1 Joonatan Nieminen Geopolitical Risk and the U.S. Stock Returns Vaasa 2023 School of Accounting and Finance Master’s thesis in Finance Master’s Degree Programme in Finance 2 UNIVERSITY OF VAASA School of Accounting and Finance Author: Joonatan Nieminen Title of the Thesis: Geopolitical Risk and the U.S. Stock Returns Degree: Master of Science in Economics and Business Administration Programme: Master’s Degree Programme in Finance Supervisor: Professor Sami Vähämaa Year: 2023 Pages: 103 ABSTRACT: The conspicuousness of the geopolitical risk has grown considerably in recent years among ex- perts from various fields. Especially late in the 21st century, the topic has been approached in the academic literature, for example, through oil, tourism, national defense, and financial mar- kets. However, studies that specifically has been dealt with the connection of different sectors of the United States stock market to the geopolitical risk are rare when writing this thesis. This study examines the impact of the Ukraine War 2022, the Iraq War 2003, and September 11, 2001, terrorist attacks between stock returns from Information Technology, Consumer Staples, and Energy sectors of the U.S. S&P 500 index, as well as the effect of the geopolitical risk on the stock returns of various industries in the United States. The theoretical framework of the research has been built from the development of geopolitics, geopolitical risk, and its relationship to various industries in the United States and other risk indicators such as the VIX and the EPU indices, as well as the efficient market hypothesis with focusing on event study. The data for the study consists of returns from the U.S. S&P 500 index, returns of various industries in the years 1985–2023, and geopolitical information from the same period. In the empirical part of the research, the event study methodology is implemented in the review period of the 21st century and the ordinary least squares estimation method in the timespan from 1985 to 2023. The research results of this thesis show that geopolitical risk affects stock market returns in the United States with both empirical methods. The results obtained through the OLS estimation method indicate that the U.S. value-weighted stock returns of different industries react more significantly than the equally weighted returns. The companies in the Consumer Staples and Information Technology sector lose relative to the market in the event of a geopolitical incident. In addition, the results show that the GPA index is statistically the most significant in terms of stock returns. The results obtained through the event study, in turn, indicate that companies in the Energy sector gain from the geopolitical risk in longer event windows at the start of the war in Ukraine. Moreover, the results show that stock returns in the Consumer Staples sector are increasing in connection with the Iraq War and the Information Technology sector in shorter event windows around September 11, 2001. In other event windows and events, all three sectors perform neg- atively during geopolitical tension as measured by the stock returns. As a side note, in the em- pirics of the study, it is found that investors can utilize a hedging strategy with the put options of the Information Technology sector firms during a geopolitical threat. The variation of the re- search results between different geopolitical incidents and event windows of the event study indicates the possibility of further future research on the subject area. KEYWORDS: geopolitical risk, stock returns, the United States, sectors, S&P 500 index, event study, OLS methodology 3 VAASAN YLIOPISTO Laskentatoimen ja rahoituksen akateeminen yksikkö Tekijä: Joonatan Nieminen Tutkielman nimi: Geopoliittinen riski ja Yhdysvaltojen osaketuotot Tutkinto: Kauppatieteiden maisteri Oppiaine: Rahoituksen maisteriohjelma Työn ohjaaja: Professori Sami Vähämaa Valmistumisvuosi: 2023 Sivumäärä: 103 TIIVISTELMÄ: Geopoliittisen riskin tunnettuus on kasvanut huomattavasti viime vuosien aikana eri alojen asi- antuntijoiden keskuudessa. Varsinkin lähimenneisyydessä 2000-luvulla akateemisessa kirjalli- suudessa aihealuetta on lähestytty esimerkiksi öljyn, matkailualan, maanpuolustuksen, sekä ra- hoitusmarkkinoiden kautta. Kuitenkin tutkimukset, joissa on käsitelty erityisesti Yhdysvaltojen osakemarkkinan eri sektoreiden yhteyttä geopoliittiseen riskiin ovat harvinaisia tämän opinnäyt- teen tekohetkellä. Tässä tutkimuksessa tarkastellaan Ukrainan sodan 2022, Irakin sodan 2003, sekä vuoden 2001 syyskuun 11. päivän terroristi-iskujen vaikutusta Yhdysvaltojen S&P 500 in- deksin informaatioteknologia-, päivittäistavara-, ja energiasektoreiden osaketuottoihin, sekä myös geopoliittisen riskin vaikutusta eri teollisuudenalojen osaketuottoihin Yhdysvalloissa. Tutkimuksen teoreettinen viitekehys on rakennettu geopolitiikan kehityksestä, geopoliittisesta riskistä ja sen indekseistä, geopoliittisen riskin suhteesta Yhdysvaltojen eri teollisuudenaloihin ja muihin riski-indikaattoreihin kuten VIX- ja EPU-indekseihin, sekä tehokkaiden markkinoiden hypoteesista keskittyen tapahtumatutkimukseen. Tutkimuksen aineisto koostuu Yhdysvaltojen S&P 500 indeksin yritysten ja kokonaisindeksin, sekä eri teollisuudenalojen tuotoista vuosina 1985–2023 sekä myös geopoliittisesta informaatiosta samalta aikajaksolta. Tutkimuksen empii- risessä osassa toteutetaan tapahtumatutkimuksen metodologiaa 2000-luvun tarkastelujaksolla sekä pienimmän neliösumman estimointimenetelmää aikavälillä 1985–2023. Tämän opinnäytetyön tutkimustulokset esittävät, että geopoliittinen riski vaikuttaa osakemark- kinoiden tuottoon Yhdysvalloissa kummallakin empiirisellä menetelmällä. Pienimmän neliösum- man estimointimenetelmän kautta saadut tulokset osoittavat, että Yhdysvaltojen eri teollisuu- denalojen arvopainotetut osaketuotot reagoivat merkittävämmin kuin tasapainotetut tuotot, ja että yritykset päivittäistavara- ja informaatioteknologiasektorissa häviävät suhteessa markkinoi- hin geopoliittisen tapauksen sattuessa. Lisäksi tulosten kautta havaitaan, että GPA-indeksi on tilastollisesti merkittävin osaketuottojen kannalta. Tapahtumatutkimuksen kautta saadut tulokset puolestaan osoittavat, että yritykset energiasek- torissa reagoivat positiivisesti geopoliittiseen riskiin pidemmissä aikaikkunoissa Ukrainan sodan alkaessa, päivittäistavarasektorissa Irakin sodan yhteydessä, sekä informaatioteknologiasekto- rissa lyhemmissä aikaikkunoissa syyskuun 11. päivän ympärillä vuonna 2001. Muissa aikaikku- noissa ja tapahtumissa kaikki kolme sektoria suoriutuvat negatiivisesti geopoliittisen jännitteen aikana osaketuotoilla mitattuna. Sivuhuomiona tutkimuksen empiriassa löydetään, että sijoitta- jat voivat hyödyntää suojausstrategiaa informaatioteknologiasektorin yritysten myyntioptioilla geopoliittisen uhan aikana. Tutkimuksen tulosten vaihtelu eri geopoliittisten tapahtumien sekä tapahtumatutkimuksen eri aikaikkunoiden välillä kertoo aihepiirin lisätutkimuksen mahdollisuu- desta tulevaisuudessa. AVAINSANAT: geopoliittinen riski, osaketuotot, Yhdysvallat, sektorit, S&P 500 indeksi, tapah- tumatutkimus, pienimmän neliösumman estimointimenetelmä 4 Contents 1 Introduction 8 1.1 Research problem and purpose of the thesis 9 1.2 Research gap and intended contribution 10 1.3 Structure of the thesis 12 2 Literature review and hypothesis development 13 2.1 Studies related to the U.S. market 13 2.2 Studies related to the emerging markets 17 2.3 Studies related to the advanced economies, MENA-, and OPEC-countries 21 2.4 Summary of the previous studies 24 2.5 Hypothesis development 25 3 Theoretical framework 31 3.1 Development of the geopolitics 31 3.2 Geopolitical risk and index 32 3.3 Geopolitical threats and acts 36 3.4 Geopolitical risk by industries 39 3.5 Geopolitical risk vs. other risk indicators 42 3.6 Efficient market theory in the light of event studies 45 4 Data and methodology 52 4.1 Data gathering and description 52 4.2 Research methods of the thesis 57 4.3 Reliability of the research and robustness of the results 63 5 Empirical results 68 5.1 Results of the OLS regressions 68 5.2 Results of the event study calculations 74 6 Conclusions 83 6.1 Practical Implications and contributions 85 6.2 Restrictions and future research directions 87 References 89 5 Appendices 101 Appendix 1. The search query for the benchmark GPR index 101 Appendix 2. Industry exposure to GPT, GPR, and GPA with the monthly average value- weighted stock returns 102 Figures Figure 1. Geopolitical threats and its involved aspects (Brown, 2022). 11 Figure 2. Recent GPR index from January 1985 through June 2022. 33 Figure 3. Daily GPR index from 1985 through end-2020 (Caldara & Iacoviello, 2022). 34 Figure 4. Historical GPR index from January 1900 through June 2022. 36 Figure 5. GPRT and GPRA subindexes of the Geopolitical Risk Index. 39 Figure 6. Comparison between benchmark GPR index and EPU index. 44 Figure 7. Comparison between benchmark GPR index and VIX index. 44 Figure 8. The timeline for the event study. 59 Figure 9. Exposure to the geopolitical acts by industries with the monthly stock returns. 73 Tables Table 1. Examples of phrases sought to build the GPR indexes (Caldara & Iacoviello, 2019). 37 Table 2. Descriptive statistics focused on the event windows’ averages. 56 Table 3. Industry exposure to GPT, GPR, and GPA with the monthly average equal- weighted stock returns. 69 Table 4. Results of the event study calculations (values in the percentage form). 77 Table 5. Comparison with the CAAR values to Fossung et al. (2021). 81 Table 6. Market daily returns around geopolitical events. 82 Abbreviations AAR = Average abnormal returns Aero = Aircraft 6 Agric = Agriculture APT = Arbitrage pricing theory AR = Abnormal returns Autos = Automobiles and Trucks Banks = Banking Beer = Alcoholic Beverages BldMt = Construction Materials Books = Printing and Publishing Boxes = Shipping Containers Bps = Basis point(s) BRICS = Brazil, Russia, India, China, and South-Africa BusSv = Business Services CAAR = Cumulative average abnormal returns CAPM = The capital asset pricing model CAR = Cumulative abnormal returns Chems = Chemicals Chips = Electronic Equipment Clths = Apparel Cnstr = Construction Coal = Coal CRSP = The Center for Research in Security Prices DCC = The dynamic conditional correlation model DCC-MVGARCH = Dynamic conditional correlation multivariate GARCH Drugs = Pharmaceutical Products EconUnc = Economic uncertainty ElcEq = Electrical Equipment EMH = Efficient market hypothesis ENSO = El Niño-Southern Oscillation EPU = Economic policy uncertainty FabPr = Fabricated Products Fin = Trading FMOLS = Fully modified ordinary least square model Food = Food Products FS = Financial stress indicator Fun = Entertainment GARCH = Generalized autoregressive conditional heteroskedasticity GDP = Gross domestic product Gold = Precious Metals GPR = Geopolitical risk GPRA (GPA) = Geopolitical acts GPRT (GPT) = Geopolitical threats Guns = Defense 7 Hardw = Computers Hlth = Healthcare Hshld = Consumer Goods Insur = Insurance IT = Information Technology LabEq = Measuring and Control Equipment Mach = Machinery MacroUnc = Macroeconomic uncertainty Meals = Restaurants, Hotel, and Motel MedEq = Medical Equipment MENA = Middle East and North African MGARCH = Multivariate GARCH MIDAS = Mixed data sampling Mines = Nonmetallic Mining Misc = Miscellaneous NARDL = A nonlinear autoregressive distributed lag OIL = Brent crude oil prices Oil = Petroleum and Natural Gas OLS = Ordinary least squares regression Paper = Business Supplies PerSv = Personal Services RlEst = Real Estate Rtail = Retail Rubbr = Rubber and Plastic Products Ships = Shipbuilding and Railroad Equipment Smoke = Tobacco Products Soda = Candy and Soda Softw = Computer Software Steel = Steel Works, Etc. Telcm = Telecommunications Toys = Recreational Products Trans = Transportation TVP-VAR = Time-varying parameter vector autoregression model Txtls = Textiles U.S. = The United States UK = The United Kingdom USD index = US dollar index Util = Utilities VAR = Variance VAR-BEKKGARCH = Vector autoregressive Baba, Engle, Kraft, & Kroner GARCH VIX = Cboe Volatility Index® Whlsl = Wholesale 8 1 Introduction Caldara and Iacoviello (2019) present the geopolitical risk index in their first paper, de- scribing a new concept of geopolitical risk using the news-based method. European Cen- tral Bank, World Bank, and International Monetary Fund are among the institutions that use the geopolitical risk index (Caldara & Iacoviello, 2022). The Bank of England notes geopolitical risk as a massive threat to business economics that can be kept in the uncer- tainty circle together with political and economic uncertainty (Carney, 2016). Addition- ally, Wells Fargo’s (2017) survey states that 75% of the investors (from more than 1,000) fear possible diplomatic and military tensions worldwide. These issues demonstrate awakened consciousness of the geopolitical risks among the market participants. Concerning geopolitical risk, the latest and maybe the best-known episode in the past few years escalated from the spring of 2022 onwards in Ukraine when Russia attacked the country. World Population Review (2023), ACLED (2023), and Wikipedia (2023) webpages show that besides the Russo-Ukrainian War, there are two other ongoing large wars in the world located in Ethiopia and Myanmar. Approximately one-third of the world’s surface area witnesses current terrorist acts or wars. Since trade relations and political placements are nowadays global, the countries' internal tensions can affect the peaceful course of international relations or its permanence. Examples are the Burma Act which U.S. President Joe Biden signed for Myanmar’s democracy and humanitarian aid, and his comment on the defense support for Taiwan in case of China’s invasion (Eleven Myanmar, 2022; Brunnstrom & Hunnicutt, 2022). Additionally and according to Enuma (2021), the Epiphany 2021 events in the United States dramatically shaped the nation's internal order and the political field of the U.S. In their paper, Caldara and Iacoviello (2019) argue that there is a lack of measures that count for geopolitical risk because the measurements are inconsistent over time. The lack of empirical research on geopolitical risk also arises from the too-broad estimate of the geopolitical risk components or the definition of geopolitical risk. The previous measures are tough to replicate since the determination and data are not publicly 9 provided and consist of subjective analysis without mentioning the methodology. How- ever, the geopolitical risk index by Caldara and Iacoviello (2019; 2022) fills that gap in the macroeconomic field. They use real-time newspaper measure, which catches press and public discussion by the policymakers and the investors. Their index systematically cap- tures the latest crisis in Ukraine and all other geopolitically big and risky events (e.g., 9/11). The following sub-chapter shows the purpose of the thesis regarding geopolitical risk (GPR from this onwards). 1.1 Research problem and purpose of the thesis The thesis examines the relationship between the GPR and stock market returns in the United States. Specifically, this thesis looks at the Information Technology, Consumer Staples, and Energy sectors of the S&P 500 index in the US. Fossung et al. (2021) con- ducted that sort of study earlier, and the thesis follows their approach in the empirics. Consequently, a researcher divides the firms into sectors using the sectors by the GICS® (Global Industry Classification Standard) procedure and standards that MSCI (2020) pre- sents. This thesis also follows the event study methodology with four event windows described by Fossung et al. (2021). Delimitation of the thesis goes as follows; first, the geopolitical risk index by Caldara and Iacoviello (2019) provides data from 1900 onwards, while their recent index provides data from 1985. However, to keep the thesis manageable, this thesis focuses on the 21st century of geopolitical events. Second, Fossung et al. (2021) focus on 58 geopolitical events in the GPR index from 1962 to 2020. With that number of occasions, they perform more than 17,000 regressions. However, that number of estimations goes beyond the purpose of the thesis. In detail, this thesis focuses on the three (3) most significant geo- political events (spikes in the GPR-index) in the 21st century, titled 9/11 in 2001, Iraq War in 2003, and Ukraine War in 2022. Third, the delimitation means that a researcher calculates the regressions in the Infor- mation Technology sector with event studies as follows: 3 (number of events) *4 10 (number of event windows) *75 (number of firms in the IT sector) = 900. The total num- ber of regressions needed with the event studies is 1,572. In addition, to calculate the industry exposure of the GPR with the OLS methodology, a researcher follows the first equation by Caldara and Iacoviello (2019). That approach allows a broader empirical part in the thesis while simultaneously, a researcher seeks to compare the results between different methodologies. Research problems for the thesis are the following: 1. Does the geopolitical risk affect the stock market returns in the U.S., 2. Does the effect of the geopolitical risk differ between the Information Technology, Consumer Staples, and Energy sectors of the S&P 500 index, 3. Do the results differ between methodologies, 4. Do the results differ between geopolitical events? In addition, a researcher compares the results between different event windows in this thesis and the papers by Caldara and Iacoviello (2019; 2022) and Fossung et al. (2021) to the final results. 1.2 Research gap and intended contribution Even though geopolitical risk is quite a new concept in academic research, studies are increasing. However, research on the relationship between geopolitical risk and stock market returns is rare, especially with U.S. data. Another gap in the literature is a lack of studies with an event study approach, sector incorporation, and the S&P 500 index. Ad- ditionally, there is a lack of studies that use daily data on the stock returns and the GPR, therefore, not allowing to consider the impact of the GPR at a higher frequency. Figure 1 shows that geopolitical risk reaches wide dimensions in everyday business and life. For example, a few terms regarding geopolitical risk are the U.S.-China trade, the London 2005 bombings, the Paris 2015 attacks, the Iraq invasion, national political affairs, 11 uncertainties with the future, the threat of wars and terrorism, global tensions and ten- sions within states, and cyber-attacks. Specifically, this thesis aims to focus on the market dynamics part of Figure 1 regarding geopolitical risk. Figure 1. Geopolitical threats and its involved aspects (Brown, 2022). The primarily intended contribution comes from the concept of geopolitical risk itself. The index by Caldara and Iacoviello (2019; 2022) offers an opportunity to focus on some specific areas of the financial markets together with the GPR because there is room for more studies with this novel concept. In this study, the contribution comes by focusing on the U.S. data of the S&P 500 index, investigating how geopolitical events affect the stock market returns, and examining the effect between different sectors of the index. When Fossung et al. (2021) use only event study methodology, a thesis researcher also uses the OLS methodology from Caldara and Iacoviello (2019). Therefore, this study con- nects data from the Fama and French (1997) website, titled 49 industry portfolios 12 (monthly), for the thesis to compare the results of the OLS regressions to the event study method. Furthermore, when Fossung et al. (2021) focus on the Information Technology, Commu- nication Services, and Consumer Staples sectors of the S&P 500 index, in this thesis, a new industry, the Energy sector, is brought along to the research which replaces the Communication Services sector. Because of that, it is possible to compare part of the results of the thesis to their previous study but simultaneously permits producing a piece of new information from the Energy sector of the S&P 500 index. Lastly, this study con- centrates on the 21st century’s geopolitical events while taking well-known 9/11 from 2001 and Iraq War from 2003 to the study, but also the latest geopolitical event of the world, the Ukraine War from 2022. The extension to the timeline of the events makes it possible to compare the newest event to the other ones the GPR index is capturing. 1.3 Structure of the thesis This thesis contains six different chapters covering the project. First, a brief introduction starts the thesis with the purpose of the research and research questions, research gap, and intended contribution. The second chapter moves into the literature review of geo- political risk and defines the hypotheses. The third chapter presents a detailed theoret- ical framework of geopolitical risk and the efficient market theory. In chapter four, the study moves to the empirical part of the project showing data and methodology, along with the reliability and robustness of the research. Chapter five delivers the empirical results of the thesis, and Chapter six concludes with practical implications and contribu- tions, as well as restrictions and future research directions. 13 2 Literature review and hypothesis development This chapter takes a more detailed look at the previous studies related specifically to geopolitical risk, financial markets, and stock returns. This chapter aims to make a basis for formulating a theoretical framework later in Chapter 3, increase knowledge of the relevant research, and critically analyze the previous literature. Lastly, the hypothesis development takes place at the end of the chapter. 2.1 Studies related to the U.S. market The first study shows how specific events can cause uncertainties regarding the Con- sumer Staples sector in the U.S. Seo et al. (2013) analyze the impact of food safety events on the market value of food-related firms in the U.S. between 1993–2012. They calculate abnormal and cumulative abnormal returns and find that firm-specific factors and media attention affect the significance of the impact of food safety events. Furthermore, the results show that abnormal returns are significantly negative during t1 and t2. Cumulative abnormal returns stay significantly negative 57 days after the event, staying negative 254 days as a whole, meaning that negative returns are related to the event's appearance and that it takes approximately one year to recover from the event entirely. The article from Antonakakis et al. (2017) examines the relationship between geopoliti- cal risk and the oil-stock nexus from 1899–2016. The monthly data for their paper comes from the S&P 500 index, WTI oil index, and historical GPR index. They use VAR-BEKK- GARCH and multivariate GARCH models to capture the results. The findings show that the GPR significantly affects the oil returns and its volatility, and the impact is not positive. Also, the covariance between oil and stock markets decreases considerably when the time lag of the GPR index is in the regressions. Regarding stock returns, the impact of the GPR is statistically insignificant. The lack of a newfound connection between the GPR and stock returns may be because Antonakakis et al. (2017) use monthly data. 14 Evidence shows that the daily data works better to find the connection between the GPR and stock returns or volatility (e.g., Caldara & Iacoviello, 2019; Balcilar et al., 2018; Yang et al., 2021; Apergis et al., 2018). Another issue may be that discounting and incorporat- ing the exogenous GPR news into the U.S. market is more efficient. However, the fre- quency of the data better explains the unfound connection between variables in this case since other studies prove that the connection exists in the U.S. market between the GPR and the stock returns (e.g., Caldara & Iacoviello 2019; Fossung et al., 2021; Yang & Yang, 2021, Smales, 2021; Salisu et al., 2021). Caldara and Iacoviello (2019) find that geopolitical risk negatively affects the U.S. stock market returns, but the result depends on the industry. They show that when the GPR changes and spikes, the stock market returns decline, and economic activity decreases. They collect the industry data from Fama and French's (1997) website, titled 49 industry portfolios (daily). They find that exposure to the GPR differs between industries when positive exposure means declining stock returns and negative exposure means growing stock returns. Among positive exposure are, for instance, fabricated products, medical equipment, and tobacco products. Negative exposure includes petroleum and natural gas, precious metals, and alcohol products. Moreover, they indicate that fixed investments in the U.S. market decline when the GPR spikes. Indeed, Caldara and Iacoviello (2019) demonstrate that the possible threat af- fects the markets more than the realized act. They comment that GPR can act as a sup- port for previous theories predicting economic uncertainty. Thus the variations in GPR and other macroeconomic factors can push the economic and business cycles. Atems et al. (2020) concentrate on a specific industry, the U.S. food and agricultural in- dustry, and its stock returns in terms of El Niño-Southern Oscillation (ENSO) shocks be- tween 1980–2018. They use monthly data of 12 companies with a vector autoregression model and find that ENSO shocks affect significantly and positively seven of the twelve companies. All twelve companies belong to the S&P 500 index, and ten are in the 15 Consumer Staples sector. However, the results significantly decrease close to zero after three to six months, meaning that the effect of the shock is short to the stock returns. They also find that the predictive explanatory power of the ENSO shocks is small and that historically other shocks impact the U.S. food and agricultural stock returns. Baur and Smales (2020) investigate the relationship between geopolitical risk and asset prices. They find that geopolitical risk is a new and exogenous risk that cannot be straightly compared to the other possible risks, such as financial or political (cf. Das et al., 2019). Their study shows that the precious metals industry serves as a hedge against geopolitical risk, and gold and silver demonstrated the best cover for the crisis during geopolitical threats. Vice versa, stocks and bonds negatively correlate with geopolitical risk. The data used for stock prices is from the S&P 500 index. Additionally, they find that stock prices respond more to GPR threats than GPR acts and that after 9/11 and the market reopening, the S&P 500 index was 5.31% lower than before the terrorist attacks. The results which Bauer and Smales (2020) find indicate that investors can reduce the risk involved in geopolitical risk when hedging it with precious metals. This risk is not in economic, financial, or other political indexes. Triki and Maatoug (2021) examine the relationship between gold and the S&P 500 index under geopolitical risk from 1985 to 2018. Specifically, they use monthly data for GPR, gold, and the stock market. The model they utilize in their paper is a multivariate GARCH model to capture moving variances and covariances over time. The results of their work show that the average geopolitical risk is lastingly higher after the 9/11 terrorist attacks. In addition, gold correlates highly with the S&P 500 index during high-ranking geopoliti- cal tensions and lowly during small GPR, and gold hedges significantly against S&P 500 index volatility, particularly during geopolitical tensions. Moreover, they find that S&P 500 index monthly returns are negatively skewed, meaning expected future losses (and small gains). Gold and GPR show positive asymmetry representing an improved proba- bility of positive returns. Their results indicate that the gold and materials industry can 16 differ from the other industries of the S&P 500 index during geopolitical threats and that GPR affects the stock market returns in the U.S. Yang and Yang (2021) examine in their article the relationship between mixed-frequency geopolitical risk and stock market returns with OLS and MIDAS regression models during the sample period of 2000–2019. They find that the S&P 500 and Dow Jones index stock returns decline significantly during higher GPR spikes when using quarterly, monthly, and weekly frequency. They also calculate the economic significance of the results with weight coefficients finding that if the GPR increases by 1% in one month, the quarterly stock market returns will fall by 3.74%. Similarly, if the weekly GPR increases by 1% in one week, the quarterly stock market returns will decrease by 3.22%. Additionally, they find that the MIDAS model with mixed frequency generates better β and R2 compared to the traditional OLS regression meaning that the MIDAS model explains better the var- iation of the stock returns. Lee et al. (2021) investigate the association between oil prices, geopolitical risk, and the green bond index with monthly U.S. data. Their methodology is the Granger-causality in quantile analysis within 2013–2019. The results tell that the GPR and the oil prices have positive skewness. Lee et al. (2021) state unidirectional relationships exist between the GPR and oil prices at the extreme quantiles. The relationship between the GPR and the green bond index exists at the lower quantiles. Specifically, at the quantiles 0.80 and 0.90, a p-value is 0.029 with a 5% significance level and one lag between the GPR and the oil prices. The results remain approximately the same, with two or three lags. In his paper, Smales (2021) studies the role and effect of geopolitical risk in oil and stock markets from 1986 to 2018. The author uses OLS, univariate and multivariate GARCH models to calculate the results and the daily data of the S&P 500 index, the WTI index, and the GPR. The results uncover that the growth in the GPR correlates positively with oil returns and negatively with stock market returns. In detail, the OLS regressions reveal that the change in the GPR predicts 20 (0.2%) basis points higher oil returns and –7 (– 17 0.07%) basis points lower stock returns. The results are significant at the 1% level. Eco- nomic significance follows when, for instance, four standard-deviation changes in the GPR relate to –15.6 (55.8*4*–0.07) percent change in stock returns when all else is equal. Since the mean daily stock price return is 3.1%, the four standard-deviation change in the GPR has a multiple of 5.03 towards it. Additionally, the GPR predicts higher variations in oil price volatility than in stock price volatility. The local environment of some geopolitical events may explain this issue when some circumstances – such as disruptions in oil supply from terrorist attacks to oil fields – immediately change oil production. Finally, the author finds a positive connection be- tween the GPR and VIX index and a negative correlation with the GPR and EPU index. 2.2 Studies related to the emerging markets Balcilar et al. (2018) analyze the connection between geopolitical risk, stock market re- turn, and volatility in the BRICS countries. They use monthly GPR and stock return data with a nonparametric causality-in-quantiles test. However, the stock returns have daily data to calculate realized volatility. The sample period under study varies from 1985 to 2016. The results reveal that the impact of the GPR is heterogeneous regarding stock returns in BRICS countries, but the volatility reacts more than the returns. Additionally, the GPR affects the stock return quantiles below the median. The results show that using daily data is justifiable when researching this area of the GPR to get more accurate re- sults. The second study regarding geopolitical risk and emerging markets comes from Bouras et al. (2019). Specifically, they study the relationship between GPR, stock returns, and volatility with the panel GARCH model under the sample period of 1998–2017. They use monthly data and find that neither the country-specific GPR nor global GPR index has significant touch on stock returns and that the effect of global GPR is more substantial on the volatility of the returns than the impact of domestic GPR factor. In detail, the 18 coefficient for the country-specific GPR is 0.691, and for the global GPR, 0.650 with a 5% significance level, but the country-specific GPR variable is insignificant. Hoque and Zaidi (2020) find in their 2003–2017 sample that the effect of the GPR is nonlinear with emerging economies, named by Brazil, India, Indonesia, South Africa, and Turkey, yielding negative and positive stock market returns when the crucial factor was volatility. However, political uncertainty generates negative stock market performance, except in India, where the performance is positive (cf. Hoque et al., 2019). However, none of their study’s countries are big-running economies worldwide. With a more ex- tended period, Apergis et al. (2018) demonstrate similarly to Hoque and Zaidi (2020) that volatility is the primary explanatory factor relating to stock returns. Still, regarding the prediction of stock returns, there is no evidence that the GPR would explain future re- turns. Similar results come from Rawat and Arif (2018). Hedström et al. (2020) research the emerging market contagion under geopolitical inse- curity. They use monthly data between 1995–2016 with variance autoregression and the GARCH model to capture spillovers between markets. They find that the possibility of contagion with ten emerging markets is high and that the GPR does not affect the stock market returns and has a weak or no spillover effect on the emerging markets. The GPR spillover varies from 0.03% to 1.15% between the markets, having a value of 0.51% in the U.S. The results uncover that the GPR correlates positively with the VIX index, the standard risk measure for stock market risk in financial markets. They also find that when comparing the GPR to the EPU index, the results show a more significant spillover influ- ence revealing the importance of policy and economic choices. Hasan et al. (2020) scrutinize the relationship between geopolitical risk and tourism stock returns in emerging markets. They use non-parametric causality-in-quantiles and cross-quantilogram models with monthly data from 13 emerging markets. Their findings tell that the global and country-specific GPR has significant predictive power to explain the average stock returns in the tourism sector of emerging markets under normal 19 market circumstances. Still, exceptions include South Korea and Columbia, where the tourism stock trade is limited or is in the emerging phase. Additionally, the global GPR has more potent power than the country-specific GPR when discussing future stock re- turns. The results indicate that the sector under study can matter when finding the con- nection between stock returns and the GPR. Furthermore, the stock returns of the tour- ism stocks are more negatively than positively skewed. Yang et al. (2021) take the geopolitical component with China’s stock market volatility and returns. They use the GARCH-MIDAS model and CSI 300 index for China’s daily stock returns to calculate and discover that global and country-specific GPR significantly influ- ences China’s stock market volatility. In detail, they study the impact of the global GPR index, GPR action and threat indexes, GPR broad and narrow indexes, and country-spe- cific indexes on China’s stock market returns. Yang et al. (2021) find that the GPR actions have a most noticeable predictive effect on stock returns between 2011–2020, but oth- erwise, the GPR threat is more significant than other indexes. Another study that examines the effect of political uncertainty and the GPR on the Chi- nese financial markets comes from Chiang (2021). He uses the DCC model with monthly stock data between 2000–2020. He finds that stock-bond return correlations are ad- versely associated with changes in the EPU, and stock-gold return correlations correlate positively with changes in the GPR in the Chinese market. Specifically, Chiang (2021) finds that changes in the EPU and the GPR create negative stock returns. This divergence with the GPR’s effect from Yang et al. (2021) can be from different methodologies, data, and sample periods, and Chiang (2021) uses change (Δ) in indexes. In their article, Saadli et al. (2021) explore the relationship between Turkish stock returns, geopolitical risk, and investor sentiment between 2004 and 2017. They use monthly data for stock returns and the GPR index with MGARCH methodology to calculate the results. The results show that the GPR and the investor sentiment adversely affect the Turkish stock market returns of the BIST 100 index and its volatility. Furthermore, results tell that 20 the GPR index is more volatile than the Turkish stock market or the investor sentiment captured by the Consumer Confidence Index. A related study by Erdoğan et al. (2022) examines the Turkish stock markets and the effect of economic policy uncertainty, geopolitical risk, and oil prices on the returns. They employ the arbitrage pricing theory model with NARDL methodology between 1997 and 2020. They find that the country-specific GPR positively affects the Turkish stock market results, that the EPU negatively affects the stock returns, and that the oil prices affect stock returns positively if the real oil price change is not positive. The results of these last two studies differ regarding the GPR and the Turkish stock market returns. Still, the difference can be because Saadli et al. (2021) use a recent monthly GPR index. In contrast, Erdoğan et al. (2022) use a country-specific GPR index for Turkey because of a different methodology and sample period. Finally, these results indicate that the global GPR factor affects the stock returns more than the domestic factor. A paper from Zaremba et al. (2022) studies the impact of geopolitical risk on emerging stock market returns under a sample period of 1990 to 2020 with asset pricing models. The results show that the change in GPR is associated with positive forecasts of future stock market returns and that the countries with the greatest exposure to geopolitical risk surpass their equivalents with the lowest exposure by up to 1% in a month. Moreover, the results indicate that the GPR does not massively correlate with the other regression variables. The country-specific idiosyncratic risk pushes the change in GPR more than the global systematic geopolitical risk. They also find that abnormal returns tend to have relative asymmetry depicted from the long and short strategies. That means the alphas are larger in high GPR countries than in the low GPR countries, unveiling evidence that the GPR rises matter more than the GPR reductions. 21 2.3 Studies related to the advanced economies, MENA-, and OPEC-coun- tries In their article, Nikkinen and Vähämaa (2010) find that terrorism negatively correlates with stock market sentiment and that the expected outcome of the FTSE 100 index de- cline during the damaging episode. The FTSE 100 index locate in London, so two of the three terrorist attacks that took place outside of the United Kingdom (in the U.S. and Madrid) within the study prove that the impact of the geopolitical events is interconti- nental. In general, it makes sense that shocking events increase people's uncertainty. In their paper, Nikkinen and Vähämaa (2010) find that implied kurtosis is higher after the adverse events, meaning that investors expect a more extreme positive or negative im- pact for the FTSE 100 index options after the attacks. The research from Apergis et al. (2018) investigates the association between GPR and stock returns of top defense companies with a nonparametric causality test. Their sam- ple period runs from 1985 to 2016, and the data is at monthly frequency. Under consid- eration are 24 global defense companies whose results show no proof that the GPR would forecast stock returns. Caldara and Iacoviello (2019) find the opposite in their pa- per that the defense industry is negatively exposed to the GPR, meaning in their article that defense companies gain more than the market when the GPR spikes. Apergis et al. (2018) find that the GPR predicts 50% of the realized volatility of leading defense com- panies by using the daily stock price data, which can prove that results within stock re- turns may differ because of the frequency of the data. Contradictory results with stock returns may also be due to different data samples, periods, and methodology. Especially the frequency in the data by Caldara and Iacoviello (2019) is quarterly, while Apergis et al. (2018) have a monthly frequency. In addition, Yang and Yang (2021) show in their article that setting the data to a minor frequency may provide more accurate results. A paper from Bouoiyour et al. (2019) examines the link between geopolitical risk and oil prices. In detail, they research whether the GPR threats or acts would be the main driver behind the higher oil prices. The authors use monthly GPR and oil price data with 22 dynamic conditional correlation, copula, and multifractal detrended fluctuation analysis models. Their findings show that the realization of the GPR acts has a powerful and pos- itive effect on oil prices, while the impact of the GPR threats remains small or non-sig- nificant. Specifically, the result of the GPR acts on the oil price is positive and greater during high-level quandary times. In addition, the level of oil prices does not affect the relationship when the association between the changes in the GPR and the oil price re- turns remains significant and positive under a larger probability of a crisis period. The correlation is forceful and positive within the countries which are large suppliers or con- sumers of oil. Similar results with oil return performance come in studies from Alqahtani et al. (2020) and Liu et al. (2019); in oil-dependent countries, there is some delicate evi- dence that the GPR predicts crude oil returns. Finally, Bouoiyour et al. (2019) find that in specific events, the oil price and the GPR correlate positively, for instance, after 9/11, during the Iraq Invasion in 2003, and the 2014 Russia-Ukraine crisis. Alqahtani and Taillard (2020) chart in their article the changes in geopolitical risk regard- ing the returns of oil prices. They use monthly data with vector autoregression, GARCH, and OLS models within the sample period of 2004 to 2018 and find that the shock in the GPR is not in connection with the oil prices and that the GPR does not produce the oil returns. However, the authors conclude that the GPR can be beneficial when decreasing the uncertainty of the regressions from the oil price predictions when incorporating the change in the GPR with a two-month lag into the models, which will develop the returns of oil indexes. An article from Salisu et al. (2021) explores the relationship between historical geopolit- ical risk and stock returns in advanced economies, specifically G7 countries and Switzer- land. They use monthly data for stock indexes, GPR between 1899 and 2020, and the historical average (constant return) model. Their results show that an increase in the GPR index produces smaller stock returns in all countries and that a 10% increase in geopo- litical threat (GPRT) produces lower stock returns, but when the GPRT decreases by 10%, then stock returns increase. Furthermore, they find that the GPR predicts stock returns 23 in all countries except Italy, and the GPRT predicts more minor returns. For instance, the GPR predicts with controls –6.87% lower stock returns in the U.S. and its S&P 500 index, while the GPRT predicts –7.77% lower stock returns in the U.S. during the latest month. In their article, Elsayed and Helmi (2021) map the association between geopolitical risk and stock return volatility in MENA (Middle East and North African) countries. They use daily stock price data between 2005 and 2018 with ADCC-GARCH and vector autoregres- sion models for spillover effects. The results show that the GPR does not promote the spillovers of returns in the MENA’s financial markets. Still, the dynamic analysis reveals that the total spillover index responds to large political events. They also find that corre- lation coefficients between countries' stock returns and the GPR are roughly zero, mean- ing that the returns are independent of the GPR index. Moreover, the results tell that the GPR explains fewer than 0.5% of the forecast-error variance in the MENA countries, meaning that stock returns of the MENA countries may already reflect historically shaky events and tensions. Related research from Abdel-Latif and El-Gamal (2021) investigates the relationship be- tween geopolitical risk, economic growth, and investment in MENA countries. They use quarterly data from 1979 to 2017 with the global vector autoregression model and find that increased GPR has adverse effects on GDP and investment and that the influence of investment declines during geopolitical tension. They also discover that the negative im- pact of the GPR on the GDP is greater in countries that are exporting oil. The effect of increased GPR on investments lasts around 5 to 10 quarters in nearly all countries. Aloui and Hamida (2021) examine the association between the oil-stock nexus and geo- political risk in Saudi Arabia, an oil-rich territory. The authors utilize the bivariate, partial, and multivariate wavelet coherency model with monthly data and a sample period from 1989 to 2019. The findings disclose that the GPR spikes produce a higher relationship to the stock prices at high-frequency with stock price as a lagging variable meaning that the GPR spikes impact Saudi Arabia’s stock market over the short-term time-scale. The 24 longer the time scale produces an elevated impact of the GPR on oil prices. Moreover, the higher GPR reduces the oil-stock connectedness in a short time horizon and de- creases the volatility correlation of the oil-stock nexus, which is in line, for instance, with Smales (2021). The article from Singh and Roca (2022) studies the connection between China’s geopo- litical risk and Canada’s equity markets. They use GARCH, a cointegration-based fully modified ordinary least square model (FMOLS), and monthly data from 2000 to 2018 for the GPR to calculate the results. The results show that China’s GPR continuously affects Canada’s stock market returns and volatility. The effect is most profound on Canada's stock market's Energy and Resources sectors, which depend on the trade relationship with China. Additionally, the impact of China’s GPR is more significant than that of the global GPR on Canada’s national stock index. Jarque-Bera test reveals that all the sectors (not including Consumer Staples) of Canada’s national index desert the normality condi- tions. The results tell that Canada is open to the GPR because of its international con- nections meaning that its national markets are vulnerable to geopolitical tensions and that the GPR affects globally due to the linkages between economies of the world’s coun- tries. 2.4 Summary of the previous studies Previous U.S. studies' findings have yielded similar results in geopolitical risk, financial markets, and stock returns. Antonakakis et al. (2017) find that the impact of the GPR is insignificant to the S&P 500 index stock returns, whereas Caldara and Iacoviello (2019) show that higher GPR risk lowers the stock returns. However, Caldara and Iacoviello (2019) do not show the significance of their values related to Figure number six. Never- theless, Baur and Smales (2020) find that stock returns decline due to the GPR, but gold is a hedge during geopolitical tensions. Similarly, Triki and Maatoug (2021) discover the same results as Baur and Smales (2020). Yang and Yang (2021) report that stock returns fall significantly in the U.S. during the elevated geopolitical tensions. Smales (2021) also finds a negative correlation between the GPR and stock market returns. Therefore and 25 based on the literature review, in the U.S., the GPR effects with negative and mainly sig- nificant touch on the stock market returns. In emerging markets, Hoque and Zaidi (2020) find that the GPR alone cannot predict future stock prices, and Hedström et al. (2020) report that the GPR does not impact the stock returns in emerging countries. On the contrary, Hasan et al. (2020) reveal that the GPR predicts both positive and negative stock returns in emerging countries when the stock market sector is specified. Chiang (2021) presents that the change in the GPR cre- ates negative stock returns in China, and Saadli et al. (2021) show the negative impact of the GPR on Turkish stock returns. However, Erdoğan et al. (2022) reveal GPR’s positive impact on the Turkish stock market. Zaremba et al. (2022) find that the positive future returns in emerging markets are associated with the change in the GPR. To conclude, the results are non-linear in emerging markets regarding the impact of the GPR. Previous studies on advanced economies, MENA, and OPEC countries have also yielded varied results. First, Apergis et al. (2018) indicate that the GPR does not predict the fu- ture stock returns of top global defense companies. Second, Bouoiyour et al. (2019) find that the GPR positively affects oil prices, while Alqahtani and Taillard (2020) show that the GPR does not correlate with the oil price returns. Third, Salisu et al. (2021) report decreased stock returns in G7 countries and Switzerland due to increased GPR. Singh and Roca (2022) indicate that China’s GPR adversely affects Canada’s national stock mar- ket returns because of Canada’s dependence on China due to trade linkages. Thus, the impact of the GPR varies from non-significant to positive and negative, subject to the methodology, country, time, and sector, for instance. 2.5 Hypothesis development With the development of the hypotheses, the expectation is that geopolitical risk influ- ences the stock market returns. For instance, Caldara and Iacoviello (2019; 2022) prove in their paper that a higher GPR adversely affects the stock market returns in the United States. In his article, Lee (2018) discover that the GPR affects the world’s stock market 26 returns statistically significantly. Caldara and Iacoviello (2019; 2022) also reveal that pos- itive exposure to the GPR leads at the industry level to greater and more consistent de- clines in stock prices. Vice versa, the industries with negative exposure to the GPR tend to outperform the market in terms of stock returns during the GPR spikes. Trade open- ness, the business's cyclicity, and the company's leverage level affect the exposure level. In addition, the firm's location, the nature of the industries and business, and the logistic network can raise imbalance with the magnitude effect of the exposure for the GPR (Cal- dara & Iacoviello, 2019). Thus there are variations in the stock prices. Another rationale under the statement that a relationship between GPR and stock prices should exist is the empirical evidence that terrorism affects stock market returns. Memdani and Shenoy (2019) find in their article that terrorist attack impacts the stock market indices either positively or negatively in Japan, the UK, China, and Germany. Ad- ditionally, the article from Aslam and Kang (2015) shows that terrorism strikes decrease the stock returns in the Pakistani market and that the effect is short-breathed, lasting only one day. This finding justifies using the event study as a methodology with different event windows to capture the impact of the GPR on stock returns. However, Aslam and Kang (2015) reveal that the attack's magnitude correlates with negative KSE-100 stock market returns. This finding shows that the thesis's decision to focus on big GPR spikes is reasonable. Moreover, Aslam and Kang (2015) find that the threat of terrorism can affect through rumors and warnings to the stock market when one day before the attack, the stock market returns decrease –0.24% (–0.32% on the attack day), which is in line with Caldara and Iacoviello (2019). The word terrorism is a subclass for geopolitical risk. In their papers, Caldara and Iacoviello (2019; 2022) show how they construct the GPR index using the word terrorism in search queries. Hence terrorism news is part of the risks that their indexes capture. Previous research also expresses the fact that there is an association between geopolit- ical risk and stock returns in the U.S. market (e.g., Fossung et al., 2021; Yang & Yang, 2021; Smales, 2021; Salisu et al., 2021). Therefore, it is a logical prediction that the GPR 27 affects the stock prices in the U.S. However, the direction of the stock returns in the GPR index is hard to predict, although the previous studies find an existing link between the stock returns and the GPR in the U.S. The studies, for example, from Baur and Smales (2020), Hoque and Zaidi (2020), Zaremba et al. (2022), Saadli et al. (2021), and Erdoğan et al. (2022) show that the GPR may affect either positively or negatively to the stock prices. Other studies present that the effect on stock returns does not exist (Bouras et al., 2019; Balcilar et al., 2018; Elsayed & Helmi, 2021; Apergis et al., 2018; Alqahtani & Taillard, 2020). Consequently, in a hypothesis setting, the direction of the relationship between the GPR and the stock market returns is not specified. H1: The geopolitical risk affects the stock market returns in the U.S. The outcome and impact of the geopolitical risk can also vary between industries and sectors, similar to what Caldara and Iacoviello (2019; 2022) find. Ntatis et al. (2021) find that geopolitical risk negatively affects the Information Technology sector, which aligns with Fossung et al. (2021). Comparably, Khan et al. (2022) result in their article that there is a two-way connectedness between the GPR and the IT sector; in detail, they find that there are both positive and negative influences from the GPR on the Technology sector. Baur and Smales (2020) show in their paper that precious metals respond positively to geopolitical tension, which is the same as what Caldara and Iacoviello (2019) results in their research. Demiralay and Kilincarslan (2019) reveal that when the industry performs weakly, the impact of the GPR is higher on the travel and leisure stock returns (not including Asia and Pacific index). Additionally, the actual GPR spikes correlate with the decreasing travel and leisure stock returns, and the GPR threat affects only during times of decreasing travel and leisure stock returns. Collaterally, Hailemariam and Ivanovski (2021) discover that the GPR negatively impacts the demand for tourism service export. One standard deviation surprise in the GPR explains approximately 12.6% of the fluctuation in tourism net service exports. Moreover, Atems et al. (2020) result that ENSO shocks bring food 28 and agricultural stock prices higher in the U.S., while Seo et al. (2013) uncover that food safety events lead to negative stock returns of food-related firms. Results also vary with defense companies in studies by Caldara and Iacoviello (2019) and Apergis et al. (2018). These articles reveal a variation between different sectors regarding stock market re- turns during various shocks. Hence, the direction of the GPR’s influence is not straight- forward. In addition to what Caldara and Iacoviello (2019; 2022) find relating to the variation be- tween industries regarding GPR effectiveness, studies present that the crisis period can distinctly affect different sectors. For example, in his article, Thorbecke (2020) studies how COVID-19 impacts 125 industries and their stock returns in the U.S. The author finds that the COVID-19 virus negatively affects the industries such as oil, funerals, aerospace, airlines, brewers, retail apparel, tourism, real estate, and airlines. Under the positive im- pact of the virus are industries such as electronic entertainment, biotechnology, com- puter hardware and software, diversified retailers, and nondurable household goods. These findings form a rationale under the second hypothesis, that the impact of the crisis period, which in this study is the GPR, results in a variation between different sector’s stock returns of the S&P 500 index. H2: The effect of the geopolitical risk differs between Information Technology, Consumer Staples, and Energy sectors Empirical findings from the past also show that results can differ between methodolo- gies regarding the effectiveness of the GPR. For instance, Caldara and Iacoviello (2019) reached different results with their OLS regression compared to Fossung et al. (2021) with the event study methodology. In detail and as an example, Caldara and Iacoviello (2019) get the results where the GPR negatively affects the communication industry. In contrast, Fossung et al. (2021) find that the GPR affects the Communication Services sec- tor mainly positively. The period under study and the different datasets also affect the results. However, a foreconceived assumption can still be that different methodologies 29 might give variation to the results, and therefore the basis for the third hypothesis is established. Another example is that Zaremba et al. (2022) find with asset pricing models that the GPR positively predicts the stock market returns in emerging markets. However, Bouras et al. (2019) do not find any effect of the panel GARCH model in emerging countries. The same difference exists between Saadli et al. (2021) with MGARCH and Erdoğan et al. (2022) with NARDL in Turkish stock market returns, but the sample period and frequency of the data are again distinct. In China’s market, the outcomes are also prone to variation. Yang et al. (2021) find with the GARCH-MIDAS model that the GPR predicts positive stock market returns, but Chiang (2021), with the DCC model, presents that the GPR creates negative stock prices. Apergis et al. (2018), with a nonparametric causality test, and Cal- dara and Iacoviello (2019), with OLS regressions, also compose contrary conclusions when considering defense companies. The results also differ under the oil concept; for instance study from Alqahtani and Taillard (2020) with OLS, GARCH, and vector autore- gression models versus a study from Antonakakis et al. (2017) with VAR-BEKK-GARCH and multivariate GARCH models. The latter finds that the GPR significantly affects oil prices, and the former does not. H3: The results differ between methodologies Finally, the results can vary between different geopolitical events. Fossung et al. (2021) study 58 different geopolitical events in their article and show that the impact of the geopolitical risk varies between different geopolitical events and event windows (–3,3; 0,5; 0,1; –10,10). Specifically, they reveal that historical average CAARs per sector are separate and that the percentage of events with statistically significant CAARs varies be- tween sectors and different event windows. The finding also aligns with the level of risk from the GPR index by Caldara and Iacoviello (2019). For instance, their monthly GPR index gave a value of 498.65 for September 2001, 358.71 for Iraq Invasion in March 2003, and 330.78 for Ukraine War in March 2022. The value differs according to the percentage 30 of the real-time news relating to the GPR their search query captures. Moreover, the literature review shows that the results can vary even within the same country if the sample period differs. One example is the Turkish stock market returns (Saadli et al., 2021; Erdoğan et al., 2022), where the former find that the GPR affects the stock market returns adversely, and the latter finds that the GPR affects positively to the stock market returns. The commonsense support this since the more extended period under study also means more geopolitical events to the sample period and thus more variation in the results because the level of the GPR varies in the index. H4: The results differ between geopolitical events 31 3 Theoretical framework This chapter aims to provide a theoretical framework for the thesis, increase knowledge of the core finance theories relevant to this study, and link the theoretical framework to the research problem. The framework starts with handling the progress of geopolitics, then moving to different parts of the geopolitical risk index, GPR by industries, and its capability to capture a novel exogenous risk. The efficient market theory concludes the chapter. 3.1 Development of the geopolitics Geopolitics is better known and longer than geopolitical risk. They mean different things; however, geopolitics and its history form the foundation of the new concept, which in this study is geopolitical risk. Chapter 3.2 provides a review in more detail of the GPR- concept. According to Hagan (1942), geopolitical situations did not begin any further than those of the Greeks and Romans when people tried to find their place in a geo- graphical environment. In his article, there is already a connection between economic, social, and political dimensions within geographical locations. Hagan gives the honor of formulation of modern geopolitics to Ratzel (1898), which combines the state with dif- ferent surroundings and things. In more recent literature, Woodley (2015) describes terms uncertainty, globalism, and the U.S. as today’s shaky international leader together with geopolitics. Cohen (2014) says that geopolitics aims not to offer specific information about future events, threats, or crises but to concentrate on the policymakers and how they can influence them. In addition, Dodds (2004) and Agnew et al. (2003) regard geopolitics as a common interna- tional and global issue. Caldara and Iacoviello (2019; 2022) describe geopolitics as the exercise of states and organizations to monitor and race for region and state that media often refer to geopolitics when talking about global disasters and brutality. Finally, Dodds (2004, pp. 1) defines geopolitics as a ‘’study of the state, its borders, and its relations with other states.’’ 32 3.2 Geopolitical risk and index Caldara and Iacoviello (2019) define the term geopolitical risk as ‘’the risk associated with wars, terrorism, and tensions among states that affect the normal course of inter- national relations.’’ Their latest paper defines the following: ‘’the threat, realization, and escalation of adverse events associated with wars, terrorism, and any tensions among states and political actors that affect the peaceful course of international relations’’ (Cal- dara & Iacoviello, 2022). They describe the geopolitical risk also through ‘’immaterial tensions’’ as power battles, such as Cuban Missile Crisis, or pressures between countries, for instance, between the U.S. and Iran or North Korea. Their constructed index captures the risks from the acts and the new risks that could escalate from the threat of current actions. However, their index covers only the events that are usually considered geopo- litical. Hence, events such as Brexit, the global financial crisis, or climate change are not included in their index since they are not always considered geopolitical. This approach allows them to exclude the tiny weight from what those events are creating to the index based on the algorithm's results and to exclude the events separate from the words war and terrorism. The authors (Caldara & Iacoviello, 2019) created the geopolitical risk index from 11 news- papers' data with an algorithm that captures the percentage of articles associated with geopolitical risk. In detail, the authors ran an automated text-search query in ProQuest Newsstream starting from 1985 to find the number of articles using geopolitical terms conversing risks and events related to the GPR. Then they split it with the total number of articles per month. They use a dictionary-based methodology to capture the existence of words relating to geopolitical risk in the newspapers. That method allows capturing the preliminary information about exact words such as military or terror concerning the final sample. Their sample covers about 25 million news articles resulting in approxi- mately 30,000 articles per month in a recent sample and 10,000 articles per month in a historical sample. The newspapers included in the index are The Boston Globe, The Chi- cago Tribune, The Daily Telegraph, The Financial Times, The Globe and Mail, The Guard- ian, The Los Angeles Times, The New York Times, The Times, The Wall Street Journal, and 33 The Washington Post. The index includes one Canadian newspaper, four UK newspapers, and six U.S. newspapers. Therefore, the GPR index captures the risk from the U.S. and simultaneously catches the global threat but concentrates on these three countries. Figure 2 plots the recent monthly geopolitical risk index from 1985 built on ten newspa- pers, also named the benchmark GPR index (Caldara & Iacoviello, 2019). When concen- trating on the 21st century, the three most giant spikes in the index are 9/11, Iraq Invasion, and Ukraine War. These three spikes are also the events under examination in this thesis. The index spikes for the first time during the U.S.’s bombing of Libya in April 1986 after the terrorist happenings. Subsequent spikes happen during the U.S. Invasion of Panama, the Iraq Invasion of Kuwait, and the Gulf War. In 1994, the summer Bosnian War and tensions in Iraq from airstrikes executed by the Turkish air force spiked up the GPR index, following the 1999 spike from the Iraq Disarmament Crisis. Then the index spikes from the well-known 9/11 in the U.S., and other spikes during the 21st century are, for instance, the London bombings, Russia’s annexation of Crimea, and the Paris attacks. After 9/11, the mean of the index is higher than before, indicating heightened news reporting of geopolitical concerns and terror realizations. Figure 2. Recent GPR index from January 1985 through June 2022. Figure 3 plots the GPR index at a daily frequency, which is understandably noisier than the monthly equivalent index. Caldara and Iacoviello (2019) show that the green dots U.S. bombing Libya U.S. Invasion of Panama Iraq invades Kuwait Gulf War Bosnian War and airstrikes on Iraq Iraq Disarmament Crisis 9/11 Iraq Invasion London bombings Iran Nuclear tensions Libya Intervention Russia annexes Crimea ISIS Escalation Paris attacks U.S. vs. North Korea tensions U.S. vs. Iran tensions Ukraine War 0 100 200 300 400 500 600 34 depict the daily observations, whereas the solid blue line represents the monthly GPR index. The red dots display spikes in the index together with descriptions. According to the authors, the daily index captures some events which are not apparent in the monthly GPR index because the reported news relating to the specific events are short-lived. For Figure 3. Daily GPR index from 1985 through end-2020 (Caldara & Iacoviello, 2022). 35 example, the index plots escalated tensions in former Yugoslavia in August 1991 and the tried attack on the Soviet Union. The index also captures NATO’s air strikes in Kosovo in March 1999. Moreover, Caldara and Iacoviello (2019) say that the daily index can show how prolonged tensions in the daily index can lead to big spikes in the monthly GPR index, as in the case of the Gulf War. Another finding is that one climax event can produce increased plots in the daily GPR index following higher average spikes in the after-effects, as after 9/11. Additionally, in the daily GPR index, the slow-moving geopolitical conflicts in news coverage can create improved spikes in the monthly GPR index, as in the 2017– 2018 North Korea crisis (Caldara and Iacoviello, 2019). The daily GPR index correctly cap- tures the risk – revealing the valuable capability of the index in high frequencies, such as days and weeks – leading to increased efficacy and practicality in short-time effects, such as the reaction of the stock prices during geopolitical spikes. Figure 4 shows the historical monthly GPR index starting from 1900. In this index, Caldara and Iacoviello (2019) use three newspapers to construct the monthly historical index: The Chicago Tribune, The Washington Post, and The New York Times. The authors state that the correlation with the benchmark index is 0.95 from 1985 onwards. The index's spikes depict growing geopolitical pressures around the increased conflict events, similar to the benchmark index. At the beginning of the sample, the index spikes during World War I and II staying high during the time of both wars. From the 1950s to 1980s index shows high-ranking stages, indicating different wars and crises, but also a threat of nu- clear weapons and rising geopolitical pressures between countries. In the 21st century, terrorism occurs in the index with increased two-sided conflicts between nations, most recently in Ukraine. The most significant spikes in the index are the beginning of WWI and WWII, the Pearl Harbor attacks and D-Day, and the well-known 9/11. 36 Figure 4. Historical GPR index from January 1900 through June 2022. 3.3 Geopolitical threats and acts Table 1 presents examples of the words used to build the GPR indexes, divided into six categories. According to Caldara and Iacoviello (2019), the phrase selection is based on a pilot audit of newspapers most likely discussing the geopolitical risk. Chapter 4.3 dis- cuss the pilot audit in more detail. The authors state that another selection criterion for the words is the most general uni-grams and bi-grams from the geopolitical textbooks. For example, a book from Dalby et al. (2003) – named as The Geopolitics Reader – totals 91,210 bigrams. It covers docker of 39 essays from geopolitics, and the most frequent of those bigrams are “world war,” “gulf war,” “unit(ed) states,” “nation secur(ity),” “war II,” “world order,” “cold war,” “nation(al) state,” “nuclear weapon,” and “foreign polic(y).” Similarly, with uni-grams, Flint’s (2016) textbook Introduction to Geopolitics includes the most usual word roots, such as “terror,” “geopolit,” “polit,” “nation,” “war,” “global,’’ and “countri.” In their recent paper, Caldara and Iacoviello (2022) report that the goal is to make an index that can catch different features of the GPR and share geographically and conceptually, which serves their part as a reason for selecting the words. They have also increased the number of words in search phrases and excluded words in the search query in their latest (2022) paper. Boxer Rebellion Russia vs. Japan War WWI begins WWI escalation Occupation of the Ruhr Shagnai Incident Italy vs. Ethiopia War Germany invades Czechia WWII begins Pearl Harbor D-Day Korean War Suez Crisis Berlin Problem Cuban Missile Crisis Six Day War Yom Kippur War Afghan Invasion Falklands War Iraq invades Kuwait Gulf War Bosnian War 9/11 Iraq War Paris terror attacks Ukraine War 0 100 200 300 400 500 600 37 Table 1. Examples of phrases sought to build the GPR indexes (Caldara & Iacoviello, 2019). Search Category Examples of Search Terms 1. Geopolitical Threats geopolitical AND (risk* OR concern* OR tension* OR uncertaint*) AND “United States’’ AND (coup OR guerrilla OR warfare) AND (“Latin America’’ OR “Central America’’ OR “South America’’ OR Europe OR Africa OR “Middle East’’ OR “Far East’’ OR Asia) 2. Nuclear Threats (“nuclear war’’ OR “atomic war’’ OR “nuclear conflict’’) AND (fear* OR threat* OR risk* OR peril* OR menace*) 3. War Threats “war risk*’’ OR “war fear*’’ OR “military threat*’’ 4. Terrorist Threats “terrorist threat*’’ OR “terrorism menace*’’ 5. War Acts ((beginning OR outbreak OR start OR escalation) “of the war’’) 6. Terrorist Acts “terrorist act’’ OR “terrorist acts’’ NOTE: This table shows a subclass of phrases explored in building the GPR indexes, arranged by categories. The precise search query is in the appendix. The asterisk (*) symbol indicates a wildcard character. In Table 1, Caldara and Iacoviello (2019) define different geopolitical threats and con- cerns covering categories from 1 to 4, and the last two categories include different geo- political actions and happenings. The words' division allows the construction of two sep- arate subindexes under the GPR index, GPR acts (GPRA/GPA) and GPR threats (GPRT/GPT). The authors state that category number one divides the areas where geo- political risks are happening in the United States and extensive territories. The articles discussing GPR contain either immediate United States participation (e.g., the 2003 In- vasion of Iraq) or district conflicts between the group of countries with the political in- volvement of the United States. Additionally, category number one defines the search terms from the articles that distinctly refer to GPR and military-related words regarding conflicts of the U.S. and other parts of America, Europe, Africa, Asia, and the Middle as well as the Far East. Caldara and Iacoviello (2019) show in category number two of Table 1 the nuclear tense- ness-related words searched from the articles. Category number three contains terms which are portraying war tensions. Similarly, category number four represents words re- lating to terrorism tensions. Category five turns to search the words from the articles containing war actions – the real happenings – contrasted to only risks. Ultimately, 38 category number six involves phrases describing terrorist acts. According to the authors, the words represent the negative GPR affairs instead of positive ones, increasing the ac- curacy of the index. For example, the search query terms include the war's beginning rather than the war's closing. Moreover, Caldara and Iacoviello (2022) show an updated search query for the construc- tion of the GPR index, which contains eight search categories in their recent paper. The threat part of the index is now divided into five categories and acts as part of the index into three categories. The difference with the 2019 search query is that the “Geopolitical Threats’’ search category consists of “Peace threats’’ and “Military buildup’’ search cat- egories, and the “War Acts’’ search category includes “Beginning of war’’ and “Escalation of war’’ search categories. That allows allocating the parts of the index more accurately. In their article, Caldara and Iacoviello (2019) split the GPR index into two subindexes, GPR threats (GPRT) and GPR acts (GPRA). They show the GPR index as a measure of the threat of possible future risks and tensions in sight of happenings or actual materialized risks related to geopolitical events. According to the authors, the articles containing the search terms in categories 1 to 4 in Table 1 create the GPRT index. Similarly, the search terms in the categories from 5 to 6 in Table 1 produce the GPRA index. Thus, the two subindexes split the search terms; in the GPRA, the actual realized events define the index, whereas in the GPRT, the possible future risks and tensions are more prevalent. One example of how the two subindexes can provide separate information is that the actualized GPR can serve as a launch of elevated GPR. For instance, during Ukraine War in 2022 spring, the threat of war may increase in other parts of Europe, or after the terrorist assault, the threat of future terrorist assaults may be higher. The authors state that the division of the GPR index performs an essential part of the analysis as the accu- racy of the timing of the GPRA events can be distinct from the benchmark GPR index. Figure 5 draws the GPRT and the GPRA indexes. Caldara and Iacoviello (2019) have found a correlation between these indexes is 0.51. Figure 5 shows that almost all spikes in the 39 GPRA index overlap with the GPRT index's spikes. However, there is still a massive vol- ume of individual discrepancies between the indexes. Authors state that, in some cases, the GPRT index rises sometime before the actual geopolitical event, such as before Gulf War, Arab Spring, Iraq Invasion, and Ukraine War. This feature may catch information and associated news from future geopolitical risks and happenings. At some point, the GPRT index rises without existing incidents, such as across 2017–2018 U.S. vs. North Korea tensions when the two subindexes move in different directions. In some cases, the GPRT spikes, but the GPRA does not, such as during the Kuwait Invasion in August 1990. Cal- dara and Iacoviello (2019) profess this feature is due to the media coverage and its re- porting, the associated news about the possibilities versus the actual realized geopoliti- cal events. Figure 5. GPRT and GPRA subindexes of the Geopolitical Risk Index. 3.4 Geopolitical risk by industries In their article, Caldara and Iacoviello (2019) study the role of industry exposure to geo- political risk and find that positively exposed industries to aggregate GPR tend to have smaller investments than firms with negative exposure. In this thesis, the primary inter- est is whether the effect of the geopolitical risk differs between the Information Tech- nology, Consumer Staples, and Energy sectors of the S&P 500 index and whether the exposure to the GPR varies between industries in the United States. Caldara and Kuwait Invasion Iraq Disarmament Crisis Russia annexes Crimea ISIS Escalation U.S. vs. North Korea tensions U.S. vs. Iran tensions Ukraine War U.S. bombing Libya U.S. Invasion of Panama Gulf War 9/11 Iraq Invasion London bombings Iran Nuclear tensions Arab Spring Syrian War Paris attacks 0 100 200 300 400 500 600 700 800 900 GPR Threats GPR Acts 40 Iacoviello (2019) find that industries such as petroleum and natural gas, beer and liquor, and electronic equipment are under negative exposure to the GPR, meaning that the stock returns of these industries outperform the market during the GPR pressures. Un- der positive exposure to the GPR are industries such as tobacco products, coal, and com- puter software, meaning that their stock returns underperform the market stock returns for the period of higher geopolitical conflicts. Daily data is essential when measuring the industry's exposure to the GPR. In their article, Caldara and Iacoviello (2019) study that after 9/11, the first trading day on September 17th in the United States revealed –13% loss in the transportation industry, opposite to the precious metals industry, which gained +7.4%. Thus, daily data is justified since stock market returns fluctuate rapidly due to responding to the news. The authors state that the daily geopolitical risk data also aims to find a more accurate connection between variables since some GPR events last only some days, as with the stock return fluctua- tions. Chapter 4.2 shows the equation for the industry exposure measure by Caldara and Iacoviello (2019). The meaning of the method is that throughout elevated geopolitical pressures – the industries which expose positively to the GPR are losing concerning stock returns when compared to the cumulative returns of the market – therefore denoting the connection between rising GPR and negative stock returns (of the positively exposed industries). Vice versa, the industries which are negatively exposed to the GPR are gain- ing in terms of stock market returns relative to the market, consequently uncovering the correlation between the GPR spikes and positive stock price returns (of the negatively exposed industries). According to Caldara and Iacoviello (2019), raised exposure to GPR follows industries where commerce is higher, business is more cyclically-unprotected, and leveraging num- bers are higher. The authors read 100 transcripts of earning calls between firms and in- vestors from the conversations of the GPR and its effect on a firm’s business and find, for instance, that the transportation industry may be more exposed to the GPR because of terrorism, natural disasters, and high fuel prices. The same conclusion happened on 41 the first trading day on September 17th after 9/11 (Caldara & Iacoviello, 2019), when the transportation industry lost –13 %. Another finding from the literature is that the sectors and industries more dependent on the international markets and demand can be more exposed to the GPR because the GPR can lower oversea demand for the firms operating internationally (Boutchkova et al., 2012). For example, agricultural manufacturing com- panies can suffer from geopolitical situations because of Russia’s annexation of Crimea and the Ukraine War since Ukraine is one of the biggest corn and wheat producers in the world (World Population Review, 2022; Index Mundi, 2022), and therefore a great user of the agricultural products. Indeed, Caldara and Iacoviello (2019) show that weakening trade, lack of credit availa- bleness, and injunction of financial penalties to Russia impact apparatus sales of inter- national companies. Finally, according to Caldara and Iacoviello (2019), one example of high leverage is that geopolitical risk can increase the effectiveness of more forceful dis- aster shocks due to financial tightness. Thus, the effect of the reinforced credit can strengthen the impact of the disaster with high GPR companies (Gourio, 2013). Caldara and Iacoviello (2022) say that other factors influencing the exposure include ge- ographical locality, the logistics network of the industries, diplomatic relationships, and risk-controlling procedures. Examples provided by the authors contain the general loca- tion of the significant oil sources, which could impact petroleum (and defense) compa- nies during suspensions in the Middle East, and terrorism assaults could have more im- pact on industries such as entertainment, airlines, and transportation. More fresh evi- dence comes from the geopolitical situation in Ukraine, which has raised the price of food and energy together with increased inflation and the risk of stagflation since Ukraine is a giant player in the agriculture industry, and Russia is one of the biggest oil exporters (World Bank, 2022; World Population Review, 2022; Index Mundi, 2022; Worldometer, 2022). 42 3.5 Geopolitical risk vs. other risk indicators Caldara and Iacoviello (2019) argue in their paper that the GPR is a novel and exogenous risk for the U.S. economic changes, which differs from the Cboe Volatility Index® (VIX) and Economic Policy Uncertainty (EPU) indicators. They find, for instance, that the GPR is not Granger-caused by the economic advancements in financial, economic uncertainty, or macroeconomic factors. Economic uncertainty involves the log of the VIX and the EPU indexes, and the macroeconomic variable consists of the log differences of private em- ployment, U.S. industrial production, and the log of the WTI value deflated by the U.S. consumer price index. Financial factor includes the gains of the two-year Treasury Yield and S&P 500 index. Hence, the authors state that the GPR can depict and segregate risks from geopolitical tensions, for example, terrorism and escalation of wars, which are not measured and fully understood by the other risk indicators or the economic develop- ments in the U.S. economy at the frequency of the commercial cycle. Even so, the GPR index by Caldara and Iacoviello (2019) still correlates at some amount with the EPU index by Baker et al. (2016). For instance, a two-standard deviation shock in the GPR index produces a brief decline in the median impulse response of the EPU index from approximately 15% to 3% after one year. It vanishes close to zero after eight quarters (cf. –20 basis point drop in the yield of the U.S. two-year treasuries from the same shock). By contrast, the identical effect with the S&P 500 index lasts more than 12 quarters, bottoming at –2.5% after two quarters, together with statistical and economic magnitude. Authors continue that, on the other hand, the division of the GPR index to the GPRT and the GPRA indexes shows that the two-standard-deviation growth in the GPRA index creates a negative response to the EPU index from quarters one to eight. The same effect from the GPRT index indicates the positive reaction of the EPU index from quarters zero to eight. This finding means that the increased threats in the GPRT index can also raise the EPU index when actual materialized events can move the EPU index the other way, revealing the exogenous property of the GPR index. 43 Additionally, the authors show that a remarkable quantity of separate and supernumer- ary fluctuation follows from what the GPR index exhibit and produces, leading to a fore- cast of diminished economic operations. Moreover, the GPR index is uncorrelated with well-known episodes such as the global financial crisis or recessions. This finding reveals an uncorrelation with the VIX and EPU indexes, which capture the global financial crisis. Thus, the GPR index highlights episodes not captured in the VIX and EPU indexes, which are mostly self-standing of the economy in terms of the U.S. enterprise cycle frequency. Figure 6 plots the relationship between the benchmark GPR index and the EPU index by Baker et al. (2016). Figure 7 depicts the association between the benchmark GPR and VIX indexes, where the latter denotes financial volatility. In both figures, the indexes spiked during Gulf War in 1991 and over the 9/11 terrorist strikes in the U.S. The corre- lation in Figure 6 is 12.68%, and the correlation between GPR and VIX indexes is 7.17%. Taken together, Caldara and Iacoviello (2019) claim in their paper that it is reasonable that the correlation between indexes rushes from the GPR to the other indexes in exam- ples of the Gulf War and 9/11. Correspondingly, Iraq Invasion by the U.S. in 2003 spiked the EPU index even though the impact on the VIX index is more moderate or practically nonexistent. Caldara and Iacoviello (2019) state that the indexes' figures show a considerable extent of independent alteration. For instance, the Asian financial crisis at the end of the 1990s or the euro crisis beginning of the 2010s does not follow the GPR index, opposite what the VIX and EPU indexes capture. Similarly, the bankruptcy of Lehman Brothers, the dot- com bubble in the late 1990s, political elections, or Black Monday does not induce spikes in the GPR index. Also, COVID-19 is independent of the GPR index. The figures support the authors finding that the GPR is not Granger-caused by financial, economic uncer- tainty, or macroeconomic factors. Vice versa, the spikes in the GPR index, such as Syrian tensions, the Paris attacks, the rise of ISIS, or the geopolitical situation in Ukraine in 2014, do not cause spikes in the VIX or the EPU indexes. These findings are similar which Baur 44 Note: This figure compares the benchmark GPR index from Caldara and Iacoviello (2019) and the economic policy uncertainty index from Baker et al. (2016). The EPU index in this figure is the three-component index. The GPR index is normalized to 100 from January 1985-June 2022. Figure 6. Comparison between benchmark GPR index and EPU index. Note: This figure plots the comparison between the benchmark GPR index from Caldara and Iacoviello (2019) and option-implied volatility from Cboe Volatility Index® according to S&P 500 firms. The data of 1986–1990 within the VIX index is matched to the old VXO index. The GPR index is normalized to 100 from January 1985-June 2022. Figure 7. Comparison between benchmark GPR index and VIX index. U.S. bombs Libya Gulf War 9/11 Iraq Invasion Madrid bombings London bombings Transatlantic aircraft plot Russia annexes Crimea Ukraine and ISIS Paris attacks Ukraine War Black Monday Kuwait Invasion Clinton Election Russian Crisis/LTCM Bush Election Stimulus Debate Lehman Failure and TARP Euro Crisis Debt Ceiling Debate U.S. Fiscal Cliff Govt. Shutdown Brexit U.S. Election COVID-19 0 50 100 150 200 250 300 350 400 0 100 200 300 400 500 600 EPUGPR GPR EPU U.S. bombs Libya 9/11 Iraq Invasion ISIS Escalation Paris attacks Syrian Tensions Ukraine War Black Monday Kuwait Invasion Asian Financial Crisis LTCM 2002 correction Lehman Euro Crisis COVID-19 0 10 20 30 40 50 60 70 80 90 0 100 200 300 400 500 600 VIXGPR GPR VIX 45 and Smales (2020) results in their article that the GPR is separate from other risk measures related to financial volatility (VIX), economic policy uncertainty (EPU), eco- nomic uncertainty (EconUnc) by Bekaert et al. (2022), and macroeconomic uncertainty (MacroUnc) by Jurado et al. (2015). The GPR index also uncorrelates with the USD index and U.S. 10-year treasury notes. Additionally, Das et al. (2019) show that the effect of EPU, GPR, and financial stress (FS) by Püttmann (2018) is heterogenous in emerging mar- kets regarding causation and intensity. Lastly, Sharif et al. (2020) show that the GPR moves in a different direction during COVID-19 than the EPU index, which is a similar finding in Figure 6 above. The authors (Caldara & Iacoviello, 2019) conclude that the GPR index has two independ- ent features compared to the other risk indexes. First, it can portray the cases which might upsurge accumulated financial instability and political insecurity. Second, it can catch events that are probably exogenous from the primary business and financial fre- quency. In their recent paper, Caldara and Iacoviello (2022) state that the GPR positively correlates with U.S. military spending (Ramey, 2011) and the human cost of conflict fac- tors such as war deaths. But like in the case of risk indicators, the GPR still captures much evidence not revealed by other indicators handled at this point. Authors research that war deaths are associated with the GPRA by 83%; in contrast, the correlation with the GPRT is 46%. The GPR index and war death indicator are close to each other during World Wars. However, only GPR remains relatively high to its mean after the World Wars, re- vealing the increased awakening to the geopolitical issues by news coverage and audi- ence after the brutal World Wars. 3.6 Efficient market theory in the light of event studies A theoretical framework continues with an efficient market theory and presents the core finance theories relevant to this study around the event study methodology. The start of the event studies can be stretched to 1969 in a paper by Fama et al. (1969). In their paper, the authors research the adaptation of standard stock prices to the new information, measured as stock splits and its inherent data. The authors discuss the extensive 46 research on the continuous price changes in common stocks, revealing that the changes are maverick and logical. However, according to the authors, the stock price changes are maverick and logical only in “efficient’’ markets – a market that quickly conforms to the latest data. Fama et al. (1969) argue that before their paper, there was a lack of studies concentrating on detailed data and its pace of adapting to the new information. Instead, studies have used detected maverick and logical, continuous stock price changes and deduced market efficiency from that. Therefore, the authors are turning the research focus to the data testing of stock price modification pace regarding market efficiency. That is the perspective also adopted in this thesis. When Fama et al. (1969) researched the extraordinary (abnormal) performance of stock price returns around stock splits, the research conducted in this thesis studies the performance of the stock price returns of three different sectors from the S&P 500 index in the U.S. around three unique geopo- litical events in the 21st century. Fama et al. (1969) manifest that the average stock price returns increase three or four months before the stock split, but after the stock split, the increase of average stock prices stops, and cumulative abnormal returns stay stable and end their growth. The authors conclude that the market uses the stock split publication to forecast future predicted stock dividends and that the stock prices contain all the in- formation available regarding the stock split without delay right after the stock split’s declaration moment or instantly after the split month. In his article, Fama (1970) continues the work from the 1969 article and presents the efficient market hypothesis. According to Fama (1970), the efficient market hypothesis means that all openly available new stacked information is reflected in the price of a given security at a particular time. Therefore, for example, the efficient market hypoth- esis is applicable within a stock market. In addition, the author claims that the level of the stock price measures the corporation's value. Thus the capitalization of the company contains all publicly available information in its stock prices defined by the market. Fama (1970) continues that the optimal market would be where investors can pick the firm's stock, which always contains all publicly available information. Firms can make choices 47 relating to investments and output because the market communicates through securi- ties that already hold all the information accessible. Therefore the communication pro- vided by the market is precise and correct. He concludes that among three different sub- groups regarding accessibility and timing of the information, the market model performs the best (based on weak form tests, i.e., on historical prices and returns). It can provide results when stock price changes or returns go through one day or more (e.g., during geopolitical events under uncertainty). However, the semi-strong tests also endorse the EMH, such as stock splits, where the security price contains all publicly offered infor- mation by the time of the stock split (Fama et al., 1969; Fama, 1970). Bowman (1983) goes on with the event studies and argues in his article that outside decisions affect the event study method, not only the decisions made within the corpo- ration. For instance, the stock split or the announcement of the annual accounting earn- ings would be examples made within the firm. Accounting standards set by the financial accounting standard board or other usual actions, such as the oil embargo, can be exam- ples of outside decisions or declarations affecting the event study modification. The au- thor focuses intensely on event studies by segregating them to separate subclasses by type and sets up the parameters for shaping event studies with five steps. The five steps in numerical order include identifying the event of interest, modeling the security price reaction, estimating the excess returns (including the market model), or- ganizing and grouping the excess returns (e.g., CARs), and analyzing the results. In addi- tion, he provides a comprehensive discourse about optional procedures for event stud- ies to manage and apprehend the topic. Finally, the author notes that the information content of the event studies published by Ball and Brown (1968) focuses on the time before and concurrent with the specific event regarding the stock price. In contrast, Fama et al. (1969) concentrate mainly on time after the event in EMH (but also on time before and concurrent with the stock split, as discussed above). 48 Brown and Warner (1985) prolong the research on the event studies by taking along the daily frequency data relating to the stock price returns under investigation and examin- ing features of that data and its influence on the event study method. The result of the article is that the event study methods built upon the OLS market model and t-tests are determined and justified well in different circumstances. The authors find a high corre- lation between the empirical and theoretical strengths of the event study practices. The authors also disclose that under excess daily returns, the dependency in the cross-sec- tion of the gains, the autocorrelation, and the variance changes can impact the outcome of the results. In addition, Brown and Warner (1985) conclude that the impact of the event study methodology is much more substantial when using daily instead of monthly data and that the nonnormality of the daily stock returns does not affect the results. Additionally, the variance changes in daily returns are tiny and not often repeated, the dependence's effect in the excess returns' cross-section is rare, and the autocorrelation has limited influence with methods. Klein and Rosenfeld (1987) investigate the event study methods under different market conditions, in detail, under the bull