Oil Price Declines Could Hurt U.S. Financial Markets: The Role of Oil Price Level.

AuthorNguyen, Ha
  1. INTRODUCTION

    In 2014, oil prices dropped sharply to the lowest level since the Great Recession (Figure 1.1). Stock prices also declined in this period. This positive correlation between stock prices and oil prices contradicts the conventional expectation that cheaper oil prices benefit oil-importing economies like the United States. Because of this abnormal correlation, many economists believe that declines in both oil prices and the stock market are driven by weak aggregate demand (Hamilton, 2014; International Monetary Fund, 2015; Bernanke, 2016; Baumeister and Kilian, 2016).

    This paper asks a complementary question: "How do changes in oil prices causally affect financial markets, controlling for shocks to aggregate demand?" To establish a causal effect of a change in oil prices on stock prices, we follow Rigobon (2003) to instrument for changes in oil prices with plausibly exogenous news that might exclusively affect current or future oil supplies. We call these news oil-specific news. We note that investor's current oil demand may increase if the oil-specific news makes them worry about the availability of future oil supplies, (this is referred to as precautionary demand shocks in Kilian, 2009). We call days with oil-specific news "event days" and days without oil-specific news "non-event days". This method by Rigobon (2003) has been widely used in different contexts (see for example, Rigobon and Sack (2004), Anderson et al (2007), Ehrmann et al (2011), Chaboud et al (2014), and Hebert and Schreger (2017)).

    We use daily data of oil, stock, and bond indices from January 2011 to October 2016. Motivated by the debate around how the sharp drop in oil price in 2014 affected the economy, we study the period before and after the drop separately. A structural break test in Section II suggests that oil price switched between a high regime and a low regime around November 13, 2014. Therefore, we split the sample on November 13, 2014.

    We find that since November 13, 2014, a decline in oil price hurt risky assets (equities and high-yield bonds) and lifted safe assets (investment-grade bonds and long-term treasury bonds). (1) This suggests that oil price declines may hurt the operation of the economy (reflected by stock prices) and credit-worthiness of riskier borrowers (reflected by high-yield bond prices). However, from January 1, 2011 to November 13, 2014, our IV estimates on the effects of oil prices on stock prices are negative and statistically insignificant.

    These findings for November 2014- October 2016 contradict the literature, which tends to observe that a decrease (increase) in oil prices either increases (decreases) or does not significantly affect stock prices. There are three possible reasons for such difference: (1) we examine the period after November 2014 whereas the literature studies the period before 2014, (2) our method is different from the structural Vector Autoregression (VAR), a widely-used framework in the literature, (2) and (3) we use daily data instead of monthly, quarterly, or annual data. (3) We note that the effect of oil prices on stock prices is negative and statistically insignificant before November 2014, which is consistent with findings in the literature. This suggests that the distinct results since November 2014 are not due to different methods or data frequency. Thus, how oil price fluctuations affect financial markets after November 2014 differs from how they did before November 2014.

    What makes post- November 2014 different? We explore a possible explanation that at low and high oil price levels, the effects of oil price fluctuation on financial markets are different, possibly due to concerns about the energy sector and its spillover to the rest of the economy. In fact, the average daily oil price from 2014 to 2016 was $63/barrel, around the break-even price for shale oil extraction. (4) We interact the change in log of oil prices with oil price level and find that this interaction term is negative and significant from November 2014, implying that the adverse effect of oil price declines on financial markets were stronger when oil prices were lower. For the period from 2011 to November 2014, the interaction term is not statistically significant. This suggests that the level of oil prices might not matter much for the relationship between oil prices and stock prices in this period when the average daily oil prices were high, around $95/barrel. Although the interaction result is only suggestive (as oil prices could be correlated with other factors), it is consistent with the explanation that at very low level of oil prices, the concern about oil companies going out of business is magnified, which could have spillover effects to other sectors.

    This paper contributes to the debate surrounding the oil price drop in 2014. As mentioned earlier, the puzzling positive correlation between oil prices and stock prices since 2014 leads many economists to believe that a weak global demand drives both declines in oil prices and the stock market. Our findings suggest that weak global demand is not the only factor responsible for the recent co-movement between oil prices and stock prices: declines in oil prices might have direct causal impact on stock prices as well. (5) A decline in the energy sector and its potential subsequent spillovers could offset or even outweigh the benefits brought about by cheaper oil prices, as pointed out in Baumeister and Kilian (2016). In section VI, we will discuss other potential transmission mechanisms in more detail.

    This paper extends a broad empirical literature on the impact of oil prices on financial markets with two novel results. First, we find that a decline in oil prices caused by exogenous oil-specific shocks could decrease stock prices, which has not been found in the literature (see Degiannakis et al (2018) for a review). The existing literature finds that an oil price decline (increase) is either associated with an increase (decrease) (6) or insignificant change in advanced countries' stock markets. (7) Second, we find that when oil prices are low, the level of oil prices could affect how changes in oil prices impact stock prices. (8) The existing literature examines how an unit change in oil prices affect stock prices on average, but does not study how this effect may differ depending on the actual value of oil prices.

    Finally, this paper also relates to the literature on oil prices and the U.S. economy. The literature generally finds that an increase in oil prices adversely affects the economy, especially during the oil crisis in the 1970s. To the extent that stock prices represent a forward-looking view about the U.S. economy, our results suggest the opposite: a decrease (not an increase) in oil prices could adversely affect the US economy when oil prices are low enough. (9)

    The rest of the paper is organized as follows. Section II explains in more details the methodology. Section III discusses data sources. Sections IV and V present the effects of oil prices after and before 2014 respectively. Section VI examines why the effect after November 2014 differs from that before November 2014 and argues for a role of oil price level. We conclude in section VII.

  2. METHODOLOGY

    We identify the effect of changes in oil prices on prices of various asset classes through a heteroscedasticity-based identification strategy, following Rigobon (2003) as well as Rigobon and Sack (2004). Consider the following system of equations:

    [DELTA][p.sub.t] = [gamma][DELTA][s.sub.t]+[beta][z.sub.t]+[[epsilon].sub.t], (1)

    [DELTA][s.sub.t]= [alpha][DELTA][p.sub.t]+[delta][z.sub.t]+[[micro].sub.t] (2)

    where [DELTA][p.sub.t] is the change in oil prices, [DELTA][s.sub.t] is the change in asset price, and [z.sub.t] is a set of common aggregate demand factors that could affect both oil prices and stock prices (such as interest rates, news about global growth, or other aggregate demand news). [[epsilon].sub.t] represents oil shocks that only affect oil prices. [[epsilon].sub.t] captures events that affect current or future oil supply, such as a North Sea storm that forces oil firms to evacuate platforms. Similarly, [[micro].sub.t] are the idiosyncratic shocks that only affect stock prices. Our goal is to estimate the value of a: the causal impacts of changes in oil prices on changes in stock prices. Note that in this framework, the effects of oil price increases and decreases are symmetric.

    We divide the days in our sample into two types: event (E) and non-event (N) days. Event days are days with important announcements and developments about the current or future oil supply. A useful feature of the approach is that it does not require the complete absence of common aggregate demand or shocks on global economy during event days. This strategy instead relies on the identifying assumption that the variances of the common aggregate demand shocks [z.sub.t] and financial shocks [[micro].sub.t] are the same on non-event days and event days, whereas the variance of oil-specific shocks [[epsilon].sub.t] is higher on event days than on non-event days:

    [[sigma].sup.2.sub.z,E] = [[sigma].sup.2.sub.z,N] (3)

    [[sigma].sup.2.sub.[mu],E] = [[sigma].sup.2.sub.[mu],N] (4)

    [[sigma].sup.2.sub.[epsilon],E] = [[sigma].sup.2.sub.[epsilon],N] (5)

    These assumptions imply the "importance" of oil-specific announcements increases on event days (E). Again, it is important to note that news on aggregate demand can take place on event days, as long as the influence of aggregate demand factors is similar to that on non-event days. As argued by Rigobon and Sack (2004), these assumptions are much weaker than those required in traditional event-study approach.

    Under such assumptions, we can identify parameter [alpha] by comparing the covariance matrices of stock price and oil price changes on event days and non-event days. In particular, for each of the two types of days j [member of] {E, N}, we...

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