Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Does Bitcoin Hedge Categorical Economic Uncertainty? A Quantile Analysis:

Does Bitcoin Hedge Categorical Economic Uncertainty? A Quantile Analysis: This paper examines the hedge and safe-haven abilities of Bitcoin against U.S. aggregate and categorical economic policy uncertainty (EPU) via the application of quantile regression model augmented with a dummy and some control variables. Using monthly data from September 2011 to December 2019, empirical results indicate that Bitcoin does not act as a strong hedge against the aggregate U.S. EPU. However, it acts as a strong safe-haven for this aggregate measure of uncertainty when the Bitcoin market is bearish. Looking deeper into the disaggregated level of the U.S. EPU data, the analyses involving categorical EPU data indicate the ability of Bitcoin to act as a strong hedge and safe-haven against specific uncertainties related to fiscal policy, taxes, national security, and trade policy. Keywords Bitcoin, EPU, categorical economic uncertainty, hedge, safe-haven, quantile the Cyprus banking crisis (2012–2013). Since its lunch, the Introduction value of this digital currency has soared from $0.09 in July Research interest in the effects of uncertainty and the econ- 2010 to around $19,000 in December 2017. Bitcoin market omy in general on financial markets goes back to the theo- capitalization surged from less than $1.6 Billion to about retical models of Bernanke (1983) and Bloom (2009). $316 billion during the same period, allowing Bitcoin to However, since the 2008 global financial crisis, policy fac- occupy more than 80% of the total market value of all cryp- tors have gained ground in shaping the economic environ- tocurrencies. Accordingly, Bitcoin gained a large interest in ment and financial markets. Lately, Baker et al. (2016) have the financial literature, given its beneficial proprieties. In proposed a news-based index of U.S. economic policy uncer- fact, Bitcoin is almost isolated from the global financial sys- tainty (EPU), which has been widely used in academic litera- tem, making it a valuable addition to portfolios containing ture. Accordingly, numerous studies try to understand the conventional assets (Corbet et al., 2018; Guesmi et al., 2019; impact of the U.S. EPU on U.S. stock market returns and Symitsi and Chalvatzis, 2019). In view of this, some aca- report evidence of a negative impact (e.g., Arouri et al., demic literature examines the hedging and safe-haven prop- 2016). Other studies focus on the impact of the U.S. EPU on erties of Bitcoin for equities (Bouri et al., 2017c; Klein et al., safe-haven assets such as gold, mostly showing that EPU has 2018; Shahzad et al., 2019, 2020), commodities (Bouri et al., the ability to predict gold prices (Raza et al., 2018) and that 2017b; Klein et al., 2018), and currencies (Urquhart & gold prices and economic uncertainty are positively related (Bilgin et al., 2018). Such findings concur with previous evi- dence arguing that, during periods of economic and political Northern Border University, Arar, Saudi Arabia Gabès University, Tunisia uncertainties, investors switch their investments from risky Lebanese American University, Beirut, Lebanon assets to less risky or safe-haven assets such as gold (Baur & University of Economics Ho Chi Minh City, Vietnam Lucey, 2010). Prince Sattam bin Abdulaziz University, Al-Kharj, Saudi Arabia Remarkably, the most popular and largest cryptocurrency, 6 Manouba University, Tunisia Bitcoin, has emerged as a substitute instrument for the inef- Corresponding Author: fectiveness of traditional economic and financial systems, Khaled Mokni, College of Business Administration, Northern Border especially during stress periods (Bouri et al., 2017a, 2017c), University, Arar 91431, Saudi Arabia. such as the European sovereign debt crisis (2010–2013) and Email: kmokni@gmail.com Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 SAGE Open Zhang, 2019). Other studies consider the effect of EPU on in general, with some exceptions.). Second, we contribute to the Bitcoin market and make inferences regarding hedging the literature dealing with the debatable role of Bitcoin as a and acting as a safe-haven (Cheng & Yen, 2019; Demir et al., hedge and safe-haven against uncertainty (Bouri et al., 2018; Panagiotidis et al., 2019; Wang et al., 2019b; Wu et al., 2017a; Cheng & Yen, 2019; Panagiotidis et al., 2019) during 2019). The findings are, however, mixed and inconclusive. stress periods. Unlike previous studies (Cheng & Yen, 2019; Some studies show that the impact of the U.S. EPU is insig- Demir et al., 2018; Panagiotidis et al., 2019; Wang et al., nificant (Cheng & Yen, 2019; Wang et al., 2019b), while oth- 2019b; Wu et al., 2019), we find convincing evidence for ers find a positive impact (e.g., Panagiotidis et al., 2019) that Bitcoin being a hedge and safe-haven for specific categorical varies across the lower, middle, and upper quantiles (Demir U.S. EPU indices such as fiscal policy, taxes, national secu- et al., 2018; Wu et al., 2019). Notably, related studies limit rity, and trade policy uncertainty. These findings are useful to their focus to the aggregate EPU index, which would mask a variety of economic players such as policymakers, finan- any potential heterogeneity in the impact of the categorical cial advisors, and investors. U.S. EPU on the Bitcoin market. Therefore, extending the The rest of the paper is organized into four sections. above literature to the categorical level of the U.S. EPU data Section 2 reviews the related literature dealing with Bitcoin. (Baker et al., 2016) would help uncover potential heteroge- Section 3 describes the data and methodology. Section 4 neity in the role of Bitcoin as a hedge and safe-haven against reports the empirical results and offers a discussion in light the various 11 components of U.S. economic uncertainty of previous studies. Section 5 concludes and opens paths for (monetary policy, fiscal policy, taxes, government spending, further research. health care, national security, entitlement programs, regula- tion, financial regulation, trade policy, and sovereign debt Previous Studies and currency crises). This is important for at least two rea- sons. First, specific economic uncertainties and develop- The release of Bitcoin as a genuine and fast payment mecha- ments, such as those related to monetary, fiscal, political, and nism has marked the last decade. Some studies examine the trade policies in the United States, have been in the financial safety and legal aspects of the Bitcoin market (Anceaume news. For example, the trade war between the United States et al., 2017; Teomete Yalabık & Yalabık, 2019). As the market and China and the impeachment action in the United States for Bitcoin grows, Bitcoin becomes a leading digital asset, have moved financial markets, and some press articles have attracting the attention of many investors. Accordingly, numer- tried to establish a link to Bitcoin. Second, a more nuanced ous studies focus on the economic and financial implications analysis that accounts for the heterogeneity in the composi- of Bitcoin by considering price discovery (Baur & Dimpfl, tion of the U.S. EPU helps investors make inferences regard- 2019), volatility (Bouri et al., 2019), and speculative nature ing portfolio management. It also helps financial advisors (Baur et al., 2018). Notably, Bitcoin is segmented from the make risk management and hedging decisions regarding the global financial system, offering valuable diversification ben- ability of Bitcoin to hedge various components of economic efits (Bouri et al., 2017c, 2017b; Corbet et al., 2018; Guesmi uncertainty, especially during market downturns. et al., 2019; Klein et al., 2018; Selmi et al., 2018; Shahzad In light of the above, we examine in this study the impact et al., 2019, 2020; Symitsi & Chalvatzis, 2019; Urquhart & of categorical U.S. EPU indices on Bitcoin returns. Using Zhang, 2019). Bouri et al. (2017a) report that Bitcoin repre- monthly data from September 2011 to December 2019, we sents a hedging tool against stock market uncertainty. Other apply a quantile-based regression augmented with dummy studies focus on the impact of EPU on the Bitcoin market. variables and some other control variables to account for the Demir et al. (2018) use a quantile-based approach to examine heavy-tails of asset returns and various dependent variable the prediction power of EPU on Bitcoin prices. They suggest distribution levels. Such an examination allows us to differ- that Bitcoin can be used as a hedging tool against EPU. Wu entiate between the impact of categorical U.S. EPU indices et al. (2019) employ a GARCH model and quantile regression at various quantiles of Bitcoin return distribution (i.e., bear- to compare the hedge and safe-haven roles of gold and Bitcoin ish, normal, and bullish periods) (Wu et al., 2019). against EPU. They show the inefficacity of these two assets in Our contributions are on various fronts. First, we contrib- acting as a hedge or safe haven against EPU. Fang et al. (2019) ute to the strand of literature dealing with EPU and Bitcoin apply multivariate GARCH models to investigate the impact returns (Cheng & Yen, 2019; Demir et al., 2018; Panagiotidis of aggregate EPU on Bitcoin and other assets. They report that et al., 2019; Wang et al., 2019b; Wu et al., 2019) by moving Bitcoin can be considered a hedge under specific economic the debate to the categorical level of the U.S. EPU. In fact, uncertainty conditions. Also, using a quantile-based approach our paper is the first to examine the impact of various com- and causality tests, Wang et al. (2019b) find that Bitcoin has ponents of the U.S. EPU on Bitcoin returns during bearish the propriety of a safe-haven and a diversifier for the extreme and bullish market states, which extends previous studies shocks of economic uncertainty. Cheng and Yen (2019) inves- focusing only on trade policy uncertainty (e.g., Gozgor et al. tigate the relationship between cryptocurrency volatility and (2019) apply wavelets methods and show that the relation- EPU. They indicate that Bitcoin and Litecoin are useful hedg- ship between trade uncertainty and Bitcoin prices is positive ing tools against EPU. Mokni et al. 3 Table 1. Description of the Different EPU Categorical Uncertainty Indices. Categorical index Description Monetary policy Is a sub-index based on news data incorporating the mention of terms related to monetary policy. as examples “federal reserve”, “money supply”, “monetary policy”, “overnight lending rate”, . . . Fiscal Policy Is a sub-index to measure uncertainty related to fiscal policy. It refers to the mention of terms like “government spending”, “budget battle”, “military spending”, “fiscal stimulus”, . . . Government spending Is a sub-index of uncertainty regarding government spending. It is based on news data mentioning terms like “government spending”, “defense spending”, “military spending”, “entitlement spending”, . . . Taxes Is a sub-index calculated based on the news data related to taxes and mentioning the terms like “taxes”, “tax”, “taxation”, . . . Health care Is a sub-index of uncertainty related to health, Medicaid, Medicare, health insurance, malpractice tort reform, malpractice reform, prescription drugs, drug policy, food and drug administration, FDA, medical malpractice, prescription drug act, medical insurance reform, medical liability, part d, affordable care act, Obamacare National security Is a sub-index based on the news data related to national security and mentioning terms like: “national security”, “military conflict”, “war”, “terrorism”, . . . Entitlement programs Is a sub-index of search results from news related to entitlement programs incorporating terms like “program”, “government entitlements”, “social security”, . . . Regulation Is a sub-index to measure uncertainty regarding regulation. It is based on the research from news citing terms in relation with regulation like: “banking supervision”, “bank supervision”, “financial reform”, . . . Financial Regulation Is a news based index based on the number of citation of terms regarding financial regulation loke “banking supervision”, bank supervision”, “house financial services committee”, . . . Trade policy Is a sub-index of uncertainty related to trade activity. It is based on news incorporating terms related to trade like “import duty”, “government subsidies”, “world trade organization”, . . . Sovereign debt Is a news based sub-index related to sovereign debt and currency crisis. It is based on the mention of uncertainty terms like “currency crisis”, “currency devaluation”, “currency manipulation”, “exchange rate” Other studies focus on the hedging and safe haven propri- Regarding the diversification potential, the results report the eties of cryptocurrencies. For example, Wang et al. (2019a) dominance of Bitcoin over both gold and commodities. investigate the spillover effects between Bitcoin and major In this paper, we contribute to the academic literature by financial assets based on the VAR-GARCH-BEKK frame- investigating the potential hedging and safe-haven effects of work. Their results show that Bitcoin can serve as a hedge Bitcoin with respect to various categories of U.S. EPU. To against some assets, including stocks and bonds. Besides, the best of our knowledge, no previous research has focused Bitcoin can act as a safe haven against extreme changes in on various categories of EPU to address whether Bitcoin is a monetary markets. Shahzad et al. (2019) compare the hedg- hedge or safe-haven against categorical U.S. EPU indices at ing and the safe haven proprieties of Bitcoin compared to the various states of the Bitcoin market (i.e., bullish and bearish commodities and gold against stock market investments dur- market periods). Methodologically, we use quantile regres- ing bear and bull market conditions. They report that Bitcoin, sions augmented with dummy variables to address various gold, and the commodity index can serve as weak safe-haven situations of markets and various lower and upper quantiles assets in some cases. of Bitcoin return distributions. More recently, Wang et al. (2020) investigate the propri- eties of stablecoins for traditional cryptocurrencies. They Data and Methodology show the following. First, USD-pegged stablecoins have bet- ter risk-dispersion abilities for traditional cryptocurrencies The Dataset than gold-pegged ones. Second, Tether plays the role of a strong hedge for traditional cryptocurrencies. Third, gold is a In this study, we use a monthly dataset covering the closing better hedge than stable coins, and the USD is a better hedge prices of Bitcoin against the U.S. dollar and the U.S. EPU. To than two of the USD-pegged stablecoins. However, gold as a eliminate aggregation bias, we use 11 categorical uncertainty safe haven is not as good as the stablecoins it backs. Third, indices covering monetary policy, fiscal policy, taxes, gov- the gold-pegged stablecoin is less efficient than the USD- ernment spending, health care, national security, entitlement pegged stablecoins in terms of risk reduction. Bouri et al. programs, regulation, financial regulation, trade policy, and (2020b) compare the safe-haven properties of Bitcoin, gold, sovereign debt. Following Baker et al. (2016), these indices and the commodity index against the world, developed, are solely based on news data and constructed using the emerging, the United States, and Chinese stock market indi- Access World News (AWN) database of about 2000 U.S. ces based on the wavelet coherency approach analysis. They newspapers. Each index is constructed with respect to the indicate a weak dependence between Bitcoin/gold/commod- mention of categorical policy terms related to uncertainty ities and the considered stock markets at various time scales. terms. Table 1 presents the different terms used to construct 4 SAGE Open (a) (b) 80 150 100 EPU Fiscal policy Monetary policy 40 100 50 0 50 0 -40 0 -50 -80 -50 -100 -120 -100 -150 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Govermentspending Health care Taxes 100 50 50 0 -50 -100 -50 -100 -200 -100 -150 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Entitlementprograms National security Regulation -50 -50 -100 -40 -100 -150 -150 -80 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 200 300 FinancialRegulation Tradepolicy Sovereigndebt 200 200 0 0 -100 -100 -100 -200 -200 -200 -300 -300 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Figure 1. (a) Bitcoin returns and (b) aggregate and categorical economic policy uncertainty changes. each categorical EPU index, For more details, see Baker and represent the price of Bitcoin against the U.S. dollar et al. (2016) and the website: https://www.policyuncertainty. from Bitstamp, the leading exchange. Uncertainty indices com.) are sourced from the website https://www.policyuncer- Besides the availability of categorical U.S. EPU data, tainty.com/. Notably, the starting date depends on the we focus on U.S. economic uncertainty for at least two availability of Bitcoin prices. We employ the logarithmic other reasons. First, the United States is the country run- returns of Bitcoin and aggregate EPU (Demir et al., 2018), ning most Bitcoin nodes in the world (21.49% in January and categorical economic uncertainty indices, that is., 2020). Second, the Bitcoin price is mainly traded against Rx =× 100 log(/ x ), where is the level of Bitcoin it ,, it it , −1 it , the U.S. dollar. The dataset covers the period from price or the uncertainty index of category . Figure 1 pres- September 2011 to December 2019, yielding 100 monthly ents the plot of returns for Bitcoin, the aggregate EPU, and observations Data for Bitcoin price are from DataStream the categorical uncertainty indices. Mokni et al. 5 Table 2. Descriptive Statistics and Stationarity Tests. Mean Minimum Maximum STD Skewness Kurtosis J-B ADF PP *** *** *** Bitcoin 10.203 −49.215 174.000 34.917 1.629 8.251 0.000 −8.065 −8.073 *** *** *** EPU −0.496 −81.285 74.117 25.529 0.120 3.518 0.013 −8.937 −19.512 *** *** Monetary −0.191 −86.853 145.453 46.285 0.454 3.235 0.133 −8.210 −24.248 policy *** *** *** Fiscal Policy −1.012 −125.245 91.556 33.261 −0.241 4.392 0.007 −12.299 −16.245 *** *** Taxes −0.999 −86.757 98.919 32.144 0.015 3.388 0.707 −12.381 −16.523 *** *** *** Government −1.649 −196.458 132.850 53.846 −0.278 4.069 0.036 −13.019 −18.428 spending *** *** Health care −1.162 −136.539 80.704 38.509 −0.291 3.613 0.195 −12.891 −16.709 *** *** National −0.109 −122.504 122.488 47.273 0.209 3.109 0.652 −10.575 −32.140 security *** *** Entitlement −1.516 −142.644 106.680 46.992 −0.042 2.841 0.928 −13.151 −19.973 programs *** *** *** Regulation −1.110 −72.979 93.907 31.809 0.532 3.623 0.031 −8.448 −29.883 *** *** Financial −1.534 −158.271 144.694 62.976 −0.080 2.913 0.927 −7.500 −26.375 Regulation *** *** Trade policy 1.816 −198.783 202.726 74.609 0.024 3.176 0.927 −10.924 −23.805 *** *** Sovereign −0.072 −279.765 230.444 103.345 0.002 2.998 0.999 −8.588 −54.930 debt Note. The sample is September 2011—December 2019, covering the monthly returns series. STD (standard deviation) J-B denotes the p-value of the Jarque-Bera normality test. ADF and PP are test statistics for the augmented Dickey-Fuller (ADF) and Phillip-Perron, respectively. (***) indicates the statistical significance at 1% level. EPU = economic policy uncertainty. Table 2 provides descriptive statistics and stationarity plays a more important role than others for the Bitcoin mar- tests of Bitcoin returns and the returns of EPU and 11 cate- ket (Bouri et al., 2020a). Conversely, other sources such as gorical uncertainty indices. Bitcoin experiences the highest entitlement programs are of domestic influence and may be positive average return (10.203%). Conversely, all the uncer- segmented from the Bitcoin market. This is also relevant as tainty indices display negative average variations, except the previous findings provide evidence on the adverse impact of trade policy uncertainty index. The standard deviation fluc- some categorical policy uncertainties on stock market indi- tuates between 25.529 and 103.345, the figures for the aggre- ces (Chiang, 2020). Accordingly, we conjecture that Bitcoin gate EPU index and the sovereign debt uncertainty index, exhibits heterogeneity in its relationship with categorical respectively. Bitcoin and most of the U.S. categorical uncer- EPU indices and thus in its hedging ability. tainty indices have excess kurtosis and non-zero skewness. Based on the results of the Jarque-Bera test, the normality of Methodology Bitcoin returns is rejected. The normality hypothesis is rejected for the returns of the aggregate EPU index and three To determine the ability of Bitcoin to act as a hedge or safe- categorical uncertainty indices. Using both the augmented haven against aggregate and categorical EPU under Bitcoin’s Dickey-Fuller (ADF) and Phillip-Perron (PP) tests, all return various market conditions, we follow Wu et al. (2019) by series are found to be stationary. considering a quantile regression augmented with dummy The various categorical EPU indices capture policy uncer- variables as follows: tainty related to monetary and fiscal policies, taxation, finan- QU τμ =+ βτ + βτ DU U () () () cial regulation, trade policy, government spending, national () BTC ii ,, 01 ti,, iq09 . 0 it security, debt and crises, and entitlement programs. In other +βτ DU U () (1) () words, information regarding categorical uncertainty indices ii ,, 20 q ..95 it , comes from different terms and sources. Thus, an in-depth +βτ DU ()U () ii ,, 30 qi ., 99 t analysis of the association between categorical EPU and a global cryptocurrency currency such as Bitcoin helps iden- where is the log-difference of the uncertainty index of it , tify the specific source of risk in the EPU that can be hedged the category i. DU , DU , and DU () () () iq ,. 090 iq ,. 095 iq ,. 099 by Bitcoin returns. Some of the source (e.g., trade uncer- denote the dummy variables that assume value 1 if the log- tainty) is of global influence on financial markets and thus difference of the categorical uncertainty index i exceeds the 6 SAGE Open 0.9th, 0.95th, and 0.99th quantiles, respectively, and 0 else- Results and Discussion where. Q τ is the τth quantile of Bitcoin returns (BTC) () BTC U We present coefficient estimates from Equation (3) at 9 defined as: quantiles τ∈ (. 0050 ,.10 ,.20 ,.., ., 90.) 95 , representing three τ∈ (. 0050 ,.10 ,.2) market conditions: bearish ( ), normal −1 QF ττ =∈ ();[ τ 01 .] (2) () BTC BTC ( ), and bullish (τ∈ (. 08,. 09,. 095)). τ∈ (. 04,. 05,. 06) U U F (.) is the cumulative conditional distributional func- Hedging Property Analysis BTC tion of the Bitcoin returns given uncertainty level Uu = . Table 3 provides the estimation of the hedging parameter To account for other factors affecting Bitcoin returns, βτ for the aggregate EPU index and the 11 categorical () i,0 we add some control variables, namely the S&P500 rep- indices under Bitcoin’s various market conditions. resenting the effect of the U.S. stock market, the U.S. Considering the aggregate EPU index, we note that an dollar index (USDX) to represent the currency market, increase in the level of EPU has a negative impact on Bitcoin and the gold price (Gold), and oil price (WTI) to account returns. However, the parameter is negatively insig- βτ () i,0 for commodity market effects. We also add the U.S. nificant at almost all quantiles considered, which indicates CBOE VIX index to take into account the volatility of the that Bitcoin cannot act as a hedge against the U.S. policy- U.S. stock market. Then, the augmented Equation (1) related economic uncertainty. This finding is not in line with becomes: Wu et al. (2019), who find that Bitcoin can act as a weak hedge against EPU during extreme bearish and bullish mar- QS τα = τα + τα PU + τ SD () () () () BTC 01 tt 2 ket conditions. To give a more comprehensive picture of the hedging +ατ Gold + ατ Oil + ατ VIX () () (() 34 tt 5 t ability of Bitcoin against U.S. economic uncertainty, we (3) + βτ UD + βτ UU () () () focus on the coefficient estimates of Equation (3) for the 11 ii ,, 01 ti,, iq09 . 0 it categorical economic uncertainty indices. Notably, the esti- +βτ DU U () (() ii ,, 20 q .95 it , mates provide more mixed and nuanced results than those involving the aggregate EPU index. In fact, Table 3 shows +βτ DU U + () () ii ,, 30 qi ., 99 tit βτ () that the parameter i,0 is positive and statistically sig- nificant generally at high Bitcoin return quantiles for some We run Equation (3) for the 5, 10-, 20-, 40-, 50-, 60-, 80-, uncertainty indices, including fiscal policy, taxes, national 90-, and 95-th quantile. Following the line of previous stud- security, and trade policy. These results imply that Bitcoin ies (Baur & Lucey, 2010, who study the gold and stock mar- is a strong hedge against these categorical economic uncer- ket indices; Bouri et al., 2017a, 2017c, who study Bitcoin tainties during bullish states in the Bitcoin market. These and conventional assets), especially Wu et al. (2019) who findings are comparable to those of Gozgor et al. (2019), study gold/Bitcoin and aggregate EPU, we conjecture that who find that trade policy uncertainty impacts Bitcoin Bitcoin is a strong (weak) hedge against categorical eco- returns during periods of regime change. However, when th nomic uncertainty at the quantile (under a given i τ the Bitcoin market is bearish or normal, Bitcoin generally Bitcoin market condition) if the parameter βτ is posi- () i,0 acts as a weak hedge against all categorical uncertainty tively significant (insignificant). Conversely, when βτ () i,0 indices. is negative and significant, Bitcoin is not a hedge against On the other hand, under bullish market condition, the categorical uncertainty . Moreover, the safe-haven prop- negative and significant parameter βτ () for EPU, govern- i,0 erty of Bitcoin against the uncertainty of category is tested ment spending, health care, and entitlement programs show based on the dummy variable parameters (during stress peri- 1 2 3 that Bitcoin does not play the role of hedging assets against ods). If the sum βτ ,, βτ and is pos- () () βτ () these categorical uncertainty indices. In addition, Bitcoin can ij ,, ij ij , ∑∑ ∑ j== 0 j 0 j=0 act as a weak hedge against monetary policy, especially itively significant (insignificant), Bitcoin is a strong (weak) under normal conditions but it can be a strong hedging asset against fiscal and taxes policy uncertainty under bullish mar- safe-haven against economic uncertainty for category i at ket condition. Besides, the insignificant parameter βτ for () i,0 90%, 95%, and 99% quantile respectively of this categorical regulation and sovereign debt shows that this digital cur- index. A negative value for a given sum indicates that rency fails to hedge these two categorical policy uncertain- Bitcoin is not a safe-haven against uncertainty. ties regardless of market conditions. Mokni et al. 7 Table 3. Estimation Results of the Bitcoin Hedging Parameter (β ) 0 Against Categorical Uncertainty. Bearish market Normal market Bullish market 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.114 0.133 −0.088 −0.042 −0.158 −0.040 −0.176*** −0.607*** −1.630*** (0.433) (0.276) (0.193) (0.214) (0.220) (0.254) (0.053) (0.291) (0.291) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.046 −0.126 −0.032 0.047 0.116 0.089 0.177 −0.007 −0.099 (0.175) (0.144) (0.139) (0.109) (0.112) (0.114) (0.120) (0.182) (0.165) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.161 0.178 0.071 0.031 0.037 0.066 −0.190 −0.505 1.022*** (0.246) (0.222) (0.205) (0.156) (0.162) (0.162) (0.374) (0.418) (0.180) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes 0.335 0.459** 0.134 0.075 0.092 0.076 −0.266 0.434 1.022** (0.248) (0.231) (0.213) (0.143) (0.144) (0.152) (0.315) (0.334) (0.393) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.067 0.079 0.007 0.064 0.015 0.000 −0.424 −0.592*** −0.424 (0.120) (0.106) (0.107) (0.119) (0.099) (0.095) (0.347) (0.121) (0.347) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.048 −0.086 0.068 0.000 −0.084 −0.037 −0.198 −0.327 −0.834*** (0.187) (0.158) (0.138) (0.122) (0.137) (0.137) (0.258) (0.231) (0.182) R 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.002 −0.056 −0.040 0.001 −0.012 0.047*** 0.313*** 0.960*** 0.945*** (0.250) (0.078) (0.073) (0.077) (0.082) (0.010) (0.127) (0.183) (0.273) 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 0.071 0.026 0.013 −0.054 −0.094 0.013 −0.108 −0.334 −0.783*** (0.155) (0.090) (0.095) (0.092) (0.099) (0.108) (0.147) (0.289) (0.175) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation −0.170 −0.132 −0.069 −0.002 0.023 0.035 −0.056 −0.176 −0.645 (0.269) (0.194) (0.133) (0.140) (0.143) (0.168) (0.210) (0.286) (0.421) 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation −0.039 −0.121 0.058 0.043 0.062 0.019 −0.120 −0.099 −0.211* (0.221) (0.107) (0.079) (0.063) (0.066) (0.071) (0.077) (0.078) (0.112) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.062* −0.034 −0.040 0.012 −0.024 0.027 0.065 0.078** 0.147*** (0.039) (0.042) (0.043) (0.047) (0.050) (0.054) (0.057) (0.035) (0.065) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.131 −0.036 −0.027 0.016 0.010 −0.010 −0.016 −0.141 −0.145 currency crises (0.260) (0.149) (0.122) (0.111) (0.116) (0.125) (0.134) (0.169) (0.177) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. Overall, it seems that Bitcoin is characterized by a strong acts as a hedging tool in the face of the inefficacity of tradi- ability to act as a hedging tool against some categorical tional assets to hedge portfolios against economic and finan- uncertainty, mainly when it is Bullish. This result could be cial uncertainty conditions (Guesmi et al., 2019; Symitsi and explained by the fact that during periods of price increases Chalvatzis, 2019). Corbet et al. (2018) show that cryptocur- for Bitcoin, investors take more long positions in this digital rencies are somewhat segmented from stock market shocks asset, which makes it more attractive during high uncertainty and dissociated from popular financial assets, which points (stress) periods (Bouri et al., 2017a, 2017c). In fact, Bitcoin to the hedging ability of Bitcoin against uncertainty. 8 SAGE Open Figure 2. Hedge coefficients of Bitcoin with 95% confidence bands. Furthermore, our empirical results show that the hedging Safe-haven Property Analysis ability of Bitcoin is limited to financial (fiscal and taxes), Here, we investigate the safe-haven ability of Bitcoin by national security, and trade uncertainty and not to regulatory focusing on the coefficients of the dummy parameters. uncertainty. This new finding could be explained by the fact Results are provided in Tables 4 to 6 at 90%, 95%, and 99% that the cryptocurrency markets generally operate away from quantiles, respectively. Figures 3 to 5 graphically present the government regulation systems. parameter estimations. Figure 2 presents a graphical illustration of the estimated Table 4 shows that, at the 90% quantile, the parameter parameters, showing the shape of the parameter against the quantile order. Different forms and patterns of the hedging sum βτ is positive and significant for the aggregate () ij , parameter are observed. The value of the parameter tends to j=0 increase with quantile order for financial policy uncertainty τ= 01 . EPU only at the low quantile of Bitcoin distribution ( ), (fiscal and taxes), national security, and trade uncertainty. indicating that Bitcoin can be a strong, safe haven against However, there is generally a decrease in the effect of the aggregate EPU during bearish market states. In addition, other categorical uncertainty indices on Bitcoin returns start- Bitcoin acts as a strong safe-haven against uncertainty related ing at positive values at lower quantiles and finishing at neg- to monetary policy and entitlement programs during the ative and insignificant values at high quantiles. This confirms same bearish market state of Bitcoin. Bitcoin’s ability to hedge uncertainty during Bitcoin’s bull Our above findings nicely complement the related lit- market periods, while this ability is weak during Bitcoin’s erature dealing with Bitcoin and various measures of bear markets. uncertainties (Bouri et al., 2017b), especially previous Mokni et al. 9 Table 4. Estimation Results of the Safe-haven Parameter Sum at 90% βτ (). i =0 Bearish market Normal market Bullish market τ 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.676 0.963** 0.429 0.361 0.331 0.271 0.424 0.971 0.319 (0.618) (0.478) (0.414) (0.432) (0.448) (0.493) (0.614) (0.801) (0.614) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.544* 0.384 0.234 0.120 0.153 0.041 −0.053 −0.396 −0.455* (0.279) (0.263) (0.274) (0.211) (0.224) (0.221) (0.258) (0.258) (0.237) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.621 0.603 0.402 0.220 0.211 0.228 0.597 0.791 2.907*** (0.416) (0.426) (0.361) (0.387) (0.433) (0.577) (1.412) (1.236) (0.823) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes 0.530 0.258 0.107 −0.112 0.154 −0.051 −0.101 0.851 1.061* (0.414) (0.462) (0.388) (0.334) (0.379) (0.378) (0.483) (0.639) (0.593) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.269 0.054 0.141 0.148 −0.088 −0.126 −0.266 −0.443** −0.266 (0.241) (0.231) (0.202) (0.223) (0.201) (0.214) (0.512) (0.221) (0.512) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.413 0.495 0.063 −0.096 −0.195 −0.255 0.091 −0.184 −0.705 (0.363) (0.424) (0.288) (0.279) (0.309) (0.288) (0.678) (0.550) (0.558) 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.293 −0.120 0.150 0.014 −0.054 −0.123 0.258 1.425*** 1.371*** (0.420) (0.664) (0.251) (0.259) (0.272) (0.290) (0.351) (0.471) (0.419) R 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 0.474* 0.220 0.183 0.116 0.015 0.015 −0.281 −0.173 −0.627** (0.255) (0.195) (0.213) (0.205) (0.221) (0.299) (0.288) (0.487) (0.271) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.032 −0.735 −0.747 −0.328 0.148 0.098 −0.250 −0.169 −0.448 (1.100) (0.863) (1.003) (0.506) (0.431) (0.425) (0.457) (0.504) (0.614) R 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation 0.044 −0.148 0.160 0.125 0.120 0.087 0.007 −0.071 0.107 (0.394) (0.390) (0.156) (0.142) (0.148) (0.156) (0.159) (0.158) (0.252) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.505*** −0.421 −0.290 −0.056 −0.093 −0.109 −0.166 0.429*** 0.681*** (0.043) (0.258) (0.267) (0.174) (0.146) (0.136) (0.122) (0.156) (0.145) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.100 0.045 −0.027 0.006 0.000 −0.014 0.067 −0.010 −0.023 currency crises (0.657) (0.647) (0.417) (0.314) (0.294) (0.292) (0.281) (0.413) (0.398) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. studies that consider the aggregate economic policy uncer- Conclusion tainty index (Cheng & Yen, 2019; Demir et al., 2018; Panagiotidis et al., 2019; Wang et al., 2019b; Wu et al., Previous studies consider the impact of U.S. aggregate eco- 2019) by providing a more detailed and nuanced analysis nomic policy uncertainty on Bitcoin returns to make a hedge of the hedging and safe-haven properties of Bitcoin and safe-haven inferences. However, they disregard the against the disaggregated measures of U.S. economic impact of categorical U.S. EPU data that covers disaggre- uncertainties. gated measures of uncertainties such as monetary, fiscal, 10 SAGE Open Table 5. Estimation Results of the Safe-haven Parameter Sum at 95% . βτ () i =0 Bearish market Normal market Bullish market 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.683 0.925 0.184 −0.111 −0.168 0.045 0.861 1.470 0.508 (0.730) (0.618) (0.550) (0.628) (0.644) (0.727) (0.859) (1.004) (0.996) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.328 0.121 0.104 0.035 −0.068 −0.024 −0.131 −0.265 −0.403 (0.311) (0.347) (0.340) (0.262) (0.283) (0.272) (0.315) (0.319) (0.273) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.674 0.235 0.400 0.433 0.340 0.171 0.391 0.375 2.062 (0.526) (0.759) (0.494) (0.565) (0.652) (0.802) (1.985) (1.858) (1.176) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.0641 0.1510 0.2856 Taxes 0.884* 0.793 0.548 0.332 0.249 0.588 0.455 1.142 1.968*** (0.518) (0.609) (0.577) (0.648) (0.605) (0.606) (0.618) (0.858) (0.676) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.468 0.264 0.161 0.069 0.032 0.014 −0.041 −0.069 −0.041 (0.290) (0.296) (0.271) (0.318) (0.307) (0.335) (0.593) (0.284) (0.593) 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.574 0.230 0.112 0.120 0.530 0.425 0.298 0.233 0.020 (0.445) (0.716) (0.407) (0.459) (0.465) (0.431) (0.946) (0.739) (0.721) R 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security −0.069 −0.248 −0.243 0.088 −0.041 −0.078 0.496 0.991 1.896*** (0.508) (0.915) (0.383) (0.404) (0.416) (0.428) (0.693) (0.729) (0.687) 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs −0.060 −0.249 −0.359 0.093 0.055 0.065 0.017 −0.238 −0.412 (0.370) (0.358) (0.385) (0.329) (0.336) (0.396) (0.377) (0.590) (0.357) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.465 0.253 0.182 0.115 0.145 0.073 0.014 −0.185 0.001 (1.429) (1.187) (1.405) (0.726) (0.589) (0.566) (0.603) (0.601) (0.746) 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.4997 Fin Regulation −0.016 −0.271 0.107 0.024 0.022 −0.005 −0.236 −0.338 −0.398 (0.612) (0.712) (0.210) (0.209) (0.227) (0.238) (0.242) (0.247) (0.357) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.121*** −0.123 −0.211 −0.152 −0.172 0.065 0.024 0.600** 0.817*** (0.046) (0.359) (0.388) (0.268) (0.237) (0.255) (0.186) (0.295) (0.169) 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.100 0.045 −0.027 0.006 0.000 −0.014 0.067 −0.010 −0.023 currency crises (0.990) (1.064) (0.593) (0.476) (0.464) (0.491) (0.428) (0.533) (0.522) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. regulatory, trade, and political uncertainties. To address this normal market conditions. However, Bitcoin is a strong literature gap and account for the aggregation effect in the hedge and safe-haven against some categorical EPU, includ- U.S. EPU, we examine the impact of categorical U.S. EPU ing fiscal policy, taxes, national security, and trade policy indices on Bitcoin returns. Specifically, we employ quantile under bullish market conditions. regressions augmented with dummy variables to take into The policy implications concern the particularity of account various states of the Bitcoin market. Bitcoin as a digital assets class that is very useful for inves- The results indicate that Bitcoin is a weak hedge against tors, financial advisors, and risk managers making decisions all the uncertainty categories considered under bearish and involving the risk of specific measures of U.S. economic Mokni et al. 11 Table 6. Estimation Results of the Safe-haven Parameter Sum at 99% . βτ () i =0 Bearish market Normal market Bullish market τ 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.782 0.684 0.317 0.187 0.247 0.204 0.438 1.177 −0.080 (0.801) (0.966) (0.656) (0.764) (0.773) (0.881) (1.098) (1.217) (1.233) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.200 0.374 0.101 0.252 0.279 0.166 0.199 −0.059 −0.061 (0.419) (0.912) (1.013) (0.395) (0.422) (0.411) (0.428) (0.497) (0.379) 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.105 0.623 −0.065 −0.467 −0.430 0.562 0.054 −0.130 1.911 (0.829) (1.440) (0.899) (1.004) (1.143) (1.122) (2.240) (2.305) (1.310) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes −0.122 −0.167 −0.304 −0.434 −0.081 −0.210 −0.553 0.295 1.170 (0.937) (1.023) (1.175) (1.556) (0.803) (0.753) (0.723) (0.924) (0.816) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 1.110 1.113 1.452 1.456 1.533 1.337 0.732 0.054 0.732 (0.806) (0.816) (0.996) (1.184) (1.204) (1.209) (1.401) (1.064) (1.401) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.192 0.647 0.396 0.143 −0.586 −0.251 −1.116 −1.151 −1.589 (0.728) (1.166) (0.764) (0.903) (0.891) (0.834) (1.183) (1.010) (1.159) 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.872 0.900 0.533 −0.187 −0.176 −0.313 1.899 3.636 2.705 (0.812) (1.118) (0.810) (0.642) (0.677) (0.654) (4.328) (3.566) (4.709) R 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 1.922*** 1.742*** 1.622*** 0.869 0.836 0.629 0.197 0.413 0.186 (0.606) (0.674) (0.684) (0.649) (0.653) (0.648) (0.684) (0.807) (0.670) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.632 0.452 0.371 0.083 −0.073 −0.135 −0.377 −0.179 −0.578 (1.467) (1.247) (1.435) (0.831) (0.722) (0.698) (0.765) (0.734) (0.954) R 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation 0.917 0.649 0.293 0.090 −0.096 −0.161 −0.204 −0.264 −0.678 (0.966) (1.185) (0.350) (0.402) (0.438) (0.451) (0.426) (0.413) (0.541) 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy 0.514*** 0.570 0.700 0.460 0.322 −0.303 −0.621 0.751*** 1.035*** (0.169) (0.432) (0.543) (0.453) (0.450) (0.781) (0.592) (0.303) (0.323) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.196 −0.047 0.172 0.264 0.301 0.302 0.046 −0.247 0.009 currency crises (1.414) (1.594) (0.888) (0.854) (0.888) (1.224) (1.007) (1.167) (0.859) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) denote indicate the statistical significance at 10%, 5%, and 1% level, respectively. uncertainties. Furthermore, the findings could be useful to periods, investors can incorporate Bitcoin in their portfolios investors operating in conventional and cryptocurrency mar- for hedging purposes. However, considering Bitcoin in such kets. They indicate that investors should monitor develop- investment decisions would be most suitable when the ments in taxes, fiscal policy, national security, and trade Bitcoin market experiences a bullish state. policy in order to be better prepared for uncertainty and While our empirical analysis involves categorical data hedge their conventional portfolios based on Bitcoin. from the United States only, it would be interesting to extend Additionally, investors can benefit from our results by speci- the analysis by considering categorical EPU data from other fying more suitable portfolio design. During high uncertainty countries and regions such as China and Europe. This might 12 SAGE Open Figure 3. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 90% quantile of the U.S. categorical uncertainty indices. Figure 4. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 95% quantile of the U.S. categorical uncertainty indices. Mokni et al. 13 Figure 5. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 99% quantile of the U.S. categorical uncertainty indices. be done when such disaggregated data becomes available. Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring eco- nomic policy uncertainty. The Quarterly Journal of Economics, Another potential path for future research could cover the 131(4), 1593–1636. role of other leading cryptocurrencies such as Ethereum and Baur, D. G., & Dimpfl, T. (2019). Price discovery in bitcoin spot or Ripple. futures? Journal of Futures Markets, 39(7), 803–817. Baur, D. G., Hong, K., & Lee, A. D. (2018). Bitcoin: Medium Declaration of Conflicting Interests of exchange or speculative assets? Journal of International The author(s) declared no potential conflicts of interest with Financial Markets, Institutions and Money, 54, 177–189. respect to the research, authorship, and/or publication of this Baur, D. G., & Lucey, B. M. (2010). Is gold a hedge or a safe article. haven? An analysis of stocks, bonds and gold. The Financial Review, 45, 217–229. Funding Bernanke, B. S. (1983). Irreversibility, uncertainty, and cyclical investment. The Quarterly Journal of Economics, 98(1), 85– The author(s) received no financial support for the research and/or authorship of this article. Bilgin, M. H., Gozgor, G., Lau, C. K. M., & Shing, X. (2018). The effects of uncertainty measures on the price of gold. ORCID iDs International Review of Financial Analysis, 58, 1–7. Bloom, N. (2009). The impact of uncertainty shocks. Econometrica, Khaled Mokni https://orcid.org/0000-0001-9817-6378 77(3), 623–685. Elie Bouri https://orcid.org/0000-0003-2628-5027 Bouri, E., Gil-Alana, L. A., Gupta, R., & Roubaud, D. (2019). Xuan Vinh Vo https://orcid.org/0000-0002-4725-6914 Modelling long memory volatility in the Bitcoin market: Evidence of persistence and structural breaks. International References Journal of Finance & Economics, 24(1), 412–426. Anceaume, E., Ludinard, R., Potop-Butucaru, M., & Tronel, F. Bouri, E., Gkillas, K., & Gupta, R. (2020a). Trade uncertainties and (2017). Bitcoin a distributed shared register. In International the hedging abilities of Bitcoin. Economic Notes, 49(3), Article Symposium on Stabilization, Safety, and Security of Distributed e12173. Systems (pp. 456–468). Springer. Bouri, E., Gupta, R., Tiwari, A., & Roubaud, D. (2017a). Does Arouri, M., Estay, C., Rault, C., & Roubaud, D. (2016). Economic Bitcoin hedge global uncertainty? Evidence from wavelet- policy uncertainty and stock markets: Long-run evidence from based quantile-in-quantile regressions. Finance Research the US. Finance Research Letters, 18, 136–141. Letters, 23, 87–95. 14 SAGE Open Bouri, E., Jalkh, N., Molnár, P., & Roubaud, D. (2017b). Bitcoin Panagiotidis, T., Stengos, T., & Vravosinos, O. (2019). The effects for energy commodities before and after the December 2013 of markets, uncertainty and search intensity on Bitcoin returns. crash: Diversifier, hedge or safe haven? Applied Economics, International Review of Financial Analysis, 63, 220–242. 49(50), 5063–5073. Raza, S. A., Shah, N., & Shahbaz, M. (2018). Does economic policy Bouri, E., Molnár, P., Azzi, G., Roubaud, D., & Hagfors, L. I. uncertainty influence gold prices? Evidence from a nonparamet- (2017c). On the hedge and safe haven properties of Bitcoin: ric causality-in-quantiles approach. Resources Policy, 57, 61–68. Is it really more than a diversifier? Finance Research Letters, Selmi, R., Mensi, W., Hammoudeh, S., & Bouoiyour, J. (2018). Is 20, 192–198. Bitcoin a hedge, a safe haven or a diversifier for oil price move- Bouri, E., Shahzad, S. J. H., Roubaud, D., Kristoufek, L., & Lucey, ments? A comparison with gold. Energy Economics, 74, 787–801. B. (2020b). Bitcoin, gold, and commodities as safe havens for Shahzad, S. J. H., Bouri, E., Roubaud, D., & Kristoufek, L. (2020). stocks: New insight through wavelet analysis. The Quarterly Safe haven, hedge and diversification for G7 stock markets: Review of Economics and Finance, 77, 156–164. Gold versus Bitcoin. Economic Modelling, 87, 212–224. Cheng, H. P., & Yen, K. C. (2019). The relationship between the Shahzad, S. J. H., Bouri, E., Roubaud, D., Kristoufek, L., & Lucey, economic policy uncertainty and the cryptocurrency market. B. (2019). Is Bitcoin a better safe-haven investment than gold Finance Research Letters, 35, 101308. https://doi.org/10.1016/j. and commodities? International Review of Financial Analysis, frl.2019.101308 63, 322–330. Chiang, T. C. (2020). Economic policy uncertainty and stock Symitsi, E., & Chalvatzis, K. J. (2019). The economic value of returns—Evidence from the Japanese market. Quantitative Bitcoin: A portfolio analysis of currencies, gold, oil and stocks. Finance and Economics, 4(3), 430–458. Research in International Business and Finance, 48, 97–110. Corbet, S., Meegan, A., Larkin, C., Lucey, B., & Yarovaya, L. (2018). Teomete Yalabık, F., & Yalabık, İ. (2019). Anonymous Bitcoin v Exploring the dynamic relationships between cryptocurrencies enforcement law. International Review of Law, Computers & and other financial assets. Economics Letters, 165, 28–34. Technology, 33(1), 34–52. Demir, E., Gozgor, G., Lau, C. K. M., & Vigne, S. A. (2018). Does Urquhart, A., & Zhang, H. (2019). Is Bitcoin a hedge or safe haven economic policy uncertainty predict the Bitcoin returns? An for currencies? An intraday analysis. International Review of empirical investigation. Finance Research Letters, 26, 145– Financial Analysis, 63, 49–57. 149. Wang, G.-J., Ma, X., & Wu, H. (2020). Are stablecoins truly Fang, L., Bouri, E., Gupta, R., & Roubaud, D. (2019). Does global diversifiers, hedges, or safe havens against traditional crypto- economic uncertainty matter for the volatility and hedging currencies as their name suggests? Research in International effectiveness of Bitcoin? International Review of Financial Business and Finance, 54, 101225. https://doi.org/10.1016/j. Analysis, 61, 29–36. ribaf.2020.101225 Gozgor, G., Tiwari, A. K., Demir, E., & Akron, S. (2019). The rela- Wang, G.-J, Tang, Y., Xie, C., & Chen, S. (2019a). Is Bitcoin a safe tionship between Bitcoin returns and trade policy uncertainty. haven or a hedging asset? Evidence from China. Journal of Finance Research Letters, 29, 75–82. Management Science and Engineering, 4(3), 173–188. Guesmi, K., Saadi, S., Abid, I., & Ftiti, Z. (2019). Portfolio Wang, G.-J., Xie, C., Wen, D., & Zhao, L. (2019b). When Bitcoin diversification with virtual currency: Evidence from bitcoin. meets economic policy uncertainty (EPU): Measuring risk International Review of Financial Analysis, 63, 431–437. spillover effect from EPU to Bitcoin. Finance Research Klein, T., Thu, H. P., & Walther, T. (2018). Bitcoin is not the New Letters, 31, 489–497. Gold —A comparison of volatility, correlation, and portfolio Wu, S., Tong, M., Yang, Z., & Derbali, A. (2019). Does gold or performance. International Review of Financial Analysis, 59, bitcoin hedge economic policy uncertainty? Finance Research 105–116. Letters, 31, 171–178. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png SAGE Open SAGE

Does Bitcoin Hedge Categorical Economic Uncertainty? A Quantile Analysis:

Mokni, Khaled; Bouri, Elie; Ajmi, Ahdi Noomen; Vo, Xuan Vinh
SAGE Open , Volume 11 (2): 1 – May 22, 2021
Download PDF
14 pages

Loading next page...
 
/lp/sage/does-bitcoin-hedge-categorical-economic-uncertainty-a-quantile-FcJkrSvj9h

References (38)

Publisher
SAGE
Copyright
Copyright © 2022 by SAGE Publications Inc, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses.
ISSN
2158-2440
eISSN
2158-2440
DOI
10.1177/21582440211016377
Publisher site
See Article on Publisher Site

Abstract

This paper examines the hedge and safe-haven abilities of Bitcoin against U.S. aggregate and categorical economic policy uncertainty (EPU) via the application of quantile regression model augmented with a dummy and some control variables. Using monthly data from September 2011 to December 2019, empirical results indicate that Bitcoin does not act as a strong hedge against the aggregate U.S. EPU. However, it acts as a strong safe-haven for this aggregate measure of uncertainty when the Bitcoin market is bearish. Looking deeper into the disaggregated level of the U.S. EPU data, the analyses involving categorical EPU data indicate the ability of Bitcoin to act as a strong hedge and safe-haven against specific uncertainties related to fiscal policy, taxes, national security, and trade policy. Keywords Bitcoin, EPU, categorical economic uncertainty, hedge, safe-haven, quantile the Cyprus banking crisis (2012–2013). Since its lunch, the Introduction value of this digital currency has soared from $0.09 in July Research interest in the effects of uncertainty and the econ- 2010 to around $19,000 in December 2017. Bitcoin market omy in general on financial markets goes back to the theo- capitalization surged from less than $1.6 Billion to about retical models of Bernanke (1983) and Bloom (2009). $316 billion during the same period, allowing Bitcoin to However, since the 2008 global financial crisis, policy fac- occupy more than 80% of the total market value of all cryp- tors have gained ground in shaping the economic environ- tocurrencies. Accordingly, Bitcoin gained a large interest in ment and financial markets. Lately, Baker et al. (2016) have the financial literature, given its beneficial proprieties. In proposed a news-based index of U.S. economic policy uncer- fact, Bitcoin is almost isolated from the global financial sys- tainty (EPU), which has been widely used in academic litera- tem, making it a valuable addition to portfolios containing ture. Accordingly, numerous studies try to understand the conventional assets (Corbet et al., 2018; Guesmi et al., 2019; impact of the U.S. EPU on U.S. stock market returns and Symitsi and Chalvatzis, 2019). In view of this, some aca- report evidence of a negative impact (e.g., Arouri et al., demic literature examines the hedging and safe-haven prop- 2016). Other studies focus on the impact of the U.S. EPU on erties of Bitcoin for equities (Bouri et al., 2017c; Klein et al., safe-haven assets such as gold, mostly showing that EPU has 2018; Shahzad et al., 2019, 2020), commodities (Bouri et al., the ability to predict gold prices (Raza et al., 2018) and that 2017b; Klein et al., 2018), and currencies (Urquhart & gold prices and economic uncertainty are positively related (Bilgin et al., 2018). Such findings concur with previous evi- dence arguing that, during periods of economic and political Northern Border University, Arar, Saudi Arabia Gabès University, Tunisia uncertainties, investors switch their investments from risky Lebanese American University, Beirut, Lebanon assets to less risky or safe-haven assets such as gold (Baur & University of Economics Ho Chi Minh City, Vietnam Lucey, 2010). Prince Sattam bin Abdulaziz University, Al-Kharj, Saudi Arabia Remarkably, the most popular and largest cryptocurrency, 6 Manouba University, Tunisia Bitcoin, has emerged as a substitute instrument for the inef- Corresponding Author: fectiveness of traditional economic and financial systems, Khaled Mokni, College of Business Administration, Northern Border especially during stress periods (Bouri et al., 2017a, 2017c), University, Arar 91431, Saudi Arabia. such as the European sovereign debt crisis (2010–2013) and Email: kmokni@gmail.com Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 SAGE Open Zhang, 2019). Other studies consider the effect of EPU on in general, with some exceptions.). Second, we contribute to the Bitcoin market and make inferences regarding hedging the literature dealing with the debatable role of Bitcoin as a and acting as a safe-haven (Cheng & Yen, 2019; Demir et al., hedge and safe-haven against uncertainty (Bouri et al., 2018; Panagiotidis et al., 2019; Wang et al., 2019b; Wu et al., 2017a; Cheng & Yen, 2019; Panagiotidis et al., 2019) during 2019). The findings are, however, mixed and inconclusive. stress periods. Unlike previous studies (Cheng & Yen, 2019; Some studies show that the impact of the U.S. EPU is insig- Demir et al., 2018; Panagiotidis et al., 2019; Wang et al., nificant (Cheng & Yen, 2019; Wang et al., 2019b), while oth- 2019b; Wu et al., 2019), we find convincing evidence for ers find a positive impact (e.g., Panagiotidis et al., 2019) that Bitcoin being a hedge and safe-haven for specific categorical varies across the lower, middle, and upper quantiles (Demir U.S. EPU indices such as fiscal policy, taxes, national secu- et al., 2018; Wu et al., 2019). Notably, related studies limit rity, and trade policy uncertainty. These findings are useful to their focus to the aggregate EPU index, which would mask a variety of economic players such as policymakers, finan- any potential heterogeneity in the impact of the categorical cial advisors, and investors. U.S. EPU on the Bitcoin market. Therefore, extending the The rest of the paper is organized into four sections. above literature to the categorical level of the U.S. EPU data Section 2 reviews the related literature dealing with Bitcoin. (Baker et al., 2016) would help uncover potential heteroge- Section 3 describes the data and methodology. Section 4 neity in the role of Bitcoin as a hedge and safe-haven against reports the empirical results and offers a discussion in light the various 11 components of U.S. economic uncertainty of previous studies. Section 5 concludes and opens paths for (monetary policy, fiscal policy, taxes, government spending, further research. health care, national security, entitlement programs, regula- tion, financial regulation, trade policy, and sovereign debt Previous Studies and currency crises). This is important for at least two rea- sons. First, specific economic uncertainties and develop- The release of Bitcoin as a genuine and fast payment mecha- ments, such as those related to monetary, fiscal, political, and nism has marked the last decade. Some studies examine the trade policies in the United States, have been in the financial safety and legal aspects of the Bitcoin market (Anceaume news. For example, the trade war between the United States et al., 2017; Teomete Yalabık & Yalabık, 2019). As the market and China and the impeachment action in the United States for Bitcoin grows, Bitcoin becomes a leading digital asset, have moved financial markets, and some press articles have attracting the attention of many investors. Accordingly, numer- tried to establish a link to Bitcoin. Second, a more nuanced ous studies focus on the economic and financial implications analysis that accounts for the heterogeneity in the composi- of Bitcoin by considering price discovery (Baur & Dimpfl, tion of the U.S. EPU helps investors make inferences regard- 2019), volatility (Bouri et al., 2019), and speculative nature ing portfolio management. It also helps financial advisors (Baur et al., 2018). Notably, Bitcoin is segmented from the make risk management and hedging decisions regarding the global financial system, offering valuable diversification ben- ability of Bitcoin to hedge various components of economic efits (Bouri et al., 2017c, 2017b; Corbet et al., 2018; Guesmi uncertainty, especially during market downturns. et al., 2019; Klein et al., 2018; Selmi et al., 2018; Shahzad In light of the above, we examine in this study the impact et al., 2019, 2020; Symitsi & Chalvatzis, 2019; Urquhart & of categorical U.S. EPU indices on Bitcoin returns. Using Zhang, 2019). Bouri et al. (2017a) report that Bitcoin repre- monthly data from September 2011 to December 2019, we sents a hedging tool against stock market uncertainty. Other apply a quantile-based regression augmented with dummy studies focus on the impact of EPU on the Bitcoin market. variables and some other control variables to account for the Demir et al. (2018) use a quantile-based approach to examine heavy-tails of asset returns and various dependent variable the prediction power of EPU on Bitcoin prices. They suggest distribution levels. Such an examination allows us to differ- that Bitcoin can be used as a hedging tool against EPU. Wu entiate between the impact of categorical U.S. EPU indices et al. (2019) employ a GARCH model and quantile regression at various quantiles of Bitcoin return distribution (i.e., bear- to compare the hedge and safe-haven roles of gold and Bitcoin ish, normal, and bullish periods) (Wu et al., 2019). against EPU. They show the inefficacity of these two assets in Our contributions are on various fronts. First, we contrib- acting as a hedge or safe haven against EPU. Fang et al. (2019) ute to the strand of literature dealing with EPU and Bitcoin apply multivariate GARCH models to investigate the impact returns (Cheng & Yen, 2019; Demir et al., 2018; Panagiotidis of aggregate EPU on Bitcoin and other assets. They report that et al., 2019; Wang et al., 2019b; Wu et al., 2019) by moving Bitcoin can be considered a hedge under specific economic the debate to the categorical level of the U.S. EPU. In fact, uncertainty conditions. Also, using a quantile-based approach our paper is the first to examine the impact of various com- and causality tests, Wang et al. (2019b) find that Bitcoin has ponents of the U.S. EPU on Bitcoin returns during bearish the propriety of a safe-haven and a diversifier for the extreme and bullish market states, which extends previous studies shocks of economic uncertainty. Cheng and Yen (2019) inves- focusing only on trade policy uncertainty (e.g., Gozgor et al. tigate the relationship between cryptocurrency volatility and (2019) apply wavelets methods and show that the relation- EPU. They indicate that Bitcoin and Litecoin are useful hedg- ship between trade uncertainty and Bitcoin prices is positive ing tools against EPU. Mokni et al. 3 Table 1. Description of the Different EPU Categorical Uncertainty Indices. Categorical index Description Monetary policy Is a sub-index based on news data incorporating the mention of terms related to monetary policy. as examples “federal reserve”, “money supply”, “monetary policy”, “overnight lending rate”, . . . Fiscal Policy Is a sub-index to measure uncertainty related to fiscal policy. It refers to the mention of terms like “government spending”, “budget battle”, “military spending”, “fiscal stimulus”, . . . Government spending Is a sub-index of uncertainty regarding government spending. It is based on news data mentioning terms like “government spending”, “defense spending”, “military spending”, “entitlement spending”, . . . Taxes Is a sub-index calculated based on the news data related to taxes and mentioning the terms like “taxes”, “tax”, “taxation”, . . . Health care Is a sub-index of uncertainty related to health, Medicaid, Medicare, health insurance, malpractice tort reform, malpractice reform, prescription drugs, drug policy, food and drug administration, FDA, medical malpractice, prescription drug act, medical insurance reform, medical liability, part d, affordable care act, Obamacare National security Is a sub-index based on the news data related to national security and mentioning terms like: “national security”, “military conflict”, “war”, “terrorism”, . . . Entitlement programs Is a sub-index of search results from news related to entitlement programs incorporating terms like “program”, “government entitlements”, “social security”, . . . Regulation Is a sub-index to measure uncertainty regarding regulation. It is based on the research from news citing terms in relation with regulation like: “banking supervision”, “bank supervision”, “financial reform”, . . . Financial Regulation Is a news based index based on the number of citation of terms regarding financial regulation loke “banking supervision”, bank supervision”, “house financial services committee”, . . . Trade policy Is a sub-index of uncertainty related to trade activity. It is based on news incorporating terms related to trade like “import duty”, “government subsidies”, “world trade organization”, . . . Sovereign debt Is a news based sub-index related to sovereign debt and currency crisis. It is based on the mention of uncertainty terms like “currency crisis”, “currency devaluation”, “currency manipulation”, “exchange rate” Other studies focus on the hedging and safe haven propri- Regarding the diversification potential, the results report the eties of cryptocurrencies. For example, Wang et al. (2019a) dominance of Bitcoin over both gold and commodities. investigate the spillover effects between Bitcoin and major In this paper, we contribute to the academic literature by financial assets based on the VAR-GARCH-BEKK frame- investigating the potential hedging and safe-haven effects of work. Their results show that Bitcoin can serve as a hedge Bitcoin with respect to various categories of U.S. EPU. To against some assets, including stocks and bonds. Besides, the best of our knowledge, no previous research has focused Bitcoin can act as a safe haven against extreme changes in on various categories of EPU to address whether Bitcoin is a monetary markets. Shahzad et al. (2019) compare the hedg- hedge or safe-haven against categorical U.S. EPU indices at ing and the safe haven proprieties of Bitcoin compared to the various states of the Bitcoin market (i.e., bullish and bearish commodities and gold against stock market investments dur- market periods). Methodologically, we use quantile regres- ing bear and bull market conditions. They report that Bitcoin, sions augmented with dummy variables to address various gold, and the commodity index can serve as weak safe-haven situations of markets and various lower and upper quantiles assets in some cases. of Bitcoin return distributions. More recently, Wang et al. (2020) investigate the propri- eties of stablecoins for traditional cryptocurrencies. They Data and Methodology show the following. First, USD-pegged stablecoins have bet- ter risk-dispersion abilities for traditional cryptocurrencies The Dataset than gold-pegged ones. Second, Tether plays the role of a strong hedge for traditional cryptocurrencies. Third, gold is a In this study, we use a monthly dataset covering the closing better hedge than stable coins, and the USD is a better hedge prices of Bitcoin against the U.S. dollar and the U.S. EPU. To than two of the USD-pegged stablecoins. However, gold as a eliminate aggregation bias, we use 11 categorical uncertainty safe haven is not as good as the stablecoins it backs. Third, indices covering monetary policy, fiscal policy, taxes, gov- the gold-pegged stablecoin is less efficient than the USD- ernment spending, health care, national security, entitlement pegged stablecoins in terms of risk reduction. Bouri et al. programs, regulation, financial regulation, trade policy, and (2020b) compare the safe-haven properties of Bitcoin, gold, sovereign debt. Following Baker et al. (2016), these indices and the commodity index against the world, developed, are solely based on news data and constructed using the emerging, the United States, and Chinese stock market indi- Access World News (AWN) database of about 2000 U.S. ces based on the wavelet coherency approach analysis. They newspapers. Each index is constructed with respect to the indicate a weak dependence between Bitcoin/gold/commod- mention of categorical policy terms related to uncertainty ities and the considered stock markets at various time scales. terms. Table 1 presents the different terms used to construct 4 SAGE Open (a) (b) 80 150 100 EPU Fiscal policy Monetary policy 40 100 50 0 50 0 -40 0 -50 -80 -50 -100 -120 -100 -150 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Govermentspending Health care Taxes 100 50 50 0 -50 -100 -50 -100 -200 -100 -150 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Entitlementprograms National security Regulation -50 -50 -100 -40 -100 -150 -150 -80 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 200 300 FinancialRegulation Tradepolicy Sovereigndebt 200 200 0 0 -100 -100 -100 -200 -200 -200 -300 -300 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 11 12 13 14 15 16 17 18 19 Figure 1. (a) Bitcoin returns and (b) aggregate and categorical economic policy uncertainty changes. each categorical EPU index, For more details, see Baker and represent the price of Bitcoin against the U.S. dollar et al. (2016) and the website: https://www.policyuncertainty. from Bitstamp, the leading exchange. Uncertainty indices com.) are sourced from the website https://www.policyuncer- Besides the availability of categorical U.S. EPU data, tainty.com/. Notably, the starting date depends on the we focus on U.S. economic uncertainty for at least two availability of Bitcoin prices. We employ the logarithmic other reasons. First, the United States is the country run- returns of Bitcoin and aggregate EPU (Demir et al., 2018), ning most Bitcoin nodes in the world (21.49% in January and categorical economic uncertainty indices, that is., 2020). Second, the Bitcoin price is mainly traded against Rx =× 100 log(/ x ), where is the level of Bitcoin it ,, it it , −1 it , the U.S. dollar. The dataset covers the period from price or the uncertainty index of category . Figure 1 pres- September 2011 to December 2019, yielding 100 monthly ents the plot of returns for Bitcoin, the aggregate EPU, and observations Data for Bitcoin price are from DataStream the categorical uncertainty indices. Mokni et al. 5 Table 2. Descriptive Statistics and Stationarity Tests. Mean Minimum Maximum STD Skewness Kurtosis J-B ADF PP *** *** *** Bitcoin 10.203 −49.215 174.000 34.917 1.629 8.251 0.000 −8.065 −8.073 *** *** *** EPU −0.496 −81.285 74.117 25.529 0.120 3.518 0.013 −8.937 −19.512 *** *** Monetary −0.191 −86.853 145.453 46.285 0.454 3.235 0.133 −8.210 −24.248 policy *** *** *** Fiscal Policy −1.012 −125.245 91.556 33.261 −0.241 4.392 0.007 −12.299 −16.245 *** *** Taxes −0.999 −86.757 98.919 32.144 0.015 3.388 0.707 −12.381 −16.523 *** *** *** Government −1.649 −196.458 132.850 53.846 −0.278 4.069 0.036 −13.019 −18.428 spending *** *** Health care −1.162 −136.539 80.704 38.509 −0.291 3.613 0.195 −12.891 −16.709 *** *** National −0.109 −122.504 122.488 47.273 0.209 3.109 0.652 −10.575 −32.140 security *** *** Entitlement −1.516 −142.644 106.680 46.992 −0.042 2.841 0.928 −13.151 −19.973 programs *** *** *** Regulation −1.110 −72.979 93.907 31.809 0.532 3.623 0.031 −8.448 −29.883 *** *** Financial −1.534 −158.271 144.694 62.976 −0.080 2.913 0.927 −7.500 −26.375 Regulation *** *** Trade policy 1.816 −198.783 202.726 74.609 0.024 3.176 0.927 −10.924 −23.805 *** *** Sovereign −0.072 −279.765 230.444 103.345 0.002 2.998 0.999 −8.588 −54.930 debt Note. The sample is September 2011—December 2019, covering the monthly returns series. STD (standard deviation) J-B denotes the p-value of the Jarque-Bera normality test. ADF and PP are test statistics for the augmented Dickey-Fuller (ADF) and Phillip-Perron, respectively. (***) indicates the statistical significance at 1% level. EPU = economic policy uncertainty. Table 2 provides descriptive statistics and stationarity plays a more important role than others for the Bitcoin mar- tests of Bitcoin returns and the returns of EPU and 11 cate- ket (Bouri et al., 2020a). Conversely, other sources such as gorical uncertainty indices. Bitcoin experiences the highest entitlement programs are of domestic influence and may be positive average return (10.203%). Conversely, all the uncer- segmented from the Bitcoin market. This is also relevant as tainty indices display negative average variations, except the previous findings provide evidence on the adverse impact of trade policy uncertainty index. The standard deviation fluc- some categorical policy uncertainties on stock market indi- tuates between 25.529 and 103.345, the figures for the aggre- ces (Chiang, 2020). Accordingly, we conjecture that Bitcoin gate EPU index and the sovereign debt uncertainty index, exhibits heterogeneity in its relationship with categorical respectively. Bitcoin and most of the U.S. categorical uncer- EPU indices and thus in its hedging ability. tainty indices have excess kurtosis and non-zero skewness. Based on the results of the Jarque-Bera test, the normality of Methodology Bitcoin returns is rejected. The normality hypothesis is rejected for the returns of the aggregate EPU index and three To determine the ability of Bitcoin to act as a hedge or safe- categorical uncertainty indices. Using both the augmented haven against aggregate and categorical EPU under Bitcoin’s Dickey-Fuller (ADF) and Phillip-Perron (PP) tests, all return various market conditions, we follow Wu et al. (2019) by series are found to be stationary. considering a quantile regression augmented with dummy The various categorical EPU indices capture policy uncer- variables as follows: tainty related to monetary and fiscal policies, taxation, finan- QU τμ =+ βτ + βτ DU U () () () cial regulation, trade policy, government spending, national () BTC ii ,, 01 ti,, iq09 . 0 it security, debt and crises, and entitlement programs. In other +βτ DU U () (1) () words, information regarding categorical uncertainty indices ii ,, 20 q ..95 it , comes from different terms and sources. Thus, an in-depth +βτ DU ()U () ii ,, 30 qi ., 99 t analysis of the association between categorical EPU and a global cryptocurrency currency such as Bitcoin helps iden- where is the log-difference of the uncertainty index of it , tify the specific source of risk in the EPU that can be hedged the category i. DU , DU , and DU () () () iq ,. 090 iq ,. 095 iq ,. 099 by Bitcoin returns. Some of the source (e.g., trade uncer- denote the dummy variables that assume value 1 if the log- tainty) is of global influence on financial markets and thus difference of the categorical uncertainty index i exceeds the 6 SAGE Open 0.9th, 0.95th, and 0.99th quantiles, respectively, and 0 else- Results and Discussion where. Q τ is the τth quantile of Bitcoin returns (BTC) () BTC U We present coefficient estimates from Equation (3) at 9 defined as: quantiles τ∈ (. 0050 ,.10 ,.20 ,.., ., 90.) 95 , representing three τ∈ (. 0050 ,.10 ,.2) market conditions: bearish ( ), normal −1 QF ττ =∈ ();[ τ 01 .] (2) () BTC BTC ( ), and bullish (τ∈ (. 08,. 09,. 095)). τ∈ (. 04,. 05,. 06) U U F (.) is the cumulative conditional distributional func- Hedging Property Analysis BTC tion of the Bitcoin returns given uncertainty level Uu = . Table 3 provides the estimation of the hedging parameter To account for other factors affecting Bitcoin returns, βτ for the aggregate EPU index and the 11 categorical () i,0 we add some control variables, namely the S&P500 rep- indices under Bitcoin’s various market conditions. resenting the effect of the U.S. stock market, the U.S. Considering the aggregate EPU index, we note that an dollar index (USDX) to represent the currency market, increase in the level of EPU has a negative impact on Bitcoin and the gold price (Gold), and oil price (WTI) to account returns. However, the parameter is negatively insig- βτ () i,0 for commodity market effects. We also add the U.S. nificant at almost all quantiles considered, which indicates CBOE VIX index to take into account the volatility of the that Bitcoin cannot act as a hedge against the U.S. policy- U.S. stock market. Then, the augmented Equation (1) related economic uncertainty. This finding is not in line with becomes: Wu et al. (2019), who find that Bitcoin can act as a weak hedge against EPU during extreme bearish and bullish mar- QS τα = τα + τα PU + τ SD () () () () BTC 01 tt 2 ket conditions. To give a more comprehensive picture of the hedging +ατ Gold + ατ Oil + ατ VIX () () (() 34 tt 5 t ability of Bitcoin against U.S. economic uncertainty, we (3) + βτ UD + βτ UU () () () focus on the coefficient estimates of Equation (3) for the 11 ii ,, 01 ti,, iq09 . 0 it categorical economic uncertainty indices. Notably, the esti- +βτ DU U () (() ii ,, 20 q .95 it , mates provide more mixed and nuanced results than those involving the aggregate EPU index. In fact, Table 3 shows +βτ DU U + () () ii ,, 30 qi ., 99 tit βτ () that the parameter i,0 is positive and statistically sig- nificant generally at high Bitcoin return quantiles for some We run Equation (3) for the 5, 10-, 20-, 40-, 50-, 60-, 80-, uncertainty indices, including fiscal policy, taxes, national 90-, and 95-th quantile. Following the line of previous stud- security, and trade policy. These results imply that Bitcoin ies (Baur & Lucey, 2010, who study the gold and stock mar- is a strong hedge against these categorical economic uncer- ket indices; Bouri et al., 2017a, 2017c, who study Bitcoin tainties during bullish states in the Bitcoin market. These and conventional assets), especially Wu et al. (2019) who findings are comparable to those of Gozgor et al. (2019), study gold/Bitcoin and aggregate EPU, we conjecture that who find that trade policy uncertainty impacts Bitcoin Bitcoin is a strong (weak) hedge against categorical eco- returns during periods of regime change. However, when th nomic uncertainty at the quantile (under a given i τ the Bitcoin market is bearish or normal, Bitcoin generally Bitcoin market condition) if the parameter βτ is posi- () i,0 acts as a weak hedge against all categorical uncertainty tively significant (insignificant). Conversely, when βτ () i,0 indices. is negative and significant, Bitcoin is not a hedge against On the other hand, under bullish market condition, the categorical uncertainty . Moreover, the safe-haven prop- negative and significant parameter βτ () for EPU, govern- i,0 erty of Bitcoin against the uncertainty of category is tested ment spending, health care, and entitlement programs show based on the dummy variable parameters (during stress peri- 1 2 3 that Bitcoin does not play the role of hedging assets against ods). If the sum βτ ,, βτ and is pos- () () βτ () these categorical uncertainty indices. In addition, Bitcoin can ij ,, ij ij , ∑∑ ∑ j== 0 j 0 j=0 act as a weak hedge against monetary policy, especially itively significant (insignificant), Bitcoin is a strong (weak) under normal conditions but it can be a strong hedging asset against fiscal and taxes policy uncertainty under bullish mar- safe-haven against economic uncertainty for category i at ket condition. Besides, the insignificant parameter βτ for () i,0 90%, 95%, and 99% quantile respectively of this categorical regulation and sovereign debt shows that this digital cur- index. A negative value for a given sum indicates that rency fails to hedge these two categorical policy uncertain- Bitcoin is not a safe-haven against uncertainty. ties regardless of market conditions. Mokni et al. 7 Table 3. Estimation Results of the Bitcoin Hedging Parameter (β ) 0 Against Categorical Uncertainty. Bearish market Normal market Bullish market 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.114 0.133 −0.088 −0.042 −0.158 −0.040 −0.176*** −0.607*** −1.630*** (0.433) (0.276) (0.193) (0.214) (0.220) (0.254) (0.053) (0.291) (0.291) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.046 −0.126 −0.032 0.047 0.116 0.089 0.177 −0.007 −0.099 (0.175) (0.144) (0.139) (0.109) (0.112) (0.114) (0.120) (0.182) (0.165) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.161 0.178 0.071 0.031 0.037 0.066 −0.190 −0.505 1.022*** (0.246) (0.222) (0.205) (0.156) (0.162) (0.162) (0.374) (0.418) (0.180) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes 0.335 0.459** 0.134 0.075 0.092 0.076 −0.266 0.434 1.022** (0.248) (0.231) (0.213) (0.143) (0.144) (0.152) (0.315) (0.334) (0.393) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.067 0.079 0.007 0.064 0.015 0.000 −0.424 −0.592*** −0.424 (0.120) (0.106) (0.107) (0.119) (0.099) (0.095) (0.347) (0.121) (0.347) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.048 −0.086 0.068 0.000 −0.084 −0.037 −0.198 −0.327 −0.834*** (0.187) (0.158) (0.138) (0.122) (0.137) (0.137) (0.258) (0.231) (0.182) R 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.002 −0.056 −0.040 0.001 −0.012 0.047*** 0.313*** 0.960*** 0.945*** (0.250) (0.078) (0.073) (0.077) (0.082) (0.010) (0.127) (0.183) (0.273) 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 0.071 0.026 0.013 −0.054 −0.094 0.013 −0.108 −0.334 −0.783*** (0.155) (0.090) (0.095) (0.092) (0.099) (0.108) (0.147) (0.289) (0.175) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation −0.170 −0.132 −0.069 −0.002 0.023 0.035 −0.056 −0.176 −0.645 (0.269) (0.194) (0.133) (0.140) (0.143) (0.168) (0.210) (0.286) (0.421) 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation −0.039 −0.121 0.058 0.043 0.062 0.019 −0.120 −0.099 −0.211* (0.221) (0.107) (0.079) (0.063) (0.066) (0.071) (0.077) (0.078) (0.112) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.062* −0.034 −0.040 0.012 −0.024 0.027 0.065 0.078** 0.147*** (0.039) (0.042) (0.043) (0.047) (0.050) (0.054) (0.057) (0.035) (0.065) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.131 −0.036 −0.027 0.016 0.010 −0.010 −0.016 −0.141 −0.145 currency crises (0.260) (0.149) (0.122) (0.111) (0.116) (0.125) (0.134) (0.169) (0.177) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. Overall, it seems that Bitcoin is characterized by a strong acts as a hedging tool in the face of the inefficacity of tradi- ability to act as a hedging tool against some categorical tional assets to hedge portfolios against economic and finan- uncertainty, mainly when it is Bullish. This result could be cial uncertainty conditions (Guesmi et al., 2019; Symitsi and explained by the fact that during periods of price increases Chalvatzis, 2019). Corbet et al. (2018) show that cryptocur- for Bitcoin, investors take more long positions in this digital rencies are somewhat segmented from stock market shocks asset, which makes it more attractive during high uncertainty and dissociated from popular financial assets, which points (stress) periods (Bouri et al., 2017a, 2017c). In fact, Bitcoin to the hedging ability of Bitcoin against uncertainty. 8 SAGE Open Figure 2. Hedge coefficients of Bitcoin with 95% confidence bands. Furthermore, our empirical results show that the hedging Safe-haven Property Analysis ability of Bitcoin is limited to financial (fiscal and taxes), Here, we investigate the safe-haven ability of Bitcoin by national security, and trade uncertainty and not to regulatory focusing on the coefficients of the dummy parameters. uncertainty. This new finding could be explained by the fact Results are provided in Tables 4 to 6 at 90%, 95%, and 99% that the cryptocurrency markets generally operate away from quantiles, respectively. Figures 3 to 5 graphically present the government regulation systems. parameter estimations. Figure 2 presents a graphical illustration of the estimated Table 4 shows that, at the 90% quantile, the parameter parameters, showing the shape of the parameter against the quantile order. Different forms and patterns of the hedging sum βτ is positive and significant for the aggregate () ij , parameter are observed. The value of the parameter tends to j=0 increase with quantile order for financial policy uncertainty τ= 01 . EPU only at the low quantile of Bitcoin distribution ( ), (fiscal and taxes), national security, and trade uncertainty. indicating that Bitcoin can be a strong, safe haven against However, there is generally a decrease in the effect of the aggregate EPU during bearish market states. In addition, other categorical uncertainty indices on Bitcoin returns start- Bitcoin acts as a strong safe-haven against uncertainty related ing at positive values at lower quantiles and finishing at neg- to monetary policy and entitlement programs during the ative and insignificant values at high quantiles. This confirms same bearish market state of Bitcoin. Bitcoin’s ability to hedge uncertainty during Bitcoin’s bull Our above findings nicely complement the related lit- market periods, while this ability is weak during Bitcoin’s erature dealing with Bitcoin and various measures of bear markets. uncertainties (Bouri et al., 2017b), especially previous Mokni et al. 9 Table 4. Estimation Results of the Safe-haven Parameter Sum at 90% βτ (). i =0 Bearish market Normal market Bullish market τ 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.676 0.963** 0.429 0.361 0.331 0.271 0.424 0.971 0.319 (0.618) (0.478) (0.414) (0.432) (0.448) (0.493) (0.614) (0.801) (0.614) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.544* 0.384 0.234 0.120 0.153 0.041 −0.053 −0.396 −0.455* (0.279) (0.263) (0.274) (0.211) (0.224) (0.221) (0.258) (0.258) (0.237) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.621 0.603 0.402 0.220 0.211 0.228 0.597 0.791 2.907*** (0.416) (0.426) (0.361) (0.387) (0.433) (0.577) (1.412) (1.236) (0.823) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes 0.530 0.258 0.107 −0.112 0.154 −0.051 −0.101 0.851 1.061* (0.414) (0.462) (0.388) (0.334) (0.379) (0.378) (0.483) (0.639) (0.593) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.269 0.054 0.141 0.148 −0.088 −0.126 −0.266 −0.443** −0.266 (0.241) (0.231) (0.202) (0.223) (0.201) (0.214) (0.512) (0.221) (0.512) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.413 0.495 0.063 −0.096 −0.195 −0.255 0.091 −0.184 −0.705 (0.363) (0.424) (0.288) (0.279) (0.309) (0.288) (0.678) (0.550) (0.558) 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.293 −0.120 0.150 0.014 −0.054 −0.123 0.258 1.425*** 1.371*** (0.420) (0.664) (0.251) (0.259) (0.272) (0.290) (0.351) (0.471) (0.419) R 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 0.474* 0.220 0.183 0.116 0.015 0.015 −0.281 −0.173 −0.627** (0.255) (0.195) (0.213) (0.205) (0.221) (0.299) (0.288) (0.487) (0.271) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.032 −0.735 −0.747 −0.328 0.148 0.098 −0.250 −0.169 −0.448 (1.100) (0.863) (1.003) (0.506) (0.431) (0.425) (0.457) (0.504) (0.614) R 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation 0.044 −0.148 0.160 0.125 0.120 0.087 0.007 −0.071 0.107 (0.394) (0.390) (0.156) (0.142) (0.148) (0.156) (0.159) (0.158) (0.252) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.505*** −0.421 −0.290 −0.056 −0.093 −0.109 −0.166 0.429*** 0.681*** (0.043) (0.258) (0.267) (0.174) (0.146) (0.136) (0.122) (0.156) (0.145) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.100 0.045 −0.027 0.006 0.000 −0.014 0.067 −0.010 −0.023 currency crises (0.657) (0.647) (0.417) (0.314) (0.294) (0.292) (0.281) (0.413) (0.398) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. studies that consider the aggregate economic policy uncer- Conclusion tainty index (Cheng & Yen, 2019; Demir et al., 2018; Panagiotidis et al., 2019; Wang et al., 2019b; Wu et al., Previous studies consider the impact of U.S. aggregate eco- 2019) by providing a more detailed and nuanced analysis nomic policy uncertainty on Bitcoin returns to make a hedge of the hedging and safe-haven properties of Bitcoin and safe-haven inferences. However, they disregard the against the disaggregated measures of U.S. economic impact of categorical U.S. EPU data that covers disaggre- uncertainties. gated measures of uncertainties such as monetary, fiscal, 10 SAGE Open Table 5. Estimation Results of the Safe-haven Parameter Sum at 95% . βτ () i =0 Bearish market Normal market Bullish market 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.683 0.925 0.184 −0.111 −0.168 0.045 0.861 1.470 0.508 (0.730) (0.618) (0.550) (0.628) (0.644) (0.727) (0.859) (1.004) (0.996) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.328 0.121 0.104 0.035 −0.068 −0.024 −0.131 −0.265 −0.403 (0.311) (0.347) (0.340) (0.262) (0.283) (0.272) (0.315) (0.319) (0.273) R 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.674 0.235 0.400 0.433 0.340 0.171 0.391 0.375 2.062 (0.526) (0.759) (0.494) (0.565) (0.652) (0.802) (1.985) (1.858) (1.176) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.0641 0.1510 0.2856 Taxes 0.884* 0.793 0.548 0.332 0.249 0.588 0.455 1.142 1.968*** (0.518) (0.609) (0.577) (0.648) (0.605) (0.606) (0.618) (0.858) (0.676) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 0.468 0.264 0.161 0.069 0.032 0.014 −0.041 −0.069 −0.041 (0.290) (0.296) (0.271) (0.318) (0.307) (0.335) (0.593) (0.284) (0.593) 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.574 0.230 0.112 0.120 0.530 0.425 0.298 0.233 0.020 (0.445) (0.716) (0.407) (0.459) (0.465) (0.431) (0.946) (0.739) (0.721) R 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security −0.069 −0.248 −0.243 0.088 −0.041 −0.078 0.496 0.991 1.896*** (0.508) (0.915) (0.383) (0.404) (0.416) (0.428) (0.693) (0.729) (0.687) 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs −0.060 −0.249 −0.359 0.093 0.055 0.065 0.017 −0.238 −0.412 (0.370) (0.358) (0.385) (0.329) (0.336) (0.396) (0.377) (0.590) (0.357) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.465 0.253 0.182 0.115 0.145 0.073 0.014 −0.185 0.001 (1.429) (1.187) (1.405) (0.726) (0.589) (0.566) (0.603) (0.601) (0.746) 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.4997 Fin Regulation −0.016 −0.271 0.107 0.024 0.022 −0.005 −0.236 −0.338 −0.398 (0.612) (0.712) (0.210) (0.209) (0.227) (0.238) (0.242) (0.247) (0.357) R 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy −0.121*** −0.123 −0.211 −0.152 −0.172 0.065 0.024 0.600** 0.817*** (0.046) (0.359) (0.388) (0.268) (0.237) (0.255) (0.186) (0.295) (0.169) 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.100 0.045 −0.027 0.006 0.000 −0.014 0.067 −0.010 −0.023 currency crises (0.990) (1.064) (0.593) (0.476) (0.464) (0.491) (0.428) (0.533) (0.522) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) indicate the statistical significance at 10%, 5%, and 1% level, respectively. regulatory, trade, and political uncertainties. To address this normal market conditions. However, Bitcoin is a strong literature gap and account for the aggregation effect in the hedge and safe-haven against some categorical EPU, includ- U.S. EPU, we examine the impact of categorical U.S. EPU ing fiscal policy, taxes, national security, and trade policy indices on Bitcoin returns. Specifically, we employ quantile under bullish market conditions. regressions augmented with dummy variables to take into The policy implications concern the particularity of account various states of the Bitcoin market. Bitcoin as a digital assets class that is very useful for inves- The results indicate that Bitcoin is a weak hedge against tors, financial advisors, and risk managers making decisions all the uncertainty categories considered under bearish and involving the risk of specific measures of U.S. economic Mokni et al. 11 Table 6. Estimation Results of the Safe-haven Parameter Sum at 99% . βτ () i =0 Bearish market Normal market Bullish market τ 0.05 0.1 0.2 0.4 0.5 0.6 0.8 0.9 0.95 EPU 0.782 0.684 0.317 0.187 0.247 0.204 0.438 1.177 −0.080 (0.801) (0.966) (0.656) (0.764) (0.773) (0.881) (1.098) (1.217) (1.233) R 0.1019 0.0837 0.0558 0.0491 0.0475 0.0451 0.0406 0.0629 0.0759 Monetary policy 0.200 0.374 0.101 0.252 0.279 0.166 0.199 −0.059 −0.061 (0.419) (0.912) (1.013) (0.395) (0.422) (0.411) (0.428) (0.497) (0.379) 0.1087 0.0774 0.0845 0.0635 0.0466 0.0467 0.0508 0.0611 0.0883 Fiscal Policy 0.105 0.623 −0.065 −0.467 −0.430 0.562 0.054 −0.130 1.911 (0.829) (1.440) (0.899) (1.004) (1.143) (1.122) (2.240) (2.305) (1.310) 0.1237 0.0869 0.0652 0.0411 0.0433 0.0488 0.06407 0.1510 0.285664 Taxes −0.122 −0.167 −0.304 −0.434 −0.081 −0.210 −0.553 0.295 1.170 (0.937) (1.023) (1.175) (1.556) (0.803) (0.753) (0.723) (0.924) (0.816) 0.1249 0.0838 0.0612 0.0476 0.0546 0.0659 0.0735 0.1085 0.1571 Gov spending 1.110 1.113 1.452 1.456 1.533 1.337 0.732 0.054 0.732 (0.806) (0.816) (0.996) (1.184) (1.204) (1.209) (1.401) (1.064) (1.401) R 0.1379 0.1151 0.0942 0.0661 0.0644 0.0612 0.0805 0.1408 0.2919 Health care 0.192 0.647 0.396 0.143 −0.586 −0.251 −1.116 −1.151 −1.589 (0.728) (1.166) (0.764) (0.903) (0.891) (0.834) (1.183) (1.010) (1.159) 0.1035 0.0623 0.0552 0.0433 0.0536 0.0573 0.0688 0.1217 0.2183 Natl security 0.872 0.900 0.533 −0.187 −0.176 −0.313 1.899 3.636 2.705 (0.812) (1.118) (0.810) (0.642) (0.677) (0.654) (4.328) (3.566) (4.709) R 0.0646 0.0732 0.0422 0.0304 0.0313 0.0418 0.0456 0.0488 0.1579 Ent programs 1.922*** 1.742*** 1.622*** 0.869 0.836 0.629 0.197 0.413 0.186 (0.606) (0.674) (0.684) (0.649) (0.653) (0.648) (0.684) (0.807) (0.670) R 0.1163 0.0836 0.0613 0.0570 0.0631 0.0571 0.0748 0.1114 0.2427 Regulation 0.632 0.452 0.371 0.083 −0.073 −0.135 −0.377 −0.179 −0.578 (1.467) (1.247) (1.435) (0.831) (0.722) (0.698) (0.765) (0.734) (0.954) R 0.1077 0.0694 0.0462 0.0426 0.0359 0.0394 0.0461 0.0931 0.1417 Fin Regulation 0.917 0.649 0.293 0.090 −0.096 −0.161 −0.204 −0.264 −0.678 (0.966) (1.185) (0.350) (0.402) (0.438) (0.451) (0.426) (0.413) (0.541) 0.1245 0.0732 0.0344 0.0451 0.0354 0.0741 0.0124 0.1471 0.1874 Trade policy 0.514*** 0.570 0.700 0.460 0.322 −0.303 −0.621 0.751*** 1.035*** (0.169) (0.432) (0.543) (0.453) (0.450) (0.781) (0.592) (0.303) (0.323) R 0.2059 0.1940 0.1122 0.0541 0.0536 0.0429 0.0603 0.0937 0.1037 Sovereign debt, −0.196 −0.047 0.172 0.264 0.301 0.302 0.046 −0.247 0.009 currency crises (1.414) (1.594) (0.888) (0.854) (0.888) (1.224) (1.007) (1.167) (0.859) R 0.1322 0.0916 0.0536 0.0479 0.0578 0.0557 0.0323 0.0737 0.0967 Note. EPU = economic policy uncertainty. Numbers between parentheses denote the standard error. (*), (**), and (***) denote indicate the statistical significance at 10%, 5%, and 1% level, respectively. uncertainties. Furthermore, the findings could be useful to periods, investors can incorporate Bitcoin in their portfolios investors operating in conventional and cryptocurrency mar- for hedging purposes. However, considering Bitcoin in such kets. They indicate that investors should monitor develop- investment decisions would be most suitable when the ments in taxes, fiscal policy, national security, and trade Bitcoin market experiences a bullish state. policy in order to be better prepared for uncertainty and While our empirical analysis involves categorical data hedge their conventional portfolios based on Bitcoin. from the United States only, it would be interesting to extend Additionally, investors can benefit from our results by speci- the analysis by considering categorical EPU data from other fying more suitable portfolio design. During high uncertainty countries and regions such as China and Europe. This might 12 SAGE Open Figure 3. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 90% quantile of the U.S. categorical uncertainty indices. Figure 4. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 95% quantile of the U.S. categorical uncertainty indices. Mokni et al. 13 Figure 5. Safe-haven coefficients of Bitcoin with 95% confidence bands at the 99% quantile of the U.S. categorical uncertainty indices. be done when such disaggregated data becomes available. Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring eco- nomic policy uncertainty. The Quarterly Journal of Economics, Another potential path for future research could cover the 131(4), 1593–1636. role of other leading cryptocurrencies such as Ethereum and Baur, D. G., & Dimpfl, T. (2019). Price discovery in bitcoin spot or Ripple. futures? Journal of Futures Markets, 39(7), 803–817. Baur, D. G., Hong, K., & Lee, A. D. (2018). Bitcoin: Medium Declaration of Conflicting Interests of exchange or speculative assets? Journal of International The author(s) declared no potential conflicts of interest with Financial Markets, Institutions and Money, 54, 177–189. respect to the research, authorship, and/or publication of this Baur, D. G., & Lucey, B. M. (2010). Is gold a hedge or a safe article. haven? An analysis of stocks, bonds and gold. The Financial Review, 45, 217–229. Funding Bernanke, B. S. (1983). Irreversibility, uncertainty, and cyclical investment. The Quarterly Journal of Economics, 98(1), 85– The author(s) received no financial support for the research and/or authorship of this article. Bilgin, M. H., Gozgor, G., Lau, C. K. M., & Shing, X. (2018). The effects of uncertainty measures on the price of gold. ORCID iDs International Review of Financial Analysis, 58, 1–7. Bloom, N. (2009). The impact of uncertainty shocks. Econometrica, Khaled Mokni https://orcid.org/0000-0001-9817-6378 77(3), 623–685. Elie Bouri https://orcid.org/0000-0003-2628-5027 Bouri, E., Gil-Alana, L. A., Gupta, R., & Roubaud, D. (2019). Xuan Vinh Vo https://orcid.org/0000-0002-4725-6914 Modelling long memory volatility in the Bitcoin market: Evidence of persistence and structural breaks. International References Journal of Finance & Economics, 24(1), 412–426. Anceaume, E., Ludinard, R., Potop-Butucaru, M., & Tronel, F. Bouri, E., Gkillas, K., & Gupta, R. (2020a). Trade uncertainties and (2017). Bitcoin a distributed shared register. In International the hedging abilities of Bitcoin. Economic Notes, 49(3), Article Symposium on Stabilization, Safety, and Security of Distributed e12173. Systems (pp. 456–468). Springer. Bouri, E., Gupta, R., Tiwari, A., & Roubaud, D. (2017a). Does Arouri, M., Estay, C., Rault, C., & Roubaud, D. (2016). Economic Bitcoin hedge global uncertainty? Evidence from wavelet- policy uncertainty and stock markets: Long-run evidence from based quantile-in-quantile regressions. Finance Research the US. Finance Research Letters, 18, 136–141. Letters, 23, 87–95. 14 SAGE Open Bouri, E., Jalkh, N., Molnár, P., & Roubaud, D. (2017b). Bitcoin Panagiotidis, T., Stengos, T., & Vravosinos, O. (2019). The effects for energy commodities before and after the December 2013 of markets, uncertainty and search intensity on Bitcoin returns. crash: Diversifier, hedge or safe haven? Applied Economics, International Review of Financial Analysis, 63, 220–242. 49(50), 5063–5073. Raza, S. A., Shah, N., & Shahbaz, M. (2018). Does economic policy Bouri, E., Molnár, P., Azzi, G., Roubaud, D., & Hagfors, L. I. uncertainty influence gold prices? Evidence from a nonparamet- (2017c). On the hedge and safe haven properties of Bitcoin: ric causality-in-quantiles approach. Resources Policy, 57, 61–68. Is it really more than a diversifier? Finance Research Letters, Selmi, R., Mensi, W., Hammoudeh, S., & Bouoiyour, J. (2018). Is 20, 192–198. Bitcoin a hedge, a safe haven or a diversifier for oil price move- Bouri, E., Shahzad, S. J. H., Roubaud, D., Kristoufek, L., & Lucey, ments? A comparison with gold. Energy Economics, 74, 787–801. B. (2020b). Bitcoin, gold, and commodities as safe havens for Shahzad, S. J. H., Bouri, E., Roubaud, D., & Kristoufek, L. (2020). stocks: New insight through wavelet analysis. The Quarterly Safe haven, hedge and diversification for G7 stock markets: Review of Economics and Finance, 77, 156–164. Gold versus Bitcoin. Economic Modelling, 87, 212–224. Cheng, H. P., & Yen, K. C. (2019). The relationship between the Shahzad, S. J. H., Bouri, E., Roubaud, D., Kristoufek, L., & Lucey, economic policy uncertainty and the cryptocurrency market. B. (2019). Is Bitcoin a better safe-haven investment than gold Finance Research Letters, 35, 101308. https://doi.org/10.1016/j. and commodities? International Review of Financial Analysis, frl.2019.101308 63, 322–330. Chiang, T. C. (2020). Economic policy uncertainty and stock Symitsi, E., & Chalvatzis, K. J. (2019). The economic value of returns—Evidence from the Japanese market. Quantitative Bitcoin: A portfolio analysis of currencies, gold, oil and stocks. Finance and Economics, 4(3), 430–458. Research in International Business and Finance, 48, 97–110. Corbet, S., Meegan, A., Larkin, C., Lucey, B., & Yarovaya, L. (2018). Teomete Yalabık, F., & Yalabık, İ. (2019). Anonymous Bitcoin v Exploring the dynamic relationships between cryptocurrencies enforcement law. International Review of Law, Computers & and other financial assets. Economics Letters, 165, 28–34. Technology, 33(1), 34–52. Demir, E., Gozgor, G., Lau, C. K. M., & Vigne, S. A. (2018). Does Urquhart, A., & Zhang, H. (2019). Is Bitcoin a hedge or safe haven economic policy uncertainty predict the Bitcoin returns? An for currencies? An intraday analysis. International Review of empirical investigation. Finance Research Letters, 26, 145– Financial Analysis, 63, 49–57. 149. Wang, G.-J., Ma, X., & Wu, H. (2020). Are stablecoins truly Fang, L., Bouri, E., Gupta, R., & Roubaud, D. (2019). Does global diversifiers, hedges, or safe havens against traditional crypto- economic uncertainty matter for the volatility and hedging currencies as their name suggests? Research in International effectiveness of Bitcoin? International Review of Financial Business and Finance, 54, 101225. https://doi.org/10.1016/j. Analysis, 61, 29–36. ribaf.2020.101225 Gozgor, G., Tiwari, A. K., Demir, E., & Akron, S. (2019). The rela- Wang, G.-J, Tang, Y., Xie, C., & Chen, S. (2019a). Is Bitcoin a safe tionship between Bitcoin returns and trade policy uncertainty. haven or a hedging asset? Evidence from China. Journal of Finance Research Letters, 29, 75–82. Management Science and Engineering, 4(3), 173–188. Guesmi, K., Saadi, S., Abid, I., & Ftiti, Z. (2019). Portfolio Wang, G.-J., Xie, C., Wen, D., & Zhao, L. (2019b). When Bitcoin diversification with virtual currency: Evidence from bitcoin. meets economic policy uncertainty (EPU): Measuring risk International Review of Financial Analysis, 63, 431–437. spillover effect from EPU to Bitcoin. Finance Research Klein, T., Thu, H. P., & Walther, T. (2018). Bitcoin is not the New Letters, 31, 489–497. Gold —A comparison of volatility, correlation, and portfolio Wu, S., Tong, M., Yang, Z., & Derbali, A. (2019). Does gold or performance. International Review of Financial Analysis, 59, bitcoin hedge economic policy uncertainty? Finance Research 105–116. Letters, 31, 171–178.

Journal

SAGE Open – SAGE

Published: May 22, 2021

Keywords: Bitcoin; EPU; categorical economic uncertainty; hedge; safe-haven; quantile

There are no references for this article.