activity based on the macroeconomic and microeconomic perspectives. For the first group (Kuper and Soest, 2006; Elder and Serletis, 2010; Rahman and Serletis, 2011), analysis is oriented on the impact of macroeconomic aggregates (inflation, exchange rate, GDP growth, and energy use), while the second group investigated the OPU effect based on the investment decisions of firms (Henriques and Sadorsky, 2011; Wang et al., 2017) and manufacturing production (Aye et al., 2014).

    The second strand of literature shows the impact of OPU on expected cash flows, discount rate (Ciner, 2013), and uncertainty related to equity prices (Bernanke, 1983; Pindyck, 1991). Recently, Bams et al. (2017) stipulate that OPU may be an important factor for stock valuation. Wang et al. (2017) also showed the negative effect of OPU on corporate investment expenditure in China, while Joo and Park (2017) showed that OPU negatively affects stock returns. Other studies investigated the impact of OPU on clean energy stock returns, (Dutta, Nikkinen, and Rothovius 2017; Henriques and Sadorsky, 2008). This literature postulates that oil price shocks significantly affect the financial market performance of renewable energy firms. Consequently, OPU plays an important role in the global renewable energy policy planning and the overall economy.

    Broadly, while several studies investigated the impact of OPU on the real economy and provided interesting results, we note two remarks. On the one hand, in most previous studies, the proxies for OPU are selected in an ad hoc manner. While on the other hand, previous studies refer to simple and less flexible measures of OPU, implying limited analysis of its impact on the real economy. Overall, four main approaches are used in the related literature to measure OPU. The first approach refers to the monthly standard deviation, as in Federer (1996), while the second approach stipulates the use of squares of log differences in oil price series. A third approach approximates OPU through a 13-month rolling standard deviation of the monthly oil price as a logarithm. Finally, a fourth approach approximates OPU with the conditional standard deviation computed from a GARCH model specifying the log difference in oil price series. Additionally, to the best of our knowledge, no previous study has investigated the issue of OPU forecasting.

    Our study addresses this gap and contributes to the literature in several ways. First, we propose three different measures of OPU based on the stochastic volatility models (standard stochastic volatility, stochastic volatility moving average, and leverage stochastic volatility models). The use of stochastic volatility models has several advantages. The stochastic volatility model proposes a more flexible framework than classical measures, as it embodies two separate disturbance terms (Carnero et al., 2004). In fact, unlike the GARCH family models, unexpected OPU shocks may be regarded as independent changes in oil prices because the model has its own innovation in addition to innovations in the mean equation. Furthermore, the latent specification in these models rejects any ad hoc assumptions about the specification of conditional volatility. Moreover, since volatility is typically unobservable, the stochastic volatility model embodies a main characteristic of volatility as it is formulated based on latent variables. Second, we compare OPU measures through a forecasting analysis. To the best of our knowledge, this is the first study that provides a forecasting essay of OPU. Accordingly, we show that the standard...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT