دانلود مقاله لاتین در مورد نفت شیل، قیمت های واقعی وست تگزاس اینترمیدیت – الزویر ۲۰۲۰

۲٫ The data

۳٫ Methodologies: the SVAR models and copula modeling

۴٫ Results

۵٫ Concluding remarks

Appendix A. Supplementary data

References

Following the adoption of new techniques of shale and fracking by U.S. oil companies, a structural vector autoregression model (SVAR) complements studies on why Brent and WTI started to diverge around early2011. Using monthly data from 2000 to 2018, we decompose oil supply into: world oil (excluding U.S.), U.S. conventional (non-tight) oil and U.S. tight oil. We examine the variance decomposition of stock returns for the aggregate market (S&P 500), the S&P Energy sector and Chevron and Exxon Mobil oil companies, and we further identify differences between two subsamples from 2000 to 2010 and 2011 to 2018, respectively. We find that supply considerations (especially due to tight oil) become more important in the subsample after 2011, not only for individual oil companies but also for the aggregate market and energy sector: Supply shocks due to tight oil explain in our benchmark model between 29% (S&P 500) and 31% (S&P Energy) of the variance in stock returns after 24 months and between 28% and 29% for oil companies. None of these are statistically significant in the pre-2011 subsample. Among impulse responses, tight oil production responds positively to disruptions in world oil, and U.S. stock returns respond positively to oil price shocks and respond negatively to tight oil shocks which is a further finding while being consistent with the literature. Copula modeling uncovers stronger tail dependences in the second subsample for the interactions during downturns and upturns among global demand, crude oil prices and stock markets.

Introduction

In an influential paper, Kilian (2009) proposes a structural vector autoregression (SVAR) approach to global oil supply, an index of global real demand and real oil prices. The latter are typically calculated by oil prices denominated in U.S. dollars (such as the U.S. refiner cost of oil, crude WTI or Brent oil prices) deflated by the U.S. price index. Its main message is that supply shocks are relatively less important than demand shocks, themselves captured by a global index of commodity prices, which became known as “Kilian index”. Due to its impact in the academic literature, several extensions of Kilian (2009) have been developed, usually adding one series in the SVAR to consider the effects of supply and demand not only on real oil prices but also on other financial markets, including U.S. stock prices by Kilian and Park (2009), U.S. bond prices by Kang et al. (2014), exchange rates by Chen et al. (2016a), economic policy uncertainty by Kang et al. (2017) and U.S. consumer sentiment by Güntner and Linsbauer (2018). Another stream of papers has focused on misspecifications of the basic SVAR model to allow for inventories and speculative trading put forward by Kilian and Murphy (2014).

Some authors have examined the causes of why Brent and WTI started to diverge. This is very important because the two oil prices serve as major benchmarks for the world and the U.S. For example, Scheitrum et al. (2018, p. 463) aim to “explain what happened in 2011 when the spread diverged from historical levels, sending WTI to more than a $20 discount under Brent and resulting in the WTI being viewed as a broken benchmark for a time.” Fig. 1a illustrates the divergence between Brent and WTI using monthly data making clear the shift in the spread after 2011 when advances in oil production became evident, as shown in Fig. 1b which compares the world (excluding U.S., right axis), U.S. non-tight and U.S. tight oil production over our sample period. Interestingly, the oil spread started diverging from historical levels at a time when U.S. tight oil production started rising. In this paper we will therefore refer to the first subsample running from 2000 to 2010 and the second running from 2011 to 2018. This paper merges the SVAR literature on real oil prices and the divergence in crude oil prices. Our contribution is not on the causes of the divergence but on the implications of this divergence for stock returns. We believe this is of interest because both academic studies and the financial press have suggested that the vast increase in U.S. oil production may be responsible for the divergence in oil prices, which itself may impact how oil price shocks are transmitted into the financial markets.