FACULTY OF ECONOMICS AND ADMINISTRATION
Habilitation Thesis
Brno 2022 Tomáš Plíhal
FACULTY OF ECONOMICS AND ADMINISTRATION
Habilitation Thesis
TOMÁŠ PLÍHAL
Department of Finance
Brno, Spring 2022
Bibliographic record
Author: Tomáš Plíhal
Faculty of Economics and Administration
Masaryk University
Department of Finance
Title of Thesis: Essays on Volatility Modeling: The Effect of Expected and Unexpected
Events
Field of Study: Finance
Academic Year: 2021/2022
Number of Pages: 11 + 73
Keywords: volatility forecasting, volatility modeling, news announcements,
realized volatility, implied volatility
Annotation
This collection of essays contains six published academic papers related to volatility
modeling and forecasting. It includes the introductory chapter about the importance of
volatility and different volatility models, followed by a summary of the papers. Each
section states the bibliographic record and my personal contribution to each paper,
along with the article’s contribution to academic literature, a summary of the relevant
literature, methodology, and main results, ending with potential research limits. Lastly,
the conclusion summarizes the thesis and indicates the expected future direction of my
research.
Acknowledgements
This habilitation thesis would not have been possible without the enormous support of
many people. The main credit for this goes to Štefan Lyócsa, my former Ph.D. supervisor,
who taught me a lot and became a valued co-author. Next, I really appreciate the support
of my colleagues, especially Oleg Deev, Dagmar Vágnerová Linnertová, and Filip Hampl.
They have created a perfect working environment, are always willing to help, and have
become great companions. Last but not least, my thanks go to my wife, who provided
me with enough peace and comfort at home, allowing me to deliver high-quality work.
Contents
List of Tables 11
Introduction 13
1 A Brief Review of Volatility Models 17
2 Realized Volatility Forecasting 21
2.1 FX market volatility modelling: Can we use low-frequency data? . . . . . 21
2.1.1 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 Modeling realized volatility of the EUR/USD exchange rate: Does implied
volatility really matter? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2.1 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3 News Announcements and Realized Volatility 29
3.1 Central bank announcements and realized volatility of stock markets in
G7 countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Impact of macroeconomic news, regulation and hacking exchange markets
on the volatility of bitcoin . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4 Realized Volatility Reaction During Crises 43
4.1 How to calm down the markets? The effects of COVID-19 economic policy
responses on financial market uncertainty . . . . . . . . . . . . . . . . . . 43
4.1.1 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . . . 44
4.1.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Russia’s ruble during the onset of the Russian invasion of Ukraine in early
2022: The role of implied volatility and attention . . . . . . . . . . . . . . 47
4.2.1 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Conclusion 53
Bibliography 55
A Full text of the paper collection 73
List of Tables
1 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 21
2 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 25
3 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 29
4 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 36
5 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 43
6 Paper Information and Author Contribution . . . . . . . . . . . . . . . . . . 48
Introduction
This collection of essays contains six papers from the field of volatility modeling and
forecasting. The understanding of volatility is crucial for understanding price behavior.
It can help with more efficient asset allocation, market risk management, or financial
derivatives valuation. Volatility might also be traded using options or artificial indices
linked to market volatility (Poon et al., 2003). Therefore, modeling and forecasting
volatility has received considerable attention in the financial literature (e.g. Engle, 1982;
Bollerslev, 1986; Andersen; Bollerslev, 1998; Corsi, 2009).
The volatility of a financial asset represents the fluctuation of its price. In other
words, it measures how quickly and how much the price changes. However, it does
not represents the direction of such price movements. It is also closely connected with
the trading activity of market participants, where higher volatility is usually connected
with increased trade volume indicating changes in the expectation of investors about
the future prices of a given asset.
The central role plays various expected and unexpected events that have the power
to drive the market prices and their volatility. Volatility plays a central role in finance and
is often perceived as a measure of risk because it represents an uncertainty of the future
development of asset prices. Moreover, it is often used as a proxy of asset-market-base
uncertainty (e.g. Andersen; Bollerslev, 1998; Andersen; Bollerslev; Diebold; Ebens, 2001;
Andersen; Bollerslev; Diebold; Labys, 2003; Andersen; Bollerslev; Diebold, 2007; Corsi,
2009; Corsi; Renò, 2012; Bollerslev; Hood, et al., 2018; Cascaldi-Garcia et al., 2021)
This essay collection is divided into four main chapters. Chapter 1 provides a brief
review of volatility models and their historical development. This chapter aims to motivate
the model selection in the following chapters and introduce the primary literature
related to volatility modeling and forecasting.
Chapter 2 focuses on two papers that examine the effect of implied volatility and
the data sampling frequency for realized volatility forecasting. Both papers focused on
foreign exchange markets and enriched the academic literature in this field. We provide
new empirical evidence that implied volatility from short-term options (one day and one
week) is highly relevant for forecasting realized volatility of the next day. In the paper
about data sampling frequency, we confirm that high-frequency data are superior to
low-frequency counterparts. However, this advantage is statistically significant only for
short forecast horizons (less than five days). For longer forecast horizons, low-frequency
data become competitive.
13
Introduction
Chapter 3 contains two papers that analyzed the effect of news announcements
and other events on the stock market volatility and the volatility of bitcoin. The third
essay sheds new light on the issue of the Central bank announcements and realized
volatility of stock markets in G7 countries. It focuses on the impact of policy rates and
quantitative easing announcements of domestic and foreign central banks on realized
volatility before, during, and after the event. The results show that volatility increases
on the day of an interest rate announcement by the domestic central bank. Additionally,
evidence indicates a volatility decline five days after an interest rate announcement
across all countries in the analyzed sample. The results also show that quantitative
easing announcements have no impact on stock market volatility.
The fourth essay shifts the focus to the volatility of bitcoin, which incredible price
appreciation, especially during the year 2017, attracted much attention not only from
speculative investors but also from the academic literature. The essay focuses on the
effect of news and sentiment about bitcoin regulation, the hacking of bitcoin exchanges,
and scheduled macroeconomic news announcements on the realized volatility of bitcoin
and its jump component. The results show that realized variance and its jump component
exhibit similar dynamics and react similarly to various types of news. The volatility of
bitcoin reacts most strongly to news on bitcoin regulation, positive investor sentiment
regarding bitcoin regulation extracted using Google searches, and most notably, hacking
attacks on cryptocurrency exchanges. Quantile regression reveals that hacking attacks
have a powerful impact on the upper conditional distribution of bitcoin volatility. The
analysis also provides evidence that the volatility of bitcoin is not influenced by most
scheduled US macroeconomic news announcements, such as government budget deficits,
inflation, or even monetary policy announcements. On the other hand, bitcoin responds
with increased volatility to announcements of forward-looking indicators, such as the
consumer confidence index.
Chapter 4 shows the last two essays focusing on the two most recent crises related to
the COVID-19 pandemic and the Russo-Ukrainian war in early 2022. Both events created
unprecedented volatility in financial assets and provided valuable information to help
us understand the volatility behavior during such extreme periods. The essay about
the COVID-19 pandemic analyzes the stock market realized volatility in 23 countries. It
focuses on the effect announcements of policy interventions and responses intended to
buffer the pandemic’s short-term economic impact and offset financial turmoil. Under
the augmented heterogeneous autoregressive model framework, the results indicate
that the international calming effect of COVID-19 economic policy actions originates
from the US macroprudential policy announcements.
14
Introduction
The last essay analyzed the beginning of the Russo-Ukrainian crisis in early 2022 that
led to the rapid depreciation of the Russian ruble. It models intraday price fluctuations
of the USD/RUB and the EUR/RUB exchange rates from the 1st of December 2021 to
the 7th of March 2022. It utilizes a novel approach to this issue by using intraday (highfrequency)
data of google searches and implied volatility instead of commonly used
daily data. Google searches and implied volatility represent the investor’s attention and
expectations proxies. The results indicate that both approaches help predict intraday
price fluctuations of the two exchange rates, although implied volatility encompasses
intraday attention.
At the end of this habilitation thesis is a conclusion summarizing the previous
chapters. The full texts of all essays presented in this thesis are provided in the Appendix
A.
15
1 A Brief Review of Volatility Models
A large number of realized volatility models were developed during the last several
decades. One of the stylized facts of financial time series is so-called time-varying
volatility clustering. In simple terms, it means that there are observable periods of
high volatility followed by periods of low volatility and vice versa. This stylized fact is
exploited by the popular autoregressive conditional heteroskedasticity model (ARCH)
by Engle (1982). The simplest version of this model is defined as:
σ2
t = α0 +
q
∑
i=1
αiϵ2
t−i (1.1)
where σ2
t is the conditional variance at time t, α0 . . . αi are estimated coefficients, and
ϵ2
t−i is a random error. A few years after the ARCH model, probably the most famous
extension of this model was developed, called the generalized ARCH model (GARCH)
by Bollerslev (1986). The structure is similar to the basic ARCH model but with added
lagged values of volatility σ2
t−j.
σ2
t = α0 +
q
∑
i=1
αiϵ2
t−i +
p
∑
j=1
βjσ2
t−j (1.2)
The ARCH and GARCH models were initially estimated on low frequency (usually
daily) data. Researchers continued to extend and modify these models to capture specific
stylized facts for various time series. Some of these variations include, for example,
nonlinear GARCH (NGARCH; Lanne et al., 2005), integrated GARCH (IGARCH; Engle;
Bollerslev, 1986), exponential GARCH (EGARCH; Nelson, 1991), quadratic GARCH
(QGARCH; Sentana, 1995), GJR-GARCH Glosten et al. (1993) and many others (for an
early review, see Poon et al., 2003).
Financial time series often exhibit much more stylized facts than only volatility
clustering. Among the essential characteristics usually belong long-memory, fat tails, or
self-similarity. These properties could be very challenging for econometric models, and
a standard GARCH model is not able to replicate all of these properties due to its shortmemory
nature. There exist some models that are able to capture the long-memory by
using the fractional difference operators, e.g., FIGARCH (Baillie et al., 1996) or ARFIMA
(Granger et al., 1980). However, these models are hard to fit and also contain other
complications, such as a lack of clear economic interpretation.
With the increased availability of intraday high-frequency data, the new family
of models emerged based on the Heterogeneous Autoregressive models of Realized
17
1. A Brief Review of Volatility Models
Volatility (HAR-RV) developed by Corsi (2009) which uses realized volatility estimators
of Andersen; Bollerslev (1998) and Andersen; Bollerslev; Diebold; Ebens (2001). The
basic form of the model is defined as follows:
RVt = β1 + β2RVD
t−1 + β3RVW
t−1 + β4RVM
t−1 + ϵt (1.3)
RVt is a realized volatility in day t and RVD
t−1, RVW
t−1, RVM
t−1 represent are average
volatilities over the past day, week, and month.
This model does not have typical characteristics of a long-memory model but is still
able to reproduce most of the volatility stylized facts, including long-memory (e.g.,
Andersen; Bollerslev; Diebold, 2007; Vortelinos, 2017). Moreover, the model is easy to
estimate and economically intuitive. Its ordinary least squares properties allow us to add
other regressors effortlessly and calculate their significance levels. The original HAR
model was also modified by the decomposition of realized volatility to continuous and
jump components (e.g., Andersen; Dobrev, et al., 2012), the use of semivariances and
signed jumps (Patton; Sheppard, 2015), the inclusion of measurement error (Bollerslev;
Patton, et al., 2016), the incorporation of nontrading volatility components (Lyócsa;
Molnár, 2017; Lyócsa; Todorova, 2020) or the use of hidden Markov chains (Luo et al.,
2022), to name a few. It is also possible to use a multivariate version of the HAR model
(Chiriac et al., 2011; Bauer et al., 2011; Čech et al., 2017).
There is robust empirical evidence suggesting that models of volatility based on
high-frequency estimators provide superior forecasts compared to models based on
low-frequency data (e.g., Koopman et al., 2005; Andersen; Bollerslev; Diebold, 2007;
Corsi; Pirino, et al., 2010; Busch et al., 2011; Horpestad et al., 2019). However, there is
a possible issue with the microstructure noise in the high-frequency data. As a result,
high-frequency estimators of volatility worked well only for assets with significant trade
volume and high-quality data. Otherwise, the results could be biased and inefficient
(Andersen; Bollerslev; Diebold; Ebens, 2001). Many researchers tried to solve this issue
by creating alternative estimators of volatility (e.g., Ait-Sahalia et al., 2005; Bandi et al.,
2008; Barndorff-Nielsen; Hansen, et al., 2008; Andersen; Bollerslev; Meddahi, 2011; Liu
et al., 2015).
There also exist approaches that try to merge GARCH-type models with highfrequency
volatility estimators. The primary model in this category is represented
by the realized-GARCH model from Hansen; Z. Huang, et al. (2012), followed by several
alternative models (e.g., Xie et al., 2020; Wu et al., 2021).
The selected six essays in the following chapter predominantly utilize the various
modifications of the HAR class models of volatility. The choice of this type of model
18
1. A Brief Review of Volatility Models
was based on their relatively very good in-sample as well as out-of-sample performance,
computational efficiency, and the possibility to add additional variables and measure
their statistical significance easily. The model also allows to include different volatility
estimators, various decomposition of volatility, or forecast only some specific part of
volatility, such as volatility jumps.
19
2 Realized Volatility Forecasting
The first two papers in this essay collection focus on forecasting realized volatility. The
first paper challenges the idea that high-frequency data are superior to daily data for
forecasting purposes. We contribute to the literature by providing the evidence that this
statement is valid only for a short forecasting horizon, and daily data are also able to
provide comparable forecasts and offer other additional advantages. The second paper
focuses on the information content of implied volatility for forecasting realized volatility.
Our contribution demonstrates that the quoted implied volatility on the foreign exchange
market is highly relevant for realized volatility forecasting due to its forward-looking
nature.
2.1 FX market volatility modelling: Can we use low-frequency data?
Table 1: Paper Information and Author Contribution
Reference
Lyócsa, Š., Plíhal, T., & Výrost, T. (2021).
FX market volatility modelling: Can we use low-frequency data?
Finance research letters, 40, 101776.
Journal Rankings 2021 Citations
WoS AIS: 1.245 (Q2) WoS: 3
WoS IF: 9.848 (Q1) Scopus: 3
Scholar: 4
My Contribution (33%)
Conceptualization, Methodology, Data curation, Investigation
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
As mentioned in the previous chapter about volatility models, high-frequency data
are theoretically preferred (Andersen; Bollerslev; Diebold; Ebens, 2001) and often provide
superior forecasts for one-day time horizons (e.g., Koopman et al., 2005; Andersen;
Bollerslev; Diebold, 2007; Corsi; Pirino, et al., 2010; Busch et al., 2011; Horpestad et al.,
2019). Surprisingly, the academic evidence for longer forecast horizons is mostly missing.
We assume that for longer prediction (e.g., one month ahead), the information content
of intraday price fluctuations decreases, and daily data could offer comparable forecasts.
21
2. Realized Volatility Forecasting
Daily data would provide many benefits, e.g., lower microstructure noise, lower space
and computational requirements, high availability, etc.
In this study, we fill the gap in the literature and compare the forecasting accuracy of
several volatility models as a function of the forecast horizon on the six major currency
pairs. Our results confirm the dominance of high-frequency data estimators for forecasting
one-day-ahead volatility. However, for longer forecast horizons (5-days or more),
low-frequency models can provide statistically comparable results. These findings provide
important insights for practitioners and policymakers to help them to choose lowor
high-frequency volatility models for a particular setting.
2.1.1 Data and Methodology
The study analyzes major foreign exchange market pairs: AUD/USD, EUR/USD, GBP/USD,
USD/CAD, USD/CHF, and USD/JPY. This choice was motivated by the high liquidity of
selected assets and the high quality and availability of high-frequency data necessary for
meaningful analysis. The data were collected from OANDA using a 5-minute calendar
sampling over a 24-hour trading window. The whole sample covers the period from
May 2005 to the end of September 2019.
A high-frequency volatility estimator is represented by the sum of intraday 5-minute
squared returns, which is the most common choice in academic literature (e.g., Andersen;
Bollerslev, 1998; Andersen; Bollerslev; Diebold; Ebens, 2001; Liu et al., 2015). Lowfrequency
volatility estimators are calculated from daily open, high, low, and close
prices. We use three versions of estimators according to Parkinson (1980), Garman
et al. (1980), and Rogers et al. (1991). Moreover, motivated by Patton; Sheppard (2009)
we take the average of these three estimators because the forecast combination could
increase the efficiency of the estimation process.
We choose three volatility models suitable for low- and high-frequency data and,
simultaneously, capable of replicating long memory and volatility clustering effects.
The models are the heterogeneous autoregressive model (HAR) of Corsi (2009), the
autoregressive fractionally integrated model (ARFIMA), and the realized generalized
autoregressive conditional heteroscedasticity (realized-GARCH) of Hansen; Z. Huang,
et al. (2012).
We use the standard specification of the HAR model with a few modifications. First,
we added a fourth volatility component that represents three-month realized volatility. It
is motivated by the fact that our forecasting horizon ranges from one day to three months.
Moreover, we tried the specification that utilizes positive and negative semivariances
(Patton; Sheppard, 2015) that captures the asymmetric volatility response. The last
22
2. Realized Volatility Forecasting
specification is motivated by Corsi; Reno (2009) and Horpestad et al. (2019), which
utilize absolute returns that are usually correlated with variance. All versions of the
HAR model are estimated using weighted least squares, where weights are reciprocal
values of the dependent variable (Clements et al., 2021).
Our ARFIMA-GARCH model uses the mean equation specified as:
RVt = α + ut (2.1)
1 −
p
∑
i=1
ϕiLi
(1 − L)d
ut = 1 +
q
∑
j=1
θjLj
εt (2.2)
εt = vtηt, where ηt ∼ iid(0, 1) (2.3)
where d is the differencing parameter (e.g., Granger et al., 1980), vt is the timevarying
volatility and ηt is an iid variable following a flexible distribution (Johnson,
1949b; Johnson, 1949a). The variance equation is the exponential GARCH model of
Nelson (1991):
ln v2
t = ω + αzt−1 + γ (|zt−1| − E|zt−1|) + β ln v2
t−1 (2.4)
The sign and the size effects are captured by α and γ, and zt is the standardized innova-
tion.
Finally, the realized GARCH model of Hansen; Z. Huang, et al., 2012 uses the mean
equation of daily returns:
Rt = κ + ut (2.5)
1 −
p
∑
i=1
ϕiLi
(1 − L)d
ut = 1 +
q
∑
j=1
θjLj
εt (2.6)
εt = vtηt, where ηt ∼ iid(0, 1) (2.7)
The variance and the measurement equations are:
ln v2
t = ω + α ln RVt−1 + β ln v2
t−1 (2.8)
ln RVt = ξ + δ ln v2
t + λ1zt + λ2(z2
t − 1) + wt , wt ∼ N(0, θ) (2.9)
Each of the mentioned models is estimated with low-frequency volatility estimator
and then with an estimator based on high-frequency data. The out-of-sample forecasting
procedure uses a rolling window of 1000 days. To mitigate model uncertainty, we apply a
combination of forecasts based on weighted averages with weights given by discounted
23
2. Realized Volatility Forecasting
forecast error. We combine low-frequency models, high-frequency models, and all
models. Volatility forecasts are then evaluated using two statistical loss functions (mean
square error and quasi-likelihood) and model confidence set by Hansen; Lunde, et al.
(2011).
2.1.2 Results
To compare the predictive abilities of low- and high-frequency models, we forecast
volatility for one day and up to three months ahead, calculate the loss functions, and
estimate if there is some statistically significant difference between the forecasts.
For most cases, high-frequency models could provide slightly lower values of the
loss functions, which indicates better performance. Still, this difference became quickly
statistically insignificant with a longer forecast horizon (more than five days). Moreover,
the combination of the forecasts proves to be beneficial for all forecast horizons. These
results can be generalized for all our exchange rates and loss functions.
When analyzing the results a bit deeper, we found that the low-frequency model
forecasts tend to be more biased, especially for shorter forecast horizons, compared
to high-frequency models. It suggests that resolving the bias could further improve
forecasting accuracy. Finally, we study how the difference between the accuracy of highand
low-volatility forecasting models changes over time and in high-/low-volatility
periods. The results suggest that high-frequency models tend to be superior during
periods of increased volatility.
We conclude that if high-frequency data are not available, then low-frequency data
can be used to estimate and predict long-term market volatility. These results have
implications for researchers and investors alike, as they demonstrate that low-frequency
volatility models can provide competitive performance to that of high-frequency models
under some circumstances. Our study notes that high-frequency data might not always
be worth the much higher acquisition, data management, and processing costs, especially
if the forecast horizon of interest is sufficiently long.
This paper contains several limitations. It is focused solely on the major currency
pairs from the foreign exchange market. Even though the results provide some intuition
on how the volatility forecasts behave, it is not recommended to generalize the results for
all assets without further analysis. Moreover, we focused on forecasting performance, not
the performance in real applications. The results could become more or less statistically
significant for specific applications, like risk management, portfolio management, or
valuation of assets.
24
2. Realized Volatility Forecasting
2.2 Modeling realized volatility of the EUR/USD exchange rate: Does implied
volatility really matter?
Table 2: Paper Information and Author Contribution
Reference
Plíhal, T. & Lyócsa, Š.(2021).
Modeling realized volatility of the EUR/USD exchange rate:
Does implied volatility really matter?
International Review of Economics & Finance, 71, 811-829.
Journal Rankings 2021 Citations
WoS AIS: 0.572 (Q3) WoS: 3
WoS IF: 3.399 (Q2) Scopus: 4
Scholar: 8
My Contribution (50%) - Corresponding author
Methodology, Formal analysis, Investigation
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
We can distinguish two main types of volatility. The first type is realized or historical
volatility. It is calculated from historical data and represents the past behavior of the
financial asset. The most simple estimator of this type of volatility is a standard deviation
of prices for a selected time period. However, a large body of various estimators of
historical volatility exists, and each of them can capture slightly different information
from the historical prices. The second type of volatility is implied volatility estimated
from the prices of financial derivatives. As a result, it represents the expected volatility
of the underlying asset of the specified derivative. In other words, it captures the market
view on the future development of volatility.
Because of the forward-looking nature of implied volatility, there is a question of
whether it can improve the forecasting performance of realized volatility models. This
issue was studied in the academic literature, but the results are mixed. Canina et al.
(1993), Becker; Clements; White (2006), Becker; Clements; White (2007), and Becker;
Clements (2008) found very little incremental information of implied volatility. On the
other hand, Day et al. (1992), Christensen et al. (1998), Blair et al. (2001), Busch et al.
(2011), Han et al. (2013), Kambouroudis et al. (2016), Y.-H. Wang et al. (2016), and
Kourtis et al. (2016) claimed that implied volatility is useful and contributes to realized
volatility models. Moreover, Christensen et al. (1998) and Blair et al. (2001) argue that
25
2. Realized Volatility Forecasting
implied volatility can capture all necessary information for realized volatility forecasting,
and we do not need to use realized volatility at all.
All the mentioned studies focused on the stock market, and literature about other
assets is scarce. The foreign exchange market was examined in Jorion (1995), Xinzhong
et al. (1995), Pong et al. (2004), Covrig et al. (2003), and Busch et al. (2011) but also
with the mixed results. Additionally, most studies relied on implied volatility derived
from options with 30-day maturity even though there exist options with even shorter
maturities that could provide more precise information about the near future.
Our contribution to the academic literature is threefold. First, current literature
usually works only with implied volatility representing one-month maturity. Therefore,
we focus our analysis on foreign exchange implied volatility with different maturities:
one day, one week, and one month. Second, the existing literature seems to suggest
that the role of implied and historical volatility differs concerning the forecast horizon.
However, the recent evidence is based mainly on at least one-month-ahead forecasts. We
fill this gap in the literature and provide evidence for three different forecast horizons.
Third, we employ several volatility model specifications and combination forecasts
to address the possibility that the findings from previous studies might have been
subject to volatility model choice uncertainty. We compare individual forecasting models
that: a) separate volatility into positive and negative semi-volatilities, b) separate the
continuous volatility component and the jump component, c) weight volatility based on
the estimated variance of the measurement error. We also try to improve the forecast
through model combination.
2.2.1 Data and Methodology
Our study covers 12 years of data from 2006 to the end of 2017 and focuses on one of
the most liquid markets in the world, the EUR/USD exchange rate. Realized volatility
is calculated from high-frequency data provided by DucasCopy. Implied volatility is
collected from Bloomberg and represents quoted implied volatility for EUR/USD options
traded on the over-the-counter market. Both time series are synchronized to the same
time zone, weekends are removed, and a 24-hour trading day is used.
Our methodology is based on the HAR model of Corsi (2009) and its modifications.
The simple HAR model is specified as follows:
RVH
t+1 = β1 + β2RVD
t + β3RVW
t + β4RVM
t + ϵt+1 (2.10)
26
2. Realized Volatility Forecasting
where RVH
t+1 is the average daily realized volatility over the next H days, i.e., the
forecast horizon. In this study, we consider H = 1, 5, and 22, i.e., forecasts of intra-day
volatility over the next one, five, and twenty-two trading days, respectively. RVD
t , RVW
t ,
and RVM
t are the averages of realized volatility over the previous one, five, and twentytwo
trading days, respectively. These volatility components should mimic daily (D),
weekly (W), and monthly (M) volatility trends.
Then, we create the same model but replace the right-hand side variables with
implied volatilities with corresponding maturities:
RVH
t+1 = β1 + β2IVD
t + β3IVW
t + β4IVM
t + ϵt+1 (2.11)
Both models are then combined into one model with both realized and implied
volatility components:
RVH
t+1 = β1 + β2RVD
t + β3RVW
t + β4RVM
t +
β5IVD
t + β6IVW
t + β7IVM
t + ϵt+1 (2.12)
Following a similar fashion, we tried three modifications of the basic HAR model.
First, we use the decomposition of realized volatility to its continuous component (CC)
and jump component (JC), according to Andersen; Dobrev, et al. (2012). The second
modification follows the work of Bollerslev; Patton, et al. (2016) and controls for measurement
error. The third version of the model decomposes realized volatility into positive
and negative semivariances (Patton; Sheppard, 2015). All models are estimated in the
realized volatility specifications and then augmented by implied volatility components,
as in the simple HAR model.
Lastly, we try the forecast combination of all model forecasts that do not contain
implied volatility compared to the combination of all models that include implied volatility
information. We choose the simple average of individual forecasts as a method for
combining model predictions because it proves to be a safe and effective approach (e.g.,
Genre et al., 2013; Santos et al., 2014; Lyócsa; Molnár, 2016; Lyócsa; Molnár; Todorova,
2017).
Forecasting performance is then evaluated using two loss functions, MSE and QLIKE,
that provide a consistent ranking of forecasts (Patton, 2011). Moreover, we apply the
model confidence set by Hansen; Lunde, et al. (2011).
27
2. Realized Volatility Forecasting
2.2.2 Results
Our results provide new empirical evidence suggesting that implied volatilities from
options with a one-day and one-week maturity might be more relevant when explaining
the volatility of the next one day and one week on the EUR/USD exchange rate. The
inclusion of implied volatility usually decreased the size of the realized volatility coefficients,
and the fit of the models improved by up to 29%, 27%, and 9% for the one-day,
one-week, and one-month-ahead forecasts, respectively.
The results also indicate that realized volatility is still beneficial because the models
that include both types of volatility usually provide greater fit and forecasting performance.
However, the improvements achieved by including realized volatility in the
model are smaller than in the case of implied volatility. Moreover, the forecast horizon
is also essential, and implied volatility’s importance increases with the lower forecast
horizon.
Finally, our out-of-sample analysis confirms the incremental information of implied
volatility. The model confidence set always chooses only the models that contain implied
volatility in the superior set of models. These results are supported by estimated
improvements in forecasting accuracy as measured by loss functions MSE and QLIKE.
These results are supported by all model specifications, forecast horizons, and even
model combinations.
The main limit of this paper is only one selected asset, EUR/USD. Further analysis is
needed to confirm the results for different currency pairs or other assets. The potential
assets are limited by the availability of good quality high-frequency data and liquid
over-the-counter options market with quoted implied volatility data.
28
3 News Announcements and Realized Volatility
This chapter presents two essays explaining volatility and the effects of various events.
The first paper focuses on the stock markets of the developed countries and examines the
announcements of the central banks. The paper’s main contribution is to show volatility
behavior several days before and after the announcement and provide robust evidence
in a multicountry study. The second paper analyzes the cryptocurrency Bitcoin, which
was on a large boom. It examines the connection of bitcoin to the real economy by
measuring the reaction of volatility to various macroeconomic news announcements.
The study also included other events related to policy regulation of cryptocurrencies and
hacking attacks that lead to significant financial losses of invested money and increased
uncertainty about this asset.
3.1 Central bank announcements and realized volatility of stock markets in
G7 countries
Table 3: Paper Information and Author Contribution
Reference
Lyócsa, Š., Molnár, P., & Plíhal, T. (2019).
Central bank announcements and realized volatility
of stock markets in G7 countries.
Journal of International Financial Markets, Institutions and Money, 58, 117-135.
Journal Rankings 2019 Citations
WoS AIS: 0.574 (Q2) WoS: 8
WoS IF: 2.553 (Q1) Scopus: 7
Scholar: 20
My Contribution (33%)
Conceptualization, Methodology, Data curation, Investigation,
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
Macroeconomic news announcements are one of the drivers of price changes in
financial assets. It allows market participants to update their expectations about future
asset prices and adjust their positions accordingly. Due to the importance of this topic, a
29
3. News Announcements and Realized Volatility
large body of literature assesses the impact of macroeconomic news announcements on
financial markets. The literature focused on various asset classes, for example:
• Stock markets: Flannery et al. (2002), Ehrmann et al. (2004), Bernanke et al. (2005),
Bekaert; Engstrom (2010), and Hussain (2011).
• Foreign exchange markets: Almeida et al. (1998), Andersen; Bollerslev; Diebold;
Vega (2003), Ehrmann et al. (2005), Evans et al. (2010), Ben Omrane et al. (2015),
Petralias et al. (2015), El Ouadghiri; Uctum (2016), and Neely et al. (2010).
• Government bond markets: Fleming et al. (1997), Fleming et al. (1999), Christie-David
et al. (2002), Balduzzi et al. (2001), Gürkaynak et al. (2005), Beechey et al. (2009),
and El Ouadghiri; Mignon, et al. (2016).
• Several asset classes: Boyd et al. (2005), Faust et al. (2007), and Bartolini et al. (2008).
One of the most important macroeconomic announcements is related to the central
bank, and market participants pay particular attention to these types of news (e.g.,
Thorbecke, 1997; Thornton, 1998). Studies that focused on asset price volatility reported
often increased volatility during these announcements (e.g., Harvey et al., 1991; Ederington
et al., 1993; Dominguez, 1998; Nikkinen; Sahlström, 2001; Bauwens et al., 2005;
Dominguez, 2006; Nikkinen; Omran, et al., 2006; Beine et al., 2009, and many others).
We study the impact of central bank announcements on the volatility of equity
markets. In particular, we want to determine whether monetary policy decisions have a
stabilizing or a destabilizing impact on equity markets. The previous literature does not
offer a satisfactory answer to our question. Although the previous literature documents
an increase in volatility around news announcements, this cannot be interpreted as
finding that announcements have a destabilizing impact on financial markets. The
information content of news is usually high; therefore, it is only natural that volatility
is high around earnings announcements. For example, Chae (2005) argues that if the
information content of news is high, it will necessarily increase volatility in the market
because market participants adjust their views based on the news provided. Thus, the
consensus of market participants increases on the day of the news announcement.
A possible way to analyze some announcements’ stabilizing or destabilizing impact
is to evaluate volatility several days before and after the announcement. It was addressed
before in the literature but not to a satisfactory extent. The literature used daily data
and the GARCH framework (e.g., Bomfim, 2003; S.-J. Kim et al., 2004; Bauwens et al.,
2005) or implied volatility (e.g., Ederington et al., 1993; Nikkinen; Sahlström, 2004; Äijö,
2008; Aktas, 2011; Füss et al., 2011; Jiang et al., 2012; Marshall et al., 2012; Krieger et al.,
30
3. News Announcements and Realized Volatility
2015). However, volatility models estimated from daily data cannot reliably estimate
whether and how much volatility increased on a particular day. On the other hand,
implied volatility represents expectations about future volatility and therefore decreases
artificially after the announcement day drops from the expected future window (usually
30 days).
Our contribution to the literature is three-fold. First, most papers studying the impact
of monetary policy announcements on financial markets study foreign exchange markets,
not stock markets. Second, we use a multi-country sample, five central banks and their
impact on the volatility of eight stock market indices while controlling for the news
announcements of other central banks and news related to quantitative easing policies.
Third, and most importantly, the literature in this field is based on implied volatility,
and implied volatility mechanically declines after the scheduled announcements.
3.1.1 Data
Our sample covers the seven major advanced economies (G7) according to the International
Monetary Fund (Canada, France, Germany, Italy, Japan, the United Kingdom,
and the United States). The analyzed time period begins in January 2006 and ends in
November 2016. The dataset consists of three main parts: stock market indices, news
announcements from central banks, exchange rate and interest rate.
Stock market indices data with precalculated volatility measures are collected from
the Oxford-Man Institute of Quantitative Finance Realized Library1. The analyzed indices
are S&P 500 (United States), FTSE 100 (United Kingdom), TSX (Canada), NIKKEI 225
(Japan), STOXX 50 (Europe), DAX (Germany), CAC (France), and MIB (Italy).
The news announcements dataset focuses on the most important macroeconomic
news announcements related to the central banks: target interest rates and quantitative
easing. The data were downloaded from Bloomberg and contain interest rate decisions
from the Bank of Canada, European central bank, Bank of Japan, Bank of England, and
Federal Reserve Board. The data about quantitative easing were downloaded manually
from the official sites of the central banks, and we selected the days when the introduction
of quantitative easing or any change in this policy was announced. Quantitative
easing is an unconventional monetary policy used by central banks to stimulate their
economies when conventional monetary policy is no longer effective and usually consists
of purchasing long-term financial assets from banks and other financial institutions.
During our sample period, all the selected countries except Canada have experienced
quantitative easing.
1. http://realized.oxford-man.ox.ac.uk/data/download
31
3. News Announcements and Realized Volatility
The last part of the dataset consists of information about short-term interest rates (3
months) and effective exchange rates. The data are obtained from Bloomberg at a daily
frequency for each country.
3.1.2 Methodology
In this study, we focus on stock market volatility. However, many different volatility
estimators exist, each of which could contain slightly different information about volatility.
Therefore, we employ a methodology suggested by Patton; Sheppard (2009) who
advocates using a combination of various realized measures. As a result, we calculate a
simple arithmetic average of the eight standard realized volatility measures. In all our
calculations, we use the logarithm of the resulting mean realized variance, i.e., realized
volatility, because the distribution of the realized variances tends to be skewed to the
right and is subject to outliers.
We measure the impact of news announcements for each index separately using
a Heterogeneous AutoRegressive model with eXogenous variables and Generalized
AutoRegressive Conditional Heteroskedastic errors (DHARX-GARCH). The model
specification is as follows:
∆RVt = µ0 + µ1RVt−1 + µ2RVt−1,t−5 + +µ3RVt−1,t−22 +
S(i)
∑
s=1
κsEVs +
C(i)
∑
c=1
δcCVc + zt
(3.1)
zt = (1 + Θ1L1
)ϵt (3.2)
ϵt = σtηt, ηt ≈ iid(0, 1) (3.3)
We model differences of the RVt and level of RVt as a robustness check. The core
explanatory variables are inspired by the heterogeneous autoregressive model (HAR)
by Corsi (2009) that represents the lag of the daily, weekly, and monthly volatility. Then
we included event variables (EVs) and control variables (CVc).
Although the lagged realized volatilities and other exogenous variables capture most
of the market volatility dynamics, the error term zt may remain subject to autocorrelation
and conditional heteroscedasticity. We, therefore, model the term zt as a moving
average process (L is the lag operator), whereas we allow the evolution of σ2
t to follow
a suitable GARCH model. Two GARCH models are considered: the standard model
of Bollerslev (1986) and exponential GARCH model of Nelson (1991), each of them
32
3. News Announcements and Realized Volatility
with normal and SU-normal distribution (Johnson, 1949a; Johnson, 1949b). It leads us
to four models from which we choose the model that does not display autocorrelation
and conditional heteroscedasticity indicated by the Escanciano et al. (2009) test. If more
suitable models remain, we report a preferred specification according to the Bayesian
information criterion (BIC).
Estimating individual models allows us to observe market-level heterogeneities.
On the other hand, the realized volatilities of developed stock markets will be crosscorrelated
due to common global factors. Therefore, we also estimated the model as a
dynamic panel using the Dynamic common correlated effect model according to Chudik
et al. (2015).
Our event variables related to key interest rate announcements captures the following
information:
• volatility 5-days before, at, and 5-days after the news announcement
• a variance of analysts’ estimates of the target rate
• negative and positive surprise on the day of the announcement
• volatility one day after the announcement
• volatility 5-days after the announcement with respect to magnitude and direction
of the surprise
• international development of target rates
• local monetary policy indication of quantitative easing
• international monetary policy on quantitative easing
As control variables, we used other relevant events that may influence the market
volatility level on a given day. It includes, for example, the difference in short-term
interest rate, a logarithmic difference of an effective exchange rate, or day-of-the-week
effects.
3.1.3 Results
The time series of realized volatility for all the stock markets exhibit similar time patterns
with increased volatility during the financial and debt crises. The volatility’s logarithmic
transformation helped eliminate the skewness in the time series. Moreover, as is often
the case in the finance literature, realized volatility shows a high level of persistence
33
3. News Announcements and Realized Volatility
which supports our choice for modeling volatility via an autoregressive model: the
DHAR-GARCH model.
We first evaluate the results across all markets because the rate announcements and
quantitative easing announcements are relatively rare, and the per-market estimators
may not have the proper power to detect changes in realized volatility. We utilize two
aggregation techniques: the dynamic common correlated error (DCCE) model and the
simple mean group (MG) estimator.
Before the news announcement, increased uncertainty of the outcome tends to increase
volatility. At the same time, volatility before announcements tends to decrease
generally, but the effect is relatively small (approx 1.3% decrease in realized variance).
At the news announcement, there was a notable increase in realized variance (approx.
30.5%). The effect is particularly strong when coupled with large surprises. However, the
effect of surprises seems to be asymmetric, and a more substantial effect appears when
the announced rate is below expectations. Five days after the announcement, estimated
volatility declined (approx. 5% decrease in realized variance).
When foreign central banks make announcements, our estimates suggest that volatility
increases slightly before and on the news announcement days. After a news announcement,
volatility tends to decline. These effects, although occasionally significant,
are small. Interestingly, news related to quantitative easing led to a slight decrease in
realized volatility before and on the day of the announcement but only when reported
by the domestic central bank. Conversely, volatility increased if foreign central banks
reported initialization or continuation of the QE.
The key results from the individual market-level analysis of our DHAR-GARCH
models show that variables RVt−1, RVt−1,t−5, and RVt−1,t−22 explain the behavior of
volatility changes well because they are often significant with strong effects.
It appears that during the five days before the interest rate announcement, most
of the time-realized volatility decreases, except for the Japanese, Italian and European
markets. However, except for Japan, changes in volatility before news announcements do
not appear significant. On the announcement day, we observed that the stock market’s
volatility increased in all the countries. This finding was expected, and the effects are
strong. The coefficients range from a 2.2% (U.K.) to a 9.4% (U.S.) increase in the log of
the realized variance compared to the levels of volatility one day before (approx 10.9%
and 54.8% of the realized variance, respectively). Volatility five days after the interest
rate announcement has not decreased compared to the level of volatility before the event
day. Therefore, a monetary policy news announcement appears to affect volatility only
on the day of the announcement and not during the period after the announcement. This
34
3. News Announcements and Realized Volatility
result is statistically significant for most countries. We also find that surprises matter.
However, their effect differs with respect to any given market.
Realized volatility in the stock markets also appears to be increasing during the days
when central banks make interest rate announcements in other countries. However, this
effect is significant only in the U.K. and Canada. This result is very intuitive. Canada
has strong ties to the U.S., and the U.K. has strong ties to the EU; therefore, one would
expect these countries to respond to the announcement of foreign central banks. Most
likely, Canada is strongly responding to FED announcements, and the U.K. is strongly
responding to ECB announcements.
Lastly, we focused on the effect of announcements about quantitative easing. The
result is in contrast to the results achieved via the DCCE model. Therefore, it appears
that when controlling for common factors, the effect of the QE on market volatility
diminishes. We did not observe a significant increase or decrease in volatility after
quantitative easing announcements; the only difference was the reaction of the U.S. stock
market.
The research provides robust results but still has some limits. Generalizing results in
a multicountry study is not easy because each country behaves slightly differently, and
the results could be sensitive to the selected sample of countries. Therefore, we analyzed
each country separately and also in a panel setting. We used approximately ten years of
data, including the financial crisis, the longest available sample from our data sources,
and we think it is a reasonable sample. Still, the results could differ and be time-varying
for longer or other periods.
3.2 Impact of macroeconomic news, regulation and hacking exchange
markets on the volatility of bitcoin
This paper focuses on bitcoin volatility and tries to identify its drivers. Bitcoin is a
fully decentralized cryptocurrency with no central authority responsible for its value,
and it is not linked to any particular country. These differences distinguish bitcoin
from traditional fiat currencies and create a gap in our knowledge about the bitcoin
reactions to some fundamental determinants. Despite the endless discussion about the
nature of bitcoin, most research papers agree that bitcoin is not a currency or a long-term
investment. However, it should be viewed as a form of speculative, high-risk investment.2
2. See, e.g., Yermack (2013), Kristoufek (2013), Bouoiyour; Selmi (2015), Cheah et al. (2015), Bouoiyour;
Selmi, et al. (2016), Ciaian et al. (2016), Corbet; Lucey, et al. (2019), Baur; Dimpfl (2018), Baur; Hong, et al.
(2018), Smales (2018), Symitsi et al. (2019), Shahzad et al. (2019), Kliber et al. (2019), and Charfeddine
et al. (2019)
35
3. News Announcements and Realized Volatility
Table 4: Paper Information and Author Contribution
Reference
Lyócsa, Š., Molnár, P., Plíhal, T., & Širaňová, M. (2020).
Impact of macroeconomic news, regulation and hacking exchange markets
on the volatility of bitcoin.
Journal of Economic Dynamics and Control, 119, 103980.
Journal Rankings 2020 Citations
WoS AIS: 1.062 (Q2) WoS: 19
WoS IF: 1.588 (Q3) Scopus: 19
Scholar: 45
My Contribution (25%)
Conceptualization, Methodology, Data curation, Investigation,
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
Extensive research has been done to identify the main factors influencing bitcoin
price formation. The main drivers of bitcoin price include:
• the interaction between supply and demand (Ciaian et al., 2016)
• market microfundamentals (e.g, the velocity of bitcoin, the exchange trade ratio)
(e.g., Kristoufek, 2013; Bouoiyour; Selmi, 2015; Bouoiyour; Selmi, 2017)
• the price of gold and attention paid to bitcoin news (Bouoiyour; Selmi, 2017)
• market sentiment (Cretarola et al., 2017)
• the network hash rate (e.g., Kristoufek, 2013; Ciaian et al., 2016; Bouoiyour; Selmi,
2017)
• news about regulatory actions (Auer et al., 2018)
• global financial development, oil prices, and the EUR/USD exchange rate (Wijk,
2013)
• output as an important long-term factor (Kristoufek, 2013)
• news related to unemployment and durable goods associated with bitcoin returns
(Corbet; Larkin, et al., 2018)
36
3. News Announcements and Realized Volatility
We fill the gap in the literature by studying the role of a broad set of macroeconomic
news announcements on the bitcoin price volatility and its jump component. We divide
the scheduled macroeconomic news announcements into eight categories. Moreover,
we explore news related to regulation, sentiment, and hacking of exchange markets.
3.2.1 Data
Our dataset contains 2152 observations from January 2013 until December 2018. We collected
several types of data representing, for example, bitcoin volatility, macroeconomic
news announcements, news and sentiment related to the cryptocurrency market, and
security breaches of cryptocurrency exchanges. Our data contains entire calendar days,
including weekends, and are synchronized to the UTC timezone.
Bitcoin volatility is calculated from BTC/USD individual trades from the Bitstamp
exchange. The recent advances in financial econometrics have led to the development
of many volatility estimators. Moreover, because we have data for individual trades,
it enables a wide range of sampling frequencies and schemes (calendar or business
sampling). The actual data generating process is unknown. Therefore, we follow the
advice of Patton; Sheppard (2009) and create a simple average of different estimators
and sampling frequencies.
In total, we used four volatility estimators:
• standard realized variance that is defined as the sum of intraday squared returns
• first-order adjusted realized variance that controls for first-order serial dependence
in intraday returns (French et al., 1987; Patton; Sheppard, 2009; Liu et al., 2015)
• the bipower estimator of Barndorff-Nielsen; Shephard (2004) that leads to consistent
estimates in the presence of jumps
• the median realized variance estimator of Andersen; Dobrev, et al. (2012) that also
accounts for jumps in the price process, which is likely in the highly volatile bitcoin
time series
We use seven different frequencies for calendar and business sampling for each estimator.
Our final volatility estimator is then calculated as a simple average of these values.
We also analyze the jump component of volatility estimated according to Andersen;
Dobrev, et al. (2012) as the difference between the realized variance and its continuous
component.
Logarithmic transformation is applied to all our volatility estimators, jump, and
continuous components. It is a common practice in the literature, leading to a more
37
3. News Announcements and Realized Volatility
symmetric distribution that is much closer to the normal distribution than the raw
realized volatility series, which is more suitable for standard time-series modeling
purposes, e.g., autoregressive volatility models (Taylor, 2017; Andersen; Bollerslev;
Diebold; Ebens, 2001; Andersen; Bollerslev; Diebold; Labys, 2003; Andersen; Bollerslev;
Diebold, 2007).
Macroeconomic announcement data are collected from Bloomberg. In total, we used
34 different news divided into eight categories (real economic activity, household consumption
decisions, firm investment decisions, government finances, external balances,
price evolution, monetary policy decisions, and forward-looking). We focus only on
the news related to the US economy because the exchange rate for bitcoin is usually
quoted against the US dollar. We build our database based on relevant studies focusing
on the effect of US macroeconomic news. The existing literature also consistently reports
that US macroeconomic announcements often have a stronger impact on the behavior
of asset prices than national surprises (e.g., Andersen; Bollerslev; Diebold; Vega, 2003;
Jaggi et al., 2016).
Macroeconomic news announcements are represented by dummy variables that
indicate the date of the announcement. Because all our news is scheduled, it was known
in advance that some announcement happened on this date. To decrease the number of
parameters in the model, we decided to aggregate the information about news announcements
for each category. Our created variable measures how many news announcements
from each category happen daily from the total number of possible announcements.
These variables help us to answer the research question of whether the volatility of the
bitcoin price series reacts to economic fundamentals or if its behavior is unrelated to the
condition of the US economy.
Cryptocurrency markets tend to react to news related to possible regulatory actions
(Auer et al., 2018) and cybercrime events related to the hacking of cryptocurrency exchanges.
Therefore, we control for these effects in our model using Financial Times
articles, Google Trends sentiment, and recorded days of hacking attacks on cryptocurrency
exchanges. We manually selected the most important news from the Financial
Times related to bitcoin regulation. We used the ProQuest newspaper database and
ended up with 55 news for the period from January 2013 to December 2018. We recorded
the date when the authorities or journalists discussed the regulatory action from each
news item, and we used three dummy variables to capture the event. To control for
possible information leakage and lagged effect, we also created a dummy variable for
one day before and one day after the article release date.
38
3. News Announcements and Realized Volatility
Several recent studies have used volume data from Google searches to explain behavior
in the bitcoin exchange market (Kristoufek, 2013; Garcia et al., 2014; Cheah et al.,
2015; Urquhart, 2018; Aalborg et al., 2019). We used a similar methodology to estimate
general sentiment about cryptocurrencies divided into three categories: cryptocurrency
supporting sentiment, neutral, and nonsupporting sentiment. To obtain daily data for
such an extended period, we follow the method used by Bijl et al. (2016) and N. Kim
et al. (2019): we apply a rolling window and calculate the standardized Google Trends
(SGT) value.
Cryptocurrency cyber attacks were retrieved from Hackmageddon website 3. According
to Corbet; Cumming, et al. (2019), suspicious price behavior on cryptocurrency
exchanges occurs prior to the announcement of hacking. Therefore, we create a variable
Hackt that equals the percentage of the estimated loss from the total market capitalization
of bitcoins for the day of the official announcement of the hacking and one day
prior, 0 otherwise.
The last control variables are related to bitcoin derivative contracts. The first is the
linear time trend, which captures the long-term effect of the changing volatility. The
second is a linear time trend, which returns a value of 0 prior to 10 December 2017, thus
prior to the introduction of derivatives on the CBOE and CME (18 December 2017), and
a time trend value of 1 for 11 December 2017, 2 for 12 December 2017, etc.
3.2.2 Methodology
Our model is based on the standard realized volatility heterogeneous autoregressive
model of Corsi (2009). Even though this model is not a long memory in nature, it is
known to capture the long memory property of the time series well. Its outstanding
relative performance was used in studies from, e.g., Trucíos (2019), Chan et al. (2018),
and Lyócsa; Molnár; Plíhal (2019). Another possible alternative is some GARCH class
model that was also often used for modeling bitcoin volatility (e.g., Chu et al., 2017;
Katsiampa, 2017; Conrad et al., 2018; Baur; Dimpfl; Kuck, 2018). However, we choose to
stick with the HAR model because adding other regressors and estimating the model is
easy, which makes it more suitable for our analysis.
Our full model specification with all control variables looks as follows:
3. https://www.hackmageddon.com/category/security/cyber-attacks-timeline/
39
3. News Announcements and Realized Volatility
RVt =β1 + β2RVD
t−1 + β3RVW
t−1 + β4RVM
t−1+
RVD
t−1 × (δ1FTNt−1 + δ2FTNt + δ3FTNt+1)+
RVD
t−1 × (δ4NosTt−1 + δ5NeuTt−1 + δ6SupTt−1)+
RVD
t−1 × δ7Hackt + δ8Trendt + δ9Trendt × I(t > 10th
Dec 2017)+
RVD
t−1 ×
8
∑
i=1
γiDi,t−1 + ϵt (3.4)
RVt is a realized volatility in day t that we are trying to explain. RVD
t−1, RVW
t−1, RVM
t−1
are the components of the standard HAR model and represent are average volatilities
over the past day, week, and month. All RV variables are in logarithmic form. FTNt−1,
FTNt, FTNt+1 are variables related to Financial Times articles. NosTt−1, NeuTt−1, and
SupTt−1 stand for non-supporting, neutral, and supporting sentiment respectively. Hackt
represents hacking attacks, and Trendt are two trend variables. The parameters of interest
are γi, i = 1, 2, ..., 8, which correspond to the effect of the macroeconomic news
announcement on the volatility of bitcoin. Specifically, the γi coefficients indicate how
the next day’s volatility, at time t, is anticipated to change if a given (ith) macroeconomic
news item is announced on the next day.
Compared to the standard HAR model, our specification uses an interaction of the
lagged realized volatility with non-volatility components. We are motivated by our
expectation that the effect of the news announcement and other non-volatility variables
might differ concerning the current level of market volatility. Therefore, the interaction
estimates changes in the next day’s volatility relative to the previous day’s level of
volatility.
Our second model follows the same structure, but we replaced all RV variables on
both sides of the equation with the jump component of volatility JC. It allows us to focus
only on the jump component of volatility and provide evidence of how it is affected by
news announcements. Both models are estimated via OLS with heteroskedasticity and
autocorrelation corrected standard errors (Newey et al., 1994).
Moreover, we estimate both models using the non-crossing quantile regression
approach because volatility drivers’ absolute and relative importance might differ
across bitcoin volatility (jump component) distribution quantiles. The models are estimated
using the procedure described in Bondell et al., 2010 and considering quantiles
0.05, 0.25, 0.50, 0.75, 0.95.
40
3. News Announcements and Realized Volatility
3.2.3 Results
As a first step, we checked the bitcoin descriptive statistics and characteristics of its time
series. Bitcoin exhibited an unprecedented rise and subsequent fall from the end of 2017
until the end of our series in 2018. Moreover, bitcoin volatility peaked earlier in 2013,
when relative price changes were more significant. The average value of annualized
realized volatility of bitcoin is 8.529 in our sample. It corresponds to an annualized
standard deviation of 71.12%. For example, Bollerslev; Hood, et al. (2018) reports levels
of annualized standard deviation for commodities at 25.40%, equities at 20.60%, fixed
income at 3.10% and foreign exchange at 10.30%. It indicates that bitcoin is a highly
risky asset with enormous volatility compared to other assets. We also found a strong
long-memory property for bitcoin volatility time series even at the lag 100. Therefore,
our selection of the HAR model seems to be reasonable.
The persistence of the main HAR components does not seem to change substantially
across quantiles except for the monthly component, which shows higher coefficients
for lower quantiles. The overall level of volatility has not changed over time. However,
the volatility in lower quantiles increased, and on the other hand, volatility in higher
quantiles decreased. Moreover, introducing derivatives has only led to an increase in the
extreme quantiles of volatility distribution. Financial Times article effects also provide
statistically significant results. One day after the publication of selected news, the volatility
rises quite substantially (by 1.9%), and the effect is almost two times stronger on lower
quantiles of volatility. Google Trends sentiment and hacking attacks. Hacking attacks
can potentially increase volatility, especially in the higher quantiles. The substantial
difference between coefficients in high and low quantiles suggests that cryptocurrency
hacking events have the potential to lead to periods of extremely high volatility. Lastly,
neutral sentiment, which can be interpreted as general attention, increases the overall
level of realized volatility. A similar effect is observed for the supporting (positive)
sentiment. The results across quantiles show that the effects tend to increase for extreme
quantiles.
For our main variables of interest, macroeconomic announcements, we find that only
releases of forward-looking components tend to increase realized volatility of bitcoin.
This finding is not surprising because most empirical studies highlight the speculative
nature of bitcoin, which is more sensitive to exogenous market disturbances (crashes,
regulations) or factors influencing its use as a medium of exchange in black market
transactions and tax avoidance. Our results support the hypothesis that bitcoin volatility
does not react to macroeconomic news announcements in an economically substantial
way.
41
3. News Announcements and Realized Volatility
However, the statistically significant link between bitcoin volatility and specific
forward-looking macroeconomic announcements might suggest a potential for a more
fundamental role of bitcoin rather than it being a purely speculative asset. Alternatively,
as bitcoin currently does not entirely fulfill the role of a medium of exchange in the real
economy (Horra et al., 2019), its future utility for transaction purposes will derive from
its exchange rate with a widely accepted medium of exchange, the US dollar. Hence,
investors wishing to exchange bitcoin for this international currency will ultimately need
to incorporate the arrival of new, forward-looking information into their decision-making
process.
We also repeated the analysis for the jump component of realized volatility to better
understand the drivers of the most rapid changes in bitcoin volatility. The results for
control variables are often similar to what we found when analyzing realized volatility.
One of the most exciting results is the effect of news related to the regulation of cryptocurrencies
and the hacking of cryptocurrency markets on volatility. The effect of these
events on the jump component is slightly larger than in the case of volatility, but we
expected a more substantial result. Surprisingly, the effect of news articles related to
regulation is the same across quantiles, i.e., it does not increase the expected extreme
levels of the jump volatility component. The effect of hacking events is also confirmed
for jump components which clearly shows that it might be one of the main drivers of
the jump component. Therefore, it should not be omitted when evaluating the general
risks associated with investments in cryptocurrencies. The results for macroeconomic
news announcements are broadly in line with those found for overall volatility.
Moreover, we find a partial significance for government spending in higher quantiles
of volatility distribution. As before, the size of the coefficients suggests that the overall
effect is somewhat smaller. We, therefore, conclude that macroeconomic news has a
limited effect on the volatility process of bitcoin.
To summarize our results, volatility and its jump component seem to be driven primarily
by bitcoin-specific risk factors: regulation and hacking attacks on cryptocurrency
markets. Unlike traditional assets, bitcoin is almost uninfluenced by general macroeconomic
news, thus leading us to the conclusion that bitcoin is only weakly connected to
the overall economy via the forward-looking component.
The main limit of our analysis is probably the inclusion of macroeconomic news
announcements only from the United States. It was an inevitable choice that helped to
keep the number of variables in the models manageable. Moreover, we had reasonable
arguments for this choice supported by the academic literature.
42
4 Realized Volatility Reaction During Crises
The last two papers address two recent crises. The first paper is related to the COVID-
19 pandemic and the reaction of the stock markets around the world to government
actions. The second paper analyzes intraday investor attention during the beginning of
the Russian invasion of Ukraine in early 2022.
4.1 How to calm down the markets? The effects of COVID-19 economic policy
responses on financial market uncertainty
Table 5: Paper Information and Author Contribution
Reference
Deev, O., & Plíhal, T. (2022).
How to calm down the markets? The effects of COVID-19 economic policy
responses on financial market uncertainty.
Research in international business and finance, 60, 101613.
Journal Rankings 2021 Citations
WoS AIS: 0.695 (Q3) WoS: 0
WoS IF: 6.143 (Q1) Scopus: 1
Scholar: 2
My Contribution (50%)
Conceptualization, Methodology, Software, Visualization,
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
This paper focuses on the behavior of the world stock indexes’ volatility during
the COVID-19 pandemic and their reactions to economic policy actions. The pandemic
creates an extreme level of volatility in stock markets last seen more than a decade ago
during the global financial crisis (Ali et al., 2020; Baker et al., 2020). Many academic
studies have begun to emerge about this topic and analyzed various aspects related to
the pandemic, such as the number of cases and fatalities, reproductive number, nonpharmaceutical
interventions, investor sentiment, or simply fear.1
1. For example: Zhang et al. (2020), Ashraf (2020), Baig et al. (2021), Albulescu (2021), Díaz et al. (2022),
Zaremba et al. (2020), Ashraf; Goodell (2021), Bakry et al. (2021), Huynh et al. (2021), and Lyócsa;
Baumöhl, et al. (2020)
43
4. Realized Volatility Reaction During Crises
The effects of the pandemic were devastating. Therefore, governments, central banks,
and supervisory authorities worldwide have started an unprecedented amount of policy
interventions and responses to support the economy and the financial markets. Since
March 2020, governments have intensified their fiscal policy actions to buffer the shortterm
impact of the COVID-19 economic shock. Many central banks also intervened,
decreased interest rates, and provided liquidity injections.
In our paper, we focused on the effect of fiscal, monetary, and macroprudential
actions on the realized volatility in 23 countries worldwide. The goal is to analyze the
immediate spot market reactions based on high-frequency information and evaluate a
possible calming effect of policy actions on the stock markets. Moreover, we measure the
spillover impact of US and EU authorities’ actions on other markets. To our knowledge,
the paper is the first complex study to evaluate the impact of broad government economic
actions to relieve the consequences of the COVID-19 pandemic on the stock market’s
volatility.
Our analysis is broadly related to the literature about the financial implications of
the COVID-19 pandemic2 and the impact of economic policies on stock market risk3. We
also contribute to the strand of literature focused on the spillover effect by investigating
the role of US and EU policy actions for other international stock markets4.
4.1.1 Data and Methodology
Our analysis focuses on the period from January 2020 to the end of July 2021. We more
closely examine two sub-periods. Our main results are based on the period between
January and July 2020. The main reason for this sub-sample is the large number of
unexpected policy actions during this period and unprecedented levels of volatility. On
the other hand, the second selected period (August 2020 - July 2021) consists of a much
lower number of announcements mostly related to fiscal policies, volatility stabilizes, and
the potential policy actions are more predictable, which lowers uncertainty significantly.
The analyzed dataset consists of the main parts. The first part represents the realized
volatility of 8 stock indices from 23 countries worldwide. Volatility data are based on high-
2. Related literature about the financial implications of the COVID-19 pandemic and its regional differences,
e.g., Baker et al. (2020), Miescu et al. (2021), J. Wang et al. (2021), Vera-Valdés (2021), Kizys et al.
(2021), Demir et al. (2021), Harjoto et al. (2021), Engelhardt et al. (2021), Szczygielski et al. (2021), and
Bakry et al. (2021)
3. Related literature with the evidence that macroeconomic measures and announcements have an impact
on financial markets, e.g., Caporin et al. (2017), Fiordelisi et al. (2018), X. Huang (2018), Collingro et al.
(2020), Bernanke et al. (2005), Heyden et al. (2021), Rahman et al. (2021), Klose et al. (2021), Wei et al.
(2021), Yilmazkuday (2021), Bevilacqua et al. (2021), and Pettenuzzo et al. (2021)
4. Related literature about the spillover effects, e.g., Mei et al. (2018), Bekaert; Hoerova, et al. (2013),
Miranda-Agrippino et al. (2020), Chen et al. (2016), Yang et al. (2021), and Bevilacqua et al. (2021)
44
4. Realized Volatility Reaction During Crises
frequency data from the Thomson Reuters DataScope Tick History database provided
by Realized Volatility Library from Oxford-Man Institute of Quantitative Finance5. The
second part of the dataset comprises policy actions and interventions in response to the
COVID-19 pandemic from the COVID-19 Financial Response Tracker collected by the
Yale Program on Financial Stability (YFPS)6. It contains approved actions by central
banks, fiscal authorities, and other organizations aimed at restoring financial stability.
We performed our analysis on the heterogeneous autoregressive model (HAR)
developed by Corsi (2009) augmented by additional dummy variables representing
policy actions. We tried several model specifications, but the most general model could
be described as follows:
RVt+1 =β0 + β1RVD
t + β2RVW
t + β3RVM
t + (4.1)
β4RVD
t × Actt+1 + β5RVD
t × Actt + β6RVD
t × Actt−1 + ϵt+1
RV stands for the logarithm of 5-minute realized variance, where RVD
t , RVW
t , RVM
t
are the components of the basic HAR model that represents past values of daily, weekly,
and monthly volatility, respectively. Next, we added the dummy variable Actt with a
value of 1 if any action was announced on a given day or 0 if no actions were approved.
The variable Actt+1 represents the actions performed one day after the day for which
we explain volatility. In other words, it explains what happens with volatility one day
before the action takes place. Lastly, Actt−1 shows the action declared one day before
the day of interest. It is suitable for predictive regression and indicates how the markets
reacted the next day after the action was announced.
We add policy actions announcement to the HAR model as interactive terms. All
dummy variables in the model are multiplied by (RVD
t ) to control the level of volatility.
The model is estimated using ordinary least squares, where standard errors are obtained
via heteroskedasticity- and autocorrelation-consistent estimator of Newey et al. (1994).
4.1.2 Results
As a first step, we analyze the policy actions and volatility in individual countries in
general. The main components of the HAR model, RVD, RVW, and RVM, play a role of
the control variables in our model. The coefficients of the daily and weekly volatility
components are positive and statistically significant for almost all countries, which was
5. https://realized.oxford-man.ox.ac.uk/
6. https://som.yale.edu/faculty-research-centers/centers-initiatives/program-on-financial-
stability/covid-19-crisis
45
4. Realized Volatility Reaction During Crises
an expected result due to the high persistence of volatility. Only the monthly component
provides a negative coefficient which indicates mean reversion properties. It is caused
by the relatively short period accompanied by such a large increase in volatility at
the beginning, followed by stabilization and a decrease in volatility. The results for
these control variables are pretty stable, and they are not at the center of our attention.
Therefore, we do not focus on them in the following analysis.
When we check the effect of action announcements, it is evident that volatility increases
during the day of the announcement (Actt). It is expected because the market
tends to react rapidly to new information, and market participants quickly update their
positions according to their updated expectations. The most significant effect was observed
for the S&P500 index. For the US indexes, there is also evident the calming effect
of these policy actions because the volatility decreases significantly during the following
day, even more than it increases during the announcement day. The opposite effect was
observed in some European countries, namely Belgium, Netherlands, and Norway.
On the other hand, in some counties (India, Japan, Australia), the volatility increases
even one day before the announcement day (Actt+1). It could indicate the significance
of public discussion on possible policy actions, information leakage before the official
announcement, or overall increased policy uncertainty among market participants.
As a next step, we divided policy actions into fiscal stimulus and macroprudential
policy. This analysis provides a different picture of Europe and the United States. In
Europe, the evidence indicates that the Fiscal Stimulus category is more important
than macroprudential policy, and it tends to increase volatility. The stock indexes in the
United States show the opposite behavior. There is a significant calming effect of all policy
actions on volatility. Still, the macroprudential policy seems to have the most significant
impact and helped decrease market volatility the following day after the announcement.
The rest of the stock indexes, mainly from Asia and Australia, provide mixed results. In
general, our results suggest that the effects of policy action announcements are stronger
for developed countries than in emerging markets, where the magnitude of policy effects
and their statistical significance tend to be lower.
We also analyzed the spillover effects from the US and Euro area to other countries.
We include the additional regressor US Actt−1 or EU Actt−1 into the baseline model
specification. It indicates if some policy action was announced in the US or EU during the
previous day. This analysis shows a strong and significant spillover effect of the US policy
action announcement that was able to calm the market in most European countries. The
majority of other countries seem to be unaffected by the US announcements. On the
46
4. Realized Volatility Reaction During Crises
other hand, the influence of the EU actions on other countries seems minimal and much
weaker than the US announcements.
To support our hypothesis that the policy actions announcements had the greatest
power during the first few months of the COVID-19 pandemic, we investigate the following
one-year period from August 2020 to the end of July 2021. In general, the power of
economic policy announcements to affect market uncertainty has decreased significantly.
Especially in the United States, the announcements almost lost their ability to influence
stock market volatility.
Our findings also explain decreasing and even negative volatility risk premiums in
the US in March-April 2020. Government actions in the US decreased the uncertainty,
reflected in declining prices of insurance against unexpected volatility (as shown in
Cheng, 2020), which subsequently spread to other developed markets. While our findings
largely complement previous evidence, we additionally show far greater importance
of the macroprudential policy to calm financial markets.
This paper has several limits. It analyzed relatively rare events during a short time
period. Even though we performed robustness checks, the results could be sensitive to
the selected period. Everything also depends on the quality of the government actions
dataset, which slightly changed its methodology during our research and created a
challenging task to adjust everything accordingly.
4.2 Russia’s ruble during the onset of the Russian invasion of Ukraine in
early 2022: The role of implied volatility and attention
Russia-Ukraine relations have been complicated for a long time. The tension between
these two countries began to increase in February 2014, following the Ukrainian Revolution
of Dignity, also known as the Maidan Revolution. Russia exploited the Ukrainian
internal conflicts and subsequently annexed the Crimean Peninsula in February and
March 2014, a territory still (as of March 2022) internationally recognized as part of
Ukraine. These events were closely followed by a war in Donbas (March 2014), where
Russia supported local pro-Russian anti-government separatists in an armed conflict
against the Ukrainian government forces (Mankoff, 2014; Rywkin, 2014; Shelest, 2015;
Samokhvalov, 2015).
The whole conflict continued and escalated during November and December 2021
when Russia started to increase its military troops near Ukrainian borders. On February
21, 2022, Russian President Vladimir Putin announced that Russia recognized the independence
of two pro-Russian regions, Donetsk and Luhansk, in eastern Ukraine. It
47
4. Realized Volatility Reaction During Crises
Table 6: Paper Information and Author Contribution
Reference
Lyócsa, Š., & Plíhal, T. (2022).
Russia’s ruble during the onset of the Russian invasion of Ukraine in early 2022:
The role of implied volatility and attention.
Finance Research Letters, 102995.
Journal Rankings 2021 Citations
WoS AIS: 1.245 (Q2) WoS: 0
WoS IF: 9.848 (Q1) Scopus: 1
Scholar: 2
My Contribution (50%) - Corresponding author
Conceptualization, Data curation, Visualization, Investigation,
Writing - original draft, Writing - review & editing
Note: Author contribution stated as percentage share and specified using CRediT author
statement by Elsevier. Citations were collected in August 2022 and included self-citations.
triggered the first round of economic sanctions from NATO countries in the following
days. The whole situation escalated on February 24, 2022, when Russia began a complete
invasion of Ukraine, which led to a full-fledged war that surprised most of the global
population. As a result, the economic sanctions from other countries against Russia
become even more severe. It included, for example, restrictions on Russian imports and
exports, the removal of selected Russian banks from the SWIFT interbank system, and
the prohibition of the Central Bank of Russia from access to foreign exchange reserves7.
The war and economic sanctions have dramatically impacted the Russian ruble. It
significantly weakened and lost almost 50% of its value against the USD over a few
days at the end of February and the beginning of March in 2022. Volatility is a crucial
parameter for valuing assets with uncertain future payoffs. In such a turbulent period,
an accurate volatility model is in demand by institutional investors and policymakers to
understand the value of assets that are tied to the value of the Russian ruble. Based on the
existing literature, we hypothesize that we can create a suitable prediction model based
on i) limited attention theory and ii) the forward-looking nature of option contracts.
More specifically, in this study, we are interested in predicting intraday price fluctuations
of the ruble against the USD and the EUR during the onset of the Russo-Ukrainian crisis
using the population’s attention and investors’ expectations.
7. A comprehensive list of all sanctions can be found at
https://graphics.reuters.com/UKRAINE-CRISIS/SANCTIONS/byvrjenzmve/
48
4. Realized Volatility Reaction During Crises
Our research contributes to the academic literature in three ways. First, we contribute
to the emerging literature about the effects of the Russian invasion on the financial
markets by analyzing the period from the 1st of December 2021 to the end of the 7th of
March 2022. Therefore, we expand on existing literature about this very scarce topic so
far (e.g., Halousková et al., 2022; Mamonov et al., 2021; Ozili, 2022; Polyzos, 2022).
Second, we contribute to the literature on limited attention theory (Barber et al.,
2008) that uses Google Trends data to predict future price variation (Goddard et al.,
2015; Dimpfl et al., 2016; Audrino et al., 2020; Lyócsa; Molnár; Plíhal; Širaňová, 2020). We
construct our attention measure based on Google searches because it is a good proxy for
information demand (Bleher et al., 2021), it is prevalent in many parts of the world and
well established in finance literature (e.g. Da et al., 2011; Dimpfl et al., 2016; Audrino
et al., 2020; Lyócsa; Molnár; Plíhal; Širaňová, 2020). We extend the current literature to
address whether limited attention theory also manifests in intraday trading. Using the
intraday measures of attention, unique in the literature, we show that high-frequency
attention is still valuable for predicting fluctuations in intraday prices.
Third, we use the information contained in the forward-looking implied volatility
and measure its forecasting abilities during such an unexpected crisis. The link between
implied volatility’s information content and future realized volatility in the foreign
exchange market was examined in many previous studies, e.g., Canina et al. (1993),
Poon et al. (2003), Gonzalez-Perez (2015), and Plíhal et al. (2021). We apply a novel
approach that uses intraday implied volatility of USD/RUB and EUR/RUB. Our results
demonstrate that intraday investors’ expectations drive intraday realized price fluctuations,
and we also provide strong evidence that the implied volatility likely encompasses
investors’ attention.
4.2.1 Data and Methodology
Our sample of data starts on the 1st of December 2021 and ends on the 7th of March 2022.
In early December, the first signs of this threat were publicly recognized by the White
House8, which attracted considerable attention by the media. The selected end of the
sample was the most recent data available before the start of this research.
We focused on USD/RUB and EUR/RUB currency pairs. Our exchange rate prices
and implied volatility data were retrieved from Bloomberg in a 5-minute frequency.
Implied volatilities are extracted from OTC (over-the-counter) forex options with onemonth
maturity directly quoted on Bloomberg by institutional investors. Because the
8. see an article from the 1st of December 2021, https://www.theguardian.com/world/2021/dec/01/uswarns-russia-plans-large-scale-attack-o
n-ukraine.
49
4. Realized Volatility Reaction During Crises
implied volatility data in such high frequency did not provide enough variation for
appropriate analysis, we created six 4-hour intraday windows. For each 4-hour window,
realized volatility is calculated as the sum of squared 5-minute returns, and the average
of 5-minute implied variance represents implied volatility during the selected 4-hour
window. Both volatilities were annualized and synchronized to the UTC (Coordinated
Universal Time) time zone.
We use Google Trends as a measure of attention, which is consistent with the increasing
number of studies that suggest that attention might drive stock market activity
(e.g., N. Kim et al., 2019; Bleher et al., 2019; Lyócsa; Baumöhl, et al., 2020; Aslanidis et al.,
2022). Google publishes search volume indices representing the relative popularity of a
search term(s) within a given period and across multiple search terms. To capture the
population’s attention, we retrieve various terms related to the events surrounding the
Russo-Ukrainian war. Each term is retrieved separately with no geographical restriction.
To process the Google trends data and create a continuous time series of the same
scale, we use the set of procedures according to Kristoufek (2015) and Bleher et al.
(2021). We specified three categories of search terms: i) general financial market (e.g., SP
500, VIX, FX market,...), ii) ruble (e.g., ruble, USD rub, Russian interest rate,...), and iii)
Russian economy-related (e.g., economic sanctions, asset freeze, Nord Stream 2, export
controls, British Petroleum Russia, Ikea Russia,...).
We select the heterogeneous autoregressive volatility (HAR) model of Corsi (2009)
and adjust it to our specific needs. The standard HAR model is designed to forecast
daily volatility by lagged daily, weekly, and monthly average volatility values. 9 Our
research design is characterized by an intra-day approach and a limited sample size.
Therefore, we have to adjust the baseline HAR model accordingly.
Due to our limited sample size, we removed weekly and monthly volatility patterns
from the model. The model is estimated in log-log form because the onset of the crisis is
likely to cause a non-linear response in the market behavior. Moreover, we also consider
a specification with the first differences of implied volatilities to count for a possible
unit-root-like behavior of implied volatility in our sample.
We have to account for the serial dependence of the volatility series and the possible
intraday seasonality. To achieve this, we introduce a dummy variable It(j) that takes a
value of 1 if the next period of variation belongs to intraday trading window j and 0
otherwise, for j = 1, 2, ..., 6.
9. An alternative is to use a generalized autoregressive conditional heteroscedasticity (GARCH) model
of Bollerslev (1986). Given the limited sample size, the maximum likelihood estimation of an overparametrized
GARCH class model is not recommended.
50
4. Realized Volatility Reaction During Crises
A few days before and after the Russian invasion of Ukraine, the amount of relevant
news increased considerably, making it impossible to account for specific effects. To
account for this issue, we select a period from the 21th to the 28th of February 2022, when
the most events occurred. The dummy variable It(q) takes a value of 1 if the next given
intraday period price variation falls to the day q and 0 otherwise.
In total, we used seven model specifications. The most general model with all our
variables was specified as follows:
lnVt =β0 + β1lnVt−1 + lnVt−1
5
∑
j=1
γj It(j) + lnVt−1
6
∑
q=1
λq It(q)+
η1lnGt−1 + η2lnRt−1 + η3lnEt−1 + π1∆lnIVt−1 + π2lnIVt−1 + ϵt
(4.2)
Vt−1 stands for realized volatility for a 4-hour window. Interaction terms γjlnVt−1It(j)
and lnVt−1λq It(q) corresponds to the intraday seasonality and specific day effects respectively.
Lastly ∆lnIVt−1 corresponds to the first difference of implied volatility IVt−1.
The model is estimated via OLS, and the significance of coefficients is derived via the
stationary bootstrap method with a block length randomly drawn from the geometric
distribution of the expected value, as suggested by the procedure of Politis et al. (2004),
Patton; Politis, et al. (2009), and Racine et al. (2021). After each model, we check for the
presence of the serial dependence of residuals using the test found in Escanciano et al.
(2009).
4.2.2 Results
We provide early evidence on the effects of the imposed economic sanctions and the
conflict in the relevant foreign exchange market. Our data showed how the Russian ruble
weakened tremendously after the invasion started. This weakening was accompanied
by increased price fluctuations and attention to the ruble. Moreover, these two variables
showed a remarkable synchronization during this period. In the following days, attention
declined, but price volatility remained quite extreme.
Specifically, we find that more attention is being paid toward: i) the financial market,
ii) the Russian ruble, and iii) the relationship between economic sanctions and the
Russian economy tends to precede more significant price fluctuations in the foreign
exchange market. However, investors’ expectations, as given by implied volatility, seem
to encompass information from attention measures and even beyond.
These results are similar for the USD/RUB and EUR/RUB exchange rates. However,
we find that changing investors’ expectations (i.e., changes in implied volatilities) are
51
4. Realized Volatility Reaction During Crises
also relevant for the euro, which indicates the increased sensitivity of this currency pair
to short-term changes in expectations. The geographical position of the Eurozone is
likely behind this effect, as it is much closer to the epicenter of the conflict compared to
the United States.
Finally, we applied several robustness checks, including different lengths of intraday
windows, sample period adjustments, or a moving average of our attention measures.
Despite minor changes in the estimated coefficients, our main conclusions remained
largely unchanged.
While attention measures help explain future price changes, from a practical point
of view, their timely implementation is challenging, as search terms are not known in
advance, specifically in advance of a yet unknown event. Implied volatility does not
suffer from this issue. Our results indicate that during unexpected crisis periods, such as
the Russo-Ukrainian war, an accurate model of future price fluctuations should utilize
high-frequency quoted implied volatility measures. Such data are available in real-time;
their implementation into existing models is straightforward, while they also appear to
encompass the information content of attention measures.
The primary limit of this paper is a relatively small dataset. The research reacted to
an actual issue and tried to explain what was happening in the real world. Right after
the paper’s publication, the Russian ruble appreciated dramatically, and this event is not
captured in our research. We had enough data to explain only the initial depreciation of
the ruble. Still, it could be an interesting extension of this research to include a more
extended period and cover different reactions of the exchange rate market.
52
Conclusion
As we can see, volatility is a diverse and interesting topic that offers a lot of research
questions that could help us to understand the behavior of asset prices. This collection
of essays presented six published papers on volatility modeling and forecasting divided
into three parts, each containing two papers. This section highlights the most important
contributions to the academic literature.
The first part focuses on forecasting volatility using implied volatility variables and
low-frequency data. It contributes to the literature by showing that implied volatility
from options with shorter maturities than one month is highly beneficial for one day
ahead realized volatility forecasting. Most current literature used only options with
1-month maturity or the VIX index. We claim that selecting the proper option maturity
could influence the forecasting performance and that the 1-month rule of thumb could
be misleading. The second paper challenges the assumption that realized volatility
calculated from high-frequency data is preferable to low-frequency estimators. We show
that this assumption holds but depends on the forecast horizon. For forecast horizons of
5 days or less for most analyzed foreign exchange rates, it is better to use high-frequency
data if available. However, for longer forecast horizons, the difference between the two is
not statistically significant most of the time, which could save time, effort, and processing
power. Also, high-frequency data has to be of a good quality which is an issue for some
assets, and our research could provide a guide if it is sufficient to use only daily data.
The third and fourth papers measure the effect of macroeconomic news announcements
and other events on realized volatility. The third paper focused on central bank
announcements and quantitative easing and the reaction of stock markets of seven developed
countries. Our contribution to literature lies in using a realized volatility instead
of implied volatility that, as we argue, mechanically declined after the announcement.
Moreover, we utilize a multicountry study and analyze several days before and after
the announcement and not only the day itself. The fourth paper shed new light on
the volatility of bitcoin related to macroeconomic news announcements, news about
regulation, or security breaches of major exchanges. We also analyzed not only realized
volatility but also its jump component separately to provide more insights into this issue.
The last two papers analyzed the recent COVID-19 crisis and the Russian invasion of
Ukraine in early 2022. In the paper related to the virus pandemic, we analyzed stock
market indices around the world and the reaction of their volatility to government
actions. We were among the first to examine such effects in many countries. We found
some evidence that the governments were able to calm down the markets, especially in
53
Conclusion
the United States. We were also on the leading edge of the research literature with the
paper about ruble reaction in early 2022. Our focus was on investor attention using data
from Google trends. One of the innovations was the use of intraday attention variables
and volatility data that was not present in the previous literature.
The presented six papers characterize my past career. In future research, I want to
focus more on the volatility applications on specific tasks, such as risk management, by
improving the forecasts of expected shortfall metrics. I also started cooperation on a
project focused on option pricing, where I am trying to implement my knowledge about
volatility behavior, which leads me to more interest in modeling and forecasting implied
volatility and utilizing the information contained in it.
54
Bibliography
AALBORG, Halvor Aarhus; MOLNÁR, Peter; VRIES, Jon Erik de, 2019. What can explain
the price, volatility and trading volume of Bitcoin? Finance Research Letters. Vol. 29,
pp. 255–265.
ÄIJÖ, Janne, 2008. Impact of US and UK macroeconomic news announcements on the
return distribution implied by FTSE-100 index options. International Review of Financial
Analysis. Vol. 17, no. 2, pp. 242–258.
AIT-SAHALIA, Yacine; MYKLAND, Per A; ZHANG, Lan, 2005. How often to sample a
continuous-time process in the presence of market microstructure noise. The review of
financial studies. Vol. 18, no. 2, pp. 351–416.
AKTAS, Elvan, 2011. Systematic factors, information release and market volatility. Applied
financial economics. Vol. 21, no. 6, pp. 415–420.
ALBULESCU, Claudiu Tiberiu, 2021. COVID-19 and the United States financial markets’
volatility. Finance Research Letters. Vol. 38, p. 101699.
ALI, Mohsin; ALAM, Nafis; RIZVI, Syed Aun R, 2020. Coronavirus (COVID-19)—An
epidemic or pandemic for financial markets. Journal of Behavioral and Experimental
Finance. Vol. 27, p. 100341.
ALMEIDA, Alvaro; GOODHART, Charles; PAYNE, Richard, 1998. The effects of macroeconomic
news on high frequency exchange rate behavior. Journal of Financial and
Quantitative Analysis. Vol. 33, no. 3, pp. 383–408.
ANDERSEN, Torben G.; BOLLERSLEV, Tim, 1998. Deutsche Mark-Dollar Volatility:
Intraday Activity Patterns, Macroeconomic Announcements, and Longer Run Dependencies.
The Journal of Finance. Vol. 53, no. 1, p. 219. issn 00221082.
ANDERSEN, Torben G.; BOLLERSLEV, Tim; DIEBOLD, Francis X., 2007. Roughing it
up: Including jump components in the measurement, modeling, and forecasting of
return volatility. The Review of Economics and Statistics. Vol. 89, no. 4, pp. 701–720.
ANDERSEN, Torben G.; BOLLERSLEV, Tim; DIEBOLD, Francis X.; EBENS, Heiko, 2001.
The distribution of realized stock return volatility. Journal of financial economics. Vol. 61,
no. 1, pp. 43–76.
ANDERSEN, Torben G.; BOLLERSLEV, Tim; DIEBOLD, Francis X.; LABYS, Paul, 2003.
Modeling and forecasting realized volatility. Econometrica. Vol. 71, no. 2, pp. 579–625.
55
BIBLIOGRAPHY
ANDERSEN, Torben G.; BOLLERSLEV, Tim; DIEBOLD, Francis X.; VEGA, Clara, 2003.
Micro effects of macro announcements: Real-time price discovery in foreign exchange.
American economic review. Vol. 93, no. 1, pp. 38–62.
ANDERSEN, Torben G.; BOLLERSLEV, Tim; MEDDAHI, Nour, 2011. Realized volatility
forecasting and market microstructure noise. Journal of Econometrics. Vol. 160, no. 1,
pp. 220–234.
ANDERSEN, Torben G.; DOBREV, Dobrislav; SCHAUMBURG, Ernst, 2012. Jump-robust
volatility estimation using nearest neighbor truncation. Journal of Econometrics. Vol. 169,
no. 1, pp. 75–93.
ASHRAF, Badar Nadeem, 2020. Stock markets’ reaction to COVID-19: Cases or fatalities?
Research in International Business and Finance. Vol. 54, p. 101249.
ASHRAF, Badar Nadeem; GOODELL, John W, 2021. COVID-19 social distancing measures
and economic growth: Distinguishing short-and long-term effects. Finance Research
Letters, p. 102639.
ASLANIDIS, Nektarios; BARIVIERA, Aurelio F; LÓPEZ, Óscar G, 2022. The link between
cryptocurrencies and Google Trends attention. Finance Research Letters, p. 102654.
AUDRINO, Francesco; SIGRIST, Fabio; BALLINARI, Daniele, 2020. The impact of sentiment
and attention measures on stock market volatility. International Journal of Forecasting.
Vol. 36, no. 2, pp. 334–357.
AUER, Raphael; CLAESSENS, Stijn, 2018. Regulating cryptocurrencies: assessing market
reactions. BIS Quarterly Review. Vol. September 2018.
BAIG, Ahmed S; BUTT, Hassan Anjum; HAROON, Omair; RIZVI, Syed Aun R, 2021.
Deaths, panic, lockdowns and US equity markets: The case of COVID-19 pandemic.
Finance research letters. Vol. 38, p. 101701.
BAILLIE, Richard T; BOLLERSLEV, Tim; MIKKELSEN, Hans Ole, 1996. Fractionally
integrated generalized autoregressive conditional heteroskedasticity. Journal of econometrics.
Vol. 74, no. 1, pp. 3–30.
BAKER, Scott R; BLOOM, Nicholas; DAVIS, Steven J; TERRY, Stephen J, 2020. COVIDInduced
Economic Uncertainty. 2020-04. Working Paper, 26983. National Bureau of
Economic Research.
BAKRY, Walid; KAVALMTHARA, Peter John; SAVERIMUTTU, Vivienne; LIU, Yiyang;
CYRIL, Sajan, 2021. Response of stock market volatility to COVID-19 announcements
and stringency measures: A comparison of developed and emerging markets. Finance
Research Letters, p. 102350.
56
BIBLIOGRAPHY
BALDUZZI, Pierluigi; ELTON, Edwin J; GREEN, T Clifton, 2001. Economic news and
bond prices: Evidence from the US Treasury market. Journal of financial and Quantitative
analysis. Vol. 36, no. 4, pp. 523–543.
BANDI, Federico M; RUSSELL, Jeffrey R, 2008. Microstructure noise, realized variance,
and optimal sampling. The Review of Economic Studies. Vol. 75, no. 2, pp. 339–369.
BARBER, Brad M; ODEAN, Terrance, 2008. All that glitters: The effect of attention and
news on the buying behavior of individual and institutional investors. The review of
financial studies. Vol. 21, no. 2, pp. 785–818.
BARNDORFF-NIELSEN, Ole E; HANSEN, Peter R; LUNDE, Asger; SHEPHARD, Neil,
2008. Designing realized kernels to measure the ex post variation of equity prices in
the presence of noise. Econometrica. Vol. 76, no. 6, pp. 1481–1536.
BARNDORFF-NIELSEN, Ole E; SHEPHARD, Neil, 2004. Power and bipower variation
with stochastic volatility and jumps. Journal of Financial Econometrics. Vol. 2, no. 1,
pp. 1–37.
BARTOLINI, Leonardo; GOLDBERG, Linda S; SACARNY, Adam, 2008. How economic
news moves markets.
BAUER, Gregory H; VORKINK, Keith, 2011. Forecasting multivariate realized stock
market volatility. Journal of Econometrics. Vol. 160, no. 1, pp. 93–101.
BAUR, Dirk G; DIMPFL, Thomas, 2018. Asymmetric volatility in cryptocurrencies. Economics
Letters. Vol. 173, pp. 148–151. issn 0165-1765. Available from doi: https://doi.
org/10.1016/j.econlet.2018.10.008.
BAUR, Dirk G; DIMPFL, Thomas; KUCK, Konstantin, 2018. Bitcoin, gold and the US
dollar–A replication and extension. Finance Research Letters. Vol. 25, pp. 103–110.
BAUR, Dirk G; HONG, KiHoon; LEE, Adrian D., 2018. Bitcoin: Medium of exchange
or speculative assets? Journal of International Financial Markets, Institutions and Money.
Vol. 54, pp. 177–189. issn 1042-4431. Available from doi: https://doi.org/10.1016/
j.intfin.2017.12.004.
BAUWENS, Luc; BEN OMRANE, Walid; GIOT, Pierre, 2005. News announcements,
market activity and volatility in the euro/dollar foreign exchange market. Journal of
International Money and Finance. Vol. 24, no. 7, pp. 1108–1125. issn 0261-5606.
BECKER, Ralf; CLEMENTS, Adam E., 2008. Are combination forecasts of S&P 500
volatility statistically superior? International Journal of Forecasting. Vol. 24, pp. 122–133.
issn 0169-2070.
57
BIBLIOGRAPHY
BECKER, Ralf; CLEMENTS, Adam E.; WHITE, Scott I., 2007. Does implied volatility
provide any information beyond that captured in model-based volatility forecasts?.
Journal of Banking and Finance. Vol. 31, pp. 2535–2549. issn 0378-4266.
BECKER, Ralf; CLEMENTS, Adam E.; WHITE, Scott I., 2006. On the informational
efficiency of S&P500 implied volatility. North American Journal of Economics and Finance.
Vol. 17, pp. 139–153. issn 1062-9408.
BEECHEY, Meredith J; WRIGHT, Jonathan H, 2009. The high-frequency impact of news
on long-term yields and forward rates: Is it real? Journal of Monetary Economics. Vol. 56,
no. 4, pp. 535–544.
BEINE, Michel; LAURENT, Sebastien; PALM, Franz C, 2009. Central Bank forex interventions
assessed using realized moments. Journal of International Financial Markets,
Institutions and Money. Vol. 19, no. 1, pp. 112–127.
BEKAERT, Geert; ENGSTROM, Eric, 2010. Inflation and the stock market: Understanding
the “Fed Model”. Journal of Monetary Economics. Vol. 57, no. 3, pp. 278–294.
BEKAERT, Geert; HOEROVA, Marie; DUCA, Marco Lo, 2013. Risk, uncertainty and
monetary policy. Journal of Monetary Economics. Vol. 60, no. 7, pp. 771–788.
BEN OMRANE, Walid; HAFNER, Christian, 2015. Macroeconomic news surprises and
volatility spillover in foreign exchange markets. Empirical Economics. Vol. 48, no. 2,
pp. 577–607.
BERNANKE, Ben S; KUTTNER, Kenneth N, 2005. What explains the stock market’s
reaction to Federal Reserve policy? The Journal of finance. Vol. 60, no. 3, pp. 1221–1257.
BEVILACQUA, Mattia; BRANDL-CHENG, Lukas; DANIELSSON, Jon; ERGUN, Lerby
Murat; UTHEMANN, Andreas; ZIGRAND, Jean-Pierre, 2021. The Calming of ShortTerm
Market Fears and Its Long-Term Consequences: The Central Banks’ Dilemma.
Available at SSRN 3774218.
BIJL, Laurens; KRINGHAUG, Glenn; MOLNÁR, Peter; SANDVIK, Eirik, 2016. Google
searches and stock returns. International Review of Financial Analysis. Vol. 45, pp. 150–
156. issn 1057-5219.
BLAIR, Bevan J.; POON, Ser-huang; TAYLOR, Stephen J., 2001. Forecasting S&P
100 volatility : the incremental information content of implied volatilities and
high-frequency index returns. Journal of econometrics. Vol. 105, no. 1, pp. 5–26. issn
03044076.
BLEHER, Johannes; DIMPFL, Thomas, 2021. Knitting Multi-Annual High-Frequency
Google Trends to Predict Inflation and Consumption. Econometrics and Statistics.
58
BIBLIOGRAPHY
BLEHER, Johannes; DIMPFL, Thomas, 2019. Today I got a million, tomorrow, I don’t
know: on the predictability of cryptocurrencies by means of Google search volume.
International Review of Financial Analysis. Vol. 63, pp. 147–159.
BOLLERSLEV, Tim, 1986. Generalized autoregressive conditional heteroskedasticity.
Journal of econometrics. Vol. 31, no. 3, pp. 307–327.
BOLLERSLEV, Tim; HOOD, Benjamin; HUSS, John; PEDERSEN, Lasse Heje, 2018. Risk
everywhere: Modeling and managing volatility. The Review of Financial Studies. Vol. 31,
no. 7, pp. 2729–2773.
BOLLERSLEV, Tim; PATTON, Andrew J.; QUAEDVLIEG, Rogier, 2016. Exploiting the
errors: A simple approach for improved volatility forecasting. Journal of Econometrics.
Vol. 192, no. 1, pp. 1–18.
BOMFIM, Antulio N, 2003. Pre-announcement effects, news effects, and volatility: Monetary
policy and the stock market. Journal of Banking & Finance. Vol. 27, no. 1, pp. 133–
151.
BONDELL, Howard D; REICH, Brian J; WANG, Huixia, 2010. Noncrossing quantile
regression curve estimation. Biometrika. Vol. 97, no. 4, pp. 825–838.
BOUOIYOUR, Jamal; SELMI, Refk, 2017. The Bitcoin price formation: Beyond the fundamental
sources. arXiv preprint, arXiv:1707.01284.
BOUOIYOUR, Jamal; SELMI, Refk, 2015. What Does Bitcoin Look Like?. Annals of
Economics and Finance. Vol. 16, no. 2, pp. 449–492.
BOUOIYOUR, Jamal; SELMI, Refk; TIWARI, Aviral Kumar; OLAYENI, Olaolu Richard,
2016. What Drives Bitcoin Price?. Economics Bulletin. Vol. 36, no. 2, pp. 843–850.
BOYD, John H; HU, Jian; JAGANNATHAN, Ravi, 2005. The stock market’s reaction to
unemployment news: Why bad news is usually good for stocks. The Journal of Finance.
Vol. 60, no. 2, pp. 649–672.
BUSCH, Thomas; CHRISTENSEN, Bent Jesper; NIELSEN, Morten Ørregaard, 2011. The
role of implied volatility in forecasting future realized volatility and jumps in foreign
exchange, stock, and bond markets. Journal of Econometrics. Vol. 160, no. 1, pp. 48–57.
CANINA, Linda; FIGLEWSKI, Stephen, 1993. The informational content of implied
volatility. The Review of Financial Studies. Vol. 6, no. 3, pp. 659–681.
CAPORIN, Massimiliano; POLI, Francesco, 2017. Building News Measures from Textual
Data and an Application to Volatility Forecasting. Econometrics. Vol. 5, no. 3, p. 1. issn
22251146.
59
BIBLIOGRAPHY
CASCALDI-GARCIA, Danilo; DATTA, Deepa Dhume; FERREIRA, Thiago Revil T.;
GRISHCHENKO, Olesya V.; JAHAN-PARVAR, Mohammad R.; LONDONO, Juan M.;
LORIA, Francesca; MA, Sai; GIUDICE RODRIGUEZ, Marius del; J, 2021. What is
Certain about Uncertainty? Journal of Economic Literature. Vol. Forthcoming.
ČECH, František; BARUNIK, Jozef, 2017. On the modelling and forecasting of multivariate
realized volatility: Generalized heterogeneous autoregressive (GHAR) model.
Journal of Forecasting. Vol. 36, no. 2, pp. 181–206.
CHAE, Joon, 2005. Trading volume, information asymmetry, and timing information.
The journal of finance. Vol. 60, no. 1, pp. 413–442.
CHAN, Kam F; GRAY, Philip, 2018. Volatility jumps and macroeconomic news announcements.
Journal of Futures Markets. Vol. 38, no. 8, pp. 881–897.
CHARFEDDINE, Lanouar; BENLAGHA, Noureddine; MAOUCHI, Youcef, 2019. Investigating
the dynamic relationship between cryptocurrencies and conventional assets:
Implications for financial investors. Economic Modelling, in press. issn 0264-9993.
CHEAH, Eng-Tuck; FRY, John, 2015. Speculative bubbles in Bitcoin markets? An empirical
investigation into the fundamental value of Bitcoin. Economics Letters. Vol. 130, pp. 32–
36.
CHEN, Qianying; FILARDO, Andrew; HE, Dong; ZHU, Feng, 2016. Financial crisis, US
unconventional monetary policy and international spillovers. Journal of International
Money and Finance. Vol. 67, pp. 62–81.
CHENG, Ing-Haw, 2020. Volatility markets underreacted to the early stages of the
COVID-19 pandemic. The Review of Asset Pricing Studies. Vol. 10, no. 4, pp. 635–668.
CHIRIAC, Roxana; VOEV, Valeri, 2011. Modelling and forecasting multivariate realized
volatility. Journal of Applied Econometrics. Vol. 26, no. 6, pp. 922–947.
CHRISTENSEN, B.J.; PRABHALA, N.R., 1998. The relation between implied and realized
volatility. Journal of Financial Economics. Vol. 50, no. 2, pp. 125–150. issn 0304405X.
CHRISTIE-DAVID, Rohan; CHAUDHRY, Mukesh; KHAN, Walayet, 2002. News releases,
market integration, and market leadership. Journal of Financial Research. Vol. 25, no. 2,
pp. 223–245.
CHU, Jeffrey; CHAN, Stephen; NADARAJAH, Saralees; OSTERRIEDER, Joerg, 2017.
GARCH Modelling of Cryptocurrencies. Journal of Risk and Financial Management.
Vol. 10, no. 4, pp. 1–15.
60
BIBLIOGRAPHY
CHUDIK, Alexander; PESARAN, M Hashem, 2015. Common correlated effects estimation
of heterogeneous dynamic panel data models with weakly exogenous regressors.
Journal of econometrics. Vol. 188, no. 2, pp. 393–420.
CIAIAN, Pavel; RAJCANIOVA, Miroslava; KANCS, d’Artis, 2016. The Digital Agenda of
Virtual Currencies: Can BitCoin Become a Global Currency?. Information Systems and
e-Business Management. Vol. 14, no. 4, pp. 883–919. issn 16179846.
CLEMENTS, Adam E.; PREVE, Daniel PA, 2021. A practical guide to harnessing the har
volatility model. Journal of Banking & Finance. Vol. 133, p. 106285.
COLLINGRO, Franziska; FRENKEL, Michael, 2020. On the financial market impact of
euro area monetary policy: A comparative study before and after the Global Financial
Crisis. Global Finance Journal. Vol. 45, p. 100480. issn 1044-0283.
CONRAD, Christian; CUSTOVIC, Anessa; GHYSELS, Eric, 2018. Long-and short-term
cryptocurrency volatility components: A GARCH-MIDAS analysis. Journal of Risk and
Financial Management. Vol. 11, no. 2, pp. 1–12.
CORBET, Shean; CUMMING, Douglas J; LUCEY, Brian M; PEAT, Maurice; VIGNE,
Samuel, 2019. Investigating the Dynamics Between Price Volatility, Price Discovery,
and Criminality in Cryptocurrency Markets. Price Discovery, and Criminality in Cryptocurrency
Markets. Vol. (May 3, 2019).
CORBET, Shean; LARKIN, Charles; LUCEY, Brian; MEEGAN, Andrew; YAROVAYA, Larisa,
2018. The Volatility Generating Effects of Macroeconomic News on Cryptocurrency
Returns. SSRN Electronic Journal. Available from doi: 10.2139/ssrn.3141986.
CORBET, Shean; LUCEY, Brian; URQUHART, Andrew; YAROVAYA, Larisa, 2019. Cryptocurrencies
as a financial asset: A systematic analysis. International Review of Financial
Analysis. Vol. 62, pp. 182–199. issn 1057-5219. Available from doi: https://doi.org/
10.1016/j.irfa.2018.09.003.
CORSI, Fulvio, 2009. A simple approximate long-memory model of realized volatility.
Journal of Financial Econometrics. Vol. 7, no. 2, pp. 174–196.
CORSI, Fulvio; PIRINO, Davide; RENO, Roberto, 2010. Threshold bipower variation
and the impact of jumps on volatility forecasting. Journal of Econometrics. Vol. 159, no.
2, pp. 276–288.
CORSI, Fulvio; RENO, Roberto, 2009. HAR volatility modelling with heterogeneous
leverage and jumps. Available at SSRN. Vol. 1316953.
61
BIBLIOGRAPHY
CORSI, Fulvio; RENÒ, Roberto, 2012. Discrete-Time Volatility Forecasting With Persistent
Leverage Effect and the Link With Continuous-Time Volatility Modeling. Journal of
Business & Economic Statistics. Vol. 30, no. 3, p. 368. issn 07350015.
COVRIG, Vicentiu; LOW, Buen Sin, 2003. The Quality of Volatility Traded on the Overthe-counter
Currency Market: A Multiple Horizons Study. Journal of Futures Markets.
Vol. 23, no. 3, pp. 261–284. issn 02707314.
CRETAROLA, Alessandra; FIGA-TALAMANCA, Gianna; PATACCA, Marco, 2017. A
sentiment-based model for the Bitcoin: Theory, estimation and option pricing. arXiv
preprint, arXiv:1709.08621.
DA, Zhi; ENGELBERG, Joseph; GAO, Pengjie, 2011. In search of attention. The journal of
finance. Vol. 66, no. 5, pp. 1461–1499.
DAY, Theodore E.; LEWIS, Craig M., 1992. Stock market volatility and the information
content of stock index options. Journal of Econometrics. Vol. 52, no. 1/2, pp. 267–287.
issn 03044076.
DEEV, Oleg; PLÍHAL, Tomáš, 2022. How to calm down the markets? The effects of
COVID-19 economic policy responses on financial market uncertainty. Research in
International Business and Finance, p. 101613.
DEMIR, Ender; DANISMAN, Gamze Ozturk, 2021. Banking sector reactions to COVID-
19: The role of bank-specific factors and government policy responses. Research in
International Business and Finance. Vol. 58, p. 101508. issn 0275-5319.
DÍAZ, Fernando; HENRÍQUEZ, Pablo A; WINKELRIED, Diego, 2022. Stock market
volatility and the COVID-19 reproductive number. Research in international business
and finance. Vol. 59, p. 101517.
DIMPFL, Thomas; JANK, Stephan, 2016. Can internet search queries help to predict
stock market volatility? European Financial Management. Vol. 22, no. 2, pp. 171–192.
DOMINGUEZ, Kathryn M, 1998. Central bank intervention and exchange rate volatility.
Journal of International Money and Finance. Vol. 17, no. 1, pp. 161–190.
DOMINGUEZ, Kathryn M, 2006. The European Central Bank, the euro, and global
financial markets. Journal of Economic Perspectives. Vol. 20, no. 4, pp. 67–88.
EDERINGTON, Louis H; LEE, Jae Ha, 1993. How markets process information: News
releases and volatility. The Journal of Finance. Vol. 48, no. 4, pp. 1161–1191.
EHRMANN, Michael; FRATZSCHER, Marcel, 2005. Exchange rates and fundamentals:
new evidence from real-time data. Journal of International Money and Finance. Vol. 24,
no. 2, pp. 317–341.
62
BIBLIOGRAPHY
EHRMANN, Michael; FRATZSCHER, Marcel, 2004. Taking stock: Monetary policy
transmission to equity markets. Journal of Money, Credit and Banking, pp. 719–737.
EL OUADGHIRI, Imane; MIGNON, Valérie; BOITOUT, Nicolas, 2016. On the impact of
macroeconomic news surprises on Treasury-bond returns. Annals of Finance. Vol. 12,
no. 1, pp. 29–53.
EL OUADGHIRI, Imane; UCTUM, Remzi, 2016. Jumps in equilibrium prices and asymmetric
news in foreign exchange markets. Economic Modelling. Vol. 54, pp. 218–234.
ENGELHARDT, Nils; KRAUSE, Miguel; NEUKIRCHEN, Daniel; POSCH, Peter N, 2021.
Trust and stock market volatility during the COVID-19 crisis. Finance Research Letters.
Vol. 38, p. 101873.
ENGLE, Robert F., 1982. Autoregressive Conditional Heteroscedasticity with Estimates
of the Variance of United Kingdom Inflation. Econometrica. No. 4, p. 987. issn 00129682.
ENGLE, Robert F.; BOLLERSLEV, Tim, 1986. Modelling the persistence of conditional
variances. Econometric reviews. Vol. 5, no. 1, pp. 1–50.
ESCANCIANO, J Carlos; LOBATO, Ignacio N, 2009. An automatic portmanteau test for
serial correlation. Journal of Econometrics. Vol. 151, no. 2, pp. 140–149.
EVANS, Kevin; SPEIGHT, Alan, 2010. International macroeconomic announcements and
intraday euro exchange rate volatility. Journal of the Japanese and international economies.
Vol. 24, no. 4, pp. 552–568.
FAUST, Jon; ROGERS, John H; WANG, Shing-Yi B; WRIGHT, Jonathan H, 2007. The
high-frequency response of exchange rates and interest rates to macroeconomic announcements.
Journal of Monetary Economics. Vol. 54, no. 4, pp. 1051–1068.
FIORDELISI, Franco; GALLOPPO, Giuseppe, 2018. Stock market reaction to policy
interventions. The European Journal of Finance. Vol. 24, no. 18, pp. 1817–1834.
FLANNERY, Mark J; PROTOPAPADAKIS, Aris A, 2002. Macroeconomic factors do
influence aggregate stock returns. The review of financial studies. Vol. 15, no. 3, pp. 751–
782.
FLEMING, Michael J; REMOLONA, Eli M, 1999. Price formation and liquidity in the US
Treasury market: The response to public information. The journal of Finance. Vol. 54,
no. 5, pp. 1901–1915.
FLEMING, Michael J; REMOLONA, Eli M, 1997. What moves the bond market? Economic
policy review. Vol. 3, no. 4.
63
BIBLIOGRAPHY
FRENCH, Kenneth R; SCHWERT, G William; STAMBAUGH, Robert F, 1987. Expected
stock returns and volatility. Journal of financial Economics. Vol. 19, no. 1, pp. 3–29.
FÜSS, Roland; MAGER, Ferdinand; WOHLENBERG, Holger; ZHAO, Lu, 2011. The
impact of macroeconomic announcements on implied volatility. Applied Financial
Economics. Vol. 21, no. 21, pp. 1571–1580.
GARCIA, David; TESSONE, Claudio J; MAVRODIEV, Pavlin; PERONY, Nicolas, 2014.
The digital traces of bubbles: feedback cycles between socio-economic signals in the
Bitcoin economy. Journal of the Royal Society Interface. Vol. 11, no. 99, p. 20140623.
GARMAN, Mark B.; KLASS, Michael J., 1980. On the Estimation of Security Price Volatilities
from Historical Data. The Journal of Business. Vol. 53, no. 1, p. 67. issn 00219398.
GENRE, Véronique; KENNY, Geoff; MEYLER, Aidan; TIMMERMANN, Allan, 2013.
Combining expert forecasts: Can anything beat the simple average?. International
Journal of Forecasting. Vol. 29, pp. 108–121. issn 0169-2070.
GLOSTEN, Lawrence R; JAGANNATHAN, Ravi; RUNKLE, David E, 1993. On the relation
between the expected value and the volatility of the nominal excess return on
stocks. The journal of finance. Vol. 48, no. 5, pp. 1779–1801.
GODDARD, John; KITA, Arben; WANG, Qingwei, 2015. Investor attention and FX market
volatility. Journal of International Financial Markets, Institutions and Money. Vol. 38, pp. 79–
96.
GONZALEZ-PEREZ, Maria T, 2015. Model-free volatility indexes in the financial literature:
A review. International Review of Economics & Finance. Vol. 40, pp. 141–159.
GRANGER, Clive WJ; JOYEUX, Roselyne, 1980. An introduction to long-memory time
series models and fractional differencing. Journal of time series analysis. Vol. 1, no. 1,
pp. 15–29.
GÜRKAYNAK, Refet S; SACK, Brian; SWANSON, Eric, 2005. The sensitivity of longterm
interest rates to economic news: Evidence and implications for macroeconomic
models. American economic review. Vol. 95, no. 1, pp. 425–436.
HALOUSKOVÁ, Martina; HORVÁTH, Matůš; DANIEL, Stašek, 2022. The role of investor
attention in global assets price variation during the war in Ukraine (working paper).
Available at http://arxiv.org/abs/2205.05985.
HAN, Heejoon; PARK, Myung D., 2013. Comparison of Realized Measure and Implied
Volatility in Forecasting Volatility. Journal of Forecasting. Vol. 32, no. 6, pp. 522–533.
issn 02776693.
64
BIBLIOGRAPHY
HANSEN, Peter R; HUANG, Zhuo; SHEK, Howard Howan, 2012. Realized GARCH: a
joint model for returns and realized measures of volatility. Journal of Applied Econometrics.
Vol. 27, no. 6, pp. 877–906.
HANSEN, Peter R; LUNDE, Asger; NASON, James M, 2011. The model confidence set.
Econometrica. Vol. 79, no. 2, pp. 453–497.
HARJOTO, Maretno Agus; ROSSI, Fabrizio; LEE, Robert; SERGI, Bruno S, 2021. How do
equity markets react to COVID-19? Evidence from emerging and developed countries.
Journal of Economics and Business. Vol. 115, p. 105966.
HARVEY, Campbell R; HUANG, Roger D, 1991. Volatility in the foreign currency futures
market. The Review of Financial Studies. Vol. 4, no. 3, pp. 543–569.
HEYDEN, Kim J; HEYDEN, Thomas, 2021. Market reactions to the arrival and containment
of COVID-19: an event study. Finance research letters. Vol. 38, p. 101745.
HORPESTAD, Jone B; LYÓCSA, Štefan; MOLNÁR, Peter; OLSEN, Torbjørn B, 2019.
Asymmetric volatility in equity markets around the world. The North American Journal
of Economics and Finance. Vol. 48, pp. 540–554.
HORRA, Luis de la; FUENTE, Gabriel de la; PEROTE, Javier, 2019. The drivers of Bitcoin
demand: A short and long-run analysis. International Review of Financial Analysis.
Vol. 62, pp. 21–34.
HUANG, Xin, 2018. Macroeconomic news announcements, systemic risk, financial
market volatility, and jumps. Journal of Futures Markets. Vol. 38, no. 5, pp. 513–534.
HUSSAIN, Syed Mujahid, 2011. Simultaneous monetary policy announcements and
international stock markets response: An intraday analysis. Journal of Banking & Finance.
Vol. 35, no. 3, pp. 752–764.
HUYNH, Toan Luu Duc; FOGLIA, Matteo; NASIR, Muhammad Ali; ANGELINI, Eliana,
2021. Feverish sentiment and global equity markets during the COVID-19 pandemic.
Journal of Economic Behavior & Organization.
JAGGI, Adrian; SCHLEGEL, Martin; ZANETTI, Attilio, 2016. Macroeconomic surprises,
market environment and safe-haven currencies. Swiss National Bank Working Papers.
Vol. 2016-15.
JIANG, George J; KONSTANTINIDI, Eirini; SKIADOPOULOS, George, 2012. Volatility
spillovers and the effect of news announcements. Journal of Banking & Finance. Vol. 36,
no. 8, pp. 2260–2273.
JOHNSON, Norman L, 1949a. Bivariate distributions based on simple translation systems.
Biometrika. Vol. 36, no. 3/4, pp. 297–304.
65
BIBLIOGRAPHY
JOHNSON, Norman L, 1949b. Systems of frequency curves generated by methods of
translation. Biometrika. Vol. 36, no. 1/2, pp. 149–176.
JORION, Philippe, 1995. Predicting Volatility in the Foreign Exchange Market. Journal of
Finance. Vol. 50, no. 2, pp. 507–528. issn 00221082.
KAMBOUROUDIS, Dimos S.; MCMILLAN, David G.; TSAKOU, Katerina, 2016. Forecasting
Stock Return Volatility: A Comparison of GARCH, Implied Volatility, and
Realized Volatility Models. Journal of Futures Markets. Vol. 36, no. 12, pp. 1127–1163.
issn 02707314.
KATSIAMPA, Paraskevi, 2017. Volatility estimation for Bitcoin: A comparison of GARCH
models. Economics Letters. Vol. 158, pp. 3–6.
KIM, Neri; LUČIVJANSKÁ, Katarína; MOLNÁR, Peter; VILLA, Roviel, 2019. Google
searches and stock market activity: Evidence from Norway. Finance Research Letters.
Vol. 28, pp. 208–220.
KIM, Suk-Joong; MCKENZIE, Michael D; FAFF, Robert W, 2004. Macroeconomic news
announcements and the role of expectations: evidence for US bond, stock and foreign
exchange markets. Journal of Multinational Financial Management. Vol. 14, no. 3, pp. 217–
232.
KIZYS, Renatas; TZOUVANAS, Panagiotis; DONADELLI, Michael, 2021. From COVID-
19 herd immunity to investor herding in international stock markets: The role of government
and regulatory restrictions. International Review of Financial Analysis. Vol. 74,
p. 101663.
KLIBER, Agata; MARSZAŁEK, Paweł; MUSIAŁKOWSKA, Ida; ŚWIERCZYŃSKA,
Katarzyna, 2019. Bitcoin: Safe haven, hedge or diversifier? Perception of bitcoin in the
context of a country’s economic situation — A stochastic volatility approach. Physica
A: Statistical Mechanics and its Applications. Vol. 524, pp. 246–257. issn 0378-4371.
KLOSE, Jens; TILLMANN, Peter, 2021. Covid-19 and financial markets: A panel analysis
for european countries. Jahrbücher für Nationalökonomie und Statistik. Vol. 241, no. 3,
pp. 297–347.
KOOPMAN, Siem Jan; JUNGBACKER, Borus; HOL, Eugenie, 2005. Forecasting daily
variability of the S&P 100 stock index using historical, realised and implied volatility
measurements. Journal of Empirical Finance. Vol. 12, no. 3, pp. 445–475.
KOURTIS, Apostolos; MARKELLOS, Raphael N; SYMEONIDIS, Lazaros, 2016. An
international comparison of implied, realized, and GARCH volatility forecasts. Journal
of Futures Markets. Vol. 36, no. 12, pp. 1164–1193.
66
BIBLIOGRAPHY
KRIEGER, Kevin; MAUCK, Nathan; VAZQUEZ, Joseph, 2015. Comparing US and European
market volatility responses to interest rate policy announcements. International
Review of Financial Analysis. Vol. 39, pp. 127–136.
KRISTOUFEK, Ladislav, 2013. BitCoin meets Google Trends and Wikipedia: Quantifying
the relationship between phenomena of the Internet era. Scientific reports. Vol. 3, p. 3415.
KRISTOUFEK, Ladislav, 2015. What are the main drivers of the Bitcoin price? Evidence
from wavelet coherence analysis. PloS one. Vol. 10, no. 4, e0123923.
LANNE, Markku; SAIKKONEN, Pentti, 2005. Non-linear GARCH models for highly
persistent volatility. The Econometrics Journal. Vol. 8, no. 2, pp. 251–276.
LIU, Lily Y; PATTON, Andrew J.; SHEPPARD, Kevin, 2015. Does anything beat 5-minute
RV? A comparison of realized measures across multiple asset classes. Journal of Econometrics.
Vol. 187, no. 1, pp. 293–311.
LUO, Jiawen; KLEIN, Tony; JI, Qiang; HOU, Chenghan, 2022. Forecasting realized volatility
of agricultural commodity futures with infinite Hidden Markov HAR models.
International Journal of Forecasting. Vol. 38, no. 1, pp. 51–73.
LYÓCSA, Štefan; BAUMÖHL, Eduard; VÝROST, Tomáš; MOLNÁR, Peter, 2020. Fear of
the coronavirus and the stock markets. Finance research letters. Vol. 36, p. 101735.
LYÓCSA, Štefan; MOLNÁR, Peter, 2017. The effect of non-trading days on volatility
forecasts in equity markets. Finance Research Letters. Vol. 23, pp. 39–49.
LYÓCSA, Štefan; MOLNÁR, Peter, 2016. Volatility forecasting of strategically linked
commodity ETFs: gold-silver. Quantitative Finance. Vol. 16, no. 12, pp. 1809–1822.
LYÓCSA, Štefan; MOLNÁR, Peter; PLÍHAL, Tomáš, 2019. Central bank announcements
and realized volatility of stock markets in G7 countries. Journal of International Financial
Markets, Institutions and Money. Vol. 58, pp. 117–135.
LYÓCSA, Štefan; MOLNÁR, Peter; PLÍHAL, Tomáš; ŠIRAŇOVÁ, Mária, 2020. Impact of
macroeconomic news, regulation and hacking exchange markets on the volatility of
bitcoin. Journal of Economic Dynamics and Control. Vol. 119, p. 103980.
LYÓCSA, Štefan; MOLNÁR, Peter; TODOROVA, Neda, 2017. Volatility forecasting of nonferrous
metal futures: Covariances, covariates or combinations? Journal of International
Financial Markets, Institutions and Money. Vol. 51, pp. 228–247. issn 1042-4431.
LYÓCSA, Štefan; PLÍHAL, Tomáš, 2022. Russia’s ruble during the onset of the Russian
invasion of Ukraine in early 2022: The role of implied volatility and attention. Finance
Research Letters, p. 102995.
67
BIBLIOGRAPHY
LYÓCSA, Štefan; PLÍHAL, Tomáš; VÝROST, Tomáš, 2021. FX market volatility modelling:
Can we use low-frequency data? Finance research letters. Vol. 40, p. 101776.
LYÓCSA, Štefan; TODOROVA, Neda, 2020. Trading and non-trading period realized
market volatility: Does it matter for forecasting the volatility of US stocks? International
Journal of Forecasting. Vol. 36, no. 2, pp. 628–645.
MAMONOV, Mikhail; PESTOVA, Anna, 2021. Sorry, You’re Blocked.’Economic Effects of
Financial Sanctions on the Russian Economy. CERGE-EI Working Paper Series. No. 704.
MANKOFF, Jeffrey, 2014. Russia’s latest land grab: How Putin won Crimea and lost
Ukraine. Foreign Aff. Vol. 93, p. 60.
MARSHALL, Andrew; MUSAYEV, Taleh; PINTO, Helena; TANG, Leilei, 2012. Impact of
news announcements on the foreign exchange implied volatility. Journal of International
Financial Markets, Institutions and Money. Vol. 22, no. 4, pp. 719–737.
MEI, Dexiang; ZENG, Qing; ZHANG, Yaojie; HOU, Wenjing, 2018. Does US Economic
Policy Uncertainty matter for European stock markets volatility? Physica A: Statistical
Mechanics and its Applications. Vol. 512, pp. 215–221.
MIESCU, Mirela; ROSSI, Raffaele, 2021. COVID-19-induced shocks and uncertainty.
European Economic Review. Vol. 139, p. 103893. issn 0014-2921.
MIRANDA-AGRIPPINO, Silvia; REY, Hélene, 2020. US monetary policy and the global
financial cycle. The Review of Economic Studies. Vol. 87, no. 6, pp. 2754–2776.
NEELY, Christopher J; DEY, S Rubun, 2010. A survey of announcement effects on foreign
exchange returns. Federal Reserve Bank of St. Louis Review. Vol. 92, no. 5, pp. 417–463.
NELSON, Daniel B, 1991. Conditional heteroskedasticity in asset returns: A new approach.
Econometrica: Journal of the econometric society, pp. 347–370.
NEWEY, Whitney K; WEST, Kenneth D, 1994. Automatic lag selection in covariance
matrix estimation. The Review of Economic Studies. Vol. 61, no. 4, pp. 631–653.
NIKKINEN, Jussi; OMRAN, Mohammed; SAHLSTRÖM, Petri; ÄIJÖ, Janne, 2006. Global
stock market reactions to scheduled US macroeconomic news announcements. Global
Finance Journal. Vol. 17, no. 1, pp. 92–104.
NIKKINEN, Jussi; SAHLSTRÖM, Petri, 2001. Impact of scheduled US macroeconomic
news on stock market uncertainty: A multinational perspecive. Multinational Finance
Journal. Vol. 5, no. 2, pp. 129–148.
68
BIBLIOGRAPHY
NIKKINEN, Jussi; SAHLSTRÖM, Petri, 2004. Scheduled domestic and US macroeconomic
news and stock valuation in Europe. Journal of multinational financial management.
Vol. 14, no. 3, pp. 201–215.
OZILI, Peterson K, 2022. Global economic consequence of Russian invasion of Ukraine.
Available at SSRN.
PARKINSON, Michael, 1980. The extreme value method for estimating the variance of
the rate of return. Journal of business, pp. 61–65.
PATTON, Andrew J., 2011. Volatility forecast comparison using imperfect volatility
proxies. Journal of Econometrics. Vol. 160, no. Realized Volatility, pp. 246–256. issn
0304-4076.
PATTON, Andrew J.; POLITIS, Dimitris N; WHITE, Halbert, 2009. Correction to “Automatic
block-length selection for the dependent bootstrap” by D. Politis and H. White.
Econometric Reviews. Vol. 28, no. 4, pp. 372–375.
PATTON, Andrew J.; SHEPPARD, Kevin, 2015. Good volatility, bad volatility: Signed
jumps and the persistence of volatility. Review of Economics and Statistics. Vol. 97, no. 3,
pp. 683–697.
PATTON, Andrew J.; SHEPPARD, Kevin, 2009. Optimal combinations of realised volatility
estimators. International Journal of Forecasting. Vol. 25, no. 2, pp. 218–238.
PETRALIAS, Athanassios; DELLAPORTAS, Petros, 2015. Volatility prediction based on
scheduled macroeconomic announcements. Canadian Journal of Statistics. Vol. 43, no. 2,
pp. 199–223.
PETTENUZZO, Davide; SABBATUCCI, Riccardo; TIMMERMANN, Allan, 2021. Outlasting
the Pandemic: Corporate Payout and Financing Decisions During Covid-19.
CEPR Discussion Paper No. DP16145.
PLÍHAL, Tomáš; LYÓCSA, Štefan, 2021. Modeling realized volatility of the EUR/USD
exchange rate: Does implied volatility really matter? International Review of Economics
& Finance. Vol. 71, pp. 811–829.
POLITIS, Dimitris N; WHITE, Halbert, 2004. Automatic block-length selection for the
dependent bootstrap. Econometric reviews. Vol. 23, no. 1, pp. 53–70.
POLYZOS, Efstathios Stathis, 2022. Escalating Tension and the War in Ukraine: Evidence
Using Impulse Response Functions on Economic Indicators and Twitter Sentiment.
Available at SSRN 4058364.
69
BIBLIOGRAPHY
PONG, Shiuyan; SHACKLETON, Mark B.; TAYLOR, Stephen J.; XU, Xinzhong, 2004.
Forecasting currency volatility : a comparison of implied volatilities and AR[FI]MA
models. Journal of banking & finance. Vol. 28, no. 10, pp. 2541–2563. issn 03784266.
POON, Ser-Huang; GRANGER, Clive WJ, 2003. Forecasting volatility in financial markets:
A review. Journal of economic literature. Vol. 41, no. 2, pp. 478–539.
RACINE, Jeffrey S.; HAYFIELD, Tristen, 2021. Nonparametric Kernel Smoothing
Methods for Mixed Data Types. R package version 0.60-11. Vol. https://CRAN.R-
project.org/package=np.
RAHMAN, Md Lutfur; AL MAMUN, Mohammed Abdullah, 2021. How resilient are the
Asia Pacific financial markets against a global pandemic? Pacific-Basin Finance Journal.
Vol. 69, p. 101656. issn 0927-538X.
ROGERS, L Christopher G; SATCHELL, Stephen E, 1991. Estimating variance from high,
low and closing prices. The Annals of Applied Probability, pp. 504–512.
RYWKIN, Michael, 2014. Ukraine: between Russia and the West. American Foreign policy
interests. Vol. 36, no. 2, pp. 119–126.
SAMOKHVALOV, Vsevolod, 2015. Ukraine between Russia and the European Union:
triangle revisited. Europe-Asia Studies. Vol. 67, no. 9, pp. 1371–1393.
SANTOS, Douglas G.; ZIEGELMANN, Flavio A., 2014. Volatility Forecasting via MIDAS,
HAR and their Combination: An Empirical Comparative Study for IBOVESPA. Journal
of Forecasting. Vol. 33, no. 4, pp. 284–299. issn 1099-131X.
SENTANA, Enrique, 1995. Quadratic ARCH models. The Review of Economic Studies.
Vol. 62, no. 4, pp. 639–661.
SHAHZAD, Syed Jawad Hussain; BOURI, Elie; ROUBAUD, David; KRISTOUFEK,
Ladislav; LUCEY, Brian, 2019. Is Bitcoin a better safe-haven investment than gold
and commodities?. International Review of Financial Analysis. Vol. 63, pp. 322–330.
SHELEST, Hanna, 2015. After the Ukrainian crisis: Is there a place for Russia? Southeast
European and Black Sea Studies. Vol. 15, no. 2, pp. 191–201.
SMALES, L.A., 2018. Bitcoin as a safe haven: Is it even worth considering?. Finance
Research Letters. Vol. in press. issn 1544-6123.
SYMITSI, Efthymia; CHALVATZIS, Konstantinos J., 2019. The economic value of Bitcoin:
A portfolio analysis of currencies, gold, oil and stocks. Research in International Business
and Finance. Vol. 48, pp. 97–110. issn 0275-5319.
70
BIBLIOGRAPHY
SZCZYGIELSKI, Jan Jakub; BWANYA, Princess Rutendo; CHARTERIS, Ailie;
BRZESZCZYŃSKI, Janusz, 2021. The only certainty is uncertainty: An analysis of the
impact of COVID-19 uncertainty on regional stock markets. Finance research letters,
p. 101945.
TAYLOR, Nick, 2017. Realised variance forecasting under Box-Cox transformations.
International Journal of Forecasting. Vol. 33, no. 4, pp. 770–785.
THORBECKE, Willem, 1997. On stock market returns and monetary policy. The Journal
of Finance. Vol. 52, no. 2, pp. 635–654.
THORNTON, Daniel L, 1998. The information content of discount rate announcements:
What is behind the announcement effect? Journal of Banking & Finance. Vol. 22, no. 1,
pp. 83–108.
TRUCÍOS, Carlos, 2019. Forecasting Bitcoin risk measures: A robust approach. International
Journal of Forecasting. Vol. 35, no. 3, pp. 836–847.
URQUHART, Andrew, 2018. What causes the attention of Bitcoin? Economics Letters.
Vol. 166, pp. 40–44.
VERA-VALDÉS, J Eduardo, 2021. The persistence of financial volatility after COVID-19.
Finance Research Letters, p. 102056.
VORTELINOS, Dimitrios I, 2017. Forecasting realized volatility: HAR against Principal
Components Combining, neural networks and GARCH. Research in international
business and finance. Vol. 39, pp. 824–839.
WANG, Jingjing; WANG, Xiaoyang, 2021. Covid-19 and financial market efficiency:
Evidence from an entropy-based analysis. Finance Research Letters, p. 101888.
WANG, Yaw-Huei; WANG, Yun-Yi, 2016. The Information Content of Intraday Implied
Volatility for Volatility Forecasting. Journal of Forecasting. Vol. 35, no. 2, pp. 167–178.
issn 02776693.
WEI, Xiaoyun; HAN, Liyan, 2021. The impact of COVID-19 pandemic on transmission of
monetary policy to financial markets. International Review of Financial Analysis. Vol. 74,
p. 101705.
WIJK, Dennis van, 2013. What can be expected from the BitCoin. Erasmus Universiteit
Rotterdam.
WU, Xinyu; XIE, Haibin, 2021. A realized EGARCH-midas model with higher moments.
Finance Research Letters. Vol. 38, p. 101392.
71
BIBLIOGRAPHY
XIE, Haibin; YU, Chengtan, 2020. Realized GARCH models: Simpler is better. Finance
Research Letters. Vol. 33, p. 101221.
XINZHONG, Xu; TAYLOR, Stephen J., 1995. Conditional volatility and the informational
efficiency of the PHLX currency options market. Journal of Banking & Finance. Vol. 19,
no. 5, pp. 803–821. issn 03784266.
YANG, Jianlei; YANG, Chunpeng, 2021. Economic policy uncertainty, COVID-19 lockdown,
and firm-level volatility: Evidence from China. Pacific-Basin Finance Journal.
Vol. 68, p. 101597. issn 0927-538X.
YERMACK, David, 2013. Is Bitcoin a Real Currency? An Economic Appraisal. NBER
Working Paper. Vol. 19747.
YILMAZKUDAY, Hakan, 2021. COVID-19 and Monetary Policy with Zero Bounds: A
Cross-Country Investigation. Finance Research Letters, p. 102103.
ZAREMBA, Adam; KIZYS, Renatas; AHARON, David Y; DEMIR, Ender, 2020. Infected
markets: Novel coronavirus, government interventions, and stock return volatility
around the globe. Finance Research Letters. Vol. 35, p. 101597.
ZHANG, Dayong; HU, Min; JI, Qiang, 2020. Financial markets under the global pandemic
of COVID-19. Finance Research Letters. Vol. 36, p. 101528.
72
A Full text of the paper collection
This part is excluded from the public version due to the copyright restrictions.
73