Aino Dahlbom The relationship between oil prices and the BRICS stock markets before and after the onset of Covid- 19 pandemic Differences in oil importing and oil exporting countries Vaasa 2023 School of Accounting and Finance Master’s thesis in Finance Programme 2 UNIVERSITY OF VAASA School of Accounting and Finance Author: Aino Dahlbom Title of the Thesis: The relationship between oil prices and the BRICS stock markets before and after the onset of Covid-19 pandemic Degree: Master’s degree in Finance Programme: Finance Supervisor: Vanja Piljak Year: 2023 Pages: 79 ABSTRACT: The purpose of this study is to examine the relationship between the BRICS stock markets and global oil prices. The paper investigates how the fluctuations in oil prices impact the stock re- turns in BRICS countries before and after the onset of Covid-19 pandemic. The BRICS countries have become an important part of the global financial markets, and they are among the fastest growing economies in the world. Hence it is essential for market participants to understand the influence of different global factors, such as oil price, on these markets. In addition to their im- portance for the global financial markets, the BRICS countries include some of the most im- portant oil exporters and oil importers in the world. Therefore it will be interesting to examine whether there are differences in the oil price-stock market relationship between the oil import- ing and oil exporting BRICS economies. The study uses the daily MSCI index returns of each BRICS stock market as well as the Brent and WTI oil prices. The daily returns are analyzed over a period from November 19, 2014 to January 30, 2023. This study utilizes a multiple linear regression model in order to empirically examine whether there is a significant relationship between the global oil prices and the BRICS stock re- turns during the study period. In order to control the macroeconomic circumstances, exchange rate and interest rate changes are added to the model as control variables, in addition to the global market returns. The empirical analysis finds evidence that, in general, there is a significant relationship between the BRICS stock returns and global oil prices. The degree of the relationship is not consistent through the whole study period, and it varies between oil importing and oil exporting countries. Oil exporting countries have a positive and generally stronger relationship with the oil price movements, whereas in oil importing countries, the relationship is mostly negative. Additionally, in contrast to previous studies, not much evidence is found of a stronger relationship since the onset of the Covid-19 pandemic. KEYWORDS: Oil prices, stock returns, BRICS, emerging economies, Covid-19 pandemic 3 VAASAN YLIOPISTO Laskentatoimen ja rahoituksen yksikkö Tekijä: Aino Dahlbom Tutkielman nimi: The relationship between oil prices and the BRICS stock markets before and after the onset of Covid-19 pandemic Tutkinto: Kauppatieteiden maisterin tutinko Oppiaine: Rahoitus Työn ohjaaja: Vanja Piljak Vuosi: 2023 Sivumäärä: 79 TIIVISTELMÄ: Tämän tutkimuksen tarkoituksena on tutkia BRICS osakkemarkkinoiden ja globaalien öljyn hin- tojen välistä yhteyttä. Tutkimus selvittää, kuinka heilunnat öljyn hinnassa vaikuttavat osaketuot- toihin BRICS maissa ennen koronapandemian alkua sekä sen jälkeen. BRICS maista on tullut tär- keä osa globaaleja rahoitusmarkkinoita ja ne ovat maailman nopeiten kasvavien talouksien jou- kossa. Sen vuoksi markkinaosapuolille on tärkeää ymmärtää globaalien muuttujien, kuten öljyn hinnan, vaikutus näillä markkinoilla. Sen lisäksi, että BRICS mailla on merkittävä vaikutus globaa- leihin rahoitusmarkkinoihin, nämä maat ovat yksiä maailman tärkeimpiä öljyn viejiä ja tuojia. Niinpä on mielenkiintoista tutkia onko öljyn hinnan ja osakemarkkinoiden välisessä suhteessa eroja niiden BRICS maiden välillä, jotka ovat öljyn tuojia ja niiden, jotka ovat öljyn viejiä. Tutkimus käyttää päivittäisiä MSCI indeksituottoja jokaiselta BRICS osakemarkkinalta ja lisäksi Brent ja WTI -öljyjen hintoja. Päivittäisiä tuottoja analysoidaan ajanjaksolla, joka alkaa 19. mar- raskuuta, 2014 ja päättyy 30. tammikuuta, 2023. Tämä työ hyödyntää usean muuttujan lineaa- rista regressioanalyysia tutkiakseen empiirisesti, onko globaalien öljyn hintojen ja BRICS osake- tuottojen välillä merkittävää suhdetta tutkimusajanjakson aikana. Jotta makroekonomisia olo- suhteita voidaan kontrolloida, valuuttakurssi- ja korkokurssimuutokset lisätään regressioon kontrollimuutujiksi, globaalien markkinatuottojen lisäksi. Empiirisen analyysin tulokset osoittavat, että yleisesti BRICS osaketuottojen ja globaalien öljyn hintojen välillä on merkittävä suhde. Suhteen aste ei ole yhtenäinen koko tutkimusjakson ajan ja se vaihtelee öljyn tuonti- ja öljyn vientimaiden välillä. Öljyn viejämailla on positiivinen ja ylei- sesti vahvempi suhde öljyn hintojen kanssa, kun taas öljyn tuojamailla suhden on enimmäkseen negatiivinen. Lisäksi, verrattuna aiempiin tutkimuksiin, vahvemmasta suhteesta koronapande- mian alkamisen jälkeen ei juurikaan löydetty todisteita. AVAINSANAT: Oil prices, stock returns, BRICS, emerging economies, Covid-19 pandemic 4 Contents 1 Introduction 6 1.1 Purpose of the study 9 1.2 Research hypothesis 10 1.3 Structure of the study 10 2 Financial theory background 12 2.1 The efficient market hypothesis 12 2.2 Stock market pricing 13 2.2.1 Capital asset pricing model 15 2.2.2 Factor models 17 3 World oil markets 21 3.1 Oil price determination 23 3.1.1 Supply 24 3.1.2 Demand 27 3.1.3 Inventories 29 3.1.4 Financial markets 29 4 Theoretical relationship between oil prices and stock returns 31 5 Literature review 34 6 Data and summary statistics 45 7 Methodology 50 8 Empirical results 52 9 Conclusion 63 Appendix 67 References 72 5 Figures Figure 1. World Primary Energy Consumption by fuel in 2019 21 Figure 2. Change oil demand from 2019 to 2020 22 Figure 3. Changes in Saudi Arabia crude oil production and WTI crude oil prices 25 Tables Table 1. Descriptive statistics 48 Table 2. Descriptive statistics for the two crisis sub-periods 49 Table 3. Results for the multiple regression model with Brent in pre-crisis period. 53 Table 4. Results for the multiple regression model with WTI in pre-crisis period. 54 Table 5. Results for the multiple regression model with Brent in full crisis-period. 54 Table 6. Results for the multiple regression model with WTI in full crisis-period. 55 Table 7. Results for the multiple regression model with Brent in first crisis-period. 55 Table 8. Results for the multiple regression model with WTI in first crisis-period. 56 Table 9. Results for the multiple regression model with Brent in second crisis-period. 56 Table 10. Results for the multiple regression model with WTI in second crisis-period. 57 6 1 Introduction Oil has been the world’s most traded commodity since 2000 and its importance in our society is well known. Oil is one of the key facilitators of our life as it is needed for nu- merous different purposes. Crude oil is used, for example, in transportation, heating and electricity generation as well as in the production of different plastics, chemical products and synthetic materials. Since oil is such an essential commodity, it has a significant in- fluence on the global economy, and it is considered as an important driver of the eco- nomic and industrial activity. Therefore, movements in oil prices are closely followed by both policymakers and investors. (Ali, Mensi, Anik, Rahman and Kang, 2020; Aloui, Ngu- yen and Njeh, 2012.) Furthermore, some previous studies provide evidence that oil prices negatively affect economic growth through different channels, such as inflation, rising production cost, and investor confidence (Lardic and Mignon, 2008; Killian, 2008; Hamilton 2003). Therefore, it could be assumed that there is at least some degree of interdependence also between oil prices and stock market returns. The correlation between oil prices and stock returns might be either negative or positive. For example, because oil is an important input for many firms, increasing oil prices are reflected in companies as increased production costs, which leads to decreasing stock returns. However, crude oil prices often reflect market expectations regarding future macroeconomic variables, like for example, aggregate demand, implying that increases in oil prices lead to increases in demand (Heinlein, Legrenzi and Mahadeo, 2021). Kollias et al. (2013:744) also note that investors often associate higher oil prices with a booming economy, implying stronger business performance. These assumptions, on the other hand, are drivers of a positive correlation between stock returns and oil prices. When assessing the relationship between oil prices and stock returns, one should distin- guish between oil importing and oil exporting countries. Wang et al. (2013) have found that the magnitude, duration, and even direction of stock markets’ response to oil price shocks are highly dependent on whether the country is oil importer or exporter. In- creases in oil prices could be expected to have positive effects on stock returns in 7 countries which are oil exporters, because higher oil prices will increase the country’s income. On the other hand, in oil importing countries, increased oil prices could be ex- pected to have a negative effect on stock returns, as oil is one of the main inputs of production. For global investors, who are seeking for international diversification benefits by includ- ing assets of emerging markets into their portfolios, new information about the effects of oil on the assets’ risk-return profile in these markets is especially important as they can better manage the risk incorporated to their portfolios (Aloui et al., 2012). Further, assessing the impacts of oil price fluctuations on emerging stock market returns is im- portant for number of reasons. As a result of financial liberalization and increased inte- gration, emerging markets, and especially the BRICS markets, have become a significant part of global financial markets. Therefore, it is important for market participants to un- derstand the influence of various global financial and economic factors on these markets, given especially that investment, risk diversification and speculation opportunities may arise. (Mensi et al., 2014; Kocaarslan et al., 2017.) The BRICS markets are the fastest growing economies, China and India being the top two countries in the world, measured by the GDP (Brennan, 2020). During the 2008-09 global financial crisis, economic growth of emerging countries outperformed economic growth in developed countries. The statistics of International Monetary Fund show that in 2009 the GDP of developed countries decreased -3.6% on average, whereas the GDP in emerg- ing economies increased by 1.7% on average. Furthermore, it has been predicted that emerging economies will account for approximately 50% of the global GDP by 2050 and be the primary driving force for the global economic growth. (Cheng et al., 2007; Aloui et al., 2012.) In addition to this rapid economic growth in emerging economies, espe- cially in the BRICS countries, global oil demand grew annually by an average of 1,153 million barrels per day over the period 2001-2010. 8 The impact of higher oil prices is found to be stronger in emerging stock markets than in those of developed markets. Bhar and Nikolova (2009) explain that this is mainly because of the energy intensiveness of emerging economies as they experience rapid economic growth and, in general, use energy less efficiently. Therefore, given the oil intensity of the BRICS economies and the growing capital inflow and international investment in these markets, it is essential for global investors and other market participants to have a better understanding of the relationship between oil prices and the BRICS stock markets. On April 20th, 2020, price of the West Texas Intermediate (WTI) crude oil fell into negative levels for the first time in history. Likewise, the price of Brent dropped under 20 dollars per barrel against the average price of 64 dollars in 2019. Before this, on March 11th, 2020, the World Health Organization had announced the outbreak of the Covid-19 pan- demic. Shortly after, strict lockdowns, travel restrictions and social distancing were im- posed by many governments around the world. These efforts to control the spread of the disease disrupted also the global supply chains and reduced aggregate demand (Vidya and Prabheesh, 2020). Consequently, oil consumption dropped substantially as well as the crude oil prices, as described above. After the oil demand had returned to its normal levels, another crisis took place. First, the rapid economic growth after the pandemic outpaced energy supply globally. Then, in 2022, Russian invasion of Ukraine began on February 24, and it has been since ongoing. As a result, the Western countries’ demand of Russian oil dropped, and ultimately also sanctions towards Russia were issued, including an embargo on Russian seaborne crude oil. On the other hand, Russia reduced its gas deliveries to Europe. Together all these actions caused energy prices to reach record high levels, leading to a global energy crisis in 2022. In addition to its impacts on energy prices as well as on oil demand and supply, the Rus- sian-Ukrainian war has notable effects on financial markets. As described by Adekoya et al. (2023), the war has increased the price of oil, the most traded commodity in the world, 9 which is further impacting the economic performance of both oil exporting and oil im- porting countries. Therefore policymakers and market participants want to understand the changes happening in market dynamics due to the war in order to hedge their in- vestments and maximize future opportunities in such cases. (Adekoya et al., 2023.) 1.1 Purpose of the study This study aims to investigate the relationship between oil prices and the BRICS stock markets before and after the Covid-19 pandemic. Furthermore, this study distinguishes between oil exporting and oil importing countries when assessing that relationship. The relationship between oil prices and stock markets has been studied before, but a vast majority of these studies are focusing on developed countries. Moreover, within these previous studies, the results have been rather inconclusive, and that might be due to country specific factors, such as, different industrial structures, the state of countries’ financial markets and levels of dependence on oil imports and exports. Therefore, it is essential to distinguish between these country specific factors, like for example, the fact whether the country is an oil importer or oil exporter. Considering these country specific factors and their importance in assessing the relationship between oil prices and the stock markets, not many previous studies differentiate between oil exporting and im- porting countries or do a comparison between them. Since the onset of the Covid-19 pandemic, the world faced an unexpected situation when lockdowns were announced worldwide and transportation, especially air traffic, plunged. This had obvious consequences on oil demand and further oil prices. Previous literature has shown that global financial crisis has had a significant influence on the correlations between oil prices and the BRICS stock markets. Therefore, this study aims to find whether there are similar changes in the correlations between these markets, caused by the onset of the Covid-19 pandemic. Furthermore, it will be interesting to examine what kind of differences there might be between the oil exporting and oil im- porting countries, as the BRICS countries include both, oil exporters and oil importers. Finally, this study’s scope will also include the Russian-Ukrainian war. At the beginning of 10 the war, as a result of the disruption of crude oil supply from Russia, global oil prices reached the highest levels in 8 years as. As a critical factor in the global financial markets, this significant increase in oil prices could have an impact on the performance of other financial assets, such as stocks (Adekoya et al., 2022). 1.2 Research hypothesis The main objective of this thesis is to investigate the relationship between the BRICS stock markets and oil price in periods before and after the Covid-19 pandemic. Further- more, the aim is to investigate the impacts of the Covid-19 pandemic on the dependence between the oil prices and the BRICS stock markets. And finally, this thesis investigates whether the degree and/or structure of the dependence differs for oil importing and oil exporting countries. Therefore, the hypothesis of this thesis are: H1: There is significant relationship between oil prices and the BRICS stock markets dur- ing the study period. H2: The onset of the Covid-19 pandemic has made the relationship between oil prices and the BRICS stock markets stronger. H3: The degree and/or structure of the dependence is different for oil importing and oil exporting countries. 1.3 Structure of the study This paper is divided into ten main chapters. The first chapter gives an introduction and presents the purpose of the study as well as the research hypothesis. That is followed by a chapter that presents the most relevant finance theories, including the efficient market theory as well as some common stock market pricing models. Next, the study moves on to discuss about the oil markets and their growing complexity and introduce the main factors affecting in oil price fluctuation. 11 In the fourth chapter the theoretical relationship between oil prices and stock returns are introduced. The chapter will discuss transmission mechanisms, such as the cash flow hypothesis, through which the oil price movements can influence the behavior of stock returns. Next, the fifth chapter will go through previous studies and discuss about their main findings. Following that, the thesis continues by introducing the data and providing the summary statistics of the data. This is followed by the introduction of the used meth- odology in chapter seven. Then the results of the empirical research are introduced and discussed in more detail. Finally, this study ends with a conclusion, which shortly dis- cusses about the main topics covered in the study and presents ideas for future research as well as the most relevant findings and implications of the results. 12 2 Financial theory background 2.1 The efficient market hypothesis In 1953 Maurice Kendall examined the behavior of stock market prices over time and found that there are no predictable patterns in stock price movements. On the contrary, prices seemed to move randomly. At first, Kendall’s findings were ignored in the finance world, but later on they became labelled as the random walk model. (Bodie et al., 2013:350; Dimson and Mussavian, 1998.) The argument that stock prices should follow a random walk, relies on the notion that price movements should be random and unpre- dictable. Moreover, if prices are rationally determined, only new information will cause them to move. Consequently, prices that always reflect all current and available infor- mation would follow the random walk. (Bodie et al., 2013:350-351.) The efficient market hypothesis was introduced by Fama (1970), and it has since become one of the cornerstones of financial theories. According to Fama (1970), markets are efficient when all available information is reflected in the asset prices. In his seminal paper (1970) Fama reviews the early random walk literature and introduces the three forms of market efficiency, which are the weak form, semi-strong form and strong form. These three forms of market efficiency differ by their notions of what is meant by the term “all available information.” (Bodie et al., 2013:352-353.) According to the weak form efficiency, all historical information of stock prices is incor- porated in the current stock prices. This type of information includes, for example, data of past prices, trading volumes and short interest. The weak form of the hypothesis is the most tested one and has received a wide support in the academic world (Fama, 1970). It relies on the notion that past stock price data is publicly available and virtually free and therefore, all investors have already exploited the possible signals of future perfor- mance (Bodie et al., 2013: 253). 13 The semi-strong form of efficiency states that all publicly available information is already reflected in the stock prices. Furthermore, it indicates that stock prices respond to new information, such as earnings announcements and stock splits, rapidly and accordingly. (Brealey et al., 2020: 343; Bodie et al., 2013: 354.). Ball and Brown (1968) and Fama (1969) study the reaction of stock markets to company announcements and find that the market seems to be anticipating the information, and the majority of price adjustment is done even before the news is released to the market (Dimson and Mussavian, 1998). Finally, the strong form of the efficient market hypothesis denotes that stock prices fully reflect all relevant information to the firm, including also the private information availa- ble only to company insiders (Bodie et al., 2013: 354). In theory, this means that there should not be a way that one could outperform the market consistently, because all available information is incorporated into the stock prices. All versions of the efficient market hypothesis states that, at given time, using current information, it should not be possible outperform the market due to the fact that stock prices already reflect all available information. It is pointed out, however, by Bodie et al. (2013), that the degree of efficiency may vary between various markets. For instance, emerging markets are often less efficient due to the fact that their financial markets are not as developed, they are not as heavily analyzed compared, e.g., to the US markets, and they usually have not as strict accounting disclosure requirements. Therefore, in these markets, some stocks might be less efficiently priced than others, or they might not react to new information as quickly. (Bodie et al., 2013: 352.) 2.2 Stock market pricing In order to examine the impact of oil price on stock market returns, we need to under- stand the concept of share valuation, i.e., how the stock is priced. Valuation of stocks is one of the most essential aspects of investing, making it also a popular subject for exam- ination and analysis. Fundamentally, stock valuation is a process of determining the in- trinsic value of a stock. Intrinsic value refers to a value that is theoretical in nature, 14 meaning that the value is not affected by its market price. The importance of stock val- uation develops from the fact that the intrinsic value of a stock can be different from its current market price. By determining a stock’s intrinsic value, an investor can define whether the stock is over- or undervalued at its current market price. Essentially, stock valuation methods can be divided into two different types, which are absolute valuation and relative valuation. Absolute stock valuation relies only on the company’s fundamental information, and it is used to calculate the intrinsic value of the stock. Absolute valuation methods utilize various financial information, such as, the com- pany’s cash flows, growth rates and dividends, which can be found from the company’s financial statements. Popular absolute stock valuation techniques include, for example, the discounted cash flow model (DCF) and the dividend discount model (DDM). Relative stock valuation, on the other hand, uses ratio analyses and compares the po- tential investment to other companies. This method calculates multiples of similar com- panies and tries to derive a target company’s stock’s value through these multiples. The two main types of valuation multiples are equity multiplies and enterprise value multi- ples. The most often used equity multiple is price-to-earnings ratio, also known as P/E ratio, and it represents the company’s profit-making capability. Another commonly used multiple is price-to-book ratio, or P/B ratio, which is used to compare a company’s cur- rent market value to its book value. It should be considered, however, that companies have different amounts of debt which ultimately influence equity multiplies. Therefore, it is sometimes more appropriate to use enterprise value multiples, as they eliminate the impact of debt financing. The most commonly used enterprise value multiple is the EV/EBITDA ratio. It compares the company’s enterprise value (EV) to its earnings before interest, taxes, depreciation and amortization (EBITDA) and it is often used as a valuation method to compare the relative value of different companies. (CFI, 2022.) Above, only the few most commonly used valuation methods and ratios are discussed. In addition to those, valuation methods can consider, for instance, debt, operating 15 performance, liquidity and other profitability and cash flow measures. In addition, when valuing stocks, one needs to be able to filter the relevant information from all of that information available and use a variety of investment valuation ratios in order to achieve the most accurate results. Next, this paper moves on to present some of the most com- monly used asset pricing models in order to provide more in detail information about the essential financial theory background for this study. 2.2.1 Capital asset pricing model The capital asset pricing model, often referred to as the CAPM, is a fundamental piece of modern financial theory. Even though it is not fully supported by empirical tests, it is widely used because of its simplicity and the insight it offers in a variety of situations. It gives us an accurate prediction of the relationship we should expect to see between a systematic risk of an asset and its expected return. (Bodie, Kane and Marcus, 2013:291.) Using Harry Markowitz (1952) portfolio theory as a foundation, William Sharpe (1964) and John Lintner (1965) developed the capital asset pricing model, which marks the birth of asset pricing theory (Bodie et al., 2013; Fama and French, 2004.) Decades later, the CAPM is still broadly used in different applications, like for example, evaluating the per- formance of managed portfolios and estimating the cost of capital for companies. The popularity of the CAMP relies on the fact that it offers robust predictions about how to measure risk and the relationship between risk and expected return. (Fama and French, 2004.) The model considers the asset’s sensitivity to market risk, also known as non-diversifia- ble or systematic risk, which is often represented by beta (β), in addition to the expected return of the market and the expected return of a theoretical risk-free asset. In other words, the CAPM is based on the relationship between an asset’s beta, the equity risk premium, and the risk-free rate. In the CAPM, risk and return go hand in hand: higher risk leads to a higher return, whereas lower risk comes with lower returns. The results of the capital asset pricing model are plotted in the security market lane, SML, for all the different betas. 16 In Markowitz’s model of portfolio choice, an investor chooses a portfolio at time t – 1 that produces a stochastic return at t. The model presumes that investors are risk averse and, when choosing between portfolios, they care only about the mean and variance of their one-period investment return. Therefore, investors choose “mean-variance-effi- cient” portfolios, in the sense that the portfolios 1) minimize the variance of portfolio return, given the expected return, and 2) maximize expected return, given variance. The portfolio model generates an algebraic condition on asset weights in mean-variance-ef- ficient portfolios. In the CAPM, this algebraic condition is further developed into a test- able prediction about the relationship between risk and expected return by identifying a portfolio that must be efficient if asset prices are to clear the market of all assets. (Fama and French, 2004.) In order to derive conditions for equilibrium in the capital market, Sharpe (1964) and Lintner (1965) add two key assumptions to the Markowitz’s model. First, there is bor- rowing and lending at risk-free rate, which is the same for all investors and is not de- pendent on the amount borrowed or lent. Secondly, investors are assumed to have ho- mogeneous expectations: investors agree on the prospects of different investments, such as the expected values, standard deviations and correlation coefficients. (Sharpe, 1964.) Given the assumption that there is risk-free borrowing and lending, the expected return on assets that are uncorrelated with the market return 𝐸(𝑅!), must equal the risk-free rate 𝑅" . Consequently, the relationship between expected return and beta then be- comes the Sharpe-Lintner CAPM equation, which can be denoted as follows (Sharpe, 1964; Fama and French, 2004): 𝐸(𝑅#) = 𝑅" + 𝛽#(𝐸(𝑅!) − 𝑅"* (1) where: 17 𝐸(𝑅#) = expected return of an asset 𝑅" = risk-free rate 𝛽# = beta of an asset 𝐸(𝑅!) = expected return of the market (𝐸(𝑅!) − 𝑅"* = market risk premium Even though the CAPM is widely used, it has also been a subject of criticism since its release, mostly due to the unrealistic assumptions it requires to be derived. These as- sumptions include, for example, market equilibrium, the ability to borrow or lend at a common risk-free rate, no transaction costs, all information is publicly available, and that investors are rational, mean-variance optimizers and have homogeneous expectations (Brealey et al., 2020: 212; Bodie et al., 2013:304; Levy, 2010). One example of theoretical criticism is the prospect theory by Kahneman and Tversky (1979), which shows that typ- ical investors are not always rational and risk-averters (Levy, 2010). In addition, in their regression approach Fama and French (1992) find that size and book-to-market ratio contribute to the explanation of average stock returns associated with the market beta. Based on their findings, Fama and French (1993, 1996) suggest a new model for asset pricing. This model will be presented in the next section of this paper. 2.2.2 Factor models One of the simplest asset pricing models is the single-index model (SIM), developed by William Sharpe in 1963. It is commonly used to measure both the risk and the return of an asset. The single-index model assumes that only one macroeconomic factor causes systematic risk and therefore affects the stock returns. More precisely, this model uses the market index, such as the S&P 500, to proxy for the common macroeconomic factor. (Bodie et al., 2013:256-259.) To estimate systematic risk, the single-index model utilizes past return rate data on mar- ket indexes, like for example, the S&P 500. The model denotes the market index by M, 18 with excess return of 𝑅! = 𝑟! − 𝑟", and standard deviation of 𝜎!. Given the linearity of the single-index model, the sensitivity (or beta) coefficient of a security on the index can be estimated by using a single-variable linear regression. The excess return of a security, 𝑅# = 𝑟# − 𝑟", can be regressed on the excess return of the index, 𝑅!. In order to estimate the regression, a historical sample of paired observations, 𝑅#(𝑡) and 𝑅!(𝑡), where t de- notes the time of each observation, will be collected. Consequently, the regression equa- tion of the single-index model can then be denoted as follows (Bodie et al., 2013:259) 𝑅#(𝑡) = 𝛼# + 𝛽#𝑅!(𝑡) + 𝑒#(𝑡) (2) where: 𝑅#(𝑡) = the excess return of a security at time t. 𝛼# = the expected excess return of a security when the market excess return is zero. 𝛽# = the security’s sensitivity to the index: It is the amount by which the security return increases or decreases for every 1% increase or decrease in the return on the index. 𝑅!(𝑡) = the excess return of the index at time t. 𝑒#(𝑡) = the unexpected residual: Firm specific surprise in the security return at time t. Given that 𝐸(𝑒#) = 0, the expected return-beta relationship of the single-index model can be obtained as follows: 𝐸(𝑅#) = 𝛼# + 𝛽#𝐸(𝑅!) (2.1.) From the above equation it can be seen that a security’s risk premium, 𝐸(𝑅#), consists of two parts. Part of it is due to the risk premium of the index. The market risk premium, 𝐸(𝑅!), is multiplied the relative sensitivity, or beta, of the individual security. It is called also the systematic risk premium, because it is derived from the risk premium that char- acterizes the entire market, and which is used as a proxy for the condition of the full 19 economy. The other part of the risk premium is denoted by the first term of the equation, alpha (𝛼), which is a nonmarket premium. (Bodie et al., 2013:260.) As mentioned earlier, Fama and French (1992) pointed out some failures of the CAPM in their analysis. Fama and French (1993) argue that while the size and book-to-market fac- tors are not state variables themselves, the high book-to-market stocks and higher aver- age returns on small stocks reflect unidentified state variables that generate undiversifi- able risks in returns, which are not captured by the market return and are separately priced from the market betas (Fama and French, 2004). Based on their findings, Fama and French (1993, 1996) propose a new asset pricing model, the three-factor model, which incorporates three factors that appear to determine a security’s expected returns. These factors are the market factor (the systematic risk factor), the size factor and the book-to-market factor. (Brealey et al., 2013; Fama and French, 2004.) The three-factor model for expected returns can be denoted as follows: 𝐸(𝑅#$) = 𝑅"$ + 𝛽%(𝐸(𝑅!$) − 𝑅"$* + 𝛽&𝑆𝑀𝐵$ + 𝛽'𝐻𝑀𝐿$ (3) Where: 𝐸(𝑅#$) = the expected return for an asset 𝑖 𝑅"$ = the risk-free return rate 𝑅!$ = the market portfolio return (𝐸(𝑅!$) − 𝑅"$* = the market risk premium 𝑆𝑀𝐵$ = the size factor (small minus big) 𝐻𝑀𝐿$ = the value factor (high B/M minus low B/M) 𝛽%,&,' = the beta coefficients of the factors One implication of the three-factor model’s expected return equation presented by Fama and French (2004) is that the intercept 𝛼# is zero for all assets in the time-series regression: 20 𝑅#$ − 𝑅"$ = 𝛼# + 𝛽%7𝑅!$ − 𝑅"$8 + 𝛽&𝑆𝑀𝐵$ + 𝛽'𝐻𝑀𝐿$ + 𝜀#$ (3.1) In the above equation the error term is denoted by 𝜀#$. With the criterion on this equa- tion, that the intercept is zero, Fama and French (1993, 1996) find that model captures a lot of the variation in average return for portfolios that are formed on book-to-market equity, size and other price ratios that are found to be problematic for the CAPM. The three-factor model has become widely used equation in empirical research when a model of expected returns is needed. (Fama and French, 2004.) 21 3 World oil markets Oil has been the world’s most important source of energy for many years now. In 2019 oil accounted for 33% of the global energy needs, with coal 27% and natural gas 24% as its nearest rivals (BP, 2021). Figure 1. World Primary Energy Consumption by fuel in 2019 (BP Statistical review, 2021) As a result, oil has become the world’s largest traded commodity, measured both by volume and value. For instance, the physical crude oil market would be worth about 3.3 trillion US dollars per year if we were to assume a constant reference price of USD 100 per barrel applied to 2012 global demand of approximately 90 million barrels per day. Accordingly, for a market with this wide global reach and financial/physical size, the oil price reflects the interaction of countless considerations around supply and demand fun- damentals as well as other risk factors. Moreover, it must be noted that in the recent years the market has become a lot more complex, due to the role of financial investors in particular, leading to debate the relationship between physical fundamentals and commodity prices. (Deutsche Bank, 2013.) 33 % 24 % 27 % 4 % 7 % 5 % World Primary Energy Consumption by fuel in 2019 Oil Natural gas Coal Nuclear energy Hydro-electricity Renewables 22 In addition to the growing complexity, oil markets have been through a notable turmoil due to the Covid-19 pandemic. The combination of the pandemic and the actions that occurred in order to limit its impact had an unprecedented effect in the global oil de- mand, as lockdowns around the world decimated the transport-related oil demand. The total primary energy consumption fell by -4.5% in 2020, which is the largest decline since the second world war. The drop was driven mostly by oil, which accounted for almost three-quarters of the net decline. Oil consumption alone declined by a record 9.1 million barrels per day, or total -9.3% to its lowest level since 2011. Similarly, also the oil pro- duction saw its largest drop since the second world war, -6.6 million barrels per day over the whole year. Oil prices followed these declines accordingly. Brent reached a low of under $20 per barrel, whereas the US WTI prices dropped to negative levels for the first time in history. The oil price (Dated Brent) averaged at its lowest level since 2004, at USD 41.84 per barrel. (BP, 2021.) Figure 2. Change oil demand from 2019 to 2020 (BP Statistical review, 2021.) It is essential to note that China was virtually the only country in the world where oil demand increased in 2020. The above figure shows the percentage change in oil demand in 2020 compared to 2019 in the BRICS countries as well as in the US and Europe. It is also worth noting that on average, the oil demand in the BRICS countries fell by -6.3% -6,1% -4,9% -9,7% 2,0% -12,8% -12,4% -13,90% -16,0% -14,0% -12,0% -10,0% -8,0% -6,0% -4,0% -2,0% 0,0% 2,0% 4,0% Brazil Russia India China South Africa US Europe Change in oil demand from 2019 to 2020 23 on average, whereas in the US and Europe the average decline was over -13%. This can be explained, at least partly, with the energy intensiveness of the BRICS countries and emerging economies in general – these countries have high demand for oil and other energy sources due to their rapid economic growth and less efficient use of energy. Fur- thermore, OPEC forecast states that non-OECD countries, which includes all the BRICS countries, will drive oil demand growth in the future, whereas the OECD oil demand is expected to decrease by over 10 million barrels per day between 2021 and 2045 (OPEC, 2022). Looking at the year 2021, when Covid-19 restrictions began to loosen and economic ac- tivity recovered, also the oil demand increased, however, remaining still below 2019 lev- els. This is mostly explained by the aviation-related oil demand, which remained 33% below 2019 levels in 2021, initially due to its slower recovery from the Covid-19 pan- demic (BP, 2022; OPEC, 2022). In 2021 also the oil prices increased to the second highest level since 2015, averaging at USD 70.91 per barrel (BP, 2022). The Covid-19 pandemic had an unseen impact on the global oil markets, as the above shortly describes the main effects. Given the significant fluctuations in the oil market, including the consumption as well as the production and price fluctuations, it is essential to examine in more detail the numerous different components of oil markets and how they ultimately impact the oil price. Next section of this thesis will introduce and go through the most important factors as well as some recent literature examining the in- fluence of these factors on the oil price changes. 3.1 Oil price determination As mentioned, oil is the world’s most important commodity. As an essential source of energy as well as a financial investment tool, oil has a vital role in both modern industry and economic development. In addition, crude oil prices are the most widely monitored commodity price indicator in the world. The notable fluctuations in oil prices have a sig- nificant impact on the world economy. (Zhao, 2022; Ji and Guo, 2015.) As a result, there 24 is a large amount of academic research that focuses on the crude oil prices and the fac- tors affecting them. Recent papers by Razek and Michieka (2019) and Zhao (2022) examine the influence of some of the main factors, such as, supply, demand, inventories and the US dollar ex- change rate on the crude oil price volatility. Zhao (2022) states that supply and demand are still the most influential factors of oil price changes. Razek and Michieka (2019) ex- amine the OPEC’s influence on the global oil market as the recent decrease in OPEC’s spare capacity has resulted analysts to question OPEC’s ability to influence oil prices. Their empirical results reveal that OPEC still has an essential role in balancing oil markets, and especially in longer time horizons, it explains a large portion of the fluctuations in oil prices. These results are similar to those of Coleman (2012), who finds that OPEC market share has the most economically significant impact on long-term oil price. Re- garding the other factors, Zhao (2022) finds that inventories and US dollar exchange rate both are relatively influential factors of crude oil price volatility. 3.1.1 Supply Oil markets and oil production in particular cannot be discussed without highlighting the importance of the Organization of the Petroleum Exporting Countries, familiarly known as the OPEC (EIA, 2022). OPEC crude oil production can be described as centrally coordi- nated and controlled mainly by national oil companies. Non-OPEC production on the other hand, is performed primarily by independently operating, international investor- owned oil companies. These non-OPEC oil production companies consider economic fac- tors when making investment decisions, aiming to increase shareholder value. Therefore, their investment and future supply capabilities are mainly functions of market conditions and are able to respond to changes in market conditions more readily. Even though some OPEC oil companies operate in a similar manner, their objectives also include building infrastructure, increasing revenues and providing employment to benefit their econo- mies in a broader sense. (EIA, 2022; Razek & Michieka, 2019.) 25 Crude oil production by OPEC has a significant impact on oil prices, mostly due to the fact that it sets targets to manage oil production in its member countries. Historically, crude oil prices have increased in times when OPEC’s production targets are mitigated (EIA, 2022). Below figure shows how changes in Saudi Arabia’s crude oil production result in changes in WTI crude oil prices. Figure 3. Changes in Saudi Arabia crude oil production and WTI crude oil prices OPEC member countries account for about 40 percent of the world’s crude oil produc- tion, whereas OPEC’s oil exports comprise about 60 percent of the total global crude oil exports. Market shares this large, both in production and exports, give OPEC an influen- tial role in the world’s oil market. Projections of OPEC production fluctuations, especially in Saudi Arabia, cause changes in oil prices, mirroring the role of expectations in driving oil prices. Unexpected interruptions, the extent and duration of disruption as well as the uncertainty of reverting the output have significant impact on oil prices. It is also worth noting that OPEC adjusts its production to changes in market conditions with a delay to ensure it responses to permanent shocks rather than to momentary shocks. 26 Furthermore, regardless of OPEC’s aims to manage production and maintain target price levels, member countries do not always comply with the production targets set by the organization. Such unwillingness to maintain certain production levels can also affect the oil prices. (EIA, 2022.) The level of utilization of OPEC member countries’ available production capacity reflects the global oil market’s ability to react to possible oil supply disruptions and the extent to which OPEC is exploiting its upward influence on prices. Spare capacity is defined by EIA (2022) as “the volume of production that can be brought on within 30 days and sustained for at least 90 days”. In the past, Saudi Arabia had the largest spare capacity of about 1.5- 2 million barrels per day, which, together with other OPEC countries’ spare capacities, could be used to respond to potential crises that reduce oil supplies. Hence, a rising risk premium is often incorporated into the oil prices when OPEC spare capacity reaches low levels. Between 2003 and 2008, OPEC’s total spare capacity remained approximately at 2 million barrels per day or under, thus limiting its capability to increase its supply in case of quickly rising demand. (EIA, 2022; Razek & Michieka, 2019.) Oil production from non-OPEC countries currently represents about 60 percent of global oil production. The most important producers within non-OPEC include North America, former Soviet Union regions and the North Sea, in addition to South American countries Brazil, Columbia and Argentina. Typically, non-OPEC producers are price takers, meaning that they respond to market prices instead of trying to influence them. Hence, they usu- ally produce at or near full capacity and have limited spare capacity. Decreasing non- OPEC supply has an upward pressure on prices by decreasing total global supply and intensifying the demand for OPEC supply and hence further increasing OPEC’s ability to influence prices. On the other hand, increases in non-OPEC production contribute to lower oil prices whereas unplanned outages in non-OPEC production add the uncertainty in the markets, causing increasing levels of price volatility. (EIA, 2022.) 27 In comparison to OPEC production, non-OPEC production often occurs in rural areas, such as deep water offshores, that have relatively high finding and production costs, whereas the majority of lower cost conventional oil resources are located in OPEC mem- ber countries. Therefore, non-OPEC producers have pursued unconventional sources, like for example, oil sands. In order to reduce their production cost disadvantage com- pared to the OPEC production, non-OPEC producers are developing new production technologies. In the short term this may result in the development of higher-cost sup- plies whereas in the long term, when technology advances, costs often fall which ulti- mately leads to lower oil prices. (EIA, 2022.) 3.1.2 Demand Another important factor driving the oil prices is demand. Similar to supply, also demand is often viewed from the perspective of two groups of countries. Countries can be di- vided into developed and developing countries, or OECD and non-OECD countries. The Organization of Economic Cooperation and Development (OECD) consists of the United States, big part of Europe and other developed countries. (EIA, 2022.) OECD countries can be grouped together because their economic infrastructure is fundamentally more advanced and extensive in comparison to the non-OECD countries. In addition, differ- ences in energy infrastructure and in the amount of energy that the countries consume, are seen as divisive factors between OECD and non-OECD countries. Non-OECD countries include, for example, all the BRICS countries and other emerging economies. Oil consumption in OECD countries has been decreasing constantly since the early 2000s whereas consumption in non-OECD countries has been increasing considerably in recent years. Between 2000 and 2010, non-OECD oil consumption increased over 40 percent, with China, India and Saudi Arabia having the greatest impact on this growth. Increasing oil consumption is driven by rapid economic growth in non-OECD countries. Both current and expected levels of economic growth have an impact on global oil demand as well as in oil prices. (EIA, 2022.) 28 Furthermore, each economy’s structural conditions influence the relationship between economic growth and oil prices. Usually, non-OECD economies are more driven by the manufacturing industries, using more oil and hence being more energy intensive than service industries. Even though transportation sector is much smaller in non-OECD coun- tries than in OECD countries, the use of transportation oil is expected to increase rapidly as growing economies increase the need to move people and goods. In addition, vehicle ownership per capita is strongly correlated with higher incomes, and is expected to grow in the future in non-OECD countries as incomes increase. Finally, it must be noted that also energy policies have an impact in the relationship between oil consumption and economic activity. For example, many developing countries subsidize end-use price, therefore limiting consumer response to changes in market prices and increasing the importance of economic growth as the primary driver of non-OECD demand. Together with high population growth, these factors implicate the significance of non-OECD coun- tries in driving the oil prices. (EIA, 2022.) On a contrary to non-OECD countries, even in times of strong economic growth, the growth in oil consumption often tends to be slower in OECD countries, mostly due to economic conditions and policies. In OECD countries, oil use in transportation sector is usually much larger than in non-OECD countries, and therefore economic conditions and policies that affect transportation have a significant impact on total oil consumption in OECD countries. For instance, many OECD countries have relatively high fuel taxes and policies that increase the use of biofuels. This notably slows down the increase in oil consumption in developed countries. Furthermore, OECD countries’ economies have larger service sectors relative to manufacturing, which requires less oil and therefore strong economic growth does is not reflected in oil consumption in a similar manner as it is in non-OECD countries. Lastly, as OECD countries usually have fever subsidies on end- use prices, changes in market oil prices are usually quickly reflected in consumer prices. (EIA, 2022.) 29 3.1.3 Inventories Inventories are maintained as a balancing point between supply and demand. During periods when production exceeds demand, oil can be stored in refineries and storage terminals. Inventories help to prepare for seasonal fluctuations, unexpected weather and refinery maintenance periods. Since inventories satisfy both current and future de- mand, their level responds to the relationship between the current and future oil prices. If market expectations indicate relative increase in future demand or lower future supply, prices for future contracts will increase, creating a contango effect. On the other hand, a positive unexpected shock to current consumption associated with a negative shock to current supply will often push up spot prices relative to futures prices, encouraging in- ventory draw downs in order to meet the current demand, creating a backwardation effect. (EIA, 2022; Razek & Michieka, 2019.) Additionally, the relationship between inventories and prices allows for effects in either direction. If futures prices increase to a higher level relative to the current spot prices, there is an incentive to store oil and sell it later at higher expected price. On the contrary, increasing inventories could indicate that current supply exceeds current demand which would cause current spot prices to decrease in order to rebalance demand and supply. Therefore, physical inventory levels and price spreads over time are used as signals be- tween current market participants and those with long-term exposures. However, it is worth noting that inventory data is not always available on timely basis, or not available at all. The lack of complete inventory information creates additional uncertainty in the world’s oil market, which can further influence oil prices. (EIA, 2022.) 3.1.4 Financial markets In addition to physical quantities, market participants also trade future oil contracts and other energy derivatives in order to hedge against risk and inflation, profit from price changes and diversify portfolios. As measured by the New York Mercantile Exchange (NY- MEX), open interest on exchange-traded crude oil futures has been increasing 30 significantly over the past decades. Given the notable size and the role of futures mar- kets in price discovery, they have influence also in oil prices. Traditionally, stocks have been the largest investment market. In response to economic conditions, stock and commodity (i.e., oil) prices move together, which has an impact on corporate earnings and further, also on the demand for commodities as raw materials. Between 2008 and 2010, the level and appetite for risk changed dramatically. Over the past few decades, crude oil has shown similar risk and return characteristics to stocks. Hence, in times where risks are significantly increasing (in financial crisis) and then adapting (risks are decreasing) stocks and crude oil prices might move in the same direc- tion. (EIA, 2022; Razek & Michieka, 2019.) The negative correlation between crude oil prices and the US real effective exchange rate (REER) may be a reflection from the fact that oil benchmarks are traditionally priced in US dollars. When the value of US dollar decreases, the effective price of crude oil out- side the US drops, hence potentially increasing consumer demand and creating upward pressure on prices. In addition, since depreciation of the US dollar decreases the returns on dollar-denominated assets once converted into foreign currencies, foreign market participants may aim to maintain higher oil prices. (EIA, 2022.) During downturns, there is a deeper negative relationship between the exchange rate and the oil prices due to risk shocks and the financialization of oil prices (Fratzscher et al., 2014). 31 4 Theoretical relationship between oil prices and stock returns Literature has investigated and tested the different transmission mechanisms by which oil price fluctuations can affect the behavior of stock returns. One of the most commonly tested is the cash flow hypothesis, suggested by Jones and Kaul (1996), which assumes that stock prices are determined by expected discounted cash flows. In other words, oil prices influence stock markets directly through the stock valuation channel. To simply explain this requires two equations: firstly we define stock returns (𝑅#,$) as the first log difference as follows: 𝑅#,$ = log ( )!,# )!,#$% ) (4) where 𝑃#,$ denote the firm’s 𝑖 stock price at time 𝑡. Secondly, finance theory assumes that current stock prices reflect the discounted future cash flows of a certain stock (Degiannakis, Filis and Arora, 2018). This can be denoted as: 𝑃#,$ = ∑ @ *(,-&) (%/*(0))& A1 12$/% (5) where 𝐶𝐹1 denotes the cash flow at time 𝑛 and 𝑟 is the discount rate. 𝐸 is the expecta- tion operator. Equations 4 and 5 show that stock returns are impacted by the expected cash flows and/or discount rate, which in turn are affected by factors such as oil prices. (Degiannakis et al., 2018.) According to the cash flow hypothesis, oil price changes can affect a firm’s future cash flows either negatively or positively, depending on whether the firm is an oil producer or an oil consumer. For an oil producer, increasing oil prices will result in higher profit margins and therefore, higher expected cash flows and stock returns. Two channels imply a negative relationship. First, since for an oil-consuming firm, oil is one of the major inputs, higher oil prices will increase the production costs, which will lower profit levels and therefore future cash flows, earnings and dividends, and lastly, 32 stock returns. (Degiannakis et al., 2018; Smyth and Narayan, 2018.) Secondly, increasing oil prices can lead to an overestimation of expected inflation and higher real interest rates. As mentioned, increasing oil prices lead to higher production costs for oil consum- ing firms. Consequently, these costs will be transferred to consumers through higher re- tail prices and therefore, higher inflation expectations. The discount rate, that is used to discount expected future cash flows, is at least partially composed of expected real in- terest rates and expected inflation (Mohanty and Nandha, 2011). Therefore, higher oil prices can decrease future cash flows, and ultimately the stock returns, through inflation and interest rates. Another possible reason for a positive relationship between oil prices and stock returns is that, as noted by Kollias, Kyrtsou and Papadamou (2013), investors may associate higher oil prices with a booming economy. Therefore, increasing oil prices could be seen as a reflection of stronger business performance and the concomitant influence on stock markets (Smyth and Narayan, 2018). Furthermore, Hamilton (2009) suggests that, before the global financial crisis, increasing oil prices reflected growth in developing markets in addition to high level of business confidence. Chen, Cheng and Demirer (2017), argue that stock market momentum and oil price volatility are positively correlated. They use China as a case study and argue that this correlation is driven by time-varying investor sentiment, in which investors react to oil price volatility associated with uncertainty by putting upward demand pressure on winner stocks. On a contrary to Kollias et al., (2013), Brown and Yucel (2002), suggest that increasing oil prices will lead to higher uncertainty in the real economy, due to the effect on inflation and production costs, among others. Therefore, according to Brown and Yucel (2002), higher oil prices will decrease firms’ demand for indelible investments, which in turn, will lower the expected cash flows. It is argued that uncertainty is transmitted also to households, which will reduce their consumption of durable goods (Pindyck, 2004). Ac- cording to Edelstein and Kilian (2009), increasing uncertainty regarding the future oil prices will increase the incentives of households to save rather than consume. 33 Degiannakis et al. (2018), add that as uncertainty increases due to higher oil prices, the value of postponing consumption and investment decisions will get higher, whereas the incentive to consume or invest today decreases, which hence weakens the economic growth prospects and thereby stock market returns. 34 5 Literature review Even though fluctuations in the crude oil prices are considered as an essential factor for understanding changes in stock prices, and the relationship between oil price and stock markets has been studied quite a lot, there is still no consensus among economists about the relationship between crude oil prices and the stock markets. One reason for this could be the fact that oil prices did not float freely until 1973, when oil price was liber- ated from regulations. For this reason, the academic world has not had that long of a time to fully research the relationship between oil prices and stock markets. Since then, during the past four decades, a growing literature has appeared, analyzing the impacts of oil price fluctuations on different macroeconomic and financial variables. Especially the 2008 surge in oil prices, when the price of oil reached the 100 dollars per barrel mark for the first time in history, began an extensive interest in oil market research. (Smyth and Narayan, 2018.) This section of the thesis will go through the most relevant previous studies in order to provide a better understanding about the various methods used and the inconsistencies between different papers and their results. Bhar and Nikolova (2009), analyze the price discovery and volatility transition between the oil market and BRIC stock markets over a period of 1995-2007. They were one of the first ones to distinguish between oil exporting and oil importing countries when exam- ining the oil price-stock market relationship. Accordingly, the authors conclude that the relationship between oil prices and stock markets depends on the extent to which those countries are oil exporters or oil importers. Unlike the other BRICS countries, Russia has historically been an oil exporter and therefore the Russian economy is strongly depend- ent on oil exports, making it vulnerable to oil price fluctuations. Bhar and Nikolova (2009) find in their analysis, that there is a strong relationship between Russian equity returns and global oil prices. For the other BRIC countries there are no significant correlations found. Bhar and Nikolova (2009) note, however, that in case Brazil becomes a net oil exporter, similarly to Russia, it would make Brazil’s stock markets more responsive to the changes in the global oil prices. 35 Fang and You (2014), study the dynamic interactions between oil prices and stock returns of three large emerging countries, namely Russia, China and India, over the period of 2001-2012. Their study utilizes the structural vector autoregressive model (SVAR) origi- nally proposed by Kilian and Park (2009), and examines how precise structural shocks affect the stock market returns in India, China and Russia. To represent the oil demand and supply shocks, Fang and You (2014) use the change rate of global oil production, global real activity and real oil price. They justify the use of the change rate of variable with the fact that it has more economical meanings and is thus more meaningful from an economics perspective. For instance, the import oil price change can be associated with the specific oil demand shock, which is driven mostly by the preventive demand for crude oil. From an investor perspective this means that the economy is expected to weaken, and the firms’ profit rates are expected to decrease as the oil price increases. This leads investors to sell their stocks in order to prevent capital losses. On the other hand, the export oil price is used to capture the oil supply specific effect, which from the investor perspective means that the oil exporter’s, in this case the Russian, economy will grow and due to the higher export oil prices, the firms’ profits will also increase. The basic statistics of Fang and You’s (2014) study indicate that Russian stock returns are higher than those for India and China. In addition, Russia’s oil supply change rate is a lot higher than the global oil supply change rate, and therefore it can be stated that a large proportion of Russian stock returns are caused by the high change rate of Russian oil export prices. In contrast to Russia, China and India are net oil importers and increasing oil prices often lead to higher costs of non-oil-producing companies and furthermore decrease the stock returns due to the reduced profits. Hence, this is one of the reasons for notably lower stock returns in India and China than those in Russia, due to the in- creasing oil prices in the time of Fang and You’s (2014), study period. Fang and You (2014) find that the impact of global oil production shock is insignificant in comparison to the impact of oil-specific demand shock and the impact of aggregate de- mand shock. This finding is in line with the argument of Kilian and Park (2009) that the 36 response of aggregate stock returns might be different depending on the causes of the oil shock, and global oil production shocks are less significant. Even though Russia is a major oil exporter in the world and the whole energy industry has an essential role in the Russian economy, Fang and You (2014), find that global oil demand shocks lead to a significant decrease in the Russian stock returns in the first period. The authors explain this with the notion that even though Russia is the main oil supplier in the world, not every oil company always benefits from high oil prices. Since the oil companies will bear a heavy tax burden as the oil prices increase, they usually prefer oil prices to remain stable or have only small increase. Notable oil price fluctuations might cause the oil com- panies to fail to react to the changes in time. Consequently, the global demand oil shocks have only a temporary significant negative effect on the Russian stock returns. On the other hand, as expected, oil price shocks driven by Russian oil supply, have a significant positive impact on the Russian stock returns. Fang and You (2014) conclude that in the case of oil exporters, like Russia, the impact on stock returns varies depending on what causes the oil shock. As mentioned earlier in this paper, the energy efficiency of emerging countries is low compared to that of developed countries. Out of the countries incorporated in the study of Fang and You (2014), China has the lowest energy efficiency and that is one of the reasons why the China-specific oil demand-driven oil shocks have a significant negative impact on the Chinese stock market. Moreover, Fang and You (2014), find that oil shock driven by global demand have insignificant impact on China’s stock returns. This could be explained by the fact that the Chinese financial market is still under heavy regulation and is characterized by a lack of transparency. A few notable features of the Chinese stock markets are that foreign investors can buy only small amounts of China’s stocks and that the shares held by legal persons and the state cannot be circulated. Accordingly, Chinese stock returns reflect only the demand from domestic investors who have only few ways to invest. Therefore, Fang and You (2014) argue that the insignificant impact of global demand-driven oil price shocks on Chinese stock returns is at least partly due to the incomplete nature of China’s stock market. 37 In the case of India, Fang and You (2014) find that Indian stock market reacts positively to its oil-specific demand shock only during the first month. This indicates that these demand shocks in India should be seen as an indicator of development speed instead of precautionary demand caused by the uncertainty about the oil supply shortfalls. In ad- dition, it is found that oil price shocks caused by the global demand have a notable neg- ative impact on the Indian stock returns only at the first month. This indicates that the Indian stock market is not able to reflect the positive spillover effect from global eco- nomic expansion. Fang and You (2014) conclude that as long as the oil price is not driven by the growing oil demand of India, the oil prices always have a negative impact on In- dia’s stock market. Another study employing the structural VAR analysis proposed by Kilian and Park (2009) is done by Wang, Wu and Yang (2013). Wang et al. (2013) study the relationship between oil prices and stock markets over a period of 1999-2011 in major oil importing and ex- porting countries, such as, Saudi Arabia, Russia, Norway, Germany, US, India, and China, among others. They section oil price shocks into oil supply shocks, aggregate demand shocks, and other oil-specific shocks. Wang et al. (2013) find that the driving force of the oil price shocks as well as the net position of the country in global oil market have an impact on the relationship between oil price changes and stock market returns. Moreo- ver, their results show that the contributions of oil price changes to stock return fluctu- ations are overall larger in oil exporting countries than in oil importing countries. Wang et al. (2013), suggest that this could be at least partly due to the greater importance of crude oil for oil exporting economies. Investigating the response of stock market returns to oil price shocks, Wang et al. (2013), find that in most cases the response to oil supply shocks is insignificant. They argue that this result can be explained by the insignificant impact of oil supply shocks on oil prices, which, however, is a contradictory to the results of some other studies (e.g., Kilian and 38 Park, 2009; Basher et al. 2012), which find that the impact of oil supply shocks on oil prices is significant and persistent. When assessing the impacts of demand shocks, Wang et al. (2013) find that the response of stock markets is much stronger and more persistent in oil exporting countries than what it is in oil importing countries. It is generally suggested that oil demand shocks lead to higher oil prices, therefore causing the industry costs to rise as well, which further affect the stock market negatively. In addition, increasing oil demand will also result in a transfer of wealth from oil importing countries to oil exporting countries. Hence, the positive effect of demand shocks is greater and more persistent on the stock returns in oil exporting countries. Wang et al. (2013) point out, however, that the positive effect of demand shocks is quite strong and persistent in China and India in comparison to the other oil importing coun- tries. They argue that this is due to the relatively high economic growth rates in these two emerging economies. On the other hand, when looking at the oil exporting countries, the positive effects of demand shocks on stock markets are less persistent in Russia, Mexico and Saudi Arabia than in the other exporting countries included in their study. A probable explanation is that these countries consume greater amounts of crude oil, which leads to a stronger negative effects of high oil price on national economy, hence offsetting the positive impacts. On a contrary to previous studies by Kilian and Park (2009) and Jung and Park (2011), Wang et al. (2013) find that the impact of precautionary demand shocks on stock returns is insignificant in oil importing countries, with China as an exception. The earlier litera- ture suggest that the effect is highly significant. Inconsistent results may be due to the more recent dataset and larger number of countries included in the study of Wang et al. (2013). Furthermore, many studies show that the impact of oil price shocks on macroe- conomic variables has become less significant during the years Wang et al. (2013) carried out their research. The impacts of precautionary demand shocks on the Chinese stock 39 market become significant after nine months. This lacking response can be explained by the inefficiency of the Chinese stock market, which causes late information transmission from crude oil market to the stock market. Nguyen and Bhatti (2012) employ nonparametric chi- and K-plots and parametric copula models in order to study the dependence structures and/or tail dependence between Chinese and Vietnamese stock markets and oil price changes. The tail dependence allows to better determine whether the two variables move together in the same or opposite directions. Nguyen and Bhatti (2012) find that there is relatively strong left tail depend- ence between the oil price changes and the Vietnamese stock market, indicating that if the oil prices decrease, Vietnam’s stock market will also decrease. In the case of China, however, Nguyen and Bhatti (2012), do not find any tail dependence between the Chi- nese stock market and oil price changes. Therefore, the results suggest that the Chinese stock market is independent from the oil price changes during the study period of 2000- 2009. Nguyen and Bhatti (2012) suggest that this might be due to the strong growth of the Chinese economy during that period, meaning that the negative impacts of oil prices are absorbed by the growth of the Chinese economy. On a contrary, the Vietnamese stock market has a smaller market capitalization and is more likely to be affected by the fluctuations in the global oil prices. Study by Gupta and Modise (2013), investigates the dynamic relationship between the South African stock market and different oil price shocks. These different shocks are oil supply shock, an aggregate demand shock and speculative demand shock. They employ the structural VAR model over a study period from 1973 to 2011. According to Gupta and Modise (2013), a negative oil supply shock, which results in declining oil production and higher oil prices, has a negative impact on the South African stock returns. This result is consistent with other studies, as oil supply shock generally has a negative impact on stock returns in oil importing countries (Park and Ratti, 2008). 40 An aggregate demand shock driven by improved global activity results in increasing oil prices, increased global real activity and rising oil production. Gupta and Modise (2013) find that this type of global demand shock has a positive impact on the South African stock markets. This result is opposite to some of the other studies and general assump- tions, as it is often assumed that oil importing countries respond negatively to oil de- mand shocks, because the importing costs will rise. Gupta and Modise (2013) argue that their differing results can be explained by South Africa being a commodity exporter, as it is one of the major exporters of platinum, gold and other mining products. Similar re- sponses are found for example in the Russian stock markets (Fang, 2010), and even though South Africa does not export oil like Russia, the general increase in global com- modity prices driven by a positive global demand shock results in the same responses for the two countries. A negative speculative demand shock leads to a general decline in world oil production, in real global activity and an overall drop in the stock returns, whereas the oil price in- creases following such a shock. Similarly to some previous studies (e.g., Gunter, 2013), Gupta and Modise (2013), find that even though impact is positive for the first months, the overall effect of a speculative demand shock is negative on South African stock re- turns. This can be explained with the increasing oil prices, with no increase in the other commodity prices, which will have an inflationary effect in South Africa and hence re- ducing household wealth. Using a quantile regression approach, introduced by Koenker and Bassett (1978), Mensi et al. (2014) examine the dependence between BRICS stock markets and different global factors, oil prices being one of them. Furthermore, they study how the global financial crisis has affected this relationship. Their results for the period from 1997 to 2013 show that the oil prices display a symmetric independence with the BRICS stock markets, ex- cept for South Africa. However, Mensi et al. (2014) find that the global financial crisis has significantly increased the co-movements of oil prices and all the BRICS stock markets. As these results differ from some of the other studies, it is essential to note that Mensi 41 et al. (2013) do not identify the driving forces behind the oil price shocks nor do they differentiate whether the country is an oil importer or exporter. In addition to the differ- ent methodology, these might be reasons for the inconclusive results. Paper by Zhu, Li and Li (2014) investigates the dynamic dependence between stock re- turns in ten countries in the Asia-Pacific region and crude oil prices over the period of January 2004 – March 2012. They use unconditional and time-varying copula-GARCH models to study the dependence structure between stock returns and crude oil prices. Furthermore, they investigate whether there are changes in this dependence in periods before and after the global financial crisis. Hence, Zhu et al. (2014) divide their study period into two subperiods, referred as pre-crisis and post-crisis. Their results show that, generally, the dependence between oil prices and stock returns is weak pre-crisis, with correlations close to zero. On the contrary, their empirical evidence for the post-crisis period indicates that the dependence between oil prices and stock returns increased significantly in all markets, with India showing the strongest dependence on crude oil prices. Caporale, Ali and Spagnolo (2015) study the time-varying effect of oil price uncertainty on sectoral stock returns in China. The authors use weekly data on ten sectoral indices over the study period from January 1997 to February 2014. In their study, they estimate a bivariate VAR-GARCH with a DCC specification, allowing for mean effects. Moreover, they take a time-varying approach, differentiating between periods characterized by dif- ferent types of oil price shocks, as introduced originally by Kilian and Park (2009). The results of Caporale et al. (2015) show, that oil price uncertainty has a positive impact on stock returns during periods characterized by demand side shocks, for all sectors except the Consumer Services, Financial and Oil and Gas sectors. The last two sectors are found to react negatively to oil price volatility during periods of supply side shocks. Further- more, the effect of oil price uncertainty seems to be insignificant during periods charac- terized by precautionary demand shocks. Overall, these results are in line with those of 42 Kilian and Park (2009) and others, who found that the impact of oil price changes on stock returns and the correlation between the two depend on the type of oil shock. Paper by Mensi, Hkiri, Al-Yahyaee and Kang (2018) examines the co-movements between the BRICS stock markets and both crude oil and gold prices. Their results based on the wavelet squared coherence analysis between the BRICS stock market and oil returns show a strong co-movement across frequencies and over time, suggesting a strong rela- tionship between the markets. Furthermore, Mensi et al. (2018) find that this tendency is stronger during the onset of the global financial crisis, as their results show a pro- nounced co-variation between the BRICS stock markets and oil prices between 2007 and 2013 at high frequencies. A recent study by Wang, Ma, Niu and He (2020), propose an extreme Granger causality analysis model to examine the relationship between extreme fluctuations in crude oil prices and BRICS stock markets. Over the study period from 2000 to 2007, Wang et al. (2020), employ weekly price data from the WTI crude oil and the BRICS stock indexes. They find that the impact of crude oil on the BRICS stock markets is heterogeneous, which is in line with previous literature. Overall, their results imply that the effect of oil price changes on the BRICS stock markets is stronger under extreme circumstances com- pared to normal circumstances. Since the onset of the Covid-19 pandemic, a growing body of literature has emerged to investigate the impacts of the pandemic on different macroeconomic variables as well as on the financial markets. Heinlein et al., (2021) examine how the onset of the Covid- 19 pandemic has impacted the relationship between crude oil and stock markets returns. They focus on a heterogeneous group of oil importing and oil exporting countries in ad- vanced, emerging and small economies. Using the local Gaussian correlation and the contagion testing with high frequency intraday data, Heinlein et al. (2021) find signifi- cantly higher correlations between oil and stock market returns for all countries in their sample since the onset of the Covid-19 pandemic. Furthermore, they find that oil 43 exporting countries’ stock markets tend to have more robust correlations with oil returns, in comparison to those of oil importing countries, both during the crisis and non-crisis period. Heinlein et al. (2021) suggest that this can be a sign of relatively higher vulnera- bility to adverse shocks as well as a decrease in portfolio diversification benefits for in- vestors holding assets both in oil and stocks of these oil-exporting countries. Heinlein et al. (2021) find the most apparent difference in the oil-stock market correla- tions between China and Russia. China experienced the lowest correlation in the Covid- 19 crisis, while Russian stock market exhibited the highest correlation. Heinlein et al. (2021) argue that it is possible to explain the high correlation between Russian stock markets and the crude oil with its country specific factors, such as that it is one of the largest oil exporters in the world. These results are also consistent with previous studies (e.g., Bhar and Nikolova, 2009.) On the other hand, Heinlein et al. (2021) find that China was the only country in their sample that did not exhibit higher correlations during the crisis. They suggest that this may be due to the comparatively better position the Chinese stock market has, in comparison to the other stock markets, to withstand oil price fluc- tuations during a crisis period. This finding is also consistent with previous evidence (e.g., Fang and You, 2014) on the resilience of the Chinese stock market to fluctuations in both global financial markets as well as in oil markets. A novel strand of literature has emerged since the Russian war in Ukraine started in Feb- ruary 2022. Izzeldin et al. (2023) investigates how the Russian-Ukrainian war impacts global financial markets and commodities markets. They compare whether the markets respond differently to the war than to the Covid-19 pandemic and global financial crisis. Overall they find that stock markets react to the Russian invasion of Ukraine almost im- mediately, whereas the reaction to the Covid-19 pandemic and global financial crisis oc- curred with a lag. Izzeldin et al. (2023) argue that this is because investors interpreted the war as “real news”. Markets had not discounted an actual invasion and it was con- sidered unlikely until it happened. On the other hand, the intensity of the Russian- Ukrainian war on the markets is found to be less severe in comparison to the other two 44 crisis. It is possible, according to Izzeldin et al. (2023), that investors are underestimating the actual implications of the war. Prior warlike experiences have been notably different as they have not been happening on the European continent and/or they often involve terrorist attacks. Hence it is possible that investors are wrongly drawing insights from these past events. Regarding the commodities, Izzeldin et al. (2023) find that specific commodities are affected differently in each of the crisis. Oil markets are most affected by the onset of the Covid-19 crisis. Another recent study by Adekoya et al. (2023) investigates the multifractality and cross- correlations between oil prices and European and non-European stock markets before and during the Russian invasion of Ukraine. Their results reveal a strong multifractal be- havior in the stock and oil markets. They find that the war has a stronger direct impact on the constancy of the oil and the European stock markets. On the other hand, it has a greater indirect impact on the constancy of the non-European stock markets, namely the US, China and Japan, through oil prices. In a prior study, Adekoya et al. (2022) study how oil connects with significant financial assets, such as stocks, bonds and US dollar, before and during the Russian-Ukrainian war. In short, their results show that overall connectedness is stronger during the war in com- parison to before it. They also find that the war changes the spillover direction of some assets and increases it for others. For instance, oil becomes a net transmitter of spillover, indicating its significant effect on other assets. In addition, Adekoya et al. (2022) find that the connectedness is time varying, being the strongest at the beginning of the war. Fi- nally, they find that the war makes oil a safe-haven instrument against risks in other mar- kets, due to its net transmitting status. 45 6 Data and summary statistics This study empirically examines the relationship between the BRICS stock markets and crude oil prices over the daily period from November 19, 2014 to January 30, 2023. The MSCI (Morgan Stanley Capital International) database is used to obtain the data for the BRICS stock markets. The data consist of daily closing index prices on each individual emerging market. This study uses the MSCI country indexes for each of the BRICS econ- omies. In addition, the MSCI world index is used to act as a proxy of global stock markets. MSCI country indexes are designed to measure the performance of the large and mid- cap segments of the corresponding market. Each index covers approximately 85 percent of the free float-adjusted market capitalization in each country. Furthermore, all data is expressed in US dollars in order to measure their homogeneous features and to avoid the possible effect of currency risks on empirical results. The stock returns are computed by taking the difference in the logarithm between two consecutive prices. As for oil data, this study employs daily spot prices of West Texas Intermediate (WTI) and Brent crude oils, available from the US Energy Information Administration (EIA). For each price series, this study computes the returns (𝑟$) by taking the difference in the logarithm between two consecutive prices: 𝑟$ = 100 𝑥 log ( )# )#$% ) (6) In order to avoid distortions in the oil-stock market relationship caused possibly by the global financial crisis, the period 2008-2010 will be excluded in this study. This will also allow the study to have a more recent dataset compared to the previous similar studies, and to emphasize the importance of the onset of the Covid-19 pandemic. As mentioned, the aim of this study is to examine whether the dependence structure between the BRICS stock returns, and the global oil prices has changed since the onset of the Covid- 19 pandemic. To achieve this, the study period will be divided into two subperiods re- ferred to as pre-crisis period (November 19, 2014 to March 11, 2020) and crisis period (March 12, 2020 to January 30, 2023), respectively. March 11, 2020 is chosen as the 46 turning point, because it is the day when the World Health Organization announced the outbreak of the Covid-19 pandemic. Furthermore, since the crisis period includes also the Russian invasion of Ukraine, which started on the 24th of February 2022, the crisis period will be further divided into two sub-periods, first one from March 12, 2020 to February 23, 2022 and the second one from February 24, 2022 to January 30, 2023. The turning point between the crisis periods is chosen to be February 24, 2022, because it is the day when Russia started the invasion of Ukraine. This way it is possible to differenti- ate between the two crisis and examine the effects of Covid-19 pandemic as well as the Russian invasion of Ukraine, separately. It is essential to assess these effects separately, since the impacts on stock and oil markets have been different during the two crises. Nevertheless, the crisis period will be examined also as a full period in order to do a proper comparison. As a result of the Russian invasion of Ukraine, the Russian equity market became not investable and MSCI reclassified its Russia indexes from emerging markets to standalone market status. Thus, the study period for Russia is shorter and ends on February 28, 2022. The crisis period therefore includes 459 daily observations for the Russian market, whereas the other markets have 673 daily observations in the crisis period. Furthermore, Russia will not be included in the second crisis period, which starts on February 24, 2022, because it would not be meaningful to study the oil price-stock market relationship based on only a few days. Descriptive statistics of the stock indexes and oil returns are reported in the table 1 for the pre-crisis period and full crisis period. In the pre-crisis period the average daily BRICS stock returns are positive in China and Russia and negative in Brazil, India and South Africa, ranging from -0.019% (South Africa) to 0.021% (China). Also the average oil re- turns are negative, -0.069% for WTI and -0.046% for Brent. It is worth noting that China and Russia’s average returns are higher than that observed for the world market index (0.015%). During the crisis period, the Chinese mean return drops into negative (- 0.021%). The lowest average return is observed within the Russian stock returns (- 47 0.079%), whereas the highest returns in the crisis period are associated to the oil mar- kets, with average daily returns of 0.307% for WTI and 0.106% for Brent. Overall, during the full study period, the stock markets in oil exporting countries (Brazil and Russia) tend to be riskier in comparison to the stock markets in oil importing coun- tries. Over pre-crisis period, the highest standard deviation, which is an indicator of the risk level, is associated to the oil markets with values of 2.60% (WTI) and 2.42% (Brent). Brazilian market is the riskiest out of the BRICS markets with a standard deviation of 2.06% followed by the South African and Russian stock markets. All the BRICS markets exhibit higher volatility than the world market index, as indicated by their high standard devia- tions compared to the world market (0.78%). During the crisis period, the standard de- viations increase for all markets. Oil markets experience the greatest growth with values of 7.57% (WTI) and 4.59% (Brent). Within the BRICS markets, the highest standard devi- ation in the crisis period is observed in Russia, 3.62%, whereas India is the least risky BRICS market with a standard deviation of 1.59%. All the return series exhibit negative Skewness in the pre-crisis period, indicating that the tail is on the left side of the distribution. During the crisis period, majority of the return series get even more negative values, with the Chinese and WTI returns as an exception, as their skewness turns positive. Highest skewness in the crisis period is asso- ciated to the WTI returns (8.93) and the lowest to the Russian stock returns (-6.15). Ex- cess kurtosis, as compared to the value of the normal distribution, is observed for all return series in the pre-crisis period, with lowest values in China and South Africa. During the crisis period, all the return series exhibit higher values of excess kurtosis compared to pre-crisis. These high degrees of kurtosis indicate that the probability of observing extreme values is much higher relative to a normal distribution. Moreover, the Jarque- Bera test for normality is performed. According to the results, the null hypothesis of nor- mality is strongly rejected for all return series in both subperiods. Finally, the null hy- pothesis of a unit root is tested using the augmented Dickey and Fuller (1979) test. The 48 null hypothesis of the presence of unit root is rejected for all return series in the pre- crisis and crisis periods, indicating that they are stationary. Table 1. Descriptive statistics Table 2 reports the descriptive statistics for the two separated crisis periods. During the first crisis period the average daily returns were positive for all the return series with Brazilian stock returns as an exception, where the average return is -0.08%. The highest returns during this period are associated to the WTI returns, with an average of 0.51%. During the second crisis period, which includes the Russian invasion of Ukraine, the av- erage returns are negative for all the series, except for the Brazilian stock returns and Brent oil returns. Russia has the lowest daily returns during the second crisis period, with an average of -1.86%. During the first crisis period, which includes the onset of the Covid-19 pandemic, Brazil- ian stock market is the only one to have negative mean returns, whereas during the sec- ond crisis period all the daily mean returns are negative. The riskiest stock markets dur- ing the Covid-19 crisis are the Brazilian and Russian stock markets, with a standard devi- ations of 2.73% and 2.16% respectively. Both of the oil returns exhibit high standard de- viations, 8.94% for WTI and 5.15% for Brent. Comparing to the second crisis period, the 49 Chinese stock market is the only one to exhibit a higher standard deviation than in the first crisis period. On the other hand, the riskiness of oil markets has decreased notably from the first crisis period. Standard deviation for both WTI and Brent is little over 3%, whereas for Chinese stock market the standard deviation is 2.45% during the second crisis period. This indicates that for the oil markets, and also for the BRICS markets with China as an exception, the Covid-19 related crisis period caused more fluctuations and made the markets riskier. On the other hand, the Chinese stock market is found to be more riskier during the second crisis period in comparison to the first period. Excess kurtosis is observed during the first crisis period for all return series, with China having the lowest and India the highest level of excess kurtosis within the BRICS markets. Both of the oil returns have very high excess kurtosis levels during the first crisis period. Over the second crisis period, highest excess kurtosis is observed in the Chines stock returns. Results of the Jarque-Bera test indicate that the null hypothesis of normality can be rejected for all returns series in the first crisis period and for all returns except for South Africa and World returns in the second period. Finally, the null hypothesis of the presence of unit root can be rejected for all return series in both periods. Table 2. Descriptive statistics for the two crisis sub-periods 50 7 Methodology In order to analyze the relationship between the BRICS stock returns and global oil price changes, this study employs a multiple linear regression model. It has been previously used in the literature to assess the impact of oil price changes on stock returns in studies such as Jones and Kaul (1996), Hammoudeh and Li (2004), Sharif et al. (2005), Nandha and Faff (2008) and Scholtens and Yurtsever (2012). The model used in this paper is based on the standard market model that is augmented with the oil price factor. Two- factor model as such can be underspecified because the exchange rate between the US dollar and the home currency as well as bond yields are not included (Sadorsky, 2001). Therefore, in this study, to better control the macroeconomic circumstances and to in- crease the explanatory power and accuracy of the model, interest rate and exchange rate changes will be added as control variables to the model, since those two factors are often measured in the dependencies between oil price and stock returns. In this study the regression model is formed so that stock returns are used as the de- pendent variable and oil price changes, global market returns, exchange rate changes, and interest rate changes as independent variables. Each country’s 3-month government bond yields are used to express the interest rates. As for the global market returns, this study uses the MSCI world index returns. The multiple linear regression model can there- fore be denoted as follows (Wooldridge, 2013): 𝑅3$ = 𝛼 + 𝛽4𝑅4$ + 𝛽5𝑅5$ + 𝛽6𝑅6$ + 𝛽#07𝑅#07$ + 𝜀$ (7) Where: 𝛼 = The constant 𝑅3$ = BRICS stock returns 𝛽4𝑅4$ = Beta and daily returns of oil 𝛽5𝑅5$ = Beta and daily returns of global stock market 𝛽6𝑅6$ = Beta and daily changes of exchange rate 51 𝛽#07𝑅#07$ = Beta and daily changes of interest rate (3-month government bond yield) 𝜀$ = The error term 52 8 Empirical results In order to ensure that there is not a too high correlation between the independent var- iables, i.e., multicollinearity, in the multiple regression model, correlation matrices have been conducted for each country and each period. It is not fully clear what is too high level of correlation, but usually correlations over 0.7 are considered strong, and those should be avoided between the independent variable. Correlations close to or over the 0.7 are observed only between the two oil prices, Brent and WTI. However, since the regressions are done separately, using WTI as the oil price variable in the other model and Brent in the other, this high level of correlation will not be an issue. The results for the correlation matrices are