Professional Documents
Culture Documents
Comments Welcome!
Muhammad A. Cheema
School of Accounting, Finance and Economics
University of Waikato
Tauranga, New Zealand
Email: muhammad.cheema@waikato.ac.nz
Robert Faff
UQ Business School
The University of Queensland
Brisbane, Australia
Email: r.faff@business.uq.edu.au
Kenneth R. Szulczyk
School of Economics and Management
Xiamen University Malaysia Campus
Sepang, Selangor, Malaysia
Email: kenneth.szulczyk@xmu.edu.my
2
Abstract
This paper compares the performance of safe haven assets during two stressful stock market
regimes – the 2008 Global Financial Crisis (GFC) and COVID-19 pandemic. Our findings
show that the character of safe haven assets has changed between the crises. The traditional
choice, gold, serves as a safe haven during the GFC but loses its safe haven status during
COVID. Silver does not serve as a safe haven during either crisis. The other traditional safe
haven assets - the US dollar, Swiss Franc and Treasuries serve as weak safe havens during both
crises. AAA-grade corporate bonds emerge as a weak safe haven during COVID, suggesting
that large cash reserves held by the US firms make them insulated from the crisis. The largest
the largest asset-backed cryptocurrency, Tether emerges as a safe haven across the sample
1. Introduction
pandemic within three months – has caused severe damage to human lives and the global
economy. The stock markets around the world have plummeted to their lowest levels since the
2008 Global Financial Crisis (GFC) (BBC, 2020). Furthermore, COVID negatively impacted
stock markets more than any previous infectious disease outbreak, including the 1918 Spanish
Unforeseen and unanticipated events such as the 1987 stock market crash and the 2008
Global Financial crisis (GFC), trigger flight to quality episodes where investors transfer their
investments from risky to safe assets (e.g. Adrian et al., 2019; Baele et al., 2020; Caballero and
Krishnamurthy, 2008). It is well documented in the literature that gold (e.g. Baur and Lucey,
2010; Hillier et al., 2006; Pullen et al., 2014); US Treasury bills and bonds (e.g. Chan et al.,
2011; Fleming et al., 1998; Hartmann et al., 2004; Noeth and Sengupta, 2010); and currencies
such as the US dollar and Swiss Franc (e.g. Grisse and Nitschka, 2015; Kaul and Sapp, 2006;
Ranaldo and Söderlind, 2010) act as safe havens during periods of stock market turmoil.
However, Baur and Lucey (2010) and Chan et al. (2011) suggest that Treasury bonds possess
better properties than gold as a safe haven during stock market crises. Moreover, Brunnermeier
et al. (2020) propose US Treasuries as a global safe asset in times of the crisis.
Several recent studies argue that cryptocurrencies act as a safe haven during market turmoil
(e.g. Stensås et al., 2019; Urquhart and Zhang, 2019); however, other studies view
cryptocurrencies as a risky asset instead of a safe haven (e.g. Klein et al., 2018; Smales, 2019).
Therefore, Baur and Hoang (2020) suggest using asset-backed cryptocurrencies (stablecoins),
such as Tether, as a safe haven against Bitcoin during extreme market movements. Most
4
recently, Conlon and McGee (2020) and Kristoufek (2020) find that Bitcoin is not a safe haven
during COVID.1
whether the traditional safe assets such as gold, US Treasury bills and bonds, US dollar, and
Swiss Franc provide protection from stock market losses given the unique nature of this twin
safe haven assets during the GFC versus the COVID-19 pandemic. For instance, we ask the
question – do traditional assets that were safe havens during the GFC (e.g. Baur and
McDermott, 2010; Low et al., 2016) maintain their safe haven status during the COVID-19
traditional cryptocurrency, Bitcoin, and the largest stablecoin, Tether, serve as a safe haven
against stock market losses. Finally, COVID also provides the opportunity to examine whether
the high rated US corporate bonds (AAA-grade) can serve as a safe haven since the largest
multinational firms are hoarding almost three times more cash reserves in recent years relative
to the 2001 levels (e.g. Faulkender et al., 2019) that could insulate them from the stock market
Another motivation of our study is to provide conclusive evidence on the efficacy of safe
haven assets during stock market crises, which differs from the existing literature in three ways.
First, the existing studies on safe haven assets generally examine one or two safe haven assets
(e.g. Baur and McDermott, 2010); whereas, we analyse five different safe haven assets to
provide robust evidence on the efficacy of the safe haven assets. Second, the existing studies
1
Bitcoin is the first and largest cryptocurrency; whereas, Tether is the first and largest asset-backed
cryptocurrency (stablecoin). Stablecoins are cryptocurrencies that are pegged to other stable assets such as gold
and the traditional currencies. According to the data obtained from coinmarketcap.com on June 27, 2020, the
market capitalisation of Bitcoin and Tether is over $167 billion and $9 billion, respectively. Bitcoin tokens are
not backed by any physical commodity or precious metals; whereas, Tether tokens are 100% backed by liquid
reserves, including the US dollar and other assets that make Tether a stable asset. For details, please refer to Lipton
et al. (2020) and Tether’s Limited website, https://tether.to/
5
use two dimensions criteria (sign and statistical significance) to differentiate between weak and
strong safe haven, whereas we use three dimensions criteria (sign, statistical significance and
economic magnitude) to distinguish between weak and strong safe havens We explain in detail
the three dimensions criteria in Section 3. Finally, the existing studies (e.g. Baur and
McDermott, 2010; Low et al., 2016) report the total effect of the stock market crisis on the safe
haven assets with incremental t-statistics which might provide inaccurate conclusions; hence,
we provide the total effect of the stock market crisis on the safe haven assets with the respective
A growing number of studies examine the impact of COVID-19 on the financial markets
and financial assets (e.g. Al-Awadhi et al., 2020; Alfaro et al., 2020; Baker et al., 2020;
Kristoufek, 2020; Ramelli and Wagner, 2020; Zhang et al., 2020). Further, several working
papers and recent articles also focus on the safe haven assets during COVID and provide
contrasting results (Conlon et al., 2020; Conlon and McGee, 2020; Corbet et al., 2020a; Corbet
et al., 2020b; Ji et al., 2020). For instance, Conlon et al. (2020) show that a portfolio consisting
of Tether and a stock market index provides downside risk benefits to investors across all the
countries in their sample (namely, the US; the UK, Italy, Spain and China), whereas, a portfolio
consisting of Bitcoin and a stock market index only reduces the downside risk for Chinese
investors. In contrast, Corbet et al. (2020a) suggest that large cryptocurrencies acted as a store
of value during COVID. Furthermore, Ji et al. (2020) compare the performance of safe haven
assets between August–December 2019 and December 2019–March 2020, and find that gold
and soybean futures remain as safe haven assets during COVID. Nonetheless, no study has
compared the performance of safe haven assets between the GFC and COVID.
In this paper, we perform a coordinated comparative examination of the safe haven efficacy
of: (a) precious metals (gold and silver); (b) currencies (US dollar and Swiss Franc); (c) US
Treasuries (T-bills and T-bonds); (d) US corporate bonds (AAA-grade); and (e)
6
cryptocurrencies (Bitcoin and Tether) from stock market losses during the GFC and COVID.
We select the stock markets of the ten largest economies; namely, the US, China, Japan,
Germany, the UK, France, India, Italy, Brazil and Canada since investors prefer to invest in
these markets. We estimate a GJR-GARCH regression since it accounts for the asymmetric
effects where the stock market returns exhibit higher volatility to bad news than to good news.
Our analysis shows that gold serves as a weak safe haven across the sample countries during
the GFC. However, notably, gold loses its safe haven status during COVID since its price has
somewhat moved in line with the stock markets. The obvious question is, why? We suggest
that gold loses its safe haven status because investors might have lost trust in gold as a stable
asset after the precious metal lost 45% of its USD value between 2011 to 2015. Somewhat in
contrast, silver does not function as a safe haven during either crisis. The US dollar, Swiss
Franc and both Treasuries, T-bills and T-bonds, act as weak safe havens during both crises. It
is interesting to note that the US dollar and Treasuries have not lost their status as safe havens
during COVID even though the US has suffered from the highest death and infection rates in
the world, and consequently reaffirms local and global investors trust in the US economy. The
high rated US corporate bonds (AAA-grade) serve as a weak safe haven during COVID,
suggesting that large cash reserves held by the US firms make them insulated from the crisis.
The largest cryptocurrency, Bitcoin has proven to be a highly speculative investment during
COVID. However, the largest stablecoin, Tether emerges as a weak safe haven during COVID
This study makes three important contributions to the literature. First, by comparing the
performance of the traditional safe haven assets during the GFC versus COVID, we uncover
new evidence that the traditional safe haven assets, gold and silver are not reliable protectors
of investor wealth in all stressful markets or settings. Second, we show that investors from both
developed and emerging markets make similar choices about safe haven assets during both
7
crises. Third, we extend the existing literature on global safe assets (e.g. Brunnermeier et al.,
2020) and propose that the Swiss Franc and Tether might also act as global safe assets, along
The remainder of the paper is organised as follows. Section 2 describes the research
methods and gives some preliminary analysis, and Section 3 presents and discusses the main
results. Section 4 offers a potential explanation of why gold is not a safe haven during the
The analysis includes stock market indices of the ten largest economies in the world,
namely, S&P500 US index, SSE composite index China, NIKKEI 225 Index Japan, MSCI
Germany Index, FTSE100 Index UK, CAC 40 Index France, NIFTY 500 Index India, FTSE
MIB Index Italy, MSCI Brazil Index, and TSX composite index Canada. All variables are
denominated in US dollars, allowing a direct and fair comparison between stock market indices
Potential safe-haven assets include precious metals (gold and silver); currencies (US Dollar
Index and Swiss Franc Index); Treasuries (S&P US Treasury bills index (T-bills) and S&P US
Treasury bonds index (T-bonds)); Corporate bonds (S&P 500 AAA-grade bonds index) and
cryptocurrencies (Bitcoin and Tether). US dollar index and Swiss Franc index represents the
value of the US dollar and Swiss Franc relative to a basket of foreign currencies, respectively.
DataStream International provides all data except data for the Swiss Franc index and the
cryptocurrencies. The data of Swiss Franc index is collected from the online database of Swiss
National Bank, while coinmarketcap.com furnishes the data for Bitcoin and Tether. The sample
period for all the assets except cryptocurrencies starts December 31, 2003; whereas the sample
period for cryptocurrencies starts September 17, 2014. We restrict the start date to December
8
31, 2003, since the aim of this study is to examine the role of safe-haven assets during the 2008
GFC and COVID-19 pandemic. The sample period for all the assets ends June 26, 2020.
Following the literature (e.g. Baur and McDermott, 2010), we estimate the model,
2 2
𝜎𝑡2 = 𝜔 + (𝛼+𝛾𝐼𝑡−1 )𝜀𝑡−1 + 𝛽𝜎𝑡−1 (2)
0 if 𝑟𝑡−1 ≥ 𝜇
where 𝐼𝑡−1 : = {
1 if 𝑟𝑡−1 < 𝜇
where 𝑅𝐴 𝑖 represents the log return of each given safe-haven asset i. 𝑅𝑆𝑗 denotes the daily log
returns in US dollars of a stock market index j, with j equal to a given one of the ten countries
in our sample. GFC is a dummy variable, which takes the value one from the designated start
date (explained shortly) and the subsequent 20 trading days of the GFC, and zero otherwise.
The dummy variable, COVID, is similarly constructed to the GFC variable. The residual term
Equation (2). Maximum log likelihood jointly estimates the parameters in (1) and (2). The
GJR-GARCH model accounts for the asymmetric effects where the stock market returns
Following the literature (e.g. Baur and McDermott, 2010), we assume that the adverse
effect of a stock market crisis occurs in the first 20 trading days since the start of the crisis.
Figure 1 shows the stock market crises for both the GFC and COVID.2 It is evident from Figure
1 that the GFC stock market crisis intensified in September 2008 with the collapse of Lehman
Brothers; whereas, the stock market crisis from COVID intensified in February 2020.
2
As a robustness check, we have identified the crisis period for the subsequent 30 and 40 trading days since
the start of the crisis, and find similar results. These results are available from the authors upon request.
9
Therefore, we define the start date for GFC on September 12, 2008, and COVID on February
20, 2020.3
The interpretation of Equations (1) – (2) to see whether asset i serves as a safe haven during
the GFC and COVID, is as follows. Parameter b1 is the safe-haven asset’s baseline “hedge”
(i.e. “normal” times, excluding GFC and COVID) beta with respect to the market in question.
Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative
and not economically “small” (zero or close to zero). Parameter b2 (b3) is the incremental safe-
haven asset beta for the GFC (COVID) and, therefore, the sum of the two parameters, b1 + b2
(b1 + b3), is the total safe-haven asset beta for the GFC (COVID). If the sum, b1 + b2, is negative
and not economically “small” (zero or close to zero), then asset i serves as a strong (weak) safe
haven from stock market losses during the GFC. Similarly, if the sum, b1 + b3, is negative and
not economically “small” (zero or close to zero), then asset i serves as a strong (weak) safe
Panel A of Table 1 summarises the descriptive statistics of the daily log-returns of all assets
in our study. The average returns (mean) of the safe haven assets except Bitcoin varies between
0.001% to 0.033% per day, while the average returns of Bitcoin are 0.173% per day. The T-
bills shows the lowest standard deviation, whereas Bitcoin, silver and gold show the highest
standard deviation. Furthermore, the negative skewness and high excess kurtosis of Bitcoin,
silver, AAA-grade bonds and gold imply a significant crash risk that counters their
effectiveness as a safe haven asset. The other safe haven assets show positive skewness and
high excess kurtosis that indicates the possibility of having extreme positive returns instead of
3
Low et al. (2016) use September 12, 2008 as a start date of the 2008 GFC. The 2020 stock market crash
started in late February 2020 from the uncertainty and threat of COVID-19 (e.g. Baker et al., 2020).
4
We elaborate how we empirically operationalize the “weak” vs. “strong” safe-haven interpretations later,
in Section 3.
10
extreme negative returns. The descriptive statistics suggest that Bitcoin, silver, AAA-grade
bonds, and gold possess characteristics of risky assets rather than safe haven assets.
The average daily returns of stock market indices range between -0.009% (Italy) to 0.03%
(India) per day. The standard deviation for each of the stock market indices is higher than all
the safe-haven assets except Bitcoin, and silver. Furthermore, stock market indices of all
countries except France exhibit negative skewness and high excess kurtosis, which indicate a
significant crash risk. In sum, the descriptive statistics in Panel A suggest that the US
Treasuries, US dollar, Swiss Franc and Tether could act as better safe havens than Bitcoin, gold
and silver.
Panel B of Table 1 shows the correlations between the assets in our study. As expected, the
correlation between gold and silver is positively correlated (0.66) and indicates that precious
metals move in tandem. The correlation between gold and the US dollar is negatively correlated
(-0.34) and indicates that these assets move in the opposite direction; thus, logically both assets
cannot act as safe havens at the same time. The correlations between other candidate safe haven
assets are generally not too distant from zero (with the exception between Treasuries and bonds
), indicating that these assets do not have a tendency to move either in the same or in the
opposite direction. Returns on the stock market indices for all ten countries are positively
correlated to each other, with strong positive correlations between the US and Europe, and
In this section, we examine the performance of safe haven assets during days of extreme
stock market losses in the S&P500, during the GFC and COVID. We use the S&P 500 stock
market index since it is the proxy of the largest economy in the world, the US. Nonetheless,
11
we find similar results for the stock markets of the other nine countries as well.5 We expect
assets to earn positive or, at worst, close to zero returns on the days of large stock market losses
Panel A of Table 2 reports the results of safe-haven assets on the ten days of the largest
losses in the S&P 500 during the period of the GFC from September 12, 2008, to June 30,
2009. The results show that gold returns are positive for six of the 10 days; silver shows positive
returns for only three days, the Swiss Franc for five days AAA-grade bonds for five days, and
the remaining safe haven assets, Treasuries and the US dollar, are positive for at least seven
out of ten days. These results imply that, with the exception of silver, the chosen candidate
assets generally exhibit the characteristics of a safe haven during days of large stock market
Panel B of Table 2 reports a counterpart analysis for candidate safe-haven assets across the
ten days of largest losses in the S&P 500 during the COVID-19 pandemic period, covering
February 20, 2020, to June 26, 2020, our current sample end date. The results show that gold
returns generally move in tandem with the ten extreme stock market losses in the S&P 500
during the COVID-19 pandemic, with six negative gold returns. For instance, gold lost 4.90%
of its value on March 12, 2020, when the S&P500 index incurred a 10% loss. Silver also moved
in tandem with extreme stock market losses during COVID, with seven out of 10 negative
silver returns. Four out of the ten US dollar returns were negative, but only two Swiss Franc
returns were negative on the days of the largest 10 losses in the S&P500. Notably, the T-bills
recorded all positive returns, while the T-bonds recorded seven positive returns. Seven out of
ten AAA-grade bonds returns were positive on the days of the largest 10 losses in the S&P500.
Bitcoin and Tether have five and seven negative returns, respectively, but the magnitude of
5
We do not report the results of the other nine countries for the sake of brevity. However, those results are
available upon request from the authors.
12
Bitcoin’s negative returns is much larger than Tether’s negative returns. For example, Bitcoin
dropped in value by 46.5% on March 12, 2020, while Tether recorded the maximum loss of
just 1.07% on March 9, 2020. In sum, the results in Panel B imply that gold, silver and Bitcoin
fail to protect the wealth of investors on those days when they needed it the most.
In this section, we examine the relationship between safe haven assets and stock market
returns using the regression model in Equations (1) and (2). Based on the preliminary analysis
shown in Section 3.2, we expect gold to act as a safe-haven asset during the GFC but not during
COVID. Furthermore, we expect Treasuries and currencies to act as safe haven assets during
both crises. Finally, while Tether might act as a safe haven during COVID; we do not expect
Bitcoin to act as a safe haven asset since it can lose extreme value during days of extreme stock
market losses.
Tables 3, 4, 5, 6, and 7 present the estimation results for precious metals, currencies,
Treasuries, corporate bonds and cryptocurrencies, respectively. The tables include the
parameter estimates of b0 (constant), b1 (hedge), the incremental GFC effect (b2), and the
incremental COVID effect (b3). The total GFC effect is the sum of b1 and b2, while the total
COVID effect is the sum of b1 and b3. We empirically operationalise the “weak” vs. “strong”
deemed a strong (weak) hedge for stock market j if the parameter b1 is negative, significant and
economically “small”). We use the cutoff of 0.1 to assess whether an estimated coefficient is
economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed
to be economically small. Parameter b2 is the incremental safe-haven asset beta for the GFC,
with the incremental t-statistics in the parenthesis. Further, the sum of the two parameters, b1
+ b2, is the total safe-haven asset beta for the GFC, with the respective t-statistics of the total
13
effect in the parenthesis. If the sum, b1 + b2, is negative, statistically significant and not
“small”), then asset i serves as a strong (weak) safe haven from stock market losses during the
Starting with gold, Panel A of Table 3 shows the parameter estimate, b1 is positive,
statistically significant and economically small for eight countries, and insignificant for the US,
showing that it effectively acts as a weak hedge. Indeed, the baseline gold betas are all of very
low magnitude, mostly in the range 0.007 to 0.1 (except for Canada, where the base gold beta
is around 0.19). These results are generally consistent with Low et al. (2016).
minimum, gold serves as an improved safe haven prospect across our sample countries during
the GFC. Indeed, the improvement in gold as a safe haven is significant for five of them (the
US; China; India; Brazil and Canada), in which the incremental GFC betas are negative,
significant and not economically small (above -0.2). Furthermore, the total safe-haven gold
GFC betas (i.e. sum of b1 + b2) are negative, significant and not economically small for the US
and Brazil – indicating a stronger GFC safe haven for gold in these settings. For the other eight
countries, the total safe-haven gold GFC betas are insignificant showing them to be weak safe
havens. Therefore, across all sample countries we see some level of safe haven performance
(mostly weak) for gold during the GFC. These findings are generally consistent with the
As already tentatively signalled in the preliminary results in Table 2, gold fails to act as a
COVID safe haven against the stock market losses from all countries, since the total safe-haven
betas (i.e. sum of b1 + b3) across the sample countries are positive, statistically significant and
not economically small. Moreover, all the incremental COVID betas except Brazil and Canada
14
are also positive and statistically significant (at the 5% level). It is particularly notable that the
total gold beta for China during COVID is relatively large at 0.73, and even for Japan it is quite
distant from zero, at 0.45. Clearly, uncharacteristically gold prices moved somewhat (and for
investors uncomfortably) in line with the stock markets in these two jurisdictions during
COVID. As such, there is no pretence at all that gold is a safe haven in China or in Japan,
during these very recent troubling times. While it is also true that even for the US, Italy, Brazil
or Canada, gold does not serve a safe haven role during COVID, their total gold betas (in the
range of 0.10 to 0.18) are not too different from zero. As such, for these countries gold is still
Panel B of Table 3 shows the counterpart analysis for silver. We see that all hedge
coefficients, b1, are positive and significant, ranging from the smallest estimate of 0.09 for
China, through to the highest of 0.56 for Canada. As such, these results indicate very strongly
that silver does not act as a hedge for any of our sample countries, consistent with the findings
of Low et al. (2016). Furthermore, silver serves as a weak safe haven only for the US, China
and Brazil during the GFC, in which the total GFC betas are insignificant. However, the
incremental silver GFC betas are negative and significant (at the 5% level), though small for
the US and Brazil, which indicate some improvement in silver as a safe haven for these two
Much more telling are the counterpart results for COVID. The total safe-haven COVID
betas are positive, significant at the 5% level and not small for eight of our sample countries
which indicate that silver does not serve as a safe haven (not even weakly) against losses from
stock markets during COVID. Only for the US and Italy, do the total safe-haven COVID betas
suggest a weak-safe haven character for silver (though positive and significant, these silver
betas are economically small). Of particular note is that for China, Japan and India, the total
silver safe-haven betas are all greater than unity (e.g. China shows an estimate of 1.68). Hence,
15
during COVID, investment in silver represents a high systematic risk in these settings, let alone
3.2 Currencies
Panel A of Table 4 reports results for the US dollar as a potential safe haven asset. 6 The
table shows that hedge coefficients, b1, are (with only one exception) negative and significant
though not sizeable, indicating that the US dollar serves as a weak hedge against the stock
market indices, with China (insignificant) the weakest case. Furthermore, while all the total
safe-haven GFC betas (i.e. sum of b1 + b2) except Brazil are negative and statistically
significant (at the 5% level), their small magnitudes convey that the US dollar only serves as a
weak safe haven for stock markets in these countries during the GFC. The US dollar also serves
as a weak safe haven for Brazil during the GFC, in which the total GFC beta is insignificant.
Interestingly, the safe haven efficacy of the US dollar is only negligibly weakened during
COVID. More specifically, the US dollar generally serves only as a weak safe haven during
the pandemic since: (a) the only negative and significant case, India (COVID-beta of -0.0395)
is negligibly close to zero; (b) the only positive and significant COVID-betas, for Brazil and
Canada, are also economically indistinguishable from zero. Indeed, all the incremental COVID
betas except China, Japan and India; though positive and statistically significant (at the 5%
level), are economically small thereby suggesting that there has been only negligible change in
the baseline safe haven (b1) relationship between the US dollar and these countries during
COVID.
Panel B of Table 4 reports results for the Swiss Franc as a potential safe haven asset. At a
very general level, collectively taking into account the sign, statistical significance and
magnitude of the estimated coefficients, the Swiss Franc results have a similar flavour as for
6
We do not examine the relationship between the US stock market and the US dollar since it is a domestic
currency for US investors.
16
the US dollar. The table shows that the parameter estimate, b1 is negative and statistically
significant for eight countries, but small in magnitude; and insignificant estimates for the
remaining two (the US and China), thereby revealing a weak hedge in all cases. Further, while
the total safe-haven Franc GFC betas (i.e. sum of b1 + b2) are negative and significant (at the
5% level) for all the countries except the US and India (which are insignificant), the small
magnitudes suggest only weak GFC safe havens for the Swiss Franc in these settings.
Moreover, while the incremental Franc GFC betas are negative and significant for five
countries (China; Japan; Germany; France and Italy), the small economic size of these
coefficients suggest that the Swiss Franc only offers a mild improvement in safe haven status
The efficacy of Swiss Franc as a weak safe haven continues during COVID. The total safe-
haven Franc COVID betas (i.e. sum of b1 + b3) are negative, significant (at the 5% level) but
again small for all our sample countries. Moreover, while all the incremental COVID betas
(except Japan, the UK and Canada) are also negative and statistically significant (at the 5%
level), the small magnitudes suggest only modest improvement, at best, (no incremental
COVID beta is smaller than -0.1) in the Swiss Franc as a safe haven during COVID.
3.3 Treasuries
Panel A of Table 5 reports our results for T-bills as a potential safe haven asset. The table
shows that hedge coefficients, b1, are negative, significant (insignificant) but very small for the
US; Germany; the UK; France and Canda (China; Japan; India; Italy and Brazil), indicating
that the T-bill serve as a very weak hedge across the sample countries. While all the total safe-
haven T-bills GFC betas (i.e. sum of b1 + b2) except Japan are negative and statistically
significant (at the 5% level), their minuscule size indicates that T-bills serves as a very weak
safe haven in these settings during the GFC. The T-bills continues to serve as a very weak safe
17
haven during COVID – the total safe-haven T-bills COVID betas are all negative, very small
Panel B of Table 5 reports results for T-bond as a potential safe haven asset. Similar, to T-
bills, Panel B shows that T-bonds serve as a weak hedge. Furthermore, the total safe-haven T-
bonds GFC and COVID betas are negative (except positive and small GFC beta for Japan) and
significant (at the 5% level), despite being of negligible magnitude, for all our sample countries,
thereby again giving a weak safe haven during both crises. That said, if anything, based on
their larger magnitudes the safe haven status of T-bonds is marginally better than the T-bills.
Table 6 reports results for AAA-grade corporate bonds as a potential safe haven asset. The
hedge coefficients, b1, are negative, significant but small for all the countries (except China,
Japan, and India) indicating that AAA-bonds acts as a weak hedge in these settings.
Nonetheless, the hedge coefficients, b1 are either insignificant or economically small for China,
Japan, and India, showing that AAA-bonds also acts as a weak hedge in these three countries.
However, the total safe-haven AAA-bonds GFC betas are positive, statistically significant (at
the 5% level), and economically not small for China, Japan, Germany, the UK, India and Italy
(betas are above +0.10), showing that AAA-bonds do not serve even as a weak safe haven in
these settings during the GFC. Interestingly, the total safe-haven AAA-bonds COVID betas for
all the countries are either insignificant or economically small (either statistically positive or
negative), showing it to be a generally very weak safe haven during COVID, suggesting that
significant increase in cash reserves held by the largest US firms in recent years make the
investment in high rated bonds less risky from the perspective of investors.
18
3.5 Cryptocurrencies
Table 7 reports results for cryptocurrencies – Bitcoin and Tether – as potential safe haven
assets, noting that given their relatively recent advent, our sample only begins post GFC. Panel
A reports results for Bitcoin. The panel shows that hedge coefficients, b1, are insignificant for
nine countries (positive, significant and small for China), thereby indicating that Bitcoin
generally seems to serve as a weak hedge during normal times. However, it is startling to see
that all the total safe-haven Bitcoin COVID betas are positive, statistically significant (at the
1% level) and very large in magnitude. As such, Bitcoin clearly does not serve as a safe haven
during COVID. Moreover, and most importantly, the total Bitcoin “safe-haven” COVID betas
are mostly substantially greater than unity (e.g. China shows an amazingly high estimate of
4.82). Therefore, during COVID, investment in Bitcoin actually proves to be a highly risky
strategy. As such, the investment challenge posed by COVID, sees Bitcoin fail it’s first real
safe haven test miserably – it has proven to be a highly speculative asset during COVID.
Panel B of Table 7 reports counterpart results for Tether as a potential safe haven asset. The
table shows that hedge coefficients, b1, are insignificant for nine countries (all, but except
Germany), thereby indicating that Tether serves as a weak hedge. But that is where the
similarity with Bitcoin abruptly ends. Specifically, the total safe-haven Tether COVID betas
are negative and significant (at the 5% level), though typically small in magnitude for all our
sample countries. This indicates that Tether generally serves as a weak safe haven against
losses from stock markets during COVID. However, it is notable that three countries, China,
Japan and India stand out as offering the prospect of a stronger COVID-safe haven via Tether
3.6 Discussion
Table 8 presents a broad summary of the performance of various potential safe-haven assets
during the GFC and COVID pandemic financial crises, as analysed and reported in Tables 3-
First, scanning down Column A1-A4 relating to the GFC crisis, we observe that, at best,
only weak safe haven evidence is on display – primarily for gold, but also to a lesser extent for
the US dollar, the Swiss Franc, T-bills and for T-bonds. Silver is a standout failure in this
regard, with most of its GFC safe-haven betas in the moderately risky range, around +0.4.
AAA-grade corporate bonds also fail to serve as a safe haven with most of the GFC safe-haven
Second, scanning down Column B1-B4 relating to the COVID crisis, we observe that gold
is quite risky in some settings, especially in China, Japan and India. Further, silver has become
much riskier, especially in China, Japan and India. And, the currencies, the treasuries still show
some weak safe-haven benefits during COVID. Furthermore, AAA-grade corporate bonds
emerge as a weak safe haven during COVID. As for the cryptos, Bitcoin and Tether deliver
greatly contrasting evidence. While Tether offers some weak safe-haven benefits during
COVID, Bitcoin is a disaster. Indeed, during COVID, Bitcoin shows substantial speculative
risk, especially for investors in China and Japan. Collectively, these COVID analyses seem to
suggest that Tether has usurped the traditional role of gold (and silver) as safe haven in China,
Japan and India. This points to a highly intriguing juxtaposition between the investment roles
of our two chosen cryptocurrencies – other things equal, investors should clearly prefer Tether
over Bitcoin. Third, scanning down Column C1-C3 relating to whether the safe-haven
character of different assets has changed during COVID compared to the GFC, we do observe
some mild evidence of a weakening in COVID (compared to the GFC), especially for the
20
traditional safe-haven assets, gold and silver. However, we observe that AAA-grade corporate
bonds have strengthened somewhat as a safe haven during COVID compared to the GFC.
4. Potential Explanations
The most surprising finding from Section 3 is that the gold has lost its safe haven status
during COVID. Traditionally, gold is considered as one of the most effective safe haven assets,
and it has exhibited safe haven characteristics during the previous crises such as the 1987 stock
market crash and the 2008 GFC (e.g. Baur and Lucey, 2010; Baur and McDermott, 2010; Low
et al., 2016).
Figure 2 plots the gold prices from January 1, 1990, to June 26, 2020. It is evident from
Figure 2 that gold attained the maximum price of $1898.25 on September 5, 2011 and lost 45%
of value by December 17, 2015. Therefore, investors might have lost their trust in the gold as
a safe haven asset since a loss of 45% over four years indicates instability in gold prices. To
the extent, gold has lost its status of a safe haven among investors due to the extreme losses
between 2011 and 2015; we expect that gold does not act at least as a strong safe haven during
extreme stock market movements. Therefore, we examine the performance of gold as a safe
haven asset during extreme stock market movements after September 5, 2011. We define
extreme stock market movements where stock market returns at time t are in a low quantile,
such as the 5%, and 1% quantile. We estimate the following regression model first proposed
where RGold represents the daily log return of gold. 𝑅𝑆𝑗 denotes the daily log returns in
US dollars of a stock market index j, with j equal to a given one of the ten countries in our
sample. The dummy variables, D, capture extreme stock market losses, taking a value of one
if stock market return at time t is in the low quantile, such as 5% and 1%, and zero otherwise.
The residual term εt is modelled as a GJR-GARCH process introduced by Glosten et al. (1993)
21
as defined in Equation (2). The gold is a strong (weak) hedge for stock market j if the parameter
of either sign, but economically “small”), and the sum of parameters from b2 to b3 are not jointly
positive exceeding the value of b1. Same as in Section 3, we use the cutoff of 0.1 to assess
lying in the range [-0.1, +0.1] is deemed to be economically small. Parameters, b2, and b3 are
the incremental safe-haven gold beta for the lowest 5%, and 1% for the stock market j returns,
respectively. Further, the sum of the two parameters, b1 + b2, is the total safe-haven gold beta
for the lowest 5% stock market returns. If the sum, b1 + b2, is non-positive, statistically
significant and not economically “small” (statistically insignificant or significant of either sign,
but economically “small”), then gold serves as a strong (weak) safe haven for the lowest 5%
stock market returns. Similar interpretation applies to b1+b2+b3, with respect to the lowest 1%
Panel A of Table 9 presents the estimation results for gold against the lowest 5% and 1%
stock market returns after reaching the maximum price on September 5, 2011.7 All the total
safe-haven 5% quantile betas (i.e. sum of b1 + b2+ b3) except the US and Japan are insignificant
(at the 5% level), indicating that gold serves as a weak safe haven in these setting against lowest
5% stock market returns. Gold serves as a strong safe haven for the US during the lowest 5%
stock market returns, in which the total safe-haven 5% quantile beta is negative, significant and
Most importantly, in stark contrast to the counterpart analysis for the lowest 5% stock
market returns, gold does not serve as a strong safe haven for any country during the lowest
1% stock market returns. In fact, it does not serve even as a weak safe haven for two countries
(Brazil and Canada), in which the total safe-haven 1% quantile betas (i.e. sum of b1 + b2+ b3+
7
For the sake of brevity, we do not discuss the hedge results since those are simailr as in Section 3.1.
22
b4) are positive, significant and economically not too small. Gold serves as a weak safe haven
for the other eight countries where betas are either insignificant or economically small.
In sum, the results in Table 9 suggest that gold serves as predominantly a weak safe haven
during the lowest stock market returns. However, the efficacy of gold as a safe haven has
weakened using the recent sample period covering 2011 to 2020, especially for the lowest 1%
stock market returns since gold betas are either insignificant or small in magnitude compared
to the gold betas for the lowest 1% stock market returns reported in Panel B for the period
earlier than September 5, 2011. As previously mentioned, it could be that gold attained its peak
value on September 5, 2011 and lost it by 45% over the next four years, and consequently,
investors lost trust in gold as a stable asset. However, we cannot rule out the possibility that
investors might have sold gold to cover the losses in the other asset classes during COVID.
5. Conclusion
This paper examines the performance of precious metals (gold and silver); currencies (US
dollar and Swiss Franc); US Treasuries (T-bills and T-bonds); corporate bonds (AAA-grade);
and cryptocurrencies (Bitcoin and Tether) from stock market losses during the 2008 GFC and
COVID-19 pandemic. Our findings show that gold serves as a weak safe haven during the
GFC; however, it fails to act as a safe haven during COVID, which indicates that gold could
lose its safe haven status during pandemics. Silver generally moves in tandem with the stock
markets during both crises, which refutes the use of silver as a safe haven during stock market
crises. Furthermore, we find that the US dollar, Swiss Franc and US Treasuries remain stable
during both crises, and consequently serve as a weak safe haven across the sample countries.
It is important to note that the US dollar and Treasuries have maintained their safe haven status
during COVID even though the US has suffered from the highest death and infection rates of
COVID in the world. Interestingly, AAA-grade corporate bonds emerge as a weak safe haven
during COVID; whereas, these bonds do not serve as a safe haven during the GFC. Regarding
23
cryptocurrencies, our results show that investment in Bitcoin is a risky strategy during COVID;
therefore, Bitcoin has failed its first real test as a safe haven miserably. However, the largest
asset-backed cryptocurrency, Tether offers safe haven benefits across the sample countries
during COVID.
Our findings indicate three major patterns among the safe haven assets. First, the same
assets act as safe havens against the stock market losses from both developed and emerging
markets across our sample countries. For example, gold, US dollar, Swiss Franc, and US
Treasuries act as weak safe havens for both developed and emerging markets during the GFC.
Whereas, the US dollar, Swiss Franc, US Treasuries, AAA-grade corporate bonds and Tether
act as weak safe havens for both developed and emerging markets during COVID. Second, our
findings indicate that Swiss Franc and US Treasuries can be classified as global assets since
they act as safe havens across all the sample countries during both crises. Nonetheless, Tether
can also be classified as a global asset since it has emerged unscathed from the pandemic.
Third, our findings indicate that the safe-haven character of traditional assets such as gold and
silver has weakened during COVID; whereas, AAA-bonds have emerged as a safe haven
We also explain why gold loses its value as a safe haven asset during COVID when,
traditionally, it acted as a safe haven asset during the previous stock market crises of 1987 and
the GFC. We suggest that investors might have lost trust in gold as a stable asset after losing
45% of its value between 2011 to 2015. Furthermore, investors now have access to additional
safe haven assets for shelter during crises, such as asset-backed cryptocurrencies (stablecoins).
The findings are useful for investors and fund managers searching for the best safe haven,
such as gold, silver, currencies Treasuries, corporate bonds, and cryptocurrencies to offset large
stock market losses. Furthermore, our results suggest that investors should prefer liquid and
stable assets such as the Tether and Treasuries during a pandemic rather than gold. Therefore,
24
central banks, financial institutions and regulatory authorities should consider supporting
financial assets that remain liquid during stock market crises. Finally, our findings suggest that
policymakers and regulatory authorities should use caution in classifying Bitcoin an alternative
to traditional currencies since it has acted more like a risky than a safe asset during COVID.
Future research endeavours should identify other safe haven assets during COVID.
25
variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19)
Gold (1) 1
Tether (9) -0.0454 0.0363 -0.0002 0.0403 0.0235 0.0093 0.0246 -0.0210 1
US (10) -0.0016 0.1321 -0.1506 -0.0991 -0.1353 -0.4034 -0.2436 0.1383 -0.0937 1
China (11) 0.0528 0.0877 -0.0451 -0.0619 -0.0283 -0.0544 0.0276 0.0216 0.0162 0.0897 1
Japan (12) 0.0811 0.1515 -0.1147 -0.1558 -0.0330 0.0113 0.0960 -0.0189 0.0082 0.0380 0.2227 1
Germany (13) 0.1207 0.2472 -0.3043 -0.1135 -0.0908 -0.2678 -0.1317 0.0994 -0.0685 0.5804 0.1576 0.1750 1
UK (14) 0.1228 0.2774 -0.2907 -0.1360 -0.1434 -0.2853 -0.1078 0.0758 -0.0774 0.5760 0.1726 0.2675 0.8053 1
France (15) 0.1110 0.2488 -0.3126 -0.1227 -0.1112 -0.2799 -0.1334 0.0960 -0.0756 0.5751 0.1603 0.1864 0.9516 0.8344 1
India (16) 0.0932 0.1668 -0.1059 -0.1545 -0.0661 -0.1124 0.0147 0.0245 -0.0794 0.2521 0.2310 0.2914 0.3800 0.4089 0.3772 1
Italy (17) 0.0950 0.2242 -0.3183 -0.1139 -0.0976 -0.2752 -0.1323 0.1160 -0.1091 0.5418 0.1384 0.1653 0.8850 0.7658 0.9197 0.3463 1
Brazil (18) 0.1136 0.2400 -0.2251 -0.1088 -0.1298 -0.2631 -0.0988 0.1125 -0.1210 0.5680 0.1809 0.1877 0.5284 0.5837 0.5443 0.3628 0.4936 1
Canada (19) 0.1947 0.3335 -0.2917 -0.1052 -0.1165 -0.2985 -0.1214 0.1373 -0.0770 0.7301 0.1513 0.1687 0.6229 0.6637 0.6350 0.3345 0.5886 0.6380 1
27
Table 2: Extreme Losses during the 2008 GFC and COVID-19 Pandemic
Panels A and B list the ten largest daily losses of S&P 500 returns and the respective returns of safe haven assets during the 2008 GFC and COVID-19 pandemic,
respectively.
Table 3: Estimation results for gold and silver as safe haven assets during the 2008 GFC and Covid-19 pandemic
This table presents the estimation results of the role of gold and silver as a hedge and safe haven asset in the periods of stock market crises, such as the 2008 GFC and COVID-
19 pandemic. The crisis duration is set to 20 trading days. The GFC starts on September 12, 2008, and ends October 10, 2008, while the COVID-19 pandemic starts on February
20, 2020, and ends March 18, 2020. Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant and not economically “small”
(statistically insignificant or significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use the cutoff of 0.1 to assess
whether an estimated coefficient is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically small. Parameter b2 is
the incremental safe-haven asset beta for the GFC. Further, the sum of the two parameters, b1 + b2, is the total safe-haven asset beta for the GFC. If the sum, b1 + b2, is negative,
statistically significant and not economically “small” (statistically insignificant or significant of either sign, but economically “small”), then asset i serves as a strong (weak)
safe haven from stock market losses during the GFC. Similar interpretations apply to b3 and b1 + b3, with respect to COVID. The respective t-statistics are provided in the
parentheses.
Panel A: Gold
Coefficients US China Japan Germany UK France India Italy Brazil Canada
0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0003 0.0002 0.0003
Const (b0)
(2.19) (2.20) (2.31) (2.32) (2.17) (2.29) (2.14) (2.29) (1.65) (1.85)
0.0067 0.0366 0.0482 0.0676 0.1008 0.0593 0.0562 0.0396 0.0640 0.1869
Hedge (b1)
(0.54) (4.29) (5.25) (7.14) (9.15) (6.30) (5.67) (4.82) (9.26) (15.75)
-0.2053 -0.2190 -0.0700 -0.0556 -0.2551 -0.1163 -0.3193 -0.0870 -0.2073 -0.2609
ΔGFC (b2)
(-2.72) (-2.26) (-0.86) (-0.46) (-1.80) (-1.09) (-2.08) (-0.80) (-5.10) (-3.27)
-0.1986 -0.1824 -0.0218 0.0120 -0.1543 -0.0570 -0.2631 -0.0474 -0.1433 -0.0740
Total GFC (b1 + b2)
(-2.67) (-1.89) (-0.26) (0.10) (-1.09) (-0.54) (-1.72) (-0.44) (-3.57) (-0.94)
0.1275 0.6970 0.4026 0.2143 0.1935 0.2009 0.2615 0.1144 0.0322 -0.0085
ΔCOVID (b3)
(4.67) (11.12) (6.99) (4.75) (4.13) (4.98) (6.19) (4.53) (1.66) (-0.30)
0.1342 0.7336 0.4508 0.2819 0.2943 0.2602 0.3177 0.1540 0.0962 0.1784
Total COVID (b1 + b3)
(5.51) (11.80) (7.94) (6.37) (6.43) (6.62) (7.69) (6.43) (5.32) (6.94)
Panel B: Silver
Coefficients US China Japan Germany UK France India Italy Brazil Canada
0.0001 0.0002 0.0002 0.0002 0.0002 0.0003 0.0002 0.0002 0.0001 0.0001
Const (b0)
(0.44) (0.63) (0.88) (0.94) (0.82) (0.98) (0.69) (0.82) (0.53) (0.42)
0.2306 0.0875 0.1378 0.2674 0.3659 0.2564 0.1652 0.1888 0.2141 0.5638
Hedge (b1)
(9.75) (5.53) (7.60) (16.60) (20.60) (16.67) (9.39) (14.17) (17.32) (27.3)
-0.2659 0.0813 0.1510 0.2109 -0.0069 0.0954 0.2027 0.1794 -0.1549 -0.1360
ΔGFC (b2)
(-1.99) (0.52) (1.72) (1.61) (-0.05) (0.78) (1.30) (1.41) (-2.41) (-1.07)
-0.0353 0.1688 0.2888 0.4783 0.3590 0.3518 0.3679 0.3682 0.0592 0.4278
Total GFC (b1 + b2)
(-0.26) (1.09) (3.36) (3.69) (2.42) (2.91) (2.37) (2.92) (0.94) (3.41)
-0.1165 1.5930 1.1627 0.5019 0.4594 0.5054 0.9602 -0.1038 0.2936 0.0926
ΔCOVID (b3)
(-3.37) (17.76) (14.01) (11.86) (9.72) (11.86) (17.30) (-3.83) (8.14) (2.61)
0.1141 1.6805 1.3005 0.7693 0.8253 0.7618 1.1254 0.0850 0.5077 0.6564
Total COVID (b1 + b3)
(4.59) (19.04) (16.10) (19.65) (18.72) (19.16) (21.41) (3.64) (14.86) (22.71)
29
Table 4: Estimation results for US Dollars and Swiss Francs as safe haven assets during the 2008 GFC and Covid-19 pandemic
This table presents the estimation results of the role of US Dollar and Swiss Franc as a hedge and safe haven asset in the periods of stock market crises, such as the 2008 GFC
and COVID-19 pandemic. The crisis duration is set to 20 trading days. The GFC starts on September 12, 2008, and ends October 10, 2008, while the COVID-19 pandemic
starts on February 20, 2020, and ends March 18, 2020. Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant and not
economically “small” (statistically insignificant or significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use the cutoff
of 0.1 to assess whether an estimated coefficient is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically small.
Parameter b2 is the incremental safe-haven asset beta for the GFC. Further, the sum of the two parameters, b1 + b2, is the total safe-haven asset beta for the GFC. If the sum, b1
+ b2, is negative, statistically significant and not economically “small” (statistically insignificant or significant of either sign, but economically “small”), then asset i serves as
a strong (weak) safe haven from stock market losses during the GFC. Similar interpretations apply to b3 and b1 + b3, with respect to COVID. The respective t-statistics are
provided in the parentheses.
Table 5: Estimation results for T-bills and T-bonds as safe haven assets during the 2008 GFC and Covid-19 pandemic
This table presents the estimation results of the role of T-bills and T-bonds as a hedge and safe haven asset in the periods of stock market crises, such as the 2008 GFC and
COVID-19 pandemic. The crisis duration is set to 20 trading days. The GFC starts on September 12, 2008, and ends October 10, 2008, while the COVID-19 pandemic starts
on February 20, 2020, and ends March 18, 2020. Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant and not economically
“small” (statistically insignificant or significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use the cutoff of 0.1 to
assess whether an estimated coefficient is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically small. Parameter
b2 is the incremental safe-haven asset beta for the GFC. Further, the sum of the two parameters, b1 + b2, is the total safe-haven asset beta for the GFC. If the sum, b1 + b2, is
negative, statistically significant and not economically “small” (statistically insignificant or significant of either sign, but economically “small”), then asset i serves as a strong
(weak) safe haven from stock market losses during the GFC. Similar interpretations apply to b3 and b1 + b3, with respect to COVID. The respective t-statistics are provided in
the parentheses.
Table 6: Estimation results for AAA-grade corporate bonds as safe haven assets during the 2008 GFC and Covid-19 pandemic
This table presents the estimation results of the role of S&P500 AAA-grade bonds as a hedge and safe haven asset in the periods of stock market crises, such as the 2008 GFC
and COVID-19 pandemic. The crisis duration is set to 20 trading days. The GFC starts on September 12, 2008, and ends October 10, 2008, while the COVID-19 pandemic
starts on February 20, 2020, and ends March 18, 2020. Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant and not
economically “small” (statistically insignificant or significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use the cutoff
of 0.1 to assess whether an estimated coefficient is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically small.
Parameter b2 is the incremental safe-haven asset beta for the GFC. Further, the sum of the two parameters, b1 + b2, is the total safe-haven asset beta for the GFC. If the sum, b1
+ b2, is negative, statistically significant and not economically “small” (statistically insignificant or significant of either sign, but economically “small”), then asset i serves as
a strong (weak) safe haven from stock market losses during the GFC. Similar interpretations apply to b3 and b1 + b3, with respect to COVID. The respective t-statistics are
provided in the parentheses.
Table 7: Estimation results for Bitcoin and Tether as a safe haven asset during Covid-19 pandemic
This table presents the estimation results of the role of Bitcoin and Tether as a hedge and safe haven assets during the COVID-19 pandemic. The crisis duration is set to 20
trading days starting on February 20, 2020, and ending March 18, 2020. Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant
and not economically “small” (statistically insignificant or significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use
the cutoff of 0.1 to assess whether an estimated coefficient is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically
small. Parameter b3 is the incremental safe-haven asset beta for COVID. Further, the sum of the two parameters, b1 + b3, is the total safe-haven asset beta for COVID. If the
sum, b1 + b3, is negative, statistically significant and not economically “small” (statistically insignificant or significant of either sign, but economically “small”), then asset i
serves as a strong (weak) safe haven from stock market losses during the COVID pandemic. The respective t-statistics are provided in the parentheses.
Panel A: Bitcoin
Coefficients US China Japan Germany UK France India Italy Brazil Canada
0.0019 0.0019 0.0020 0.0020 0.0019 0.0020 0.0020 0.0020 0.0020 0.0019
Const (b0)
(1.91) (1.99) (1.99) (2.01) (1.99) (2.00) (2.06) (2.03) (2.08) (1.99)
0.1427 0.0798 -0.1415 0.0517 0.0159 0.0076 0.0021 0.0540 0.0265 0.1184
Hedge (b1)
(1.71) (2.07) (-1.84) (0.80) (0.21) (0.12) (0.03) (0.97) (0.56) (1.54)
1.6741 4.7450 3.4004 2.1400 2.2261 2.1071 2.8641 1.6002 1.1885 1.3889
ΔCOVID (b3)
(14.03) (32.26) (18.74) (19.10) (17.53) (18.53) (24.01) (15.52) (16.48) (14.48)
1.8168 4.8248 3.2589 2.1917 2.2420 2.1147 2.8662 1.6542 1.2150 1.5073
Total COVID (b1 + b3)
(20.82) (34.70) (20.20) (23.64) (21.86) (22.49) (30.28) (18.96) (21.79) (25.51)
Panel A: Tether
Coefficients US China Japan Germany UK France India Italy Brazil Canada
0.0000 0.0000 -0.0001 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 -0.0001
Const (b0)
(0.02) (-0.56) (-1.18) (6.80) (-0.55) (-0.25) (-0.37) (-0.45) (-0.51) (-0.85)
0.0045 -0.0012 -0.0053 0.0209 -0.0017 0.0014 -0.0001 -0.0005 -0.0021 -0.0041
Hedge (b1)
(0.58) (-0.21) (-0.85) (31.97) (-0.28) (0.26) (-0.02) (-0.09) (-0.69) (-0.64)
-0.1460 -0.2176 -0.2089 -0.0830 -0.1185 -0.1115 -0.2002 -0.0961 -0.0908 -0.0615
ΔCOVID (b3)
(-7.53) (-9.40) (-7.93) (-10.48) (-8.60) (-9.46) (-10.07) (-8.81) (-6.31) (-7.31)
-0.1415 -0.2188 -0.2142 -0.0621 -0.1202 -0.1101 -0.2003 -0.0966 -0.0929 -0.0656
Total COVID (b1 + b3)
(-8.12) (-9.62) (-8.25) (-7.96) (-9.66) (-10.69) (-10.47) (-10.68) (-6.57) (-11.24)
33
Table 8: Summary of estimation results for various potential safe haven assets during the GFC vs. COVID financial crises
This table presents a broad summary of the performance of various potential safe haven assets during the GFC and COVID-19 pandemic financial crises, as analysed and
reported in Tables 3-7. There are three panels: Panel A – showing a summary of the total GFC safe-haven effect; Panel B – showing a summary of the total COVID safe-haven
effect; and Panel C – providing a brief Commentary. Panels A and B contain four columns each. Column (1): “min” – the minimum statistically significant (at the 5% level)
country-based estimate of the safe-haven beta for the asset in question. Column (2): “max” – the maximum statistically significant (at the 5% level) country-based estimate of
the safe-haven beta for the asset in question. In the case of (1) and (2), “min” and “max” 0 reflects either the estimate of the safe-haven beta that is insignificant or a significant
estimate that is close to zero.. Column (3): “#0’s” – the number out of 10 cases (out of nine cases for the case of the US dollar), of effectively zero value estimates of country-
based safe-haven betas for the asset in question. In the case of (3), “effectively zero” is taken to mean either an insignificant estimate or a significnat estimate that is economically
close to zero, deemed to be within the range [-0.1 to +0.1]. Column (4): “#<0” – the number of statistically significant (at the 5% level) and negatively signed estimates of
country-based safe-haven betas for the asset in question. Panel C organises some brief commentray into three colums. Column (C1) gives an overall “call” on the safe-haven
character for the asset in question during the GFC. Column (C2) gives an overall “call” on the safe-haven character for the asset in question during COVID. Column (C3)
makes a “call” on whether the safe-haven character of the asset in question, has changed and, if so, how?, during COVID compared to the GFC.
Table 9: Estimation results for gold as a safe haven in extreme market conditions
Table presents the estimation results of the role of gold as a hedge and safe haven asset during the periods of extreme market conditions namely, quantile 5% (b2), and 1%
(b3). Asset i is deemed a strong (weak) hedge for the stock market j if the parameter b1 is negative, significant and not economically “small” (statistically insignificant or
significant of either sign, but economically “small”). We use the 5% level to assess statistical significance. We use the cutoff of 0.1 to assess whether an estimated coefficient
is economically “small” – that is, an estimated coefficient lying in the range [-0.1, +0.1] is deemed to be economically small. Parameters, b2, and b3 are the incremental safe-
haven gold beta for the lowest 5%, and 1% for the stock market j returns, respectively. Further, the sum of the two parameters, b1 + b2, is the total safe-haven gold beta for the
lowest 5% stock market returns. If the sum, b1 + b2, is non-positive, statistically significant and not economically “small” (statistically insignificant or significant of either
sign, but economically “small”), then gold serves as a strong (weak) safe haven for the lowest 5% stock market returns. Similar interpretation applies to b1+b2+b3, with
respect to the lowest 1% stock market returns. The respective t-statistics are provided in the parentheses.
Panel A: Gold as a safe haven in extreme market conditions after reaching the maximum price on September 5, 2011
Coefficients US China Japan Germany UK France India Italy Brazil Canada
-0.0002 -0.0003 -0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 -0.0001 -0.0001
Const (b0)
(-0.85) (-1.36) (-0.45) (0.23) (-0.10) (0.06) (-0.03) (0.02) (-0.37) (-0.54)
0.0200 0.0409 0.0038 -0.0385 0.0083 -0.0350 -0.0107 -0.0225 0.0403 0.1187
Hedge (b1)
(1.00) (2.28) (0.25) (-2.69) (0.47) (-2.36) (-0.61) (-1.87) (3.69) (5.80)
-0.1748 -0.0004 -0.0574 0.0411 -0.0319 0.0036 -0.0068 0.0354 -0.0607 -0.1727
Quantile Δ5% (b2)
(-4.35) (-0.01) (-1.78) (1.59) (-0.96) (0.14) (-0.17) (1.38) (-2.46) (-4.48)
-0.1548 0.0405 -0.0536 0.0026 -0.0236 -0.0314 -0.0175 0.0129 -0.0204 -0.0540
Total (b1+ b2)
(-4.44) (1.41) (-1.97) (0.10) (-0.82) (-1.55) (0.49) (0.57) (0.94) (-1.60)
0.1981 -0.0974 0.0169 0.0327 0.1045 0.0639 0.0809 -0.0644 0.1320 0.2537
Quantile Δ1% (b3)
(5.03) (-2.74) (0.51) (1.32) (3.17) (2.69) (2.03) (-2.87) (4.74) (6.43)
0.0433 -0.0569 -0.0367 0.0353 0.0809 0.0325 0.0634 -0.0515 0.1116 0.1997
Total (b1+ b2+ b3)
(2.06) (-2.84) (-1.49) (2.34) (4.44) (2.20) (3.40) (-5.13) (6.33) (9.03)
Panel B: Gold as a safe haven in extreme market conditions before reaching the maximum price on September 5, 2011
Coefficients US China Japan Germany UK France India Italy Brazil Canada
Sample start date 1979 1992 1979 1999 1984 1987 1991 1998 1988 1979
0.0001 0.0002 0.0001 0.0006 0.0001 0.0001 0.0001 0.0004 0.0001 0.0000
Const (b0)
(1.52) (1.63) (0.84) (3.39) (1.40) (0.87) (1.24) (2.68) (1.22) (0.52)
-0.0434 0.0010 0.0261 0.0207 -0.0267 -0.0316 0.0138 0.0335 0.0014 0.1076
Hedge (b1)
(-4.55) (0.22) (3.65) (1.80) (-2.86) (-4.33) (2.01) (3.00) (0.37) (10.34)
0.0343 0.0064 -0.0305 -0.0353 0.0517 0.0110 -0.0203 -0.0542 0.0269 -0.0176
Quantile Δ5% (b2)
(1.67) (0.61) (-2.13) (-1.55) (2.87) (0.82) (-1.23) (-2.62) (3.17) (-0.85)
-0.0091 0.0074 -0.0044 -0.0146 0.0250 -0.0206 -0.0065 -0.0207 0.0283 0.0900
Total (b1+ b2)
(-0.52) (0.78) (-0.37) (-0.72) (1.65) (-1.72) (-0.44) (-1.20) (3.74) (4.99)
-0.0568 0.0010 0.0437 -0.1845 -0.1154 -0.1117 0.0227 -0.1139 -0.0403 -0.0749
Quantile Δ1% (b3)
(-3.06) (0.08) (2.47) (-8.31) (-7.45) (-7.55) (1.05) (-5.16) (-4.29) (-3.74)
-0.0659 0.0084 0.0393 -0.1991 -0.0904 -0.1323 0.0162 -0.1346 -0.0120 0.0151
Total (b1+ b2+ b3)
(-9.42) (0.90) (2.94) (-14.32) (-14.60) (-15.59) (1.03) (-10.14) (-2.12) (1.86)
35
55000
9000
50000
8000
45000
7000
40000
6000
35000
Index Level
Index Level
5000 30000
25000
4000
20000
3000
15000
2000
10000
1000
5000
0 0
201907
200401
200407
200501
200507
200601
200607
200701
200707
200801
200807
200901
200907
201001
201007
201101
201107
201201
201207
201301
201307
201401
201407
201501
201507
201601
201607
201701
201707
201801
201807
201901
202001
202007
US China Germany UK FRANCE India Brazil Japan Italy Canada
Figure 1: This figure displays the daily index level of the stock markets of all the ten largest economies in the world over the sample period. For the readers convenience, the
index level of Japan, Italy and Canada is labelled on the right vertical axis, and the index level of other seven countries is labelled on the left vertical axis.
36
2 1600
1400
1.5 1200
$1051.97 1000
1 800
600
0.5 400
200
0 0
Figure 2: This figure displays the daily gold prices in US dollars from 1990 to 2020. The gold prices in USD are labelled on the vertical axis, and date on the horizontal
axis.
37
References
Adrian, T., Crump, R.K., Vogt, E., 2019. Nonlinearity and Flight‐to‐Safety in the Risk‐
Return Trade‐Off for Stocks and Bonds. The Journal of Finance 74, 1931-1973.
Al-Awadhi, A.M., Al-Saifi, K., Al-Awadhi, A., Alhamadi, S., 2020. Death and contagious
infectious diseases: Impact of the COVID-19 virus on stock market returns. Journal of
Behavioral and Experimental Finance 27, https://doi.org/10.1016/j.jbef.2020.100326.
Alfaro, L., Chari, A., Greenland, A.N., Schott, P.K., 2020. Aggregate and firm-level stock
returns during pandemics, in real time. Covid Economics 4, 14 April, 2-24.
Baele, L., Bekaert, G., Inghelbrecht, K., Wei, M., 2020. Flights to safety. The Review of
Financial Studies 33, 689-746.
Baker, S.R., Bloom, N., Davis, S.J., Kost, K., Sammon, M., Viratyosin, T., 2020. The
unprecedented stock market reaction to COVID-19. Working paper, Northwestern University.
Baur, D.G., Hoang, L.T., 2020. A Crypto Safe Haven against Bitcoin. Finance Research
Letters, 101431.
Baur, D.G., Lucey, B.M., 2010. Is gold a hedge or a safe haven? An analysis of stocks, bonds
and gold. Financial Review 45, 217-229.
Baur, D.G., McDermott, T.K., 2010. Is gold a safe haven? International evidence. Journal of
Banking & Finance 34, 1886-1898.
BBC, N., 2020. Global shares plunge in worst day since financial crisis, Available at
https://www.bbc.com/news/business-51796806.
Brunnermeier, M.K., Merkel, S., Sannikov, Y., 2020. A Safe-Asset Perspective for an
Integrated Policy Framework.
Caballero, R.J., Krishnamurthy, A., 2008. Collective risk management in a flight to quality
episode. The Journal of Finance 63, 2195-2230.
Chan, K.F., Treepongkaruna, S., Brooks, R., Gray, S., 2011. Asset market linkages: Evidence
from financial, commodity and real estate assets. Journal of Banking & Finance 35, 1415-1426.
Conlon, T., Corbet, S., McGee, R.J., 2020. Are Cryptocurrencies a Safe Haven for Equity
Markets? An International Perspective from the COVID-19 Pandemic. Research in
International Business and Finance 54.
Conlon, T., McGee, R., 2020. Safe haven or risky hazard? Bitcoin during the COVID-19 bear
market. Finance Research Letters.
Corbet, S., Hou, G., Hu, Y., Larkin, C.J., Oxley, L., 2020a. Any Port in a Storm:
Cryptocurrency Safe-Havens during the COVID-19 Pandemic. Economics Letters 194.
38
Corbet, S., Larkin, C., Lucey, B., 2020b. The contagion effects of the covid-19 pandemic:
Evidence from gold and cryptocurrencies. Finance Research Letters,
https://doi.org/10.1016/j.frl.2020.101554.
Faulkender, M.W., Hankins, K.W., Petersen, M.A., 2019. Understanding the rise in corporate
cash: Precautionary savings or foreign taxes. The Review of Financial Studies 32, 3299-3334.
Fleming, J., Kirby, C., Ostdiek, B., 1998. Information and volatility linkages in the stock, bond,
and money markets. Journal of Financial Economics 49, 111-137.
Glosten, L.R., Jagannathan, R., Runkle, D.E., 1993. On the relation between the expected value
and the volatility of the nominal excess return on stocks. The Journal of Finance 48, 1779-
1801.
Grisse, C., Nitschka, T., 2015. On financial risk and the safe haven characteristics of Swiss
franc exchange rates. Journal of Empirical Finance 32, 153-164.
Hartmann, P., Straetmans, S., Vries, C.d., 2004. Asset market linkages in crisis periods. Review
of Economics and Statistics 86, 313-326.
Hillier, D., Draper, P., Faff, R., 2006. Do precious metals shine? An investment perspective.
Financial Analysts Journal 62, 98-106.
Ji, Q., Zhang, D., Zhao, Y., 2020. Searching for safe-haven assets during the COVID-19
pandemic. International Review of Financial Analysis, 101526.
Kaul, A., Sapp, S., 2006. Y2K fears and safe haven trading of the US dollar. Journal of
International Money and Finance 25, 760-779.
Klein, T., Thu, H.P., Walther, T., 2018. Bitcoin is not the New Gold–A comparison of
volatility, correlation, and portfolio performance. International Review of Financial Analysis
59, 105-116.
Kristoufek, L., 2020. Grandpa, grandpa, tell me the one about Bitcoin being a safe haven:
Evidence from the COVID-19 pandemics.
Lipton, A., Sardon, A., Schär, F., Schüpbach, C., 2020. 10. Stablecoins, Digital Currency, and
the Future of Money, Building the New Economy.
Low, R.K.Y., Yao, Y., Faff, R., 2016. Diamonds vs. precious metals: What shines brightest in
your investment portfolio? International Review of Financial Analysis 43, 1-14.
Noeth, B.J., Sengupta, R., 2010. Flight to safety and US Treasury securities. The Regional
Economist 18, 18-19.
Pullen, T., Benson, K., Faff, R., 2014. A comparative analysis of the investment characteristics
of alternative gold assets. Abacus 50, 76-92.
Ramelli, S., Wagner, A.F., 2020. Feverish stock price reactions to covid-19.
Ranaldo, A., Söderlind, P., 2010. Safe haven currencies. Review of Finance 14, 385-407.
39
Smales, L.A., 2019. Bitcoin as a safe haven: Is it even worth considering? Finance Research
Letters 30, 385-393.
Stensås, A., Nygaard, M.F., Kyaw, K., Treepongkaruna, S., 2019. Can Bitcoin be a diversifier,
hedge or safe haven tool? Cogent Economics & Finance 7.
Urquhart, A., Zhang, H., 2019. Is Bitcoin a hedge or safe haven for currencies? An intraday
analysis. International Review of Financial Analysis 63, 49-57.
Zhang, D., Hu, M., Ji, Q., 2020. Financial markets under the global pandemic of COVID-19.
Finance Research Letters, https://doi.org/10.1016/j.frl.2020.101528.