DOI: 10.5817/FAI2018-1-4 No. 1/2018
51
Empirical Testing of the Response of Czech Stock
Market to Downgrades of Greek Credit Rating
in the Light of the Efficient Market Hypothesis
Petr Seďa1
, Juan Antonio Jimber del Río2
, María de los Baños GarcíaMoreno
García3
1
VSB-Technical University of Ostrava
Faculty of Economics, Department of Mathematical Methods in Economics
Sokolská třída 33, 701 21 Ostrava 1, Czech Republic
E-mail: petr.seda@vsb.cz
2
University of Córdoba
Faculty of Law and Business and Economic Sciences, Department of Economics, Sociology
and Agricultural Policy, Area of Financial Economy and Accounting
Puerta Nueva S/N, 14071-Córdoba, Spain
E-mail: jjimber@uco.es
3
University of Córdoba
Faculty of Law and Business and Economic Sciences, Department of Statistics,
Econometrics, Operational Research and Applied Economics
Puerta Nueva S/N, 14071-Córdoba, Spain
E-mail: d52gagam@uco.es
Abstract: Empirical testing of the linkages between macroeconomic news and
asset price movements in terms of response to released macroeconomic
information have been a subject of many investigations using different testing
methods. The objective of this paper is to study the impact of announcements of
Greek credit rating downgrades on the prices of the most liquid assets quoted in
the Czech stock market. This issue is tightly related to semi-strong form of the
efficient market hypothesis, which is one of possible analytical approaches when
analyzing behaviour of assets in financial markets. The reaction of the Czech stock
market is assessed in relation to seven announcements of Moody´s rating agency
regarding changes of credit rating of Greek government bonds in the period 2009–
2012. For the purpose of this paper, the event study methodology is applied. The
basic idea of this statistical method is to determine values of abnormal returns,
which can be defined as a difference between actual and equilibrium returns. In
order to calculate equilibrium returns, the Capital Asset Pricing Model (CAPM) is
used. The differences between actual and equilibrium returns are then verified with
a help of selected nonparametric statistical tests. Namely, the exact sign test and
the Wilcoxon signed-rank test are utilized. Based on results of nonparametric
statistical tests, the null hypothesis of information efficiency of the Czech stock
market is conclusively rejected.
Keywords: CAPM, Czech stock market, event study, semi-strong form of
efficiency, nonparametric tests
JEL codes: C12, G14, G18
52
Introduction
The knowledge of possible impact of macroeconomic news on stock prices is
important when assessing stock market efficiency. The impact of macroeconomic
information on equity prices for assessment of stock market efficiency and possible
forecasting of stock market reactions have received a considerable attention in
academic literature (Gurgul and Wójtowicz, 2015). The issue of testing the
response of stock markets to announcement of macroeconomic news is very topical
especially if we consider the recent global financial crisis. The global financial crisis
and subsequent debt crisis significantly hit stock markets worldwide. The Czech
Republic, an export-oriented economy with substantial reliance on foreign capital,
was not an exception. The fund withdrawal led by foreign investors in the Czech
stock market exacerbated volatility in stock market and decline of the whole market
by more than 60% (Seďa, 2012).
Greece is one of the countries that were affected by the global financial crisis and
the subsequent debt crisis fatally (Švihlíková, 2010). This fact led to repeated
downgrades of Greek government bonds rating. Since mutual relations among
stock markets grew during the global financial crisis (Seďa and Jimber del Río,
2014), crucial events in the Greek financial market could lead to a significant
reaction also in the Czech stock market. An effect of increasing linkages between
markets around the world was confirmed for instance by Nikkinen et al. (2006),
Kim and Nguyen (2008) or Albuquerque and Vega (2009). Macroeconomic
announcements that concern one economy can simply influence not only the
domestic stock market but also their foreign counterparts significantly.
The objective of this paper consists in empirical testing of the impact of
announcements of Greek credit rating downgrades on the prices of the most liquid
assets quoted in the Czech stock market. A response of the Czech stock market will
be measured. This includes its strength, direction and duration to changes of credit
rating of Greek government bonds as published by Moody’s agency in the period
2009–2012. For the purpose of this paper, the event study methodology will be
utilized.
The added value of this paper consists in the following aspects. Firstly, this paper
contributes to the discussion about information efficiency of the Czech stock
market, which is still considered an emerging market. Secondly, empirical studies
focusing on testing the response of stock markets usually work with macroeconomic
news that are announced on regular basis (monthly, quarterly or yearly). This paper
works with irregular news. Thirdly, statistical evaluation of results is performed
merely via utilization of the data in the form of returns of stock indexes. In this
paper, we also utilized log returns of particular shares that form the base of the
main index PX.
The rest of the paper is organized as follows. In chapter 2, the theoretical basis of
the event study method will be discussed thoroughly. In addition, the methods of
calculating the actual, equilibrium and abnormal returns will be defined. Finally,
53
selected non-parametric tests, which are usually used for statistical evaluation of
abnormal returns, will be described. In section 3, we present the data used in this
paper and give a brief description of the Greek crisis and related development of
Greek government bonds rating. The application part of this paper is presented in
section 4. It mainly contains empirical testing of the Czech stock market reaction
to announcements of Greek credit rating downgrades and consequent statistical
evaluation of results. The final section concludes this paper and opens possible
extensions of the topic investigated.
1 Theoretical Background of Efficient Market Hypothesis
The behavior of stock prices has been a subject of many recent investigations. Most
theories claim that prices in stock markets are efficient and therefore cannot be
forecasted. However, practitioners have never believed it, and naturally tried to
maximize profit using sophisticated forecasting methods. The efficient market
hypothesis is almost certainly the right place to start when thinking about asset
price formation. The basic idea of the efficient market hypothesis is based on the
determination of equity prices by interaction of interested rational market agents.
Information efficiency depends on how market prices reflect all relevant information
(Fama, 1970).
The efficient market hypothesis says that at any given time, security prices fully
reflect all available information (Seďa and Jimber del Río, 2016). Market efficiency
is defined within a hierarchy of three nested information sets. Considering this
hierarchy of information, we can define efficiency with respect to increasing
amounts of information (Fama, 1970):
 Weak form of efficiency - means that no investor can earn excess returns
by developing trading rules based on historical prices or return information.
 Semi-strong form of efficiency - denotes that it is impossible to earn above
average profit from trading based on any publicly available information.
 Strong form of efficiency - implicates that market agents cannot earn above
average returns when using any information, whether public or private.
Figure 1 describes the below mentioned sets of information:
A – All available information,
B – Publicly available information,
C – Historical price information.
It is obvious, that the following implications hold:
𝐶 ⊂ 𝐵 ⊂ 𝐴 (1)
54
Figure 1 Information sets
Source: Authors' elaboration
Equation (1) implies that testing the semi-strong form of efficiency is usually
focused on a measurement of a speed with which new publicly available information
is absorbed in stock prices. Public announcements of important macroeconomic
data about inflation, production, profit and other economic variables cause
considerable attention in both financial literature and practically oriented analyses.
Practitioners and academicians are interested in the knowledge which
macroeconomic information are reflected in prices of financial instruments and how.
2 Literature Review
In the economic literature, there is a lot of evidence for the effect of macroeconomic
news on the stock market performance. The authors predominantly examined just
developed stock markets. Boyd et al. (2005) investigated the impact of U.S.
unemployment rate on the S&P stock index. Albuquerque and Vega (2009)
confirmed a delayed reaction of the Portuguese stock market to the U.S. news
release. Dimpfl (2011) showed that investors on the Frankfurt Stock Exchange
reacted immediately after news announcement. Moreover, significant reactions
were observed in the first ten minutes. Singh et. al (2013) utilized data from
developed stock markets and confirmed that U.S. macroeconomic announcements
influenced volatility rather than stock returns.
However, Central and Eastern European (CEE) stock markets with respect to the
semi-strong form of efficiency have been investigated only rarely. Investigators
usually examined just the weak form of efficiency of these markets. In addition,
research studies are not updated (Tran, 2007). Kiete and Uloza (2005) investigated
the semi-strong form of efficiency in CEE using the event study methodology. They
only examined the Lithuanian and Latvian stock markets and concluded that the
null hypothesis of efficiency cannot be rejected in the case of Lithuanian stock
market in the period 2001–2004. Nikkinen et al. (2006) investigated the impact of
U.S. macroeconomic news releases on 35 different stock markets including CEE
stock markets. The reaction of CEE stock markets including the Czech stock market
to the announcement of U.S. indicators was insignificant. Contrary to that,
Hanousek et al. (2009) investigated the reaction of Central European stock markets
to the announcement of news from the USA and eurozone. More precisely, they
A
B
C
55
examined how stock prices in the Czech Republic, Hungary and Poland responded
to macroeconomic news coming from the USA and eurozone. They found that the
strongest reaction to data announcements from the USA was observed in the Czech
stock market. Nevertheless, Buttner et al. (2012) found that the EU macroeconomic
news outbalanced the impact of U.S. news in the Czech equity market. Moreover,
Büttner and Hayo (2012) investigated the impact of political and fiscal news and
found that those shocks significantly affected long-term bond yields and, to a lesser
extent, other financial variables in the Czech Republic, Hungary and Poland. Égerd
and Kočenda (2014) examined the impact of macro news and central bank
communication on the emerging European forex markets. They showed that GDP
announcements had a significant impact on the CZK/EUR rate during the global
financial crisis period. Hanousek and Kočenda (2001) analyzed macroeconomic
news and spillovers in the Czech, Hungarian and Polish stock markets and proved
that interactions in those markets are strongly determined by the U.S. and EU stock
markets as well as the macroeconomic news originating thereby. Gurgul and
Wójtowicz (2014) examined reactions of the Polish stock market to U.S.
announcements. Hence, this paper contributes to the discussion on the efficiency
of newly emerged stock markets in transition economies.
3 Methodology
In this section, a brief description of experimental methods used in this paper will
be given. For the purpose of this paper, the event study method will be utilized. In
addition to this, the methods of calculating the actual, equilibrium and abnormal
returns will be explained.
3.1 Procedure of Event Study
The most commonly used approach to assess the potential impact of
macroeconomic news announcements on stock prices is based on regression with
dummy variables or quantile regression (Harju and Hussain, 2011). However, this
method has some serious imperfections because of intraday volatility patterns and
overnight returns (Gurgul and Wójtowicz, 2014). In order to avoid these
imperfections, we utilized the event study method. This method has been also
widely applied when investigating the impact of news announcements on daily data
(Corrado and Troug, 2008). Since the event methodology can be used to elicit the
effects of any type of macroeconomic event on the direction and magnitude of stock
price changes, this approach is very versatile. In general, the event study is a
statistical method to assess the impact of economic event on the value of financial
assets. This approach is thus commonly used in various research areas in
economics, such as accounting (Cready and Gurun, 2010) and finance (Trung,
2011). Some examples include announcements of macroeconomic variables or
corporate events like earnings announcements or mergers and acquisitions.
However, the event study method has some reservations in relation to the
assessment of economic policy (Beigi and Budzinski, 2013). Another imperfection
is related to the robustness of this approach to outliers (Sorokina et al., 2013).
56
History of the event study is relatively long. This approach was used for the first
time by Dolley (1933), followed later for instance by Ashley (1962). Later on, this
methodology was significantly developed by Fama (1969). In later years, there
have been defined many modifications that should eliminate complications resulting
from violations of statistical assumptions (MacKinley, 1997). When testing the
reaction to announcement of macroeconomic news, it is possible to describe the
event study method in the following five steps:
 definition of the event and timeline,
 calculation of actual returns,
 determination of equilibrium returns,
 calculation of abnormal returns,
 verification of statistical significance and economic interpretation of the
results.
The structure of an event should be defined prior to testing a response of a stock
market to new information announcement. The timing sequence is illustrated with
a timeline in Figure 2. It consists of three basic windows or periods that do not
overlap. These periods are as follows:
 estimation window,
 event window, and
 post event window.
Figure 2 The general timeline for an event study
Source: MacKinley (1997)
The estimation window is a period preceding the event window. The estimation
window represents an interval with the length of  L T T .1 1 0
The estimation
window is a period that is used for calculating the actual returns and coefficients of
financial models. It is recommended to be 100–250 days long when using daily
data. The event window represents a period when new information was published.
Even if the event is represented by an announcement on a given date, it is
recommended to set the event window length to be longer than just one day
(MacKinley, 1997). This condition facilitates the use of abnormal returns around
the event day in the analysis. The event window is defined as an interval with a
length of  L T T .2 2 1
Finally, the post event window is defined as an interval with
a length of  L T T .3 3 2
After identification of an event and definition of the term
structure, it is necessary to establish criteria for selection of companies used for
the purpose of the event study. It is also important to determine whether market
response to the announcement of new events will be tested via usage of daily data,
57
which are characterized by higher volatility over time compared to data of lower
frequency.
3.2 Calculation of Equilibrium and Abnormal Returns
The basic idea of the event study method is to compare normal or actual returns
with their equilibrium returns during the event window period. The differences
between actual and equilibrium returns represent abnormal returns. Normal returns
have to be calculated not only for the event window but also for the estimation
window period. If the rate of returns from holding the shares is to move in a
reasonable interval that is not too high, we can replace the percentage rate of
return yield by a logarithmic rate of return. Withing this paper, we defined normal
returns in the following way:
𝑅 = 𝑙𝑛 = 𝑙𝑛(𝑃 ) − 𝑙𝑛(𝑃 ) (2)
where t
R is the logarithmic return, t
P and t
P 1
are the prices of assets at time t and
t-1, respectively.
Equilibrium returns can be defined as returns that would be achieved by trading
agents if new information had not been published. Equilibrium returns are based
on values of coefficients of financial models used for the purpose of the event study.
These coefficients are necessary for calculation of equilibrium returns for the event
window period. Equilibrium or expected returns represent a benchmark actual
returns are compared with. Economic models allow precise calculation of
equilibrium returns thanks to the determination of economic constraints (Campbell
et al., 1997). The most common economic model that is used to determine
equilibrium returns in financial modeling is the Capital Asset Pricing Model (CAPM).
However, the market model can be applied as well.
The CAPM was built on diversification and modern portfolio theory. Sharpe (1964)
and Lintner (1965) introduced this model independently. If all assumptions of the
CAPM were met, it would be possible to consider a stock market to be efficient in
its strong form. Semi-strong and weak forms of market efficiency are characterized
by omitting or changing some of these assumptions (Kohout, 1995).
According to CAPM model, the expected return can be calculated as follows:
𝐸(𝑅 ) = 𝑟 + 𝛽 𝐸(𝑅 ) − 𝑟 (3)
where  iE R is the expected return of i-th share, fr is the risk-free rate, i
ˆ
represents the parameter of share sensitivity, and  ME R is the expected return
of market portfolio. When testing the reaction of the Czech stock market, a
necessary condition is to measure values of abnormal returns. Let i
AR be a sample
of L2
abnormal returns for i-th share in the event window. The sample abnormal
return is given by:
58
𝐴𝑅 = 𝑅 − 𝐸(𝑅 ) (4)
where i
R is the normal return of i-th share. In order to evaluate the reaction of the
Czech stock market and assess the absorption rate of newly published information,
it is necessary to sum up the calculated values of abnormal returns within the event
window period. The abnormal returns thus must be aggregated in order to draw
overall inferences for the event of interest. The aggregation should have two
dimensions - through time and securities. Let’s define   iCAR ,1 2 as the sample
cumulative abnormal return (CAR) from 1
to 2
where    T T .1 1 2 2
The CAR
from 1
to 2
is the sum of abnormal returns:
𝐶𝐴𝑅 𝜏 , 𝜏 = ∑ 𝐴𝑅 (5)
The abnormal returns must be aggregated for the event window. The absence of
any overlap and distributional assumptions imply that abnormal returns and
cumulative abnormal returns will be independent across securities. The abnormal
returns of individual shares can be aggregated using i
AR from equation (4) for
each event period   T ,...,T .1 2
1 Given N events, the sample average aggregated
abnormal returns for period  is given by:
𝐴𝑅 = ∑ 𝐴𝑅 (6)
where N is a number of events included. The average cumulative abnormal returns
can then be aggregated over the event window using the same approach that was
used to calculate the cumulative abnormal return for i-th security. It can be
calculated for any interval within the event window as follows:
𝐶𝐴𝑅 𝜏 , 𝜏 = ∑ 𝐴𝑅 (7)
Testing the reaction of the Czech stock market to news release will be carried out
just on the basis of iAR and   iCAR , .1 2
3.3 Nonparametric Tests
Reaction of the Czech stock market to news release regarding changes of Greek
credit rating must also be evaluated statistically. Otherwise, one cannot make any
conclusion in relation to the efficient market hypothesis. If a stock market is
efficient, an announcement of new information should be reflected in stock prices
immediately. The basic idea of semi-strong form of efficiency can be expressed as
follows: if a stock market is efficient, there should not be a statistically significant
difference between actual and equilibrium returns. In other words, abnormal
returns should not significantly differ from zero.
59
If the assumption of normality of abnormal returns is violated, selected
nonparametric statistical tests should be utilized. The most suitable and commonly
used nonparametric tests are the sign test and the Wilcoxon signed-rank test
(Hendl, 2009). When testing the reaction of a stock market in the light of efficiency,
the null and alternative hypothesis can be therefore defined as follows:
𝐻 : 𝐴𝑅 = 0,
𝐻 : 𝐴𝑅 ≠ 0.
Sign Test
The sign test, which is also known as the median test, is based on the assessment
of directions of deviations from expected value of the median  .0 Under the null
hypothesis, the number of positive and negative deviations from median 0 should
be approximately the same.
Testing the null hypothesis can be performed using the exact test or the
asymptotically valid test. When applying the exact test, the binomial distribution is
used for the Z value. Z is the number of   i i
d x 0 with positive signs (Hendl,
2009). When running the asymptotically valid test, which is based on approximation
using the normal distribution, it is necessary to calculate test statistics z according
to the following formula:
𝑧 =
√
(8)
If the test statistic z lies in the interval    / /
z , z ,2 2 where  /z 2 represent the
standardized value of the normal distribution, we accept the null hypothesis H0 and
reject the alternative hypothesis H1 (Hendl, 2009). The asymptotically valid test
can be used when the sample size is greater than the value of 25.
Wilcoxon Signed-rank Test
The Wilcoxon signed-rank test for one sample is more powerful than the sign test.
This test is based on comparison of absolute values of positive and negative
deviations from estimated values of the median  ,0 which are sorted by size. Under
the null hypothesis, the total sums of T and T should be approximately the same.
In the case of a larger sample size, an approximation by normal distribution is used.
When assuming the null hypothesis, the test statistic z is then calculated by
standardizing the variable T with a help of theoretical mean  E T and standard
deviation T . The test statistic z can be given:
𝑧 =
( )
(9)
60
Approximation by normal distribution is recommended to be used if the sample size
is greater than value of 25 only. If the test statistic z lies within the interval
   / /
z , z ,2 2 where  /z 2 represent the standardized value of the normal
distribution, the hypothesis H0 cannot be rejected (Hendl, 2009).
4 Data Sample Selection
The reaction of the Czech stock market will be investigated using data from the
Czech stock market in the period 2009–2012. We will use log-returns calculated
according to equation (2) of just 12 shares, which were included in the PX index
base during the complete testing period.
4.1 Changes in Greek Credit Rating
Changes of credit rating of Greek government bonds as published by Moody’s
international rating company in the period 2009–2012 have become a key
motivation for this paper as Greece is one of the countries that were surely
significantly affected by the global financial crisis and subsequent debt crisis. GDP
per capita in market prices decreased from 21,400 to 17,300 EUR (Mavridis, 2018).
This period is also characterized by higher volatility of stock markets (Seďa, 2012).
Critical situation of Greek economy was mainly caused by huge public expenditures,
which are related to the high level of employment in the public sector and premium
wages as well as high pension payments (Švihlíková, 2010). Changes of Greek
credit rating as published by Fitch and S&P agencies are not examined in the paper
because these companies utilize different rating methodologies.
The poor state of Greek public finances was a major reason for credit rating
downgrade from grade A1 to A2 on 22 December 2009. Rating downgrade resulted
in a fall in prices of Greek securities and decline of euro currency against US dollar.
The reason for credit rating downgrade from A3 to Ba1 was especially a high rate
of economic risks associated with the implementation of the rescue program as
proposed to Greece by the Eurozone countries together with the International
Monetary Fund. Further decline in credit rating happened on 14 June 2010 when
Moody's downgraded Greek credit ratings to a speculative level. Next, Greek credit
rating was downgraded by Moody's on 7 March 2011 by three steps deeper from
Ba1 to B1 level into the highest risk zone called "junk bonds". In that case, the
reason of credit-rating downgrade was a lack of confidence in reforms which should
help to correct deficit financing and lack of government revenue. Moody’s
downgraded the credit rating of Greece's commitments by next three steps to Caa1
level on 1 June 2011 because of increasing risk of government's inability to stabilize
its debt without restructuring. Greek rating downgraded again as a result of
extraordinary summit of euro members on 27 July 2011. Credit rating shrank to
the lowest possible grade C on 3 February 2012 as a result of private investors’
decision regarding the exchange of bonds. As a consequence, creditors wrote off
more than 50% of the nominal value of Greek bonds. The largest creditors of Greek
government bonds are especially domestic banks, e.g. National Bank of Greece,
61
Eurobank, and Alpha. The most important foreign creditors are especially French
and German financial institutions. Development of credit rating of Greek
government bonds is given in Table 1.
Table 1 Changes in credit rating of Greek government bonds
Date Rating
22. 12. 2009 A2
27. 4. 2010 A3
14. 6. 2010 Ba1
7. 3. 2011 B1
1. 6. 2011 Caa11
25. 7. 2011 Ca
3. 2. 2012 C
Source: www.moodys.com
4.2 Czech Stock Market
As has been already mentioned in Introduction, the reaction of the Czech stock
market on news release is investigated using data sample in the period 2009–2012.
Prague Stock Exchange (PSE) is the largest and oldest organizer of stock market
in the Czech Republic. After a fifty-year break caused by World War II and the
Communist regime, PSE was reopened in 1993. Trading on PSE is conducted via
licensed securities traders who are also the PSE members. These are primarily the
major banks and brokerages. If a common investor decides to invest in the
Exchange, he/she needs to contact one of the trading members. The PX index is
the main price index of the PSE and it is calculated in a real time. It is a free float
weighted price index made up of the most liquid assets. This index does not take
into account the dividend yields. Updating the base of the index is to maintain the
quality index performed each quarter. The PX contains the most liquid titles with
weights changing according to the market capitalization. At the present time, the
actual number of the basic issues is variable. However, in accordance with the
Principles of Updating the Base of the Index PX, which were approved in December
2001, the base cannot consist of more than 50 issues (Březinová, 2013). On 29
October 2007 it reached its top with 1936 points and ended the year with 1815
points. As a result of the global financial crisis, the PX index reached the value of
700 points on 27 October 2008, losing more than 50% of its value in two months.
The bottom was reached on 18 February 2009 with 629 points (Seďa, 2012). The
reaction of the Czech stock market on news announcement is tested using shares,
which are included in the PX index base. In addition, we use just the shares that
were included in the PX index base during investigated period 2009–2012. The
“modified” base of the PX index is shown in Table 2.
62
Table 2 Modified PX index base
Opening
of trading
Opening
of trading
AAA 22. 6. 1993 ORCO 13. 7. 1993
CETV 5. 2. 2008
Pegas
Nonvowens
18. 12. 2006
ČEZ 1. 10. 2002 Philip Morris 1. 2. 2005
Erste group 22. 6. 1993 Telefonica 27. 6. 2005
Komerční banka 1. 3. 1995 Unipetrol 22. 10. 2012
NWR 28. 8. 1997 VIG 22. 10. 2010
Source: Summarized by authors based on PSE website
5 Empirical Results
When testing the reaction of the Czech stock market in the light of the semi-strong
form of efficiency using the event study method, the most important and significant
periods are the estimation window and the event window. In this paper, the length
of the estimation window is one year. The event window is a period within which
the speed of absorption of new information is tested. For the purpose of this paper,
the event window consists of 15 trading days before the publication of new
information and 15 trading days after the publication of new information. “Zero
day” is set on the date of announcement of new information. To sum up, the event
window consists of 31 trading days. The timeline for our event study is graphically
presented in Figure 3.
Figure 3 Timeline for event study
Source: Authors' elaboration
5.1 Calculating the Abnormal Returns
The response of the Czech stock market to announcements of changes of Greek
credit rating will be tested through the average aggregated abnormal returns and
average cumulative abnormal returns, which are calculated according to equations
(6) and (7). In order to estimate equilibrium returns according to CAPM model, it
is necessary to set risk-free rate fr . The values of fr are determined as average
gross yields of Czech government bonds (Březinová, 2013). In this paper, the riskfree
rates for the period 2009–2012 are determined by gross monthly returns of 3year
government bonds as shown in Table 3.
63
Table 3 Value of risk-free rate (in %)
2009 2010 2011 2012
Yields of Czech government bonds 2.945 1.762 1.926 0.926
Source: Březinová (2013)
Abnormal returns are calculated as differences between normal (actual) and
equilibrium returns. Equilibrium returns have been calculated using the CAPM
model according to equation (3). In order to assess the response of the Czech
equity market to the changes of Greek credit rating, it was necessary to determine
average aggregate abnormal returns and cumulative average abnormal returns for
the event window. By averaging the abnormal returns for each day of the event
window, we obtained aggregated abnormal returns. Next, we calculated cumulative
abnormal returns. Since credit rating of Greek government bonds changed seven
times in the period 2009–2012, it was also necessary to calculate average values
of aggregated returns (Březinová, 2013).
Table 4 Values of average aggregated abnormal and cumulative abnormal
returns of the Czech equity market according to CAPM in %
Days  iAR CAPM  iCAR CAPM
-15 -1.140 -1.140
-14 -1.289 -2.429
-13 -0.907 -3.336
-12 -1.124 -4.459
-11 -1.191 -5.651
-10 -1.061 -6.712
-9 -1.712 -8.424
-8 -1.240 -9.664
-7 -1.321 -10.985
-6 -1.560 -12.545
-5 -1.674 -14.219
-4 -1.115 -15.334
-3 -0.985 -16.318
-2 -0.880 -17.198
-1 -1.109 -18.307
0 -1.248 -19.555
1 -1.366 -20.921
2 -1.190 -22.111
3 -1.013 -23.124
4 -1.160 -24.284
5 -0.907 -25.193
6 -1.488 -26.679
7 -1.218 -27.898
8 -1.382 -29.281
64
Days  iAR CAPM  iCAR CAPM
9 -1.682 -30.963
10 -1.962 -32.925
11 -2.166 -35.092
12 -0.987 -36.078
13 -1.502 -37.581
14 -0.073 -37.654
15 -0.858 -38.512
Source: Authors' elaboration
The values presented in Table 4 show that the amount of aggregated abnormal
returns according to the CAPM reached only negative values throughout the event
window. It means that the actual returns reached lower values than returns
estimated according to CAPM model on average. Regarding publication of
information on “zero day”, it can be concluded that no significant deviations from
reference values have been observed. Thus, the publication of new information did
not result in a jump drop. The values of average aggregated abnormal and
cumulative abnormal returns of the Czech stock market for the period of fifteen
days before and after announcement of seven credit rating changes are graphically
shown in Figure 3.
Figure 3 Average aggregated abnormal returns and cumulative average
abnormal returns of the Czech stock market
Source: Authors' elaboration
Figure 3 also shows the values of cumulative average abnormal returns based on
equation (6). Market response to announcement of new information is not evident.
The most striking reaction can be observed between the 7th
and the 11th
days after
the news release. Average aggregated abnormal returns achieved almost zero
value just 14 days after news announcement. The cumulative abnormal losses on
the day of announcement achieved the value of -19.56%. The abnormal returns
calculated according to CAPM model started to decrease already fifteen days before
the announcement of new information. Moreover, the average cumulative abnormal
65
loss reached the value of 38.51% on the 15th
day following the publication of new
information.
5.2 Results of Nonparametric Statistical Tests
When testing the response of the Czech stock market to changes of Greek credit
rating in the light of the efficient market hypothesis, the values of abnormal stock
returns of all companies included in PX index in the event window are compared
with zero abnormal returns that are considered a hallmark of the efficient market.
In order to accept the null hypothesis, it is necessary that values of abnormal return
for the event window on both sides of median should be approximately equal. Table
5 shows values of average abnormal returns of shares included in PX index, which
are calculated according to CAPM model. These values are also expressed in
absolute values. The ranking of average abnormal returns is indicated in column
iR.
Table 5 Average abnormal returns of shares traded on PSE in % during the event
window and their ranking
iAR iAR iR
AAA -1.323 1.323 9
CETV -1.648 1.648 11
ČEZ -0.664 0.664 1
Erste group -1.151 1.151 4
Komerční banka -1.167 1.167 5
NWR -1.354 1.354 10
ORCO -1.889 1.889 12
Pegas Nonvowens -1.268 1.268 8
Philip Morris -1.247 1.247 6
Telefonica -0.958 0.958 2
Unipetrol -0.979 0.979 3
VIG -1.256 1.256 7
Source: Authors' elaboration
In subsection 2.3, two nonparametric statistical tests were defined. Firstly, average
abnormal returns were evaluated using the sign exact test as the scale of our
sample is fewer than 25. The sign test is based on the assessment of the direction
of measurement deviations from the expected value of the median. When checking
the values of abnormal returns based on the CAPM model, none of returns reached
a positive value    Z Z0, 12 . Under the null hypothesis, positive and negative
values of abnormal returns have binomial distribution B(0.5, 12). Table 6 shows
cumulative probabilities for binomial distribution B(0.5, 12).
66
Table 6 Cumulative probabilities of binomial distribution B(0.5, 12)
Value of x 0 1 3 4 5
Probability P(X  x) 0.00024 0.00317 0.01929 0.07300 0.19385
Source: Hendl (2009)
To confirm the null hypothesis, the following condition must be hold: sum of the
sequence of positive abnormal returns T and the sum of the sequence of negative
abnormal returns T must be approximately equal. If the efficiency of the Czech
stock market is tested using the CAPM model, the null hypothesis of efficiency
  iAR 0 can be rejected and the stock market can be considered inefficient at
5% significance level (p-value = 0.00024 < 0.05).
Secondly, the Wilcoxon signed-rank test was applied. This test is more powerful
and is based on absolute values of differences between actual returns and expected
values of median. These differences are sorted by their size. The null hypothesis is
not rejected if the sum of sequences of positive abnormal returns T and the sum
of sequences of negative abnormal returns T are approximately equal. Since
abnormal returns achieved negative values only, the value of T is equal to zero
while the sum of sequences with negative values T reached the value of 78.
Testing the null hypothesis is performed by comparison of smaller value of T and
T with precise critical values w12
13 at 5% significance level. The value of
 T 0 lies within interval of critical values of W , .0 13 Thus, the null hypothesis
of efficiency   iAR 0 can be rejected again. Results of both nonparametric tests
are summarized in Table 7.
Table 7 Comparison of results of the sing test and Wilcoxon signed-rank test
Sign test Wilcoxon signed-rank test
H0
H1
H0
H1
CAPM Rejected Not rejected Rejected Not rejected
Source: Authors' elaboration
Based on outcomes of statistical tests provided, it can be concluded that results do
not depend on the type of statistical nonparametric test that we apply. When
applying the CAPM model for determination of equilibrium yields, the Czech stock
market cannot be considered efficient from the statistical point of view.
67
5 Discussion
In the case of the efficient market, a significant reaction just after the release of
negative changes in Greek government bond credit rating would be expected.
However, empirical results we achieved indicate that the Czech stock market cannot
be considered efficient in terms of semi-strong form of efficiency. Market response
to announcement of new information is not clear. The most significant reaction can
be seen between the 7th
and the 11th
days after news announcement. Average
aggregated abnormal returns achieved nearly zero value just 14 days after news
release. It seems that the Czech stock market anticipated the announcement of
negative news regarding changes in Greek credit rating and started to react already
earlier (- 15 days). Nevertheless, the adjustment was relatively slow as the Czech
market reacted significantly with a delay (+ 7–11 days). Another explanation is
that negative values of average abnormal returns were caused by other
fundamental factors and news releases about changes in Greek credit rating were
not significant in the Czech stock market.
However, our results have shown that prices in the Czech stock market do not
respond to new information immediately as it would be expected in a rational
marketplace. Our outcomes are in harmony with findings of empirical studies
performed only partially before. Tran (2007) rejected the weak form of efficiency
in the Czech market using linear and nonlinear methods, which is in concordance
with our outcomes. However, Hanousek et al. (2009) found that five-minute returns
in Czech, Hungarian and Polish markets reacted strongly to news originating in the
EU. Contrary to that, U.S. announcements have a considerable impact on the Czech
and Hungarian markets only. The results delivered for the Czech market do reflect
the fact that a significant part of traded volume in the Czech market is caused by
foreign investors that put more weight on foreign announcements. In addition,
Gurgul and Wójtowicz (2014) examined the reaction of the Polish stock market to
U.S. announcements based on intraday data. They found that the Polish market
reacted to unexpected news from the USA just one minute after a news release.
Inconsistency in these results can be caused by different data frequency used,
dissimilar data sample, different methods applied when calculating equilibrium
returns and different statistical tests used as well.
Nevertheless, our findings could have some important implications especially for
regulators and investors. For regulators, our research provides empirical
assessment of the current state of information efficiency and detects areas for
potential improvements. In bigger markets with more participants, the lower values
of abnormal returns were observed on the event day (Nikkinen et al., 2006). It
suggests that big markets are generally more efficient than small ones. Attracting
potential investors by lowering transaction fees could be an important tool for
further development of the Czech stock market. Next, the speed of information
dissemination could be supported by highlighting the upcoming news
announcements more actively for investors. For investors, fundamental analysis
can bring above average returns if share prices do not reflect their inner value.
68
Moreover, our results may also have important implications for diversification and
risk management strategies as the Czech Republic is preparing to enter the
eurozone. We can expect that the Czech market will be more sensitive to
macroeconomic shocks, especially those coming from the eurozone. Investors
should expect higher volatility when investing in the Czech market.
Our findings also bring some ideas about possible extensions of the topic
investigated. The reaction of stock markets can be further examined by separating
positive and negative news announcements as done recently for instance by Gurgul
and Wójtowicz (2015). The frequency of selected data sample may also play a
significant role when investigating the speed and strength of price reactions and
the duration of price adjustments after news release as confirmed by Hanousek et
al. (2009) and Gurgul and Wójtowicz (2015). In addition, the effects of news
announcements could be investigated using different types of companies
(according to their size, value or growth). Moreover, examining the downgrades of
Greek credit rating, as published by Fitch and S&P rating agencies, and utilizing
more current data sample would lead to results that are more valuable. Finally, the
reaction of a stock market could be theoretically related to behavioral aspects of
market participants.
6 Conclusions
Rational investors seek for mispriced stocks to make a profit on purchase or sale.
However, activities of investors on an efficient market lead to the fact that no
market participant is able to overcome the respective market and achieve aboveaverage
returns. In this paper, we examined the response of the Czech stock
market returns to publication of changes of Greek government bonds credit rating
as published by Moody's rating agency in the period 2009–2012. For the purpose
of this paper, the event study method is applied. All the computations are based
on daily log-returns of the shares included in PX index during the whole
investigation period. Based on results of the sign test and the Wilcoxon signed-rank
test, we can conclude that results of our analysis do not depend on a type of the
selected statistical nonparametric test applied. The equilibrium returns, which are
important for computation abnormal returns, are calculated with the help of the
CAMP model. Based on results of nonparametric statistical tests, the null hypothesis
of information efficiency is conclusively rejected. Regarding economic interpretation
of statistical outcomes, it can be concluded that the inefficiency of the Czech stock
market could be caused by infringement of certain assumptions of the efficient
market hypothesis (rational expectations of market participants, a large number of
investors, liquidity of market, etc.). Achieved results also have some economic
implications for investors and regulators. If equity prices do not reflect their inner
value, some investors can make excess returns, given their level of risk exposure.
In other words, announcement of news, whether good or bad, may not be reflected
in the prices for minutes, hours, days or even longer. Therefore, some investors
could purchase an asset at a bargain price before other market participants become
aware of the new information release. In addition, the risk premium, transaction
69
and information costs must be taken into account as well. Market inefficiency
generates important signals also for market regulators. They could focus on
fulfilling the assumptions of efficient markets.
Acknowledgments
This paper was supported by the Czech Science Foundation through project No. 13-
13142S and the European Social Fund within the project CZ.1.07/2.3.00/20.0296.
References
Albuquerque, R. and Vega, C. (2009). Economic News and International Stock
Market Co-movement. Review of Finance, 13(3), pp. 401–465. DOI:
10.1093/rof/rfn020.
Ashley, J. W. (1962). Stock Prices and Changes in Earnings and Dividends: Some
Empirical Results. Journal of Political Economy, 70(1), pp. 82-85.
Beigi, M. H. A. and Budzinski, O. (2013). Reservations on the Use of Event Studies
to Evaluate Economic Policy. Intereconomics, 48(3), pp. 174–179. DOI:
10.1007/s10272-013-0459-6.
Březinová, J. (2013). Testování středně silné formy efektivnosti vybraných
akciových trhů. (Unpublished master's thesis). VŠB – Technical University of
Ostrava.
Boyd, J. H., Hu, J. and Jagannathan, R. (2005). The Stock Market’s Reaction to
Unemployment News: Why Bad News is Usually Good for Stocks. Journal of
Finance, 60(2), pp. 649–672. DOI: 10.1111/j.1540-6261.2005.00742.x.
Büttner, D., Hayo, B. (2012). EMU-Related News and Financial Markets in the Czech
Republic, Hungary and Poland. Applied Economics, 44(31), pp. 4037–4053. DOI:
10.1080/00036846.2011.587775.
Büttner, D., Hayo, B., Neuenkirch, M. (2012). The Impact of Foreign
Macroeconomic News on Financial Markets in the Czech Republic, Hungary, and
Poland. Empirica, 39(1), pp. 19-44. DOI: 10.1007/s10663-010-9153-0.
Campbell, J. Y., Lo, A. W. and MacKinlay, A. C. (1997). The Econometrics of Financial
Markets. 2nd ed. New York: Princeton University Press.
Corrado C. J. and Truong C. (2008). Conducting Event Studies with Asia-Pacific
Security Market Data. Pacific-Basin Finance Journal, 16(5), pp. 493–521. DOI:
10.1016/j.pacfin.2007.10.005.
Cready, W. M. and Gurun, U. G. (2010). Aggregate Market Reaction to Earnings
Announcements. Journal of Accounting Research, 48(2), pp. 289-334. DOI:
10.1111/j.1475-679X.2010.00368.x.
70
Dimpfl, T. (2011). The Impact of US News to the German Stock Market - An Event
Study Analysis. The Quarterly Review of Economics and Finance, 51(4), pp. 389–
398. DOI: 10.1016/j.qref.2011.07.005.
Dolley, J. C. (1933). Characteristics and Procedure of Common Stock Split-Ups.
Harvard Business Review, 11(1), pp. 316-326.
Égert, B. and Kočenda, E. (2014). The Impact of Macro News and Central Bank
Communication on Emerging European Forex Markets. Economic Systems, 38(1),
pp. 73–88. DOI: 10.1016/j.ecosys.2013.01.004.
Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical
Work. Journal of Finance, 25(2), pp. 383-417.
Fama, E. F. (1969). The Adjustment of Stock Prices to New Information.
International Economic Review, 10(1), pp. 1-21.
Gurgul, H. and Wójtowicz, T. (2014). The Impact of US Macroeconomic News on
the Polish Stock Market. The Importance of Company Size to Information Flow.
Central European Journal of Operations Research, 22(4), pp. 795–817. DOI:
10.1007/s10100-014-0343-x.
Gurgul, H. and Wójtowicz, T. (2015). The Response of ATX Intraday Returns to U.S.
Macroeconomic News. Czech Journal of Economics and Finance, 65(3), pp. 230–
253.
Hanousek, J. and Kočenda, E. (2011). Foreign News and Spillovers in Emerging
European Stock Markets. Review of International Economics, 19(1), pp. 170-188.
DOI: 10.1111/j.1467-9396.2010.00939.x.
Hanousek, J., Kočenda, E. and Kutan, A. M. (2009). The Reaction of Asset Prices
to Macroeconomic Announcements in New EU Markets: Evidence from Intraday
Data. Journal of Financial Stability, 5(2), pp. 199–219.
Harju, K. and Hussain, S. M. (2011). Intraday Seasonalities and Macroeconomic
News Announcements. European Financial Management, 17(2), pp. 367–390. DOI:
10.1111/j.1468-036X.2009.00512.x.
Hendl, J. (2009). Přehled statistických metod: Analýza a metaanalýza dat, 3rd ed.
Praha: Portál.
Kiete, K. and Uloza, G. (2005). The Information Efficiency of the Stock Markets in
Lithuania and Latvia. Working Paper no. 75. Stockholm School of Economics in Riga.
Kim, S. J., Nguyen, D. Q. T. (2009). The Spillover Effects of Target Interest Rate
News from the US Fed and the European Central Bank on the Asia-Pacific Stock
Markets. Journal of International Financial Markets, Institutions and Money, 19(3),
pp. 415–31.
Kohout, P. (1995). Nelineární CAPM. Finance a úvěr, 45(8), pp. 429-442.
71
Lintner, J. (1965). The Valuation of Risk Assets and the Selection of Risky
Investments in Stock Portfolios and Capital Budgets. Review of Economics and
Statistics, 47(1), pp. 13-37.
MacKinlay, A. C. (1997). Event Studies in Economics and Finance. Journal of
Economic Literature, 35(3), pp. 13-39.
Mavridis, S. (2018). Greece’s Economic and Social Transformation 2008–
2017. Social Sciences. 7(1), pp. 1-14.DOI: 10.3390/socsci7010009.
Nikkinen, J., Omran, M., Sahlström, M. and Äijö, A. (2006). Global Stock Market
Reactions to Scheduled U. S. Macroeconomic News Announcements. Global Finance
Journal, 17(1), pp.92–104. DOI: 10.1016/j.gfj.2006.06.003.
Seďa, P. (2012). Impact of the Global Financial Crisis on Stock Market Volatility:
Evidence from Central European Stock Market. In: Mathematical Methods in
Economics, Karviná: Silesian University in Opava, pp. 787-792.
Seďa, P. and Jimber del Río, J. A. (2014). Ekonometrická analýza vzájemných vazeb
akciových trhů zemí střední Evropy v kontextu finanční krize. In: Řízení a
modelování finančních rizik. Sborník příspěvků ze 7. mezinárodní vědecké
konference, Ostrava: VŠB-Technical University of Ostrava, pp. 691-699.
Seďa, P. and Jimber del Río, J. A. (2016). Testing the Weak Form of Efficiency on
Chinese Stock Market. In: European Financial Systems 2016. Proceedings of the
13th International Scientific Conference, Brno: Masaryk University, pp. 669-677.
Sharpe, W. F. (1964). Capital Asset Prices – A Theory of Market Equilibrium under
Conditions of Risk. Journal of Finance, 19(3), pp. 425–442.
Singh, M., Nejadmalayeri, A. and Lucey, B. (2013). Do U.S. Macroeconomic
Surprises Influence Equity Returns? An Exploratory Analysis of Developed
Economies. The Quarterly Review of Economics and Finance, 53(2013), pp. 476–
485. DOI: 10.1016/j.qref.2013.05.002.
Sorokina, N., Booth, D. E. and Thornton, J. H. (2013). Robust Methods in Event
Studies: Empirical Evidence and Theoretical Implications. Journal of Data Science,
11, pp. 575-606.
Švihlíková, I. (2010). Globalizace a krize, 1st Ed. Praha: Grimmus.
Tran, V. Q. (2007). Testing the Weak Form of Efficiency on Czech Stock Market.
Politická ekonomie, 6(6), pp. 751-772. DOI: 10.18267/j.polek.622.
Trung, C. (2011). Post-Earnings Announcement Abnormal Return in the Chinese
Equity Market. Journal of International Financial Markets, Institutions and Money,
2(5), pp. 637-661. DOI: 10.1016/j.intfin.2011.04.002.