Force Majeure Events and Stock Market Reactions In Ukraine

This paper examines reactions in the Ukrainian stock market to force majeure events, which are divided into four groups: economic force majeure, social force majeure and terrorist acts, natural and technological disasters. More specifically, using daily data for the main Ukrainian stock market index (namely PFTS) over the period 01.01.1997-31.12.2018 this study investigates whether or not force majeure events create (temporary) inefficiencies and there exist profitable trading strategies based on exploiting them. For this purpose cumulative abnormal returns and trading simulation approaches are used in addition to Student’s t-tests. The results suggest that the Ukrainian stock market absorbs new information rather fast. Negative returns in most cases are observed only on the day of the event. The only exception is technological disasters, the market needing up to ten days to react fully in this case. Despite the presence of a detectable pattern in price behaviour after force majeure events (namely, a price decrease on the day of the event) no profitable trading strategies based on it are found as their outcomes do not differ from those generated by random trading.


Introduction
The Efficient Market Hypothesis (EMH, see Fama, 1970) is still the dominant theoretical paradigm for understanding the behaviour of asset markets. However, the empirical literature has provided plenty of evidence that is inconsistent with market efficiency in the form of various anomalies (including calendar and size ones), over-reactions and under-reactions, persistence, fat tails in the distribution of asset returns etc. One possible explanation for (temporary) inefficiencies is the arrival of unexpected new information; whilst earnings announcements are normally scheduled and therefore markets are ready to react to them (Foster, 1973;Chambers and Penman, 1984;Falk and Levy, 1989;Lonie et al., 1996;Cready and Gurun, 2010;Syed and Bajwa, 2018 and others), force majeure events (such as technological and natural disasters, terrorist acts, unexpected economic events etc.) are by their nature unpredictable and could have a significant impact on stock markets, especially in the case of major shocks such as the 9/11 attacks in the US or the result of the Brexit referendum. This paper analyses the specific case of the Ukrainian stock market with the aim of investigating whether or not force majeure events create (temporary) inefficiencies and there exist profitable trading strategies based on exploiting them. For this purpose, cumulative abnormal returns and trading simulation approaches are used in addition to Student's t-tests. The analysis is carried out for four different types of force majeure events: economic force majeure, social force majeure and terrorist acts, natural and technological disasters.
The layout of the paper is the following. Section 2 contains a brief literature review. Section 3 describes the data and methodology. Section 4 presents the empirical results. Section 5 provides some concluding remarks.

2.
Literature Review Event studies are often carried out to examine the impact of some specific events on stock markets (Ball and Brown, 1968;Fama et al., 1969;MacKinlay, 1997;de Jong et al., 1992;Corrado, 2011). A lot of papers focus on earning announcements on market variables (Falk and Levy, 1989;Cready and Gurun, 2010;Syed and Bajwa, 2018). Other studies analyse instead the effects of dividend announcements (Lonie et al., 1996), earning per share announcements (Foster, 1973), earnings announcements through timeliness of reporting disclosure (Chambers and Penman, 1984).
Force majeure events are unexpected and therefore cannot be incorporated into asset prices in advance. As a result, they could create conditions for obtaining abnormal profits until the new information has been completely absorbed by market participants. In addition to the traditional list of natural disasters (floods, earthquakes, hurricanes), social, military and political events (terrorist attacks, mass riots, protest actions), technological and aviation accidents, some economic events (such as unexpected bankruptcies of companies and financial institutions, cyberattacks to the commercial sector, etc.) can also be considered as force majeure events.
As for the impact of natural disasters, evidence for Australia is provided by Worthington and Valadkhani (2004) and Worthington (2010), for Canada by Laplante and Lanoie (1994), for Japan and the US by Wang and Kutan (2013).
Unlike natural disasters, which normally affect individual countries, technological disasters are industry related. These include the explosions in chemical plants and refineries worldwide in 1990-2005(Capelle-Blancard and Laguna, 2009); the nuclear disaster in Fukushima-Daiichi causing abnormal returns for Japanese, French and German nuclear utilities (Ferstl et al., 2012); 209 energy accidents in 1907-2007 without a significant impact on stock markets (Scholtens and Boersen, 2011); aviation disaster announcements causing market losses and higher stock volatility (Kaplanski and Levy, 2008).
Regarding the effects of economic events, Campbell et al. (2003) investigated negative stock market reactions to information security breaches, and Knight and Pretty (1999) the impact of 15 corporate catastrophes on stock prices their volatility, and trading volumes.
Additional studies considered the impact of various force majeure events. Hanabusa (2010) showed that the 9/11 terrorist attacks had a significant effect on the stock prices of Japanese companies, whilst the Iraq War and Hurricane Katrina did not. Baker and Bloom (2013) used a panel including stock prices and volatilities for 60 countries over the period 1970-2012 to investigate the impact of natural disasters, terrorist attacks and unexpected political shocks on economic growth through stock market proxies. Tavor and Teitler-Regev (2019) examined 344 significant effects of natural disasters, artificial disasters and terrorism on the stock market using the Pessimism Index. Karolyi and Martell (2010) compared stock market reactions to 77 terrorist attacks with those to extreme events such as financial crashes (4) and natural catastrophes (19). Below we also consider a variety of unexpected force majeure events to assess their impact on the Ukrainian stock market in particular.

Data and Methodology
The Ukrainian stock market index, namely the PFTS, is used for the empirical analysis. The sample period goes from 01.01.1997 to 31.12.2018. The frequency is daily. As already mentioned, force majeure events are divided into four categories: economic, technological and natural disasters, terrorist acts in Ukraine during 1997-2018. A full list of force majeure events and their description is provided in Appendices A-D.
The following hypotheses are then tested: -Hypothesis 1 (H1)force majeure events create temporary inefficiencies in the Ukrainian stock market. -Hypothesis 2 (H2)trading strategies based on force majeure events can generate abnormal profits. To test them we use the following methods:


A cumulative abnormal returns approach;  Student's t-tests;  A trading simulation approach. The cumulative abnormal returns approach is based on MacKinlay (1997) and is standard for event studies. Abnormal returns are defined as follows: Electronic copy available at: https://ssrn.com/abstract=3362155 where is the daily return of the PFTS index over the period t and ( ) is the average return over the sample period Returns ( ) are computed as follows: where i R returns on the і-th day in %; close price on the (і-1) day; i Closeclose price on the і-th day.
and ( ) is calculated as follows where is the sample size.
The mean abnormal return corresponding to force majeure events on day α denoted as ̅̅̅̅ is the sum of the individual abnormal returns on that day divided by the number of force majeure events, where N is the number of force majeure events.
The cumulative abnormal return for the PFTS index from day to day denoted as ( 1 , 2 ) is simply the sum of the daily abnormal returns from day 1 to day 2 : ( 1 , 2 ) = ∑ , 2 = 1 The sample average cumulative abnormal return for event observations from 1 to 2 denoted as ̅̅̅̅̅̅ ( 1 , 2 ) is the sum of the mean abnormal return from day 1 to day 2 : Electronic copy available at: https://ssrn.com/abstract=3362155 These abnormal returns are cumulated over 1, 2, 3, 5, and 10 days following the force majeure event considered by adding them up over these periods. Negative cumulative abnormal returns are evidence in favour of Hypothesis 1.
Parametric t-tests are also carried out for Hypothesis 1. The Null Hypothesis (H0) is that the data (returns after force majeure events and over the full sample) belong to the same population, a rejection of the null suggesting the presence of a statistical anomaly in the price behaviour after force majeure events. The test is carried out at the 95% confidence level, and the degrees of freedom are N -1 (N being equal to N 1 + N 2 ).
To test H2 a trading strategy based on force majeure events is used to establish whether or not it can generate abnormal profits. The trading simulation approach replicates the actions of the trader according to a given algorithm. In this particular case the algorithm is the following: open a short position in the Ukrainian stock market right after the force majeure event and hold it for a specific period of time.
The percentage result of the individual deal is computed as follows: where opening price closing price The sum of the results from each deal is the total financial result of trading. A strategy producing positive total profits implies that there exists an exploitable market anomaly.
Another important indicator of trading strategy efficiency is the percentage of successful trades: To make sure that the results we obtain are statistically different from the random trading ones t-tests are carried out. They compare the means from two samples to establish whether or not they come from the same population. The first sample consists of the trading results based on Hypothesis 2, and the second one of random trading results.
The null hypothesis (H0) is that the mean is the same in both samples, and the alternative (H1) that it is not. The computed values of the t-test are compared with the critical one at the 5% significance level. Failure to reject H0 implies that there are no advantages from exploiting the trading strategy being considered, whilst a rejection suggests that the adopted strategy can generate abnormal profits.
It should be mentioned that the trading simulation approach used in this paper does not incorporate transaction costs (spread, fees to the broker or bank, swaps etc.).

Empirical Results
Descriptive statistics for the PFST returns over the period of analysis are provided in Table 1. The series is rather volatile, as indicated by the size of standard deviation and minimum/maximum values; its mean return is consistent with random walk behaviour. First we calculate the cumulative abnormal returns in the case of economic force majeure events; the results are reported in Table 2. Mean abnormal return across force majeure event on day α -0.59% 1.24% -0.08% 0.15% 0.01% Cumulative abnormal return across event -4.73% 5.15% 4.50% 8.52% 12.49% observations from 1 to day 2 Average cumulative abnormal return across event observations from 1 to day 2 -0.59% 0.64% 0.56% 1.07% 1.56% As can be seen, there appears to be a negative reaction of the stock market only on the day of the event, when cumulative abnormal returns as well as mean abnormal returns are negative. Over longer time horizons (2, 5 and 10 days after force majeure event) there is no evidence of a market drop.
To establish whether the effect of economic force majeure events is statistically significant t-tests are carried out (see Table 3). The null hypothesis is not rejected, which implies that the behaviour of returns after economic force majeure events does not statistically differ from their usual one.
As for Hypothesis 2, Table 2 indicates a market drop only on the day of the event, but the size of returns is not statistically different from the average returns over the whole sample period. Nevertheless we simulate the action of a trader for this case and obtain the results reported in Table 4. As can be seen, the percentage of successful trades is 50% and profit per trade are close to the average return for the full sample, which suggests that these results do not differ from the random ones. As a further check a t-test is carried out (see Table 5), which again leads to the same conclusion. Therefore there is no evidence of abnormal profits based on exploiting the occurrence of economic force majeure events. The results of the cumulative abnormal returns approach for the case of social force majeure and terrorist acts in Ukraine during 1991-2018 are presented in Table 6. The results are very similar to those for the economic force majeure events: negative returns are observed only on the day of event (not for any other time horizon). The t-test results are presented in Table 7. The null hypothesis is not rejected in any case, i.e. the behaviour of returns after social force majeure and terrorist acts does not statistically differ from the usual one.
To test Hypothesis 2 the time horizon α =1 is used (this is the only case when a negative reaction was observed). The results are presented in Table 8. There is a high percentage of successful trades, namely 75%, but profit per trade is close to the mean return, which is evidence in favour of the randomness of the results. A t-test confirms that there are no statistical differences between these trading results and those from random trading, i.e. Hypothesis 2 is rejected. The results of the cumulative abnormal returns approach for the case of natural disasters are presented in Table 10. As in the previous cases (economic force majeure and social force majors/terrorist acts) Hypothesis 1 is not rejected only for the day of the force majeure event, whilst there is no evidence of negative returns for any other time horizons. A t-test indicates that H1 is rejected even for α =1 (see Table 11). The results of the trading strategy for α = 1 are presented in Table 12. Only 25% of the trades are successful, and profit per trade is less than the mean return. The t-test statistics implies that these results are not statistically different from the random ones (see Table 13). Finally, the results of the cumulative abnormal returns approach in the case of technological disasters are displayed in Table 14. In this case, unlike the previous ones, there is no pattern in price behaviour on the day of the event, and the longer the time horizon the higher cumulative abnormal returns are. To see whether the detected differences in returns are statistically significant a t-test is again performed (see Table 15). The null hypothesis is rejected only for the case of α = 10. To find out whether this abnormal behaviour provides opportunities to "beat the market" a trading simulation approach is used once more. The results for the time horizon α = 10 are showed in Table 16. The number of successful trades is close to the 60% and profit per trade is three times higher than the average return. However, the t-test statistic (see Table  17) implies that these results are not statistically different from the random ones. The results of the tests for the Hypothesis 1 and 2 are summarised in Tables  18 and 19 respectively. -+/-+/-+/-+ * "+" -Hypothesis 1 is confirmed and differences are statistically significant; "+/-" -Hypothesis 1 is not rejected but differences are not statistically significant; "-" -Hypothesis 1 is rejected. ----+/-* "+" -Hypothesis 2 is not rejected and results differ from random ones; "+/-" -Hypothesis 2 is not rejected but results do not differ from random ones; "-" -Hypothesis 2 is rejected.

Conclusions
This paper examines price behaviour in the Ukrainian stock market after four types of force majeure events (economic force majeure, social force majeure and terrorist acts, natural and technological disasters. Using daily data for the PFTS index (the main index of the Ukrainian stock market) for the period 01.01.1997-31.12.2018 two different hypotheses are tested: force majeure events create temporary inefficiencies in the Ukrainian stock market (H1), and trading strategies based on force majeure events can generate abnormal profits (H2). For this purpose a variety of methods are used, including cumulative abnormal returns and trading simulation approaches as well as Student's t-tests. On the whole, it appears that the Ukrainian stock market absorbs new information rather fast. Negative returns in most cases are observed only on the day of the event. The behaviour of returns on other days shows no sign of abnormality. The only exception is technological disasters, possibly because it is harder for agents to evaluate their consequences and incorporate them into stock prices.
Further, the trading simulation analysis implies that, despite the presence of a specific pattern in price behaviour after force majeure events (namely a decrease on the day of the event), it is not possible to devise trading strategies exploiting it that generate abnormal profits: the results from apparently successful strategies are not statistically different from the random trading ones.
Appendix A Bankruptcy of the bank "Nadra" Because of insolvency of the bank, the amount of payments from the Guarantee Fund for Individuals' Deposits is 3.6 billion UAH.

Bankruptcy of DeltaBank
Because of insolvency of the bank, the amount of payments from the Guarantee Fund for Individuals' Deposits and state loses is 24 billion UAH.

Nationalization of Privatbank
According to the decision of the government and the National Security and Defense Council, PrivatBank became the property of the state for UAH 1; the capital requirement at the time of nationalization amounted to UAH 148 billion, the amount of issued Eurobonds was USD 595 million.

18.12.2016
Hacker attack on government sites in Ukraine Due to the mass hacking attack on government sites (the State Treasury of Ukraine and others) and the network of state authorities there were mass delays in payments. To protect against hackers, the Government of Ukraine allocated 80 million UAH.

6.12.2016
Bankruptcy "PlatinumBank" Due to insufficient level of capital, the bank is declared as bankrupt, the amount of losses of state-owned enterprises -clients of the bank is 500 million UAH.

«Petya» virus attack
Due to the large-scale virus cyberattacking through the software of M.E.Doc. installed on 1 million computers, the estimated losses for the Ukrainian economy are 0.5% of GDP. The largest enterprises and institutions of Ukraine were the Boryspil airport, Ukrtelecom, Ukrposhta, Oschadbank, Ukrzaliznytsya and others.

27.06.2017
Bankruptcy of insurance company "Dominanta" Due to the insolvency of the company included in the 10 largest insurance companies of Ukraine, outstanding obligations to customers remain at the amount of UAH 98 million, 300 thousand of civil liability laws remained unprotected 12.08.2018 Bankruptcy of PJSC "Black Sea Shipbuilding Plant" According to the court decision PJSC "Black Sea Shipbuilding Plant", the largest enterprise in Ukraine and Europe, which was founded in 1897, was declared bankrupt after unsuccessful and lengthy sanation procedures.

4.07.2018
Appendix B "Orange Revolution", Kyiv, etc. regional centers Because of numerous falsifications in the second round of elections, an all-Ukrainian protest rally began, which resulted in a significant change in the political vector in the country.

22.11.2004
"Revolution of dignity", Kyiv, etc. regional centers    Because of the explosions of shells, 10 warehouses from 17 were destroyed, 66 apartment buildings and 120 private houses, five schools and three hospitals were damaged, two people were injured.

10.10.2003
Explosions of artillery shells in the warehouses of the military unit A-2985 with. Novobohdanivka Zaporozhye region.
As a result of the explosions of shells, 90 thousand tons of munitions ("Grad", "Smerch" and "Hurricane") were destroyed and 1.5 thousand people were evacuated, 5 people died, 81 persons were traumatized by various degrees of severity, 22 houses were destroyed The explosion of gas in a residential building in Dnipropetrovsk Because of the explosion, the entrance of the dwelling house was completely destroyed, and three neighboring ones were damaged.

13.10.2007
23 people were killed, 20 were injured Accident at the mine named after Zasyadko, Donetsk As a result of the explosion of air-methane mixture on the horizon of 1078, 100 people died 18.11.2007 Explosions of shells in the warehouses of the military part of 0829, 61st Arsenal of the Southern Operational Command of the Army, and fire at the gas distribution station of Lozova, Kharkiv region.
Because of explosions evacuated people in the 3-km zone, 1 person was injured.
27.08. 2008 Fire at the station "Otradnoe", Dzhankoy district (ARC Crimea) at the station for storage of pesticides Because of the fire, an area of 600 m2 burned about 160 tons of pesticides. 17.10.2009 The explosion at the hospital № 7 in the city of Lugansk As a result of the explosion of an oxygen cylinder in the intensive care unit, 16 people were killed 18.01. 2010 Explosion of the gas pipeline in Uzhgorod, Transcarpathian region.
As a result of the fire, 4 power units of TPP were destroyed by 1 person, 5 people were injured, 12 thousand residents of Svitlodarsk city were left without water and heating 29.03.2013 Accident at the chemical plant of PJSC "Stirol", city of Horlivka, Donetsk region The largest accident at chemical plants during the years of independence, which resulted in the release of ammonia, 6 people killed injured 26.