*Article* **Time-Varying Influences of Oil-Producing Countries on Global Oil Price**

### **Peter Y. Jang \* and Mario G. Beruvides**

Department of Industrial, Manufacturing, and Systems Engineering, Texas Tech University, Box 43061, Lubbock, TX 79409-3061, USA; mario.beruvides@ttu.edu

**\*** Correspondence: peter.y.jang@ttu.edu

Received: 10 January 2020; Accepted: 5 March 2020; Published: 17 March 2020

**Abstract:** This paper aims to investigate the time-varying influences of major crude oil-producing countries on Brent oil prices, with seven-panel data over the observation years of 1998 to 2018. We create seven panels with 36 monthly data for each and estimate the contributions of individual producing countries to oil price changes with a multivariate regression technique of ordinary least squares. Most existing researches have focused on identifying relationships among oil price, market fundamental factors, macroeconomic variables, and geopolitical events in broad perspectives. However, this paper undertakes a longitude/panel analysis of nine oil producers' influences, with the Organisation for Economic Co-operation and Development (OECD) consumption and the U.S. Dollar Index (USDX) on oil prices in each panel and intends to identify which producers have statistically significant influencing weights on oil prices. We believe that this research contributes to the body of knowledge in better understanding the relative impacts of major oil-producing countries. Results show empirical evidences that the Organization of the Petroleum Exporting Countries (OPEC) production stayed as the greatest negative influence on the oil price in the periods of Panel 2 (2001–2003) and Panel 7 (2016–2018) only, while the U.S. Dollar Index took over the OPEC's influencing role in most of the other periods, followed by Iran, the U.S., and China.

**Keywords:** oil producers; oil price; time-varying influence; oil market fundamentals; oil price fluctuation

### **1. Introduction**

Crude oil is a critical source for economic growth and further industrialization, and the industrialized nations import a significant portion of oil from the Persian Gulf [1]. Oil accounts for 33% of global energy consumption and trades in US dollars per barrel of 42 US gallons [2].

Major benchmarks of crude oil include the West Texas Intermediate (WTI) in the U.S., the Brent Blend and the North Sea in Europe, and the Organization of the Petroleum Exporting Countries (OPEC) Reference Basket (ORB) of fourteen blends. While the WTI and Brent benchmark prices are for oil exports to the Atlantic Basin [3], the Brent crude price index is a benchmark to set oil prices for 70% of world oil transactions [4]. Figure 1 presents a map of global crude oils and their positioning by API (American Petroleum Institute) gravity, a measure of liquid petroleum density, and sulfur content. The crude oil with high API gravity has low density and light weight, while the oil with high sulfur content is called a sweet oil [3,5,6]. The WTI oil is very sweet high-quality, and the index is the most famous benchmark in the U.S. and the Western Hemisphere, with its future's products traded on the New York Mercantile Exchange (NYMEX). The Brent oil is not as light as WTI but still a high grade from fifteen oil fields in the North Sea.

**Figure 1.** Crude oil is not a homogenous resource. Source: Reprinted with permission from Federal Reserve Bank of Dallas [6].

In 2018, top three crude oil producing countries were the U.S., Russia, and Saudi Arabia, producing over 10 million barrels a day each, followed by Iraq, Canada, Iran, and China, that produce about 4 million barrels a day each (Figure 2). The top three producers accounted for 39% of daily world production, comparable to OPEC's 41%, while the top ten producers had a 69% share with 57 million barrels [7].

**Figure 2.** Major crude oil producing countries by production volume in 2018. Data source: U.S. Energy Information Administration [7].

Global oil market has the structure of two major player categories: low-cost public producers and highly competitive private producers [8]. Public producers include national oil companies (NOCs), including OPEC and non-OPEC-producers (Russia and Mexico), and account for 60% of global oil production. As such, global oil markets are heavily political, far away from a competitive market [9], and 94% of world proven reserves are controlled by governments [2]. There are some questions how long OPEC, the most influencing oil producing organization, would last as an international cartel given the historical examples of other commodity cartels, International Tin Association (1954 to 1985), and International Coffee Association (1962 to 1989) [10].

The global oil market has invisible hands of suppliers and consumers affecting oil prices. Oil price behavior, in a social system, may follow one of several fundamental modes, exponential growth, goal seeking, and oscillation, affected by a simple feedback structure of the components as suggested by a systems theory [11]. Other modes of nonlinear behavior include S-shaped growth, S-shaped growth with overshoot and oscillation, and overshoot and collapse.

In both academia and industry, it has been the subject of debate, what intrinsically drives crude oil prices. Figure 3 is a behavior over time (BOT) chart of annual prices of Brent spot and WTI spot from 1997 to 2018. The price level of U\$20 per barrel in 1997 increased to U\$100 in 2008 and settled down at U\$65 to U\$70 in 2018. Three explanations are postulated for the causes of the price declines: those associated with oversupply, those associated with under-consumption even with under-supply, and the future oil markets' bearish views of future market fundamentals and sell-now executions [8].

**Figure 3.** Crude oil benchmark annual prices (Brent and West Texas Intermediate (WTI) spot, nominal)—1997 to 2018. Data source: U.S. Energy Information Administration [12].

The period of 1973–1996 was quoted as "the age of OPEC" and the years of 1997 to present as a new industrial age outside OPEC [13], which describes the dominant power of OPEC through the mid-1990s and then emerging new forces in the late 1990s. Previous studies focused on identifying relationships between oil price and macro variables: relationship with supply factors [2,8,10,14], demand factors [15,16], macroeconomic variables [2,17], event-driven factors [9], and the consequences of high/low oil prices [18].

Oil discovery and technical improvement takes time to keep a market at an equilibrium state, and multiple time scales, short-term, mid-term, and long-term are more appropriate for analysis [19]. In a similar context, this paper takes the approach of multiple time panels, away from a traditional single time horizon.

This paper aims to investigate the influencing weights of major crude oil-producing countries on the Brent oil price in seven individual panels with 36 monthly data each. Nonproduction-related variables are petroleum consumption and the U.S. Dollar Index (USDX). This paper takes longitude/panel study approaches, from 1998, a year after 1997, a starting year of "the new industrial era" to 2018, and intends to provide empirical evidence which oil-producing countries were relatively more influential to the Brent oil price in each of the time-varying panels. As this is the first paper with a unique approach in oil market analysis, we believe that it provides better understanding of dynamic roles of major oil-producing countries in each defined period. This paper is organized as follows: Section 2: review of the current state of the art, Section 3: data and methodology, Section 4: results and discussion, and Section 5: conclusions and policy implications.

### **2. Review of the Current State of the Art**

A review of literatures resulted in a number of studies that attempted to explain the fluctuations of crude oil prices, in terms of historical overviews, supply side (OPEC and non-OPEC), demand side, macroeconomic factors, price decline factors, oil future's market and speculation, event-driven factors, impact of oil prices on producers and consumers economies, and consequences of oil price shocks.

### *2.1. Historical Overview*

When the first oil crisis took place in 1973, the imported crude oil price to the U.S. quadrupled in 1973/1974, and the West Texas Intermediate (WTI) rose from U\$4.31 in September 1973 to U\$10.11 per barrel in January 1974. Prior to 1973, oil price fluctuations were the results of shift in demand or global economic expansion [20], but since the early 1980s, the fluctuation has reflected the disruption of the flow, exogenous political events, war, revolution, and OPEC coalition [13]. Figure 4 presents a historical overview of oil prices, with geopolitical and economic events in a chronological order, from 1970 to 2015 [2].

**Figure 4.** Geopolitical and economic effects on crude price. Source: Reprinted with permission from Solomon Adjei Marfo [2].

Baumeister & Kilian (2016) [20] describes four main episodes in the past four decades: the periods of the 1973/1974 oil crisis, the 1979/1980 crisis with Iran revolution, 1980s/1990s with Iran-Iraq war, and 2003/2008 with global financial crisis. The first episode was the Arab oil embargo where OPEC

cut production by 5% and then an additional 25%. The second episode was the Iranian Revolution, resulting in oil price skyrocketing to U\$40 per barrel in April 1980, from U\$15 in September 1978.

The third episode period 1980s/1990s was a result of a large exogenous supply disruption, and subsequently, non-OPEC countries, Mexico, Norway, and the U.K., responded by transforming themselves to oil producers. In the aftermath, OPEC's market share of 53% in 1973 dropped to 43% in 1980 and then 28% in 1985. Approximately 10 years later, the WTI price plummeted to U\$25 per barrel in 1996 due to the Asian financial crisis and then down to U\$11 per barrel in 1998. The global financial crisis occurred and ran in parallel during the fourth episode in 2003–2008, when the WTI oil traded in the wide range of U\$28 to 134 per barrel. Demand growth in India and China contributed to the rising prices throughout the mid-2008, but the global financial crisis led to a sharp drop in demand, pushing down the price to U\$39 per barrel in February 2009.

### *2.2. Supply Side—OPEC*

OPEC, founded in 1960 by five oil-producing nations, and with fourteen members as of August 2019 [21], accounts for an estimated 42% of global oil production and 73% of world oil proven reserves. OPEC started setting a target production in 1980 for each of its members to maintain stable oil prices [2].

Major producers prior to 1970 were seven western companies called "Seven Sisters": Anglo-Persian Oil Co, Gulf Oil, Standard Oil of California, Texaco, Royal Dutch Shell (RDS), Standard Oil of New Jersey, and Standard Oil of New York [22], and these producers now account for 5% of global oil reserves [2].

Oil spare production capacity could be a factor to affect oil price volatility. OPEC uses its spare production capacity to stabilize the oil market. Four core producers, Saudi Arabia, Kuwait, Qatar, and U.A.E., successfully balanced oil market and reduced price volatility to one-half of the normal level by adjusting production to offset supply and demand shocks and by maintaining the volume of spare capacity at 85% as swing producers [14].

In the "age of OPEC" prior to 1997, OPEC production quantity was accompanied by non-OPEC production quantity; for example, in 1982–1985, OPEC reacted to lower real oil prices [23]. In the "new industrial age" when OPEC's market power was getting weakened [13], OPEC production quantity just sustained the global production/consumption for GDP growth.

### *2.3. Supply Side—Non-OPEC*

Small producing countries, each with a less than 5% share of world oil output, increased their combined share from 59.4% to 65.1% between 1995 and 2010 [24]. The small producers include Angola and Algeria in Africa, Indonesia, and Malaysia in Asia, and Mexico and Venezuela in South America, and Norway in Europe. They responded to the changes in the world oil market, and their decisions indicated that there was a strong relationship between their production levels and changes in oil consumption but with a lower relationship between the production level and change in prices.

Growth of the U.S. shale oil production is notable. The shale production was close to zero in 2008 but grew to 4.25 million barrels a day in 2016, standing with a 48% share of the U.S. oil production and 5% of the global production [10]. At the end of 2018, the U.S. shale production stood at 6 million barrels a day, accounting for 57% of the nation's total production. The shale production also contributed to an increase of the U.S.'s global market share to 13% in 2018, up from 8% in 2000, potentially influencing the oil price collapses in the recent decade [25].

Shale oil production can respond to changes in oil prices much quicker than a traditional oil technology, with the competitive edges of improvement in fracking technology and its lower extraction costs. The U.S. shale oil development may be analogous to a structural change in systemic processes. Kuhn (2012) describes three ways for a paradigm to respond to crises: one, proving to solve the crisis, two, giving up with no solution and leaving for next generations, and three, ending up with a new accepted paradigm after the battle [26]; the shale oil case belongs to the third way.

### *2.4. Demand Side*

On the oil consumption side, the most influential driver is demand for refined oil products, and OECD accounts for 50% of world petroleum demand [2]. When the refining capacity is lost, it would affect the oil demand side. Notably, the Arab embargo in 1973 disrupted the oil supply and pushed up the market price, resulting in depressing the U.S. and OECD demand by 2–4% per year in 1974 and 1975 [12].

Oil price rise in 2004–2008 is also the subject of debate. While a price rise is viewed in association with increased demand, another view is that oil price is more sensitive to supply as production approaches its capacity [15]. Thus, refining utilization rates, OPEC capacity utilization, and oil future's markets affect oil price. Oil price movements may exert negative impacts on gross domestic product (GDP), consumer price index (CPI), and unemployment in oil consuming countries, in particular, based on economic data collected from 26 OECD countries [16].

### *2.5. Macroeconomic Factors*

Economic growth and energy investments lead to more available oil supply, as well as oil demand. As crude oil trades in US dollars in global markets, the depreciation of US dollars leads to more demand for oil, and vice versa [2], indicating negative correlations between oil demand and the US dollar strength. Other macroeconomic factors may include future oil markets, speculators, hedgers, brokers, and exchanges available to the global players.

There are meaningful relationships between oil price and more extensive macroeconomic variables, such as global industrial production, prices, and interest rate [17]. Among five countries in the study, the global macro factors were main drivers for the U.S., Europe, and China but minor ones for Japan and India. Other studies also examined relationships between oil price and macroeconomic variables [27–29].

### *2.6. Price Decline Factors*

Price decline in 2007–2008 was a good example of formation and collapse of an oil price bubble, going through combined effects of stagnant oil supply, unexpected economic growth in India and China, low interest rate, a weak U.S. dollar, and a consequent sharp spike in oil prices [30]. This behavior is similar to a mode of overshoot and collapse in a social system [11].

The World Bank lists causal factors to sharp drops in oil prices: supply and demand, changes in OPEC objective, geopolitics, the U.S. dollar appreciation, speculative demand and investment management, relative contribution of supply and demand, and oil price outlook [10]. Fundamental oil supply and demand set the conditions for a long-term trend in price, but in the short-term, market sentiment and expectation led to price fluctuations.

The World Bank reported that the oil price drop in June 2014 was significant but not an unprecedented event, as a 30% price decline were observed in five cases over a seven-month period in history [10]. Those five periods are 1985–1986 with strong production from non-OPEC, 1990–1991 with the U.S. economic recession, 1997–1998 with Asian financial crisis, 2008–2009 with global financial crisis, and the latest, June 2014–January 2015 with a supply glut from unconventional sources since mid-2014. In November 2014, OPEC changed its policy and abandoned an objective to set a target price.

### *2.7. Oil Future's Market and Speculation*

Crude oil future's market usually reflects the expectation of market fundamentals. Unexpected disruption in the oil market comes with large regression residuals, and the derivative paper markets contain information on the magnitude and duration of major market disruption [31]. Empirical results are derived from investigating price volatility of Brent oil (2000–2014) in relationship with the term structure of option-based implied volatilities and global macro-economy; physical market fundamentals (OPEC surplus output, capacity, and storage); and equity volatility index (VIX).

On a short run, if oil market is not price-sensitive, or relatively inelastic, then oil investment would be slow to take place [32]. Oil prices spiked in the period of 2000 to 2010, and one of the explanations was that speculative pressure exerted influence on the prices of the storable commodities.

### *2.8. Event-Driven Factors*

When political instability is prevailing in any of the oil-producing countries, oil production capacity may be disrupted and reduce available supply [2]. Roles of political economic news in oil pricing are also important [9].

The Arab embargo in the early 1970s led to technology advancements among non-OPEC nations to secure oil resources, resulting in active developments of unconventional and offshore oil sources, with the North Sea in the 1970s and the Gulf of Mexico in the 1980s. The shale oil, a tight rock formation, has a shorter life cycle of 2.5 to 3 years from shale development to full extraction and contributes to the global supply with low capital costs [10].

Wars or political tensions may disrupt oil supply to the markets, but, by their models, show no direct effect [33]. As demand is inelastic to price change, economic activity is the most significant factor to affect demand. Major wars/political events in oil markets include Arab-Israeli War (1973), Persian Gulf War (1991), Operation Desert Fox (1998), Iraq War (2003–2011), Arab Spring (2011), global financial crisis (2008), and European debt crisis (2010).

### *2.9. Impact of Oil Prices on Producers' and Consumers' Economies*

Impacts of oil price shocks vary with on oil producers and oil-consuming economies [29]. Among twelve economies analyzed for a period of 1995 to 2006, oil producers, Russia and Canada, benefited from oil price shocks, while oil importers found their economic activities suffer a slowdown in their GDP. The largest negative effects from the shocks were present in Japan, China, the U.S., Finland, and the Switzerland.

When oil price shocks take place at the markets, their consequences include shocks to key supply chains, global economic activities, national accounts, inflation, and searches for non-oil sources [10,34]. The price also shocks inflation in low oil-dependency and high oil-dependency countries [18].

There is a strong relationship between oil price shocks and the U.S. recessions [35], and Canada is affected by U.S. domestic monetary policies. Foreign disturbances, such as innovations and the U.S. interest rate, leads to significant inflation in the Canadian economy [36]. Due to the oil price volatility, even oil producers find it hard to meet their government expenditure. Bahrain implemented a policy in the late 1970s to diversify their economy by attracting finance investment in the state [37]. Oil price fluctuations also affected economic activities of small oil-producing countries, Trinidad and Tobago [38], and fiscal policies in oil-exporting countries [39].

### **3. Data and Methodology**

### *3.1. Data*

Oil price is the response (dependent) variable or a simple average monthly spot price of Brent crude oil (denoted here as BRENT), published by the U.S. Energy Information Administration (EIA). The Brent crude oil price is selected as a global oil proxy price, because it represents a benchmark price for about 70% of the global oil transactions [4]. The period for analysis is 21 years, or 252 months, from 1998 to 2018, with 1997 being the start of "the new industrial age" in the global oil market, according to Hamilton (2013) [13]. The base unit is U\$ per barrel.

Explanatory (independent) variables are monthly crude oil production data by country collected from EIA, and data for analysis covers the same period of 1998 to 2018. Producing countries for analysis are nine major countries, including five OPEC and four non-OPEC countries, based on their production ranks in 2018. They are United States (US), Russia, Saudi Arabia, Iraq, Canada, Iran, China, UAE, and Kuwait. OPEC is also included for analysis, due to its heavy share (41%) of the global production in 2018. In this paper, the capitalized names of explanatory variables represent the individual production quantity of the country or organization. The base unit for oil production is one thousand barrels per day.

Two non-production related variables are selected besides the explanatory variables. Monthly petroleum consumption data for OECD is collected from EIA and used as a proxy of world petroleum consumption, since the Old World consumption data is not available on a monthly level. The OECD's share of the world consumption accounts for an average of 56% in the years 1998–2018. OECDCON represents monthly petroleum consumption of its 36 members. Its unit is also one-thousand barrels per day. Another non-production-related variable is the monthly Trade Weighted U.S. Dollar Index (USDX), a macroeconomic variable, with January 1997 as a base month. This index represents the strength of the U.S. dollar. As oil price trades in U.S. dollars, this index influences oil demand and oil price. For example, when the U.S. Dollar Index is strong (high), then consumers tend to consume oil less. The data source is the Federal Reserve Bank of St. Louis (https://alfred.stlouisfed.org). The unit is dimensionless.

Each panel covers a 36-month period, with seven panel data over the 21-year period, with an assumption that additional production may be possible to react to a need for additional supply with a 3-year period. Conventional oil development takes 5–10 years [40], but the shale oil development requires a shorter cycle [41].

The response variable and explanatory variables are all transformed to a natural logarithm format, or LN, so a percentage change in the response variable may be estimated with respect to a percentage change in one exploratory variable, if they are statically significant at the 0.1 (10%) level.

### *3.2. Methodology*

In the whole period and each of seven panel period data, coefficients of variation (CV) are first calculated to understand a degree of fluctuations of each variable. Coefficient of variation (CV) is defined as a ratio of the standard deviation to the mean of a variable, indicating a common measure of the magnitude of variability for comparison among the variables. This measure helps identify potential variables that may influence more on the response variable, Brent oil price.

In each of seven panel data, a multivariate regression technique with ordinary least squares is used as a main tool to measure the coefficient estimates of explanatory variables, and the coefficient estimates will be compared to determine influencing weights among the variables. If p-value of the regression parameter estimates of any explanatory variables is greater than 10%, the variables will be excluded from comparisons of the coefficient estimates, because they are not statistically significant at the level.

There are a total of eight runs of regression models in this analysis; that is, one for the whole period and seven for seven panels. Each of the eight runs has individual summaries of parameter estimates, and then the coefficient estimates of statistically significant explanatory variables are compared to measure influencing weights.

Equation for multivariate regression parameter estimation in each of panels is:

LN\_BRENTt = α<sup>t</sup> + LN\_OPECt + β<sup>1</sup> × LN\_USt + β<sup>2</sup> × LN\_RUSSIAt + β<sup>3</sup> × LN\_SAUDIt + β<sup>4</sup> × LN\_IRAQt + β<sup>5</sup> × LN\_CANADAt + β<sup>6</sup> × LN\_IRANt + β<sup>7</sup> × LN\_CHINAt + β<sup>8</sup> × LN\_UAEt + β<sup>9</sup> × LN\_KUWAITt + β<sup>10</sup> × LN\_OECDCONt + β<sup>11</sup> × LN\_USDXt + ε<sup>t</sup> (1)

where

Response variable:

LN\_BRENTt: natural log of monthly Brent oil price (BRENT) at a month t; Explanatory variables:

LN\_CANADAt: natural log of monthly Canada production (CANADA) at a month t; LN\_CHINAt: natural log of monthly China production (CHINA) at a month t;

LN\_IRANt: natural log of monthly Iran production (IRAN) at a month t; LN\_IRAQt: natural log of monthly Iraq production (IRAQ) at a month t; LN\_KUWAITt: natural log of monthly Kuwait production (KUWAIT) at a month t; LN\_OECDCONt: natural log of monthly OECD petroleum consumption (OECDCON) at a month t; LN\_OPECt: natural log of monthly OPEC production (OPEC) at a month t; LN\_RUSSIAt: natural log of monthly Russia production (RUSSIA) at a month t; LN\_SAUDIt: natural log of monthly Saudi Arabia production (SAUDI) at a month t; LN\_UAEt: natural log of monthly UAE production (UAE) at a month t; LN\_USt: natural log of monthly U.S. production (US) at a month t; LN\_USDXt: natural log of monthly U.S. Dollar Index (USDX) at a month t.

### **4. Results and Discussion**

Starting with a summary for the whole period of 1998 to 2018, all the variables, responses, and explanations are summarized in the tables of descriptive summary statistics in each panel, followed by a summary table of parameter estimations. Units of each variables are also displayed on Table 1, and the units stay the same in all tables and figures of this paper, unless otherwise specified.


**Table 1.** The whole period (1998–2018): summary statistics.

### *4.1. The Whole Period: 1998–2018*

Average Brent oil spot price was U\$59.87 per barrel during the whole observation period of 21 years, with its coefficient of variation (CV) the highest (54.54) among all variables (Table 1). Producing countries with high CVs for the 21-year period are Iraq (35.05), the U.S. (26.52), Canada (25.41), and Russia (16.80). These high-CV countries could be notable contributors to such a volatile Brent oil price. Summary statistics of all the variables for each of the panels will be discussed separately.

For the whole period of 1998 to 2018, parameter estimates by the multivariate regression to explain the response variable LN\_BRENT are summarized on Table 2. Major explanatory variables with a statistical significance level of 0.1% or 10% are LN\_USDX, LN\_RUSSIA, LN\_SAUDI, LN\_CHINA, LN\_KUWAIT, LN\_IRAN, and LN\_IRAQ. Among the significant variables, USDX (the U.S. Dollar Index), the only macroeconomic variable in this analysis, is the greatest influencing factor in the whole period, with one percent change in USDX resulting in a 3.68% decline in the Brent price, while one percent change in production from Kuwait, China, and Iran negatively affected the price by 1.56%, 1.5%, and 0.63%, respectively.

Meanwhile, production of other significant producers, Russia, Saudi Arabia, and Iraq, co-moved with the price, or one percent change in production leading to 2.96%, 2.46%, and 0.12%, respectively. These co-move coefficient estimates are in contradiction with the basic economic theory; a price negatively correlates with supply quantity [10]. Some of the interpretations for this co-moving supply-price relationship are that an information asymmetry may exist due to a production location

being remote from a market place that the Brent oil index is based on, or the producer's change in supply did not simply change oil price direction.


**Table 2.** The whole period (1998–2018): LN\_BRENT prediction-parameter estimation.

Note: \* represents statistical significance at a level of 0.1. t Ratio is defined as Estimate divided by Std Error to calculate p-value, while RSquare represents the explained portion of variance by an independent variable.

In the analysis of the whole period, it is also worthwhile to note that the U.S. production, OECD consumption, or OPEC production have not influenced the BRENT price in a statistically significant way (10%).

### *4.2. Variability of Variables*

One of the ways to understand a system begins with knowledge of its variation, besides a system itself [42]. Coefficients of variations would represent how the variables experienced fluctuations over a specified period. Table 3 and Figure 5 present the coefficients of variations for each variable during each of the 36-month periods. BRENT price tops in the CV ranking in six periods of the total seven, except in Panel 2 (2001–2003) when Brent's CV of 13.1 is second to the IRAQ production variable with a CV of 40.5. BRENT price range over the 21 years is U\$123 per barrel, with a low of U\$10. IRAQ production has shown the most fluctuations among all the producers in all the past seven panels, as well as over the whole period. Over the whole period, IRAQ production tops with a CV of 35.0, followed by US (26.5), CANADA (25.4), and RUSSIA (16.8).

**Table 3.** Coefficients of variations for each of variables by panel.


**Figure 5.** Coefficients of variations for each variable by panel. CV: coefficient of variable.

Table 4 is a summary of the top four crude oil-producing countries with high CVs in each panel period. Over the whole period (1998–2018), IRAQ production experienced the highest CV (IRAQ production and ranked first for most of the panels, except for a recent Panel 7 Period (2016–2018) when U.S. production takes the top position. U.S. production recorded the CVs over 10 in the past two panel periods, indicating a swing volume of ten percent or more in recent years. One observation is that CHINA, the world's top seventh oil producer, has maintained a low CV in each of the panels, implying the country is not an oil exporter but, rather, a domestic producer-consumer.


**Table 4.** Top four countries with the highest coefficients of variations for by panel.

### *4.3. Panel 1: 1998 to 2000*

Panel 1 Period covers 36 months in 1998 to 2000, with an average BRENT price of U\$19.72 per barrel. BRENT shows the highest CV (38.09), and the producers with high CVs are IRAQ (19.09), KUWAIT (5.92), SAUDI (4.93), and UAE (4.86) (Table 5). Among the producers, IRAQ production is the most fluctuating in this period, while the US and OPEC look stable in their production volume. BRENT price range is U\$23 per barrel, with a low of U\$10.

A summary of parameter estimates in Table 6 shows that RUSSIA is the only statistically significant variable at a level of 10%, with none of the producers significant. RUSSIA production exerted influence on BRENT prices in a co-move or 1% change in RUSSIA production leading to 10.31% in BRENT prices. This could be counter-intuitive in the economic sense because of the usual negative relationship between supply volume and a market price. One interpretation is RUSSIA took a good timing of the oil market, and their additional significant volume did not push down the market price or vice versa. This panel period corresponds to the events of the Asian financial crisis (1997–1998) and Operation Desert Fox, a 4-day bombing campaign on Iraq targets.


**Table 5.** Panel 1 period (1998–2000): summary statistics.

**Table 6.** Panel 1 period (1998–2000): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.4. Panel 2: in 2001 to 2003*

Panel 2 period has an average BRENT price of U\$26.10 per barrel, up about U\$6 from the Panel 1 period. BRENT has the second-highest CV (13.07) to IRAQ production (40.48), followed by producers RUSSIA (7.43), SAUDI (7.18), and KUWAIT (6.56) (Table 7). Among the producers, IRAQ production is still the most fluctuating in this period, and the US and OPEC still look stable in their production volume, with CVs of the 2.40–3.90 range. BRENT price range is U\$14 per barrel, with a low of U\$19.

A summary of parameter estimates in Table 8 presents that five producers/organizations are influencers to BRENT prices in this period at a significance level of 10%: IRAQ, OPEC, CHINA, SAUDI, and the U.S. Negative price influencers are OPEC and the U.S., with the effects of 3.15% and 1.84%, respectively, by one percent change in their production. Co-move price influencers with their production are CHINA, SAUDI, and IRAQ, with the effects of 2.98%, 1.86%, and 0.16%, respectively, by one percent change in production. This period includes the U.S. recovery time from March 2001.


**Table 7.** Panel 2 period (2001–2003): summary statistics.

**Table 8.** Panel 2 period (2001–2003): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.5. Panel 3: 2004 to 2006*

Panel 3 period sees a 50% spike in an average BRENT price of U\$52.60 per barrel, up about U\$25 from the Panel 2 period. This period corresponds to the period when China and India were booming as emerging economies. BRENT regained the highest CV (23.90), followed again by IRAQ production (9.19) and U.S. (5.93), CANADA (4.96), and UAE (4.43) (Table 9). Among the producers, IRAQ production is still the most fluctuating in this period, and the U.S. came to play with a relatively volatile volume. BRENT price range is U\$43 per barrel, with a low of U\$31.

A summary of parameter estimates in Table 10 shows that CHINA is the only producer statistically significant at a level of 10%. However, CHINA is a co-move price influencer with the effect of 3.55% by one percent change in its production. However, OPEC and RUSSIA follow CHINA as potential price influencers, though the significance level is a little away at 11%. This period is characterized by a growth theme in China and India.


**Table 9.** Panel 3 period (2004–2006): summary statistics.

**Table 10.** Panel 3 period (2004–2006): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.6. Panel 4: 2007 to 2009*

Panel 4 period sees a 30% lift in an average BRENT price of U\$76.93 per barrel, about U\$24 from the Panel 3 period. This period corresponds to the period when China and India continued growing. BRENT is still fluctuating with the highest CV (31.69), followed by again IRAQ production (7.88) and SAUDI (5.42), UAE (5.2), and U.S. (5.07) (Table 11). Among the producers, the U.S. comes to play with a relatively volatile volume, with IRAQ, SAUDI, and UAE. BRENT price range is U\$93 per barrel, with a low of U\$40.

A summary of parameter estimates in Table 12 displays that two macro-variables (USDX and OECDCON) and two producers are BRENT influencers in this period, at a significance level of 10%. Negative price influencers are USDX, IRAN, and OECDCON, with the effects of 4.79%, 3.76%, and 1.72%, respectively, by one percent change in their production or index. Co-move price influencers include SAUDI, with the effects of 3.89%, by one percent change in production. This period is still characterized by a growth theme in China and India, which caused the WTI future's price to spike to an unprecedented U\$150 per barrel in August 2008.


**Table 11.** Panel 4 period (2007–2009): summary statistics.

**Table 12.** Panel 4 period (2007–2009): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.7. Panel 5: 2010 to 2012*

Panel 5 period also experiences a 25% spike to an average BRENT price of U\$100.81 per barrel, up about U\$24 from Panel 4 period. This period corresponds to the one with major events, such as Arab Spring and European Debt Crisis, in Portugal, Italy, Iceland, and Greece. BRENT still tops the CV (16.49) but lower from the CV of the previous period (31.69), followed by again IRAQ production (7.88) and IRAN (10.48), U.S. (8.63), and CANADA (7.51) (Table 13). U.S., IRAN, and CANADA were actively participating in the additions of oil supply with the prices moving higher. The U.S. shale oil and Canada oil sands play significant roles. BRENT price range is U\$52 per barrel, with a low of U\$74.


**Table 13.** Panel 5 period (2010–2012): summary statistics.

A summary of parameter estimates in Table 14 displays that there are four significant factors: one macro-variable (USDX) and three producers: UAE, CANADA, and IRAN. Under the average price of U\$100 per barrel this period, the only negative price influencer is USDX, with an effect of 2.29% by one percent change in the index strength, while UAE, CANADA, and IRAN exert, as co-move influencers, the effects of 3.27%, 0.47%, and 0.5%, respectively, by one percent change in their production. This period is characterized by continued price spikes. The co-move influencers may not able to lower the prices by adding supplies, failing to turn the economic theory to work.

**Table 14.** Panel 5 period (2010–2012): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.8. Panel 6: 2013 to 2015*

Panel 6 period underwent a price decline mode with an average BRENT price of U\$86.67 per barrel, now down about U\$13 from the Panel 5 period. This period corresponds to the one when oil supply glut and slow growth in the Chinese economy were observed. BRENT still tops the CV (30.96) but back up from the CV of the previous period (16.49), followed by again IRAQ production (13.92) and U.S. (10.33), CANADA (6.41), and SAUDI (3.49) (Table 15). The U.S. and CANADA continue to be active producers in the global oil market. The U.S. shale oil and Canada oil sands play significant roles. BRENT price range is U\$78 per barrel, with a low of U\$38.


**Table 15.** Panel 6 period (2013–2015): summary statistics.

A summary of parameter estimates in Table 16 displays that there are two significant factors: one macro-variable (USDX) and the only producer SAUDI. Under the average price of U\$87 per barrel this period, the only negative price influencer is USDX, with an effect of 6.03% by one percent change in the index strength, while SAUDI exerts, as a co-move influencer, an effect of 1.83%, by one percent change in their production. This period is characterized by a downward price trend, with the U.S. and Canada increasing the production.

**Table 16.** Panel 6 period (2013–2015): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.9. Panel 7: 2016 to 2018*

Panel 7 period continued a downward trend, with an average BRENT price of U\$56.29 per barrel, now collapsing about U\$30 from the Panel 6 period. This period corresponds to the one when oil supply glut and slow growth in the Chinese economy were seriously observed. The U.S. shale oil growth was remarkable in contributing to the growth of the oil supply. The shale production was close to zero in 2008 but grew to 4.25 million barrels per in 2016, standing with a 48% share of the U.S. oil production and 5% of the global production [10]. BRENT still tops the CV (23.27), followed by U.S. production (10.54) and CANADA (9.24), IRAN (7.11), and IRAQ (3.11) (Table 17). IRAQ showed the highest CV-producer position to the U.S., and the U.S. and CANADA continue to be active producers in the global oil market, with the U.S. shale oil and Canada oil sands.


**Table 17.** Panel 7 period (2016–2018): summary statistics.

A summary of parameter estimates in Table 18 displays that there are six significant influencing producers/organizations: IRAN, U.S., SAUDI, OPEC, CHINA, and UAE. Under the declining average price of U\$56 per barrel this period, there are two negative price influencers and four co-move price influencers. Negative influencers are OPEC and CHINA, with the effects of 3.98% and 1.48%, respectively, by one percent change in their production volume, while the co-move influencers are SAUDI, IRAN, US, and UAE, with the effects of 2.65%, 2.31%, 1.82%, and 1.37%, respectively, by one percent change in their production. This period is characterized by a downward price trend, with a low of U\$31 per barrel and BRENT price range of U\$50.

**Table 18.** Panel 7 period (2016–2018): LN\_BRENT prediction-parameter estimation.


Note: \* represents statistical significance at a level of 0.1.

### *4.10. Summary of Results from Seven Panel Periods*

In the CV approach to measure a level of fluctuations, the Brent oil price recorded the highest CV of 54.4 over the whole period of 21 years, among all the variables, and maintained a top position during most of the seven-panel periods. Among explanatory variables, Iraq (35) topped in the variability of the production volume, followed by the U.S. (26.5), Canada (25.4), and Russia (16.8), over the whole period and has held the position for all panel periods, except for the Panel 7 period, or 2016–2018, when the U.S. takes the top spot. Only the four countries that recorded a CV of 9.0 or above in each of the periods are Iraq, Iran, the U.S., and Canada, implying their capability to expand the production capacity, for the U.S. and Canada, or unstable production levels, for Iraq and Iran.

The multivariate regression approach was used to estimate a percentage change in the Brent oil price with respect to a percentage change in explanatory variables. Figure 6 displays a behavior over time (BOT) chart of coefficient estimates of each variable for each panel period. The coefficient estimates are present only when a variable is statistically significant at the 10% level; otherwise, the estimates are treated as zero in the chart. The U.S. Dollar Index (USDX) has been shown as the most influencing variable on the Brent oil price, with a coefficient estimate of −3.68, followed by Kuwait (−1.56), China (−1.5), and Iran (−0.62), in its magnitude.

**Figure 6.** Coefficient estimates of variables by producer and panel at a significance level of 0.1.

In line with the economic theory that a negative relationship exists between production quantity and a market price, Table 19 presents a summary of negative impact ranks based on significant coefficients at the 10% level. From the analysis, OPEC was not the first influencer in all the seven panels but just in two panels, Panel 2 (2001–2003) and Panel 7 (2016–2018).

**Table 19.** Negative impact rank, based on significant coefficient estimates at the 10% level.


While there were no significant influencing countries in Panel 1 (1998–2000) and Panel 3 (2004–2006), Panel 2 (2001–2003) has two influencers of OPEC (−3.15) and the U.S. (−1.84). In the other panels, USDX has appeared as the strongest influencer starting in Panel 4 (2007–2009), with −4.79, and then in

Panel 5 (2010–2012), with −2.29, and, in Panel 6 (2013–2015), with −6.03. Panel 7 (2016–2018) records both OPEC and China as two most influencing producers, with −3.98 and −1.48, respectively.

Contrary to the economic theory of production quantity and price, some oil producers' production quantities have shown a co-move relationship with the Brent oil price in some panel periods, indicating their increased production co-moved with higher prices or decreased production with lower prices. These producers include Russia, China, Saudi Arabia, UAE, Iran, and the U.S., with a coefficient estimate of positive 1.0 or greater, as shown in Table 20. China appeared with positive coefficients in Panel 2 (2001–2003) and Panel 3 (2007–2009) when there was a growing demand for energy due to their economic growth, and both Iran and the U.S. joined the co-move impact group in Panel 7 (2016–2018). Table 20 presents the co-move impact rank.


**Table 20.** Co-move impact rank, based on significant coefficient estimates at the 10% level.

### **5. Conclusions and Policy Implications**

We investigated the relationships between Brent oil price and major producers' volumes to identify influencing weights of major crude oil producers in the world oil market, as well as those of petroleum consumption of OECD countries and the U.S. Dollar Index (USDX), with seven panels of 36 monthly data each across the years of 1998 to 2018. This paper is the first attempt to investigate the relationship between global oil price and major individual producing countries at three-year intervals for analysis over 21 years. The results will help energy policymakers understand the variability of oil market drivers in the specific time horizons and, in particular, the influences of the producing countries on world oil prices in the ever-evolving energy environment.

In the volatility measure of the CV (coefficient of variation), the CV of Brent oil price has been greater than any of the explanatory variables in all of the seven panel periods, while Iraq production, on the production side, has shown the highest CV among all the producers in the past six periods, but the U.S. later took the top position in Panel 7 (2016–2018). Iraq, the U.S., Iran, and Canada, with high CVs of 9.0 or greater, may have potentials to continue exerting significant influences on the global oil price. Surprisingly, the OPEC production volume was not as volatile as in the individual producers in the 21-year seven-panel history.

Parameter estimates from multivariate regression models for the seven panels showed that there had been various influencing producers in each of the seven panels, at the statistical significance level of 10%, that have depressed the global oil price. OPEC appeared only twice in Panels 2 (2001–2003) and 7 (2016–2018), while the U.S., Iran, and China did once each. On the petroleum consumption side, OECD consumption also appeared once as a negative influencer in Panel 4 (2007–2009) when the crude oil price spiked to a \$140/bbl level in 2008. Notably, the U.S. Dollar Index (USDX), a macroeconomic variable, was the most prevalent influencer—three times in Panels 4, 5, and 6, or for 2007–2015. This result may be due to the fact that global oil is quoted and traded in U.S. dollars. Recent influencers in Panel 7 (2016–2018) were OPEC, the biggest producer in aggregate, and China.

We admit that there is a room for further researches to understand the dynamics of the global oil market. This paper focused on a panel of 36 months each, based on the assumption that a period of three years may be sufficient to bring a change to improve supply and demand situations in the oil markets, but with the advancement of technology, the panel period could be set shorter or longer for analysis, depending on the purpose of further research. Besides, as the OECD consumption in this paper is at the aggregate level of its member nations, further research may use individual nation's consumption data to investigate the influences of major/emerging economies in Europe, North America, and Asia.

**Author Contributions:** Conceptualization, P.Y.J.; Project administration, M.G.B.; Supervision, M.G.B.; Data curation, P.Y.J.; Formal analysis, P.Y.J.; Investigation, P.Y.J.; Methodology, P.Y.J.; Resources P.Y.J.; Software, P.Y.J.; Validation, P.Y.J. and M.G.B.; Visualization, P.Y.J. and M.G.B.; Writing - original draft, P.Y.J.; Writing- Reviewing and Editing, P.Y.J. and M.G.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The Oil Market Reactions to OPEC's Announcements**

**Yue Liu 1, Hao Dong 2,\* and Pierre Failler <sup>3</sup>**


Received: 21 July 2019; Accepted: 21 August 2019; Published: 22 August 2019

**Abstract:** Because of the crucial implications of the market power of OPEC, the aim of this paper was to investigate the oil asymmetric market reactions, such as the price and risk reactions, to OPEC's announcements. Specifically, this paper first explored the oil price reactions to OPEC's announcements and their heterogeneity to depict the directional role of OPEC based on event study methodology. Furthermore, this paper analyzed the oil risk reactions in the framework of a linear model. Our findings reveal several key results. The oil price reactions to OPEC production decisions behave quite heterogeneously in three kinds of decisions. Specifically, the reaction to announcements of a production increase shows an invert "U" shape, whereas there is a linear effect of cut announcements. Otherwise, when a maintain decision is announced, the oil prices have no obvious change over the sample period. Additionally, the oil risk reactions to OPEC's announcements are heavily related to the interaction item between OPEC decisions and its production over full sample periods. Furthermore, OPEC's role in promoting stability in crude oil markets by changing its production shows a heterogeneous condition after global financial crisis.

**Keywords:** market reactions; OPEC; oil price reactions; oil risk reactions; production decision

### **1. Introduction**

The geopolitical instability of the world supply sheds light on the dominant role of the Organization of the Petroleum Exporting Countries (OPEC) in the crude oil market [1,2]. On the one hand, the oil price reactions could be regarded as direct market reactions. Indeed, OPEC, as a cartel, influences the oil price through coordinating its members' production quotas. However, this impact is a conflicting issue among different stakeholders [3]. For one thing, the correlation between OPEC and oil price varies over a time horizon, because of the changing market share of non-OPEC oil production [4]. For another, the role of OPEC changes over time due to the market fundamentals. On the other hand, the risk reactions indirectly depict market responses to OPEC's announcements. As stated by Balcilar and Ozdemir [5], the higher oil price volatility is, the greater uncertainty and risk are for market participants. Moreover, the primary aim of OPEC is to assure the balance between producers and investors, in turn promoting market stability. In this vein, this paper explored the question of how oil price and risk react to OPEC's announcements. The logical relations among these three sides are shown in Figure 1.

**Figure 1.** The logical relations among oil price reactions and oil risk reactions to OPEC's announcement.

The oil price reactions could depict the market power of OPEC. As we all know, future oil production for OPEC members is announced by OPEC conferences, based on the outlook of the balance between oil supply and demand. Obviously, OPEC's exercise of market power restrains its production level to maintain higher oil prices and decrease its markets share of oil production related to competitors [1]. Because of the benchmark role on pricing oil and other related oil derivatives, Figure 2 presents the Brent spot price reactions to OPEC conferences from 2002 to 2018. Among them, OPEC announcements, especially those agreeing to cutting or increasing production, have diverse impacts on oil prices. Moreover, OPEC is better off making a maintain decision, for the most part, when taking the change in oil demand into account. This is due to the fact that OPEC exercises its market power to control the oil price [6]. In this sense, instead of precisely estimating OPEC's behavior, this paper investigated the oil price reactions to OPEC's announcements.

**Figure 2.** Brent spot prices and OPEC's announcements. The solid, red line represents the "cut decision", the green line shows the "increase decision", and the dotted, red line depicts the "maintain decision".

Since the seminal work by Draper [7], abundant papers have explored how announcements or major events are impounded into financial asset prices [8–12]. However, little literature deals with the price reactions to the different types of OPEC's announcements based on the event study methodology. Indeed, Draper [7] first attempted to explore the oil future price reactions between fall 1978 and 1980 and found that OPEC's announcements were impacted by investors' expectations. Recently, Ji and Guo [13] investigated that there are inconsistent reactions to OPEC's announcements. Loutia et al. [3] analyzed the crude oil prices as well as two benchmark indices over two sub-periods, distinguishing that oil price uniformly increased and were turbulent. They found that the oil price reactions to OPEC's announcements evolve over time and among decisions, and these reactions are more significant for production cut and maintaining. In addition, OPEC's announcements are also sensitive to the benchmark index. However, studies that model oil price reaction to OPEC's announcements during pre- and post-crisis periods are scarce.

Additionally, high return risks create increased uncertainty in crude oil markets. Indeed, the oil return risks could inhibit current investment for crude oil [5]. Changes in market stability have considerable impacts on economy [14–18]. First, changes in crude oil return risks essentially impact decisions made by market participants, such as oil producers, consumers, and policy makers. Otherwise, changes in oil return risks also determine investors' decisions for pursuing maximum profits. In this sense, this paper further explored the oil risk reactions to OPEC's announcements. Theoretically, there are asymmetric risk reactions to OPEC's announcements over time horizons. The announcement of OPEC's decision generally exerts a simultaneous impact on oil return risks, that could be regarded as a source of participants' attention [19]. Because of the nature of OPEC's announcements, its role is perceived differently by the global crude oil market fundamentals.

Not surprisingly, an increasing number of studies have been exploring the consequences of OPEC's power to promote market stability. For example, models are often constructed to conduct theoretical and empirical analyses on OPEC announcements' role on oil prices, using production quotas, market competition, and spare capacity [20–24]. Besides, a common belief is that the market role of OPEC changes between an oil producer and a cartel, which in turn makes oil markets deviate largely from competition [25]. Another view is related to economic, market, and geopolitical conditions, indicating that OPEC's behavior is time-varying and this effect cannot be described by a single model [21]. Moreover, the linkage between oil price volatility and OPEC's announcements has been analyzed in the literature. For instance, Schmidbauer and Rösch [26] provided evidence that there is a strong effect on oil volatility as there is an asymmetric post-announcements effect on volatility. Mensi et al. [27] argued that the "cut" or "maintain" decisions play a dominant role on the oil price volatility. Furthermore, the potential ability in OPEC members to stabilize the oil price could be related to the spare capacity [28]. However, sparse attention has been paid to the analysis of oil risks.

The main contributions, which were the primary objectives of this paper, were to explore the oil asymmetric market reactions to OPEC's announcements. First, based on an event study methodology, this paper contributes to research on the oil price reactions to OPEC's announcements during distinct periods (pre- and post-financial crisis). More precisely, this objective aims to depict the direct market reactions to OPEC's announcements as mentioned above. Our second contribution is to explore the indirect market reactions to OPEC production decision, as well as investigating the heterogeneity during two distinct periods. To achieve this goal, we investigated the asymmetric risk reactions to OPEC's announcements in a framework of a linear model. Unlike existing studies focused on the influence of OPEC announcements to oil price volatility, we explored the asymmetric impacts on oil return risks. Additionally, Lin and Tamvakis [29] did not find a significant difference between oil qualities. Thus, the potential reactions to OPEC's announcements as the regressors are explicitly specified in a linear regression framework. Moreover, we divided the sample periods into two sub-periods because of the potential shock of global financial crisis in crude oil markets [30]. Importantly, this allows us to assess the heterogeneous role of OPEC in oil markets after global financial crisis.

The remainder of this paper is organized as follows. Section 2 briefly introduces the research hypothesis supported in this paper. Section 3 presents and discusses the oil price reactions to OPEC's announcements. In Section 4, we proceed to model the oil risk reactions to OPEC's announcements. Finally, the conclusions and policy implications are described in Section 5.

### **2. Research Hypothesis**

The oil price reactions to OPEC's announcements are heterogeneous, during both pre- and post-crisis periods. It is well known that the oil price reactions can be foreseen by market participants. In other words, the oil price reactions to these decisions are asymmetric. Specifically, an increase decision in oil supply can drive price up. In terms of oil producers, the optimistic attitude held by OPEC members for an increase in production plays a dominant role in crude oil markets. For other investors, they cannot change the investment decision to an increase production because of their expectations. In this sense, the expectations of market participants promote them to pursue higher profits, which in turn lead to the oil price increase. Additionally, the oil price reaction to increase announcements is heterogeneous during pre- and post-crisis periods. This is not surprising due to the market fundamentals. For one thing, OPEC can be regarded as a marginal producer in order to partially control the imbalance between global oil demand and oil supply for non-OPEC producers

through its members' quota. For another, non-OPEC oil producers can be regarded as price takers, meaning that there are few price reactions during pre-crisis periods. Thus, the oil price reactions to increase announcements vary during the post-crisis periods. Taking all these considerations together, it is obvious that the reactions of oil price to an increase announcement could be interpreted by market participants.

The picture changes a little for quota reduction. Following a cut decision, oil prices should go up. Indeed, oil prices heavily depend on the expectations of market participants. When a cut production decision is announced, it means that the production quota incudes the crude oil market's anticipation and the information on evolution of oil price. At this time, the imbalance between oil supply and demand could be expanded by quota reduction, thus driving oil price up. However, a different story may emerge. Despite the quota reduction, the oil prices could continue to decrease. This picture is more pronounced during post-crisis periods. The possible explanation for decreasing price is that the cuts in oil production perceived in the market are not far-reaching enough, or they suggest that OPEC's exercise of market power is not enough to enforce the quota reduction on its members. This apparently counterintuitive story could also indicate expectations about oil supply and demand and market participants' concern of the market power of OPEC as a cartel, as well as the actual oil production. Moreover, the upward pressure on oil prices could be impacted by world economic growth and spare capacity in oil producers.

In the case of maintain decision, the double reactions of oil price to OPEC announcements are related to market fundamentals. On the one hand, in crude oil markets, a maintain decision could be regarded as "firm indecisiveness". In this sense, it is easier to consider that members' expectation to market fundamentals should be a negative, but different interests of OPEC members could make it difficult to change the status of oil production, which in turn flow the oil supply and drive oil price further down. On the other hand, OPEC also focuses on other economic parameters. Specifically, market fundamentals, expectations of economic growth, or geopolitics could change the OPEC production quota. In other words, that the market interprets an unchanged decision in oil production means there is a sufficient level of supply. Thus, oil prices could continue to increase. Taking all these effects together, it is obvious that the oil price reaction is moderate. However, there are heterogeneous reactions of oil price during pre- and post-crisis periods. These results could be interpreted by depicting non-conventional resources, such as shale formations and natural gas production, and the increasing role of non-OPEC oil production in crude oil markets.

Accordingly, this paper proposes the following hypothesis:

**Hypothesis 1.** *The oil price reactions to OPEC's announcements vary across production decisions during pre- and post-crisis periods. Specifically, the oil price reaction to the increase or cut in production is dramatic, whereas it is moderate when the production remains unchanged.*

There are asymmetric reactions of oil risk to OPEC's announcements and these reactions heavily relate to production decision and quota level. The key reason for oil risk reactions is the uncertainty of oil supply or demand. Obviously, financial risk captures the loss but not profit in crude oil markets [31]. Indeed, oil risks could be defined as the excessive return on oil prices during distinct periods, which is a desirable characteristic in crude oil market applications to capture the market stability [32]. The different OPEC announcements could result in conflicting expectations in the oil markets and promote different investment decisions for speculation or hedging, thereby resulting in the asymmetric reactions of the oil risks [33]. From the perspective of an unchanged decision in oil production, it means that the quota level is tolerable in the oil markets. Meanwhile, the upward pressure in the oil markets caused by economic factors and uncertainty over supply and demand could be regarded as an unstable resource of oil risks. Thus, the oil risk reactions to the maintain decisions are the largest over these decisions. Additionally, the reactions of oil risks to the production increase decisions are larger than those to the cut decisions. This asymmetric picture could be related

to the expectation of market participants. Indeed, the production hike decisions could be regarded as "good news" for market investors. Speculation is blamed. It is well known that speculative flows accentuate the oil risks. On the contrary, the production cut decisions will lead to the oil price increase and further ensure the interest of different stakeholders. However, changes in oil production could also lead to the uncertainty of oil supply or demand, which results in the uncertainty of market stability. Therefore, it can be noticed that OPEC's announcements could promote the oil risks instead of being a stabilizing force. As mentioned above, it can be noted that:

**Hypothesis 2.** *The oil risks reactions depend on the interaction between OPEC's announcements and OPEC oil production.*

Moreover, there are heterogeneous reactions of oil risks to OPEC's announcements during pre- or post-crisis periods. Obviously, the recent global financial crisis could lead to some new changes in oil markets, such as the oil pricing mechanism, oil supply and demand for several countries, and the dependence between oil markets and other financial markets [34]. Therefore, this heterogeneous story may be caused by these differences. At a more specific level, during the pre-crisis periods, the production increase decisions significantly promote the market stability, whereas the oil risk reactions to another production decision are unapparent. As shown in Figure 2, the oil price increases significantly in pre-crisis periods, and its volatility is moderate. It is also interesting to note that the oil price experienced a sharper change for production hike decisions than others. This is not a surprise but due to the determinant role of oil supply factors during pre-crisis periods [35]. Theoretically, in a relatively weak market, the production hike decisions could mean an optimistic attitude for decision makers, which will lead to reasonable expectations of market participants and promote the demand in oil markets. Furthermore, these characters are expected to decrease the oil price volatility and stabilize the crude oil markets.

As the sample moves out to post-crisis periods, the oil risk reactions to different OPEC announcements depict a different picture. Inconsistent with pre-crisis periods, there are obvious reactions of oil risks to the cut decisions, and the oil risk reactions to the production maintain decisions are heavily related to the interaction item between decision and OPEC oil supply. As we all know, the key reason for these results is the uncertainty of oil supply or demand. Additionally, another reasons of oil markets' stability are the attitudes for production change in OPEC announcements. Indeed, these different pictures during post-crisis periods are determined by the sharp increase in oil demand and the change in market share for non-OPEC countries. For another, the maintain decision is often perceived as non-decisions. In this vein, the oil price would tumble further down, leading to different expectations of market participants and promoting different capital flows in the crude oil markets [36]. Additionally, economic parameters or geopolitical events could also lead to the reactions of oil risks to OPEC's production decisions. Furthermore, the diverse interests of OPEC's members make it difficult to reach production decisions and further promote the oil risks. Therefore, this paper concludes with the following hypothesis:

**Hypothesis 3.** *The oil risk reactions to OPEC's announcements in a sense are heterogeneous during pre- and post-crisis periods.*

### **3. The Oil Price Reactions to OPEC's Announcements**

### *3.1. Event Study Methodology*

To exam the asymmetric price reactions to different OPEC announcements in crude oil markets, we used an event study methodology. It is easier for the event study methodology to assess the significance of global crude oil price reactions to OPEC announcements [37]. Furthermore, the expected return of oil price could be good for capturing the seasonality and rationality features [29]. In addition, because of the change in Brent crude oil demand of emerging countries and the controlling crude

oil export of the US government in recent years, Brent has become the international oil benchmark. Although Brent has been activated in Europe, it can also be used to assess the basic price for other grades. In this way, we examined the effects of OPEC announcements on the Brent oil spot price. In this section, we used the daily price covering the period from 1 January 2002 to 31 December 2018. Data was collected from the US Energy Information Administration.

The point of using event study was to forecast the abnormal returns of oil prices during the event window of OPEC announcements. This was measured in Formula (1).

$$AR\_t = R\_t - \mathbb{E}(R\_t) \tag{1}$$

where *ARt* is the abnormal return of the oil price at *t*; *Rt* is the log-return on crude oil price, and E(*Rt*) stands for the expect return, which assumes that the OPEC announcement is not taking place (following Ji and Guo [13] and Lin and Tamvakis [29], E(*Rt*) was equal to 0 in this paper, which indicates that there will be no change in the price over the short term without exogenous shocks).

The cumulative abnormal returns (CARs) can be regarded as the sum of the daily abnormal returns over the event window. In this paper, an event window of 11 days (five days before and five days after the event date) was set (see also Loutia et al. [3] and Ji and Guo [13]). The reasons we used this choice are as follows. Firstly, the five days before the event date could better depict the picture before the OPEC announcements. Additionally, the five days after the event date capture the information after the decisions. Furthermore, this choice can also better avoid the interactional effects from other events. Thus, the CARs can be defined as in Formula (2).

$$\text{CARS}\_{l} = \sum\_{t=-5}^{5} AR\_{l}, \ i = 1, 2, 3. \tag{2}$$

where *i* refers to the type of OPEC announcements.

Then, this paper calculated the average *CARs* for different types of OPEC announcements with Formula (3).

$$\overline{CAR}\_{\ell} = \frac{1}{N} \sum\_{c=1}^{N} CAB s\_{i\_{c}\ell} \,. \tag{3}$$

where *N* is the number of different types of OPEC announcements over the sample period, and *t* is the date of event windows, which equals from −5 to 5. In this vein, we obtained the average CARs of different types of OPEC announcements during the event windows.

### *3.2. The Reaction of Oil Price to Announcements*

In this subsection, we present the average cumulative abnormal returns for different categories. Table 1 represents the OPEC's announcements from 1 January 2002 to 31 December 2018. There were 57 OPEC regular and extraordinary conferences which can be divided into three categories according to different decisions. Of those decisions, 13 announced production cuts, 10 announced production increases, and 34 announced production maintenance. The details in Table 1, such as the dates, the production decisions, and the price series of Brent spot prices (in USD/barrel) are available upon request. We set the OPEC announcement day as *t* = 0 in Formula (2) when the actual announcement was released.

Figure 3 plots the trend in the average CARs for three types of OPEC announcements. According to existing literature, the result that the price reactions to announcements are generally not significant could not be related to the choice of commodity index [38–44]. This phenomenon could be correlated with OPEC's market power. For example, Lin and Tamvakis [29] explored the oil price reactions to OPEC's announcements from 1982 to 2008 and found diverse market reactions to OPEC's announcements under different price bands. Most generally, the oil price reactions to OPEC's announcements may be heterogeneous during distinct periods. Another argument is put forward by Loutia et al. [3]. He explained these heterogeneous results by an explanation that it is easier for market participants to agree to an increase decision, whereas the reactions of oil price to the production cut decision reflects the expectation of market participants. However, economic parameters, such as economic growth, geopolitical events, and spare capacity, can also be regarded as causes of oil price change.

**Table 1.** OPEC's announcements.

**Figure 3.** Average CARs of the Brent spot prices for OPEC's announcements from five days before the event to five days after the event.

As expected, the oil price reactions to OPEC announcements display obvious asymmetry. These results follow the expectation that an increase production announcement shows an invert "U" effect, whereas there is a linear reaction to cut decision. In addition, it seems that if the OPEC maintains its production, there is no significant reactions of crude oil prices. This indicates the role of market fundamentals on crude oil markets. On the one hand, market participants focus on digesting the newly released announcements, and the influence of the cut announcement has a strong persistence over a short period of time. On the other hand, the maintain decision often means stability of the market, thus the market responds moderately to it. Furthermore, this paper examed the heterogeneous reaction of oil price to OPEC's announcements during pre- and post-crisis periods. Figure 4 presents the average CARs for the three types of OPEC decisions.

The picture changes a little during different sample periods. As shown in Figure 4, it is clear to note that the oil price reactions to OPEC's announcements were heterogeneous during distinct periods. In particular, an increase decision will drive prices up during pre-crisis periods. However, the increase announcement could drive oil prices to drop first, and then oil prices will go up during post-crisis periods. This is not a surprise due to the fact that OPEC often acts as a marginal producer in order to offset, whereas non-OPEC is generally considered as a price taker during pre-crisis periods [37,38]. Thus, the greater the OPEC supervision is, the greater reaction oil prices have. However, the market power of OPEC could be limited by its market share.

**Figure 4.** Average CARs of Brent spot price for OPEC's announcements during two periods. Brent's reaction 1 refers to the oil price reaction during pre-crisis periods, whereas Brent's reaction 2 represents the price reaction during post-crisis periods.

When a cut decision is announced, the reaction seems more sensitive to oil price during pre-crisis periods than in the post-crisis periods. Following a cut decision, oil prices should go up, thus the average CARs should be positive. This is indeed the case for the pre-crisis periods for Brent spot price. It is particularly interesting to note that the average CAR is always negative in pre-announcement period (*t* = −5, −4, ... , −1) during pre-crisis periods. This interesting information could be depicted by putting the excess return and market power of OPEC into the reasoning. In fact, OPEC's members exercise their market power to manipulate the oil prices down through disseminating the information about production decision for extreme profits. However, during the post-crisis periods, the average CAR is negative. This obviously unreasonable result may indicate considerations about the balance between oil supply and demand and market participants' attention to the power of OPEC as a cartel, as well as the change in production [39]. Additionally, as shown in Figure 2, OPEC's cut decisions announced in 2008 sent oil prices significantly down, and before a sharpened increase in 2009 during post-crisis periods. In fact, from 2009 to 2011, economic parameters such as economic growth and oil demand were active, whereas the lower oil production and spare capacity held by oil producers put upward pressure on oil prices. Finally, it is particularly interesting to note that the oil price reaction is moderate when the production quotas remain unchanged. As mentioned above, we can confirm that Hypothesis 1 is valid.

### **4. The Oil Risk Reactions to OPEC's Announcements**

### *4.1. Linear Model Specification*

Having found an asymmetric reaction to OPEC's announcements of oil price during distinct periods, we further empirically identified the oil risk reactions to OPEC's announcements. Our dataset covered the period from January 2002 to December 2018. Considering the data availability for the control variable, we used monthly data to explore the oil risk reactions to OPEC announcements, retrieved from OPEC. For this purpose, the empirical model to be estimated in this paper is shown in Formula (4). Additionally, we also provide a brief description of all variables in Table 2.

$$\text{Risk} = \text{c} + a\mathbf{D} + \beta \mathbf{X}\_{\text{i}}^{'} + \varepsilon \tag{4}$$

where Risk refers to the oil return risks; **D** is a vector of dummy variables to measure the periods of OPEC's announcements; **X** stands for control variables, and ε is the error term.


**Table 2.** Description of variables.

The dependent variable Risk is oil return risks. As we all know, value-at-risk (VaR) is defined as the maximum loss in oil markets. A large body of studies pays attention to some alternative methods to forecast the risks [45–49]. Commonly, these methods often assume that the oil return is invariable. In general, the speculative and intra-cluster make oil return risks more dynamic [50,51]. Therefore, we use the CAViaR, proposed by Engle and Manganelli [52], to calculate the oil return risks in this paper. Specifically, compared with other specifications, we forecast the oil return risks based on the asymmetric slope from the results of a dynamic quantile (DQ)test at the 5% level (these results are available upon request).

The main explanatory variables are OPEC's announcements. The OPEC's announcements are coded based on the OPEC decisions on oil production. To solve this problem, we defined three dummy variables. Take increase (inc) as an example: the measurement of inc is shown as Figure 5 (other dummy variables are the same as inc). If an increase production decision is announced at time b, we could regard the period from a to b as an increase sample.

**Figure 5.** Example for inc: a and b represent the date of the OPEC announcement.

The control variables included commonly used background variables. Referring to the OPEC monthly oil market reports (MOMRs), we considered several main strands of potential factors, including oil supply and demand, world economy, and spillover from other markets. For the perspective of oil supply and demand, we selected OPEC oil production and OECD commercial oil stocks, respectively, because of their market power. Additionally, the growth rate of the global economy could better depict the world economy. In terms of spillover from other markets, it is well known that oil futures or product markets play a dominant role on oil spot price. Thus, we selected the log-return of NYMEX WTI and the gasoline price. All data were monthly observations and collected from OPEC.

Above all, we estimated the following empirical specifications in different periods. Since in Section 3 we found that oil price volatility was low when the production remained unchanged, this paper regarded inc and cut as the regressors. This can be seen in Formulas (5) and (6).

$$\text{Risk} = \text{c} + a\_1 \text{inc} + a\_2 \text{cut} + \beta\_1 \text{OP} + \beta\_2 \text{STO} + \beta\_3 \text{WGDP} + \beta\_4 \text{FR} + \beta\_5 \text{PRO} + \varepsilon. \tag{5}$$

$$\begin{array}{l}\text{Risk} = & \mathbf{c} + a\_1 \text{inc} + a\_2 \text{cut} + \beta\_{11} \text{OP} \times \text{cut} + \beta\_{12} \text{OP} \times \text{mai} + \beta\_{13} \text{OP} \times \text{inc} \\ & + \beta\_2 \text{STO} + \beta\_3 \text{WGDP} + \beta\_4 \text{FR} + \beta\_5 \text{PRO} + \varepsilon. \end{array} \tag{6}$$

Our interest mainly lies in the α1, α2, β11, β<sup>12</sup> and β<sup>13</sup> coefficients which provide information on the asymmetric effect between OPEC's announcements and oil return risks.

### *4.2. The Reactions of Oil Risks to Announcements*

Table 3 shows the estimated results for the full sample. In the first column, the results without the interaction item between OPEC production and announcements are given. In columns 2–5, we augmented the interaction item between OPEC production and its announcements.


Note: standard errors are shown in parentheses. \*, \*\* and, \*\*\* indicate significance at 0.10, 0.01, and 0.001, respectively.

Some interesting results, correlated with the OPEC announcement variables, are shown in Table 3. We noted that there were asymmetric risk reactions to different types of OPEC announcements. Actually, the individual OPEC production announcements had an effect, which heavily correlated with the OPEC oil supply. It is worth noting, refer to column 5 in Table 3, that the interaction item between OPEC maintain decisions and its production had the highest positive influence on oil return risks, while it, with cut decision, was the smallest. This is due to the fact that the market participants' attention to different types of production decisions was asymmetric. When a maintain oil production is announced, the OPEC oil production can promote the instability of the oil markets because of the participants' expectations [53–56]. Therefore, oil return risks will increase by impacting the market participants' expectations when the oil supply is opposite to the OPEC announcements. Additionally, oil return risks could be related to forecasting future supply, which is associated to the market conditions [57]. Taking all these considerations together, OPEC's announcements could be regarded as a sign in the oil market, signaling a greater market power of OPEC and, thus, resulting in uncertainty in the oil market.

Considering the increasing market share of non-OPEC producers and the global effect of the financial crisis in 2008, this paper further explored the heterogeneous effects of OPEC's announcements on oil return risks during distinct periods. The estimation of Formulas (5) and (6) are reported in Table 4.

A comparison of Table 4 reveals there are twofold heterogeneous risk reactions to OPEC's announcements during pre- and post-crisis periods: the single and interaction items. From the perspective of the single effect of OPEC's announcements, it is interesting to note that there are significant risk reactions to OPEC increase decisions during pre-crisis periods (this result refers to column 1 during the period 2002M01–2008M08), while the "cut decision" plays a positive role on oil risks (this result refers to the column 5 during the period 2008M09–2018M12). This is not a surprise but is due to the heterogeneous market conditions during pre- and post-crisis periods [58]. Indeed, the change in oil production enhances the demand in the crude oil markets and the participants' expectations, thereby promoting the stability of oil markets during pre-crisis periods. Contrary to this

mechanism, the cut decisions in production are more sensitive during post-crisis periods than they are in first sub-periods. Obviously, the cut decision could change the fundamental balance between oil supply and demand in crude oil markets, which can influence the oil return risks level.


**Table 4.** Estimation results during pre- and post-crisis periods.

Note: standard errors are shown in parentheses. \*, \*\* and, \*\*\* indicate significance at 0.10, 0.01, and 0.001, respectively.

In terms of the interaction item, OPEC's announcements impose their significant influence on oil return risks only when the production remains unchanged during post-crisis periods (this result refers to column 5 in Table 4). However, other periods were not significantly dependent on the OPEC oil supply. These results indicate that the unchanged decisions were often regarded as non-decisions for different reasons. Economic parameters, such as market fundamentals, economic conditions or geopolitical events could result in OPEC changing its quota. Additionally, it is generally difficult for OPEC members to reach an announcement, and market investors heavily relate to OPEC members' degree of execution [59]. Thus, it could not be concluded that the market power of OPEC to promote the market stability by changing its production level has diminished. On the contrary, its role shows a heterogeneous conditions after global financial crisis [60]. As mentioned above, we can confirm that Hypotheses 2 and 3 are valid.

### **5. Conclusions and Policy Implications**

The market reactions to OPEC's announcements can reflect the market power of OPEC and the expectations of market participants in the global crude oil market. Specifically, the empirical results with the event study methodology and with a framework of a linear model all point to the asymmetric market reactions to different OPEC announcements, as well as heterogeneous reaction during pre- and post-crisis periods. In this paper, we first explored the oil price reactions to OPEC's announcements and the heterogeneity to depict the directional role of OPEC based on the event study methodology. Furthermore, this paper analyzed the oil risk reactions in a framework of a linear model. Specific conclusions are as follows.

The oil price reactions to different OPEC announcements had stronger differences in crude oil markets, as well as during distinct periods. In general, the reactions to the announcements of a production increase showed an invert "U" shape, whereas there was a linear reaction to cut decisions. In addition, when an unchanged decision was formulated, the oil prices had no obvious change over the sample period. According to the diverse mechanisms during pre- and post-crisis periods, this paper found an increase in oil production will drive prices up during pre-crisis periods, whereas it will drive oil prices to drop and then rise again dramatically during post-crisis periods. When a cut decision was announced, the reaction seemed to be more sensitive to oil price during pre-crisis periods than it was in the post-crisis periods. Additionally, the oil price reactions were moderate when the production quotas remain unchanged.

The oil risk reactions to OPEC production decisions behaved quite heterogeneously in the three kinds of decisions. Actually, it is interesting to conclude that the oil risk reactions to OPEC's announcements were heavily related to the interaction item between OPEC decisions and its production over the full sample periods. Specifically, the interaction item between OPEC maintain decisions and its production had the highest positive influence on oil risks because of the market uncertainty. The reactions of oil risks to the production increase decisions were larger than the cut decisions. The picture changed a little during pre- and post-crisis periods. There were twofold heterogeneous reactions of oil return risks to OPEC's announcements: the single and interaction items. From the perspective of the single item, there were significant negative risk reactions to the OPEC increase decision, while the "cut decision" played a positive role on oil risks during post-crisis periods. In terms of the interaction item, OPEC's announcements imposed their significant influence on oil return risks only when the production remained unchanged during post-crisis periods. However, there were no significant reactions to other OPEC announcements.

These interesting results have implications for different stakeholders. On the one hand, the expectations of market participants play a dominant role on monitoring the instability caused by oil prices and risks. Thus, policymakers need to guide a reasonable expectation to ease the oil price volatility and decrease the oil return risks. On the other hand, investors pay attention to the maximization of profits. Thus, investors had better focus on the market conditions and the policy information. Additionally, they should also be familiar with economic parameters and make reasonable investments. Last but not least, the heterogeneous role of OPEC cannot be ignored by market participants in crude oil markets. Therefore, they should focus on the OPEC meeting decisions.

This paper was not without limitations. For example, we neglected the heterogeneous reactions to OPEC announcements during different trends in oil price, such as decrease or increase. This allows investors to make short investments to maximize their profits. Thus, we could further study this reaction by dividing the sample periods into two sub-periods. Additionally, it is interesting to note that the price gap between Brent and WTI fluctuated before and after 2011. Thus, further analysis about the heterogeneous reaction by adding more benchmark oil prices could be regarded as a valuable area. Moreover, the role of investor sentiment in crude oil markets could be further explored with the development of internet finance.

**Author Contributions:** Conceptualization, Y.L., H.D. and P.F.; Data curation, Y.L. and H.D.; Formal analysis, Y.L., H.D. and P.F.; Funding acquisition, Y.L.; Investigation, Y.L. and H.D.; Methodology, H.D.; Resources, Y.L. and P.F.; Software, H.D.; Supervision, Y.L. and P.F.; Validation, P.F.; Visualization, Y.L., H.D. and P.F.; Writing—original draft, Y.L., H.D. and P.F.; Writing—review and editing, Y.L., H.D. and P.F.

**Funding:** This research was funded by the Hunan Natural Science Foundation, grant number 2019JJ50111.

**Acknowledgments:** The authors are grateful to reviewers for helpful comments. The authors would like to thank Guangzhou University for sponsoring this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
