The impact of public holidays on stock prices in The
Netherlands, Belgium and Germany
Jeroen Pennings
10088830
Field: Behavioral Economics / Behavioral Finance Specialization: Economics
Faculty of Economics and Business University of Amsterdam
Table of Contents
I. Introduction II. Literature Review III. Methodology and data IV. Results
I. Introduction
Having bank holidays has a negative impact on the economy. At least, that is the main message of a publication from the Center for Economics and Business Research from 2012 (CEBR,2012). Looking at the UK economy, they argue that every individual public holiday costs the UK economy ₤2.3 billion. In their calculation, they take into account that some businesses close and correct for the additional household spending that is accompanied with a public holiday. They conclude that businesses could lose momentum if they are confronted with too many public holidays.
One thing they don't take into account is the influence of public holidays on individuals, and especially on the way they feel. People that are happier might be able to work more efficiently. But what if the amount of leisure someone consumes also has an impact on their decision-making process, leading them to make more optimistic choices?
Conventional economic theory argues that individuals are rational in their decisions (Fama, 1970). However, especially in the last decade increasing evidence from other disciplines such as psychology is provided to show that people deviate from rational decisions for several reasons. Emotions play a role in the decision-making process of individuals. This evaluation can be extended to show that if a group of individuals has similar emotions, this 'public sentiment' influences the decision-making process of the group. This idea has been fundamental to the idea that mood can play a role in the determination of stock prices. Previously, weather conditions (Hirschleifer and Shumway, 2003), sports results (Edmans, Garcia and Norli, 2007), lunar phases (Yuan and Zhu, 2006) and public holidays (Frieder and Subrahmanyam, 2004; Bialkowski, Etebari and Wisniewski, 2012) have served as proxies for public mood, based on the relatively small direct impact on economic circumstances and the large possible impact on people's emotions.
This thesis extends on the previous literature by taking Dutch, German and Belgian public holidays and evaluating their impact on stock market indices. The criterium to be a public holiday in this research is that the national government acknowledges this day as an official public holiday. In previous research, the use of certain public holidays was fairly arbitrary. For example, Frieder and Subrahmanyam (2004) perform a research on St. Patrick's Day, Yom Kippur and Rosh Hashanah without any argument why they've chosen these three specifically. Taking all official public holidays in the three suggested countries takes away the criticism on the made selection. The main question throughout this thesis is: To what extend do public holidays influence stock prices in the Netherlands, Germany and Belgium? Since the cultural value, and thus the emotional value, differs across public holidays, the impact of all individual public holidays will be
evaluated as well as a combination of all public holidays.
The remainder of this thesis is structured as follows: section II provides an overview of the relevant existing literature, in section III the methodology and data needed to perform the analyses is given and section IV presents the results of the analyses. In the conclusion, section V, future paths of research will be provided.
II. Literature Review
One of the most influential articles concerning the efficiency of capital markets was written by Fama (1970). His review of the theoretical and empirical work in the field of efficient capital markets, has become one of the most cited articles in the field. Main question throughout the paper is whether prices in capital markets fully reflect the information people have about the financial products available. The three main factors for market inefficiency Fama recognizes are transaction costs, information asymmetry and disagreement amongst investors about the implication of certain information. They conclude that in the literature they studied, the evidence in favor of markets being efficient is much larger than the evidence that contradicts markets being efficient.
With the rise of behavioral economics (and finance), more literature has become available on the effect of behavior on financial decisions, such as stock market transactions. Various papers argue that the way people behave affects their choices, and henceforth the outcome of these decisions. Based on the papers discussed below, one can argue that mood plays an important role in the process of making financial decisions.
As said, the idea that investor psychology plays a role in financial decisions is not new. A survey by Hirshleifer (2001) summarizes the early literature pointing in this direction. He argues that “economists often argue that errors are independent across individuals, and therefore cancel out in equilibrium.
However, people share similar heuristics, those that worked well in our evolutionary past (Hirshleifer, 2001, p. 1540).” This implies that people's decisions are subject to the same biases, such as risk aversion and loss aversion. Hirshleifer (2001, pp. 1550-1551) also recognizes the role of mood and emotions on people's perceptions. Wright and Bower (1992) use hypnosis on their test subjects, combined with the recapping of happy or sad experiences. Once they induced a certain mood on their test subjects, they clearly examine people being more optimistic or pessimistic, depending on which experiences were shown. The optimistic people expect positive events with a higher probability than the pessimistic people.
Given that mood influences people's expectations, it follows that it also influences people's choices. Measuring this effect directly is not possible, since mood is a qualitative variable. Therefore, several
proxies, one more direct than the other, have been used to estimate the effect of mood on financial decisions. A relatively direct proxy is used by Bollen, Mao and Zeng (2011). They make use of two Twitter-related tools, OpinionFinder and GPOMS, to examine whether a relationship is present between public mood and the Dow Jones Industrial Average (DJIA). Firstly, they show that 'tweets' can be used to
determine the public mood at a certain moment, by showing that the public mood changes significantly in the period around the 2008 Presidential elections as well as around Thanksgiving in 2008. Secondly, using a Granger causality analysis, they observe a statistically significant correlation between only one of the dimensions of the GPOMS tool: Calm. However, all other mood variables don't seem to correlate with the DJIA. Using a non-linear 'Self-organizing Fuzzy Neural Network'(SOFNN) model, they argue that the results
of the Granger causality analysis are not providing a complete picture. The authors expect that data from Twitter can have a more significant effect if added to a more sophisticated model.
As said, the research design Bollen, Mao and Zeng (2011) use is relatively straight-forward. Alternative measures of public mood or sentiment have been proposed in other articles. As Bollen, Mao and Zeng argue: “The accuracy of these methods is however limited by the low degree to which the chosen indicators are expected to be correlated with public mood (Bollen, Mao and Zeng, 2011, p. 1)”.
An example of one such alternative method is using sports results as a proxy. Arkes, Herren and Isen (1988) observed a football match win by the Ohio State University. Afterwards, there was a significant increase in the sale of state lottery tickets after this particular win. Another paper using sports as a proxy for mood is written by Edmans, Garcia and Norli (2007). It investigates the impact of investor mood on financial markets. Using soccer success of countries as a variable, they are able to draw conclusions on the sign and potential size of the effect of sentiment. Data on 1162 soccer matches from 39 countries were collected, as well as data on other sports such as basketball, cricket, ice hockey and rugby. The authors find a strong negative relationship between stock market returns and losses in important soccer matches. For all other sports, the effect gets smaller but tends to stay significant. One important drawback could be that the effect of a win or loss is already priced into the stock indices, but the authors find no proof of this hypothesis. They observe no stronger win/loss effect if the win or loss is against the pre-expectations based on Elo-ratings. Three factors the authors also examine, are the potential effect of a soccer result on the economy directly (through direct sales for example), the potential difference of the effect for small and large stocks, and effect of soccer results on the trading volume. Since no abnormal results are found there, the authors conclude that a change in investor mood, driven by sports results, can affect stock markets.
Next to sports results, weather conditions are another proxy that is used in several articles. Hirshleifer and Shumway (2003) examine the effect of sunlight in 26 major cities with a stock exchange on the particular stock exchange. Correcting for seasonal effects other than the weather conditions, they find a significant positive effect of hours of sunlight on stock prices in fixed effect as well as pooled regressions. This outcome confirms an earlier analysis by Saunders (1993), who analyzed the effect of the weather on the Dow Jones Industrial Average and the New York Stock Exchange Index.
Apart from using weather conditions or sports events, there has been an increasing literature concerned with the use of cultural and religious events as a proxy for mood. For example, Frieder and Subrahmanyam (2004) look at Yom Kippur, Rosh Hashanah and St. Patrick’s Day to see whether these holidays have had an effect on returns and trading volume at U.S. equity markets over the course of 1946-2000. These holidays were chosen because of their relevance in the daily life of New York citizens. Yom Kippur can be qualified as earnest, whereas Rosh Hashanah and St. Patrick’s Day can be qualified as joyous. The authors show that Yom Kippur and Rosh Hashanah are associated with a significant decline in trading volume over the entire sample period. For the period 1973-2000 the effect on trading volume is more significant, and Saint Patrick’s Day also has a significant effect on equity markets. Statistically significant
effects are only found for the two days preceding Rosh Hashanah as well as the day itself, and for the two days preceding St. Patrick’s Day. For the days surrounding Yom Kippur and the other days surrounding Rosh Hashanah, the effect is insignificant but of the right sign in 9 out of 10 instances.
Frieder and Subrahmanyam (2004) argue that their research is not subject to any other calendar anomaly, which is the biggest thread for the validity of their validity. One such anomaly could be the ‘Monday effect’ (French, 1980), which is called this way because returns on Mondays tend to be lower than on other days of the week. One possible explanation for this is the fact that bad news about markets or companies is usually provided during weekends.
One particular research concerning the influence of mood, that tackles the effects of potential calendar anomalies is performed by Yuan and Zhu (2006). They investigate whether lunar phases influence stock returns and find significant effects for 48 countries. Periods surrounding a new moon are associated with a better return than periods surrounding a full moon in multiple analyses. The authors argue that this is evidence for the proposition that mood influences stock markets directly, since there is no reason to expect that lunar phases have an effect on the economy directly. However, they give a lot of evidence that lunar cycles have an influence on people’s mood.
As said, lunar phases make up an interesting test case, because they are not correlated with other calendar anomalies. The same holds for the Ramadan, the well-known Islamic tradition, which is used by Bialkowski, Etebari and Wisniewski (2012) as a proxy for sentiment. The authors find a significant positive effect on stocks during the Ramadan period, as long as at least 50% of a nation’s population is Islamic. The authors see no other explanation for this than a change in investor sentiment.
III. Methodology and data
This research is based on the analysis performed by Frieder and Subahmanyam (2004). Therefore, the following model will be tested:
R(t) = α + β * D(t) + ε(t),
where R(t) is the percentage return at day t and D(t) is a dummy variable, that equals 1 if day t is a public holiday, as categorized below. D(t) equals 0 if day t is not a public holiday.
Table 1 summarizes the official public holidays in the countries of interest, namely the Netherlands, Germany and Belgium. These can be found on the websites of the Dutch (Rijksoverheid, n.d.), German (Bundesministerium des Innern, n.d.) and Belgian (Portaal Belgium.be, n.d.) governments. These countries in particular share relatively similar cultural, historical and religious values, which is confirmed by the fact that they share some of the same public holidays. Public holidays that always take place during the
weekend (Easter Sunday, etc.) are not in the table, and are excluded from the analysis. Furthermore, public holidays that only take place in parts of a country are excluded (for Germany this holds for Corpus Christi, Assumption of Mary, Reformation Day, All Saints’ Day, Twelfth Night and Repentance Day). As table 1 shows, the Netherlands and Germany have 9 public holidays for the entire country and Belgium has 10. Furthermore, data is collected from Datastream for the three major stock market indices of interest: For the Netherlands, data on the AEX-index is used from 1983 until 2011, for Germany data on the DAX-index is used from 1965 until 2011 and for Belgium data on the BEL20-index is used from 1990 until 2011.
The biggest complicating factor in this analysis compared to that done by Frieder and
Subrahmanyam (2004) is the choice of public holidays: during Yom Kippur, Rosh Hashanah and St. Patrick’s Day the New York stock exchange doesn’t close down, whereas stock exchanges in Belgium, Germany and the Netherlands do on (some of) the suggested public holidays. One additional problem complicating the analysis performed in this thesis is the fact that sometimes public holidays fall on a Saturday or a Sunday, on which stock exchanges automatically close.
After the merge of several European stock exchanges into Euronext in 2000, most of the public holidays were standardized across countries. Nowadays, at the stock exchanges under consideration, trade is not permitted during New Year’s Day, Good Friday, Easter Monday, Labour Day, Christmas Day, and Boxing Day. This implies that trade took place after 2000 on all other public holidays. Therefore, the impact of these public holidays after 2000 can be estimated using the original model by Frieder and
Subrahmanyam (2004). Since this method gives the most direct approximation of the relationship, this research excludes data from before 2000 on these particular public holidays.
For the public holidays during which trade was prohibited, some changes in the model have to be made. In this case, one dummy variable will be created for the trading day preceding the public holiday and
The Netherlands Germany Belgium
New Year’s Day New Year’s Day New Year’s Day
Easter Monday Easter Monday Easter Monday
Ascension Thursday Ascension Thursday Ascension Thursday
Pentecost Monday Pentecost Monday Pentecost Monday
Christmas Day Christmas Day Christmas Day
Koninginnedag
Queensday
Bevrijdingsdag
Liberation Day
Good Friday Good Friday
Boxing Day Boxing Day Boxing Day
Tag der Deutschen Einheit
German Unity Day
Labour Day Labour Day
Nationale feestdag van België
National holiday of Belgium
Maria-Tenhemelopneming
Assumption of Mary
Allerheiligen
All Saints’ Day
Wapenstilstand
Armistice Day
Sinterklaas
Sinterklaas
Table 1: Official public holidays in The Netherlands, Germany and Belgium
one dummy variable will be created for the trading day succeeding the public holiday. Although this
method doesn’t allow for directly testing the impact of a public holiday, it does give the best possible proxy. Since the size of a 'public sentiment' effect is probably at its highest on the public holiday itself, this will effectively bias results downwards.
For the Netherlands, the analysis is performed using two official public holidays from table 1, Queensday and Liberation Day. Queensday was a trading-free day until 2002, and afterwards there were years when it fell on a Saturday or a Sunday. Therefore, there are only 7 instances in the dataset with data on that particular day. For Liberation Day, there were 20 occurences in the years evaluated. Therefore, another occurance is added, namely Sinterklaas, a Dutch yearly event on December 5, where children are given presents. Culturally, it can be seen as an equivalent of what Christmas is in other countries. In the dataset, Sinterklaas took place on a trading day 21 times. Besides these three cultural holidays that are specifically known in the Netherlands, the impact of New Year's Day, Christmas, Easter, Ascension Day and Pentecost Monday will be tested.
German Unity Day, the only specific German public holiday under consideration, is only celebrated since 1990, and it was a trading-free day until 2001. Therefore, the dataset only contains 8 instances where trade took place during German Unity Day. Again, the impact of New Year's Day, Christmas, Easter,
Ascension Day and Pentecost Monday is evaluated, and Labour Day is added to that list.
In Belgium, stock markets also closed during their National Holiday of Belgium until 2000, which holds for the Assumption of Mary, All Saints’ Day and Armistice Day as well. Afterwards, there were 8 respectively 9 instances of these days. In addition to that, the holidays that Belgium shares with Germany and the Netherlands are evaluated for Belgium as well.
The impact of New Year’s Day, Good Friday, Easter Monday, Labour Day, Christmas Day, and Boxing Day is evaluated in all three countries separately. Good Friday and Easter Monday are merged into one public holiday variable – Easter-, which holds for Christmas Day and Boxing Day – Christmas - as well. Given that the data for all three countries is collected over different time periods, making a good comparison across countries is not possible. Therefore, results are presented per country in the following procedure: firstly, each holiday is evaluated separately for each country. Secondly, all public holidays are pooled per country, to estimate the effect of public holidays in general, eliminating the individual cultural
characteristics of each holiday. This means that an extra dummy is created that equals 1 if the day is a public holiday in general. Two different methods are applied: one excluding New Year’s Day, Easter, Labour Day, and Christmas and one including them. In these regressions with pooled dummies, observations before 2000 are dropped for all public holidays, because there is no data on all individual public holidays before that year.
Pooling the public holidays is not done without any controversy. Since not every public holiday has the same cultural relevance, it is impossible to expect that the economic effects of each individual public holiday are similar. However, since public holidays are fairly well spread across the calendar year, the results of this analyses could give a good counterargument for stock effects that are the result of other calendar anomalies, e.g. the Monday effect (French, 1980).
In previous research, several authors chose to also include trading volume. In this research, the choice is made to exclude this factor from the analysis. The simple fact that trading volume alone is higher or lower at a given public holiday doesn't have any influence on the expectation of stock prices returns, which in the end is the variable of interest in this thesis. Besides, the effect of a public holiday on which stock exchanges close on trading volume is hard to measure, since closing stock exchanges on one day in itself has an effect on trading volume in the days before and after.
IV. Results
Table 2 summarizes the returns of the stock indices that are evaluated in this thesis. Making an inter-country comparison is not possible, because the observed sample periods are different. It can be seen that the mean returns on nearly all public holidays are positive, with the exception of Ascension Day in the Netherlands and Belgium, and Pentecost Monday and All Saints' Day in Belgium. The mean return of public holidays in the Netherlands is at least five times as high as the average return for the full sample for 8 of the 11 days. For Germany, this holds for 9 of the 11 days measured and for Belgium, this is the case for 10 of the 14 days. At a first glance, these numbers do not lead to a deviation of the idea that mood can influence stock prices.
The Netherlands Germany Belgium
Mean return (full period) .0003474 .0002787 .0001371
New Year’s Day
before .0014964 -.000609 .0019076 after .0106555 .0057154 .0085796 Christmas before .0022704 .0017279 .0026493 after .0044807 .0041259 .0030348 Easter before .0047869 .0057452 .0020056 after .0018407 .0027132 .0021571 Labour Day before .0015412 .0009316 after .0022054 .0047949 Ascension Thursday -.0039027 .0003326 -.0015001 Pentecost Monday .0008492 .0018138 -.001089 Queensday .0022629 Liberation Day .0020924
German Unity Day .0055211
National holiday of Belgium .0057549
Assumption of Mary .0074643
All Saints’ Day -.0028042
Armistice Day .0002091
Sinterklaas .0025711
For the sake of clarity, the results of the significance tests are provided per country for each public holiday individually and for all public holidays pooled together in the two ways presented earlier: one 'public holiday-proxy' excluding and one including New Year's Day, Christmas, Easter and Labour Day.
Results of these tests for the Netherlands can be found in table 3. According to the table, 4 of the 5 public holidays that could be measured at the day itself, were associated with higher returns. The effects of the days surrounding these public holidays are also given. Most of the effects are statistically insignificant, which could be the result of a relatively low amount of occurrences for most public holidays. All significant effects were found in the days surrounding New Year's Day, Christmas and Easter, which seem to have a positive effect. None of the other public holidays are associated with significantly positive or negative effects. This could be the result of a relatively low number of occurrences in the dataset. Combining all dummies including Christmas, New Year's Day and Easter gives one significant result on the days after the public holiday. One might argue that this is due to the highly significant effect of the day after New Year's Day, but if these days are excluded from the analysis the positive effect is still present (t-statistic:=1.90).
Public holiday β (t-statistic)
t-1 t t+1
New Year's Day (NYD) .0014673 (.57) .0103457 (4.03***) Christmas .0019298 (.76) .0042458 (1.68*) Easter .0044559 (1.76*) .0014983 (0.59) Sinterklaas -.0001567 (–.05) .0022291 (.75) .0029402 (.99) Queensday -.0027484 (–.54) .0019165 (.37) .0010037(.20) Liberation Day -.0001997 (–.07) .0021581 (.71) .0019539(.64) Ascension Day .0071661(1.58) -.0026815 (-.59) -.0009474 (–.21) Pentecost Monday -.0016181 (–.38) .0005017 (.12) -.0056597(-1.32) Pooled (only after 2000)
Excluding NYD, Christmas and
Easter .0002993(.12) .0015733 (.65) .0008612(.35) Including NYD, Christmas and
Easter .000927(.51) .0036398 (2.02**)
Table 3: Regression results of public holidays in The Netherlands
From the German case in table 4, it is again clear that the strongest effects can be found on the days surrounding New Year's Day, Christmas and Easter, as was the case for the the Netherlands as well. Another interesting finding is present for the day after German Unity Day: it is associated with highly significant negative returns. There is no clear explanation for this finding. One event that could explain this
is that October 4 is the holy day of St. Francis . However, the relevance of this day for Germany can be questioned. The day before Ascension Day is also characterized by significantly higher returns. Again, a good explanation
Public holiday β (t-statistic)
t-1 t t+1
New Year's Day (NYD) -.0001768 (–.10) .0053449 (2.96**) Christmas .0014548 (.82) .0038446 (2.16**) Easter .0054875 (3.08**) .0024438 (1.37) Labour Day .0012678 (.71) .0019347 (1.08) German Unity Day -.0030316(-.70) .0052458 (1.21) -.0122627 (-2.84**) Ascension Day .0065084(1.77*) .0000545 (.01) .0022719(.62) Pentecost Monday -.0000744(-0.02) .0015371 (.44) -.0009943(-.28) Pooled (only after 2000)
Excluding NYD, Christmas,
Easter and Labour Day -.0010642(-.29) .002186(.74) -.0025677(-.87) Including NYD, Christmas,
Easter and Labour Day .0024353(1.21) .0030295 (1.63)
Table 4: Regression results of public holidays in Germany
of this phenomenon is absent.
Regression results for Belgium can be found in table 5. The sign of the effects is positive for 18 of the 26 days measured, but most of the effects are insignificant. Again, the trading day following New Year's Day is associated with significantly higher returns. This holds for the day after Labour Day as well. The days preceding the National Holiday of Belgium and Ascension Day are also associated with higher returns, which holds for the Assumption of Mary itself as well. The day following Armistice Day is associated with significantly lower returns. No clear explanation can be given based on other culturally relevant calendar events for these effects.
The relative absence of significant effects could be the result of a low amount of occurences of some of the public holidays observed. This would be an explanation for the fact that significant effects are found for New Year's Day in all three countries. One possible other explanation for this fact could be that New Year's Day influences people's mood more than the other public holidays in the sample. Obviously, the research design was imperfect because stock markets close during New Year's Day, Christmas and Easter. Looking at the results, the simple fact that stock exchanges close during these days could illustrate the importance of these days on people's sentiment.
anomalies, since they make use of public holidays that are fairly spread over the year. Of the 15 results of these pooled dummies in all 3 countries, only 3 results were of the negative sign, and 12 were of the positive sign.
Public holiday β (t-statistic)
t-1 t t+1
New Year's Day .0017635 (.69) .0083451 (3.27**) Christmas .0025218 (1.01) .0029089 (1.17) Easter .0018757 (.75) .0020278 (.81) Labour Day .0008004 (.32) .0046786 (1.88*) National Holiday of Belgium .0077567 (1.88*) .0056256 (1.36) .0023267(.56) Assumption of Mary .0049362(1.27) .0073387 (1.88*) -.0055292(-1.42) All Saints .0030729 (.79) -.002946 (–.76) .0038221(.98) Armistice Day -.0013041(-.32) .0000721 (.02) -.0079256(-1.92*) Ascension Day .0073037 (1.98**) -.00164 (-0.44) .0016027(.43) Pentecost Monday -.0036026 (–.98) -.0012283 (–.33) -.0061496(-1.66) Pooled (only after 2000)
Excluding NYD, Christmas,
Easter and Labour Day .0032232(1.73*) .0012614 (.68) -.0017877 (–.96) Including NYD, Christmas,
Easter and Labour Day .0028721(1.70*) .0017972(1.32)
Table 5: Regression results of public holidays in Belgium
One of the main drawbacks of this study is the choice of public holidays. In the end, the variable under consideration is public sentiment. The correlation between public holidays and stock prices is only as strong as the correlation between public holidays and sentiment combined with the correlation between sentiment and stock prices. The correlation between public holiday and sentiment should be maximized, which can be done by taking the public holidays with the highest cultural value. The chosen public holidays in this thesis are characterized by the fact that governments call them 'official' public holidays and although this obviously has some cultural value, it doesn't provide a full picture of the cultural relevance. In defense of this research, working with proxies for variables always implies that one can, at best, only attempt to find the best possible proxy.
In the research, the results are not corrected for the fact that a positive or negative trend could be present for stock returns before and after a trading-free day. Obviously, this problem only holds for the following public holidays: New Year's Day, Easter, Labour Day and Christmas. Due to limitations in the
available data it was not possible to statistically correct for this problem.
Another important drawback is the fact that public holidays, especially the ones that are culturally the most important, can add value to stock markets directly. Looking at for example Sinterklaas in the Netherlands, when people buy a lot of gifts for their children this does have a huge effect on household spending. If this process would lead investors to think more positive on the future, this would eliminate some of the effect now attributed to sentiment. The most important counterargument here is that people could have priced the effect in ex ante. In previous literature, some of the researches on mood effects were done taking for example the weather (Hirshleifer and Shumway, 2003) or sports games (Edmans, Garcia and Norli, 2007). The expected outcome of weather conditions or sports games is known relatively late compared to the expected outcome of whether a day is a public holiday or not, which is known years in advance. This means that if direct effects are present, these should be more influential in research using the weather and sports games than in research using public holidays.
Statistically, the results in this thesis are subject to the multiple comparisons problem. One should take into account that given the total number of regressions performed the chance of retrieving statistically significant results automatically rises. Therefore, the results reported in tables 3, 4 and 5 should be
V. Conclusion
This research was done to evaluate the impact of public holidays on stock markets. Main aim was to see whether public holidays have a significant effect on stock markets, since they can be of influence for public sentiment. Using dummies to assess the impact of all public holidays individually as well as combined in the Netherlands, Germany and Belgium, the main findings were not unambiguous. For some of the public holidays, significant effects were present although not always of the same sign. Main problem in the analysis was the lack of data over the full time period on most of the individual public holidays, because stock exchanges usually close down on these days. Because of the merge of some European stock
exchanges in 2000 into Euronext, the same closing days applied for the Dutch, Belgian and German markets afterwards. Although this opened up the possibility to measure the effect more directly, it also implies that data is only available over the years 2000 until 2011. The effects surrounding NYD, Christmas, Easter and Labour Day are measured over longer time periods, so more data-points are available. This might be a reason that the effects measured for these days are more often statistically significant.
Looking at the most relevant paper used in this thesis by Frieder and Subrahmanyam (2004), they face similar problems. For 2 of the 3 public holidays they studied, they find strong anticipatory effects, whereas these are absent for the other one. This illustrates the main difficulty in this thesis, but it holds for all papers doing the same kind of analysis: an analysis of the relationship between sentiment and stock prices is only as good as the proxy that is used. Therefore, it is important to keep looking for new potential proxies to capture this effect.
Concluding, this research could find no unambiguous effect of public holidays on stock prices due to the reasons mentioned above. Future research should focus on tackling these reasons: better proxies might result in a better isolation of the sentiment effect. One suggestion might be to look at option markets instead of stock markets. Analyzing options might give a better view of how people price in their
expectations. This way, it might be possible to more directly measure the influence of public holidays on financial decision-making.
VI. References
Arkes, H. R., Herren, L. T., & Isen, A. M. (1988). The role of potential loss in the influence of affect on risk-taking behavior. Organizational behavior and human decision processes, 42(2), 181-193.
Białkowski, J., Etebari, A., & Wisniewski, T. P. (2012). Fast profits: Investor sentiment and stock returns during Ramadan. Journal of Banking & Finance,36(3), 835-845.
Bollen, J., Mao, H., & Zeng, X. (2011). Twitter mood predicts the stock market. Journal of Computational
Science, 2(1), 1-8.
Bundesministerium des Innern (n.d.). Nationale Feiertage. Retrieved June 22, 2014, from
http://www.bmi.bund.de/DE/Themen/Gesellschaft-Verfassung/Staatssymbole/Nationale-Feiertage/nationale-feiertage_node.html
CBER (2012, April 10). Bank holidays cost. Retrieved July 1, 2014, from http://www.cebr.com/reports/bank-holidays-cost/.
Datastream International. AEX INDEX (AEX) – PRICE INDEX. Retrieved from Datastream database.
Datastream International. BEL20 – PRICE INDEX. Retrieved from Datastream database.
Datastream International. DAX 30 PERFORMANCE – PRICE INDEX. Retrieved from Datastream database.
Edmans, A., Garcia, D., & Norli, Ø. (2007). Sports sentiment and stock returns.The Journal of Finance, 62(4), 1967-1998.
French, K. R. (1980). Stock returns and the weekend effect. Journal of financial economics, 8(1), 55-69.
Frieder, L., & Subrahmanyam, A. (2004). Nonsecular regularities in returns and volume. Financial Analysts
Journal, 29-34.
Hirshleifer, D. (2001). Investor psychology and asset pricing. The Journal of Finance, 56(4), 1533-1597.
Hirshleifer, D., & Shumway, T. (2003). Good day sunshine: Stock returns and the weather. The Journal of
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work*. The journal of
Finance, 25(2), 383-417.
Portaal Belgium.be (n.d.) Nationale feestdagen in België. Retrieved June 22, 2014, from http://www.belgium.be/nl/over_belgie/land/belgie_in_een_notendop/feestdagen/
Rijksoverheid (n.d.) Welke feestdagen kent Nederland en wanneer worden ze gevierd? Retrieved June 22, 2014, from http://www.rijksoverheid.nl/onderwerpen/schoolvakanties/vraag-en-antwoord/welke-feestdagen-kent-nederland-en-wanneer-worden-ze-gevierd.html.
Saunders, E. M. (1993). Stock prices and Wall Street weather. The American Economic Review, 1337-1345.
Yuan, K., Zheng, L., & Zhu, Q. (2006). Are investors moonstruck? Lunar phases and stock returns. Journal of
Empirical Finance, 13(1), 1-23.
Wright, W. F., & Bower, G. H. (1992). Mood effects on subjective probability assessment. Organizational
