The January effect: an investigation of the
European stock markets
Abstract
This thesis examines the existence of the January effect and the development of its magnitude on the
stock markets in Norway, Germany, France, Greece and the Netherlands. This will be researched by
testing two different models. The examined period is 1985 to 2014, which contains the recent financial
crisis. The influence of the crisis is taken into account as well. The results show that only in Norway
and France, the January effect exists in the small firm stocks during the crisis and after the crisis. This
is consistent with prior research of Roll (1981).
Bas Blom
5876168
June 30, 2014
Bachelor thesis
Program: Finance and Organization
Thesis specialization: Finance
12 ECTS
2
Introduction
Last month, researcher Dirk Gerritsen was promoted on his research to the relevance of
security analysis. He claims that technical analysis, a method in which analysts try predict the
future stock price behavior based on historical prices, is useless to investors because historical
trends cannot predict the future at all (Cohen, 2014). This is consistent with the so called
random walk model in which investors believe that future stock prices are unpredictable.
Based on this theory, Eugene Fama (1970) published his research in which he introduces the
Efficient Market Hypothesis (EHM). The EHM states that all available information should be
reflected by the stock price and that seasonal patterns should not be present on stock markets.
However, there is evidence that there are stock price patterns during the year. The most
researched pattern is the January effect. This refers to a phenomenon that stock returns in
January are significantly higher compared to every other month (Gultekin & Gultekin, 1983).
The existence of the January effect is very interesting for investors, because it helps them to
determine when and in which stock they should invest.
In this paper, the stock markets in Germany, the Netherlands, Greece, France and
Norway are investigated for existence of the January effect over the period from 1984 to
2014. On top of that, the magnitude over time of this effect and the influence of the most
recent financial crisis are taken into account. The main question in this thesis is: Does the
January effect show up in Europe, and will its magnitude decrease over time?
The results of this research are moderately consistent with prior research of Roll
(1981). In Norway, the January effect exists in the complete period for the smallest 30% of
the firms and during the crisis in the for the 10 to 20% smallest firms. The January effect is
supported in France in the complete period for the smallest firms only, and after the crisis for
most of the stocks of the 50% smallest firms. The magnitude of the January effect decreases.
3
Since 1980, the behaviour of stock market returns and especially stock market
seasonality is a popular topic to research. Mostly all of the previous literature focuses on the
U.S. stock markets. The respected research on the European stock markets is scarce and
outdated. Therefore, this thesis contributes to the existing literature in a way that the January
effect is investigated for Europe over the last 20 years. Moreover, this period contains the
most recent financial crisis which has not been researched in respected literature yet.
This thesis is structured as follows. In the next section, an overview and a discussion
of the relevant literature is given. Section 3 describes the data and methodology used to
answer the main question. Thereafter, in section 4, the results are be presented and analyzed.
Lastly, in section 5, the findings are summarized and a conclusion is drawn.
2
Literature review
During the past decades, a lot of research has been done to unveil seasonal anomalies in stock
markets. However, among the researchers, there is no consensus about whether the seasonal
anomalies exists and especially the January effect. The first sub-section keynotes the Efficient
Market Hypothesis. Then, the most important investigations to the existence of the January
anomaly will be discussed in chronological order in section 2.2. Section 2.3 will provide an
overview of the possible causes for the January effect. Section 2.4 closes with a brief
summary and conclusion.
2.1
Efficient Market Hypothesis
Fama (1970) states that in ideal markets, the stock price gives precise signals about the
companies' future performance on which an investor can choose his allocation of resources.
Fama defines the situation under which stock prices fully reflect all information at any point
4
new, unpredictable information. If information could be predicted, it would be part of current
available information and therefore already priced in (Bodie et al 2011). Thus, according to
this reasoning, if one has information about the future performance or state of the economy he
is not able to generate abnormal returns. However, Fama makes a distinction in what is meant
by 'all available information'. This distinction results in three forms of the EHM: the weak
form, the semi-strong form and the strong form.
Following Bodie et al. (2011), the weak form hypothesis refers to the situation in
which stock prices reflect all information which can be extracted from historical market data
such as past prices. This implies that fundamental analysis is a fruitless undertaking because
all this data is available to anyone at almost very low or no costs. Even if one could derive a
trend from historical information, this signal would be taken over by all investors resulting in
an immediate price change (Bodie et al., 2011).
But what about seasonal anomalies in stock markets? The EHM states that markets are
efficient if prices are unpredictable. This implies that, if prices are predictable which is the
case with seasonal anomalies, markets are inefficient in the weak form hypothesis. Thus, the
January effect is evidence for market inefficiency.
2.2
Evidence of January Anomaly
One of the first researchers that provided evidence for seasonal stock returns was Wachtel.
Wachtel's research (1942), an investigation by Bullock et al.(1919) and later research by
Owens and Hardy (1925) to seasonal anomalies in stock returns, strongly concluded that there
was no abnormality in stock returns before 1925. Therefore, Wachtel examined the period
from 1927 to 1942 in the Dow Jones Industrial Average. He observed the stock price
movements of twenty stocks in December and January for 15 years. He found that in 11 of
5
significant. The 4 years with negative returns in December and January were insignificant. In
most of the positive months, the returns varied between 5-10% whereas the negative returns
were not more than 4%. Wachtel warily concluded that his results imply a seasonal anomaly
in December and January. However, his research lacks statistical tests. Therefore, this
conclusion has to be adopted with prudence.
In 1953, Kendall published a researched to the behavior of stock prices. He argued that
stock prices seem to behave randomly. Therefore, it is impossible to predict future stock
prices unless one has extraneous information. This is stock behavior can be called the random
walk model and is consistent with the EHM (Fama, 1995).
More than 20 year later, Rozeff and Kinney (1976) studied the seasonal anomalies as
well. Although there was prior research which provided some evidence for seasonalities, this
evidence was not overwhelming and not supported by testing hypotheses. Rozeff and Kinney
investigated the stock returns of the New York Stock Exchange (NYSE) in 1904 to 1974. This
time span contains the Great depression and therefore, to avoid all the tests being influenced
by this period, the complete time span was divided into four sub-periods, namely 1904-1928,
1929-1940, 1941-1974 and the entire period 1904-1971 for completeness. By testing the same
model as is used in this research, Rozeff and Kinney found that there are statistical
significantly large returns in January compared to the other months during the period studied
except for 1929-1940. Second, Rozeff and Kinney state that the simple random walk model
does not hold, but this does not necessarily mean that the markets are inefficient if the
available information is priced in. This is based on their findings that the high returns in
1941-1974 come at a higher risk. Therefore, investors are not able to make abnormal returns
without increasing their risk. Haugen and Jorion(1996) state that existence of the January
effect does not mean that market are inefficient, because transaction costs are too high for
6
risk premium earned by investors is significantly higher in January than in other months, due
to higher dispersion in these months.
Whereas most of the research is focused on stock markets in the U.S., Gultekin and
Gultekin (1983) provided international evidence of the January anomaly, as one of the first, in
most of the researched countries. They used the data from the Capital International
Perspective (CIP) in the period from January 1959 to December 1979. The CIP is a monthly
and quarterly magazine containing 1100 share prices which have a listing in 18 developed
countries around the world. The stock market returns are calculated from the monthly closing
prices of the value-weighted indices as percent changes. Dividend yields are excluded from
the calculations, as they state that dividend yields do not influence the seasonality in the stock
returns. Gultekin and Gultekin found that in 12 of the 17 countries the null hypothesis that
stock returns are not related with time can be rejected with a significance level of 10%. A
second noteworthy finding is that, when using CIP data, there is no evidence for seasonality
in the U.S. stock market. This differs from the research of Rozeff and Kinney (1976).
According to Gultekin and Gultekin (1983), the different results are due to the composition of
the indices used. CIP indices are value-weighted whereas the NYSE equally weighted.
Therefore, in the NYSE the small firms are given a bigger weight. And as Keim (1983)
argued, small firm returns are an explanation of seasonality in stock returns. Concluding,
there is a seasonality around the turn of the year, but Gultekin and Gultekin (1983) are not
drawing strict conclusions about this relation.
2.3
Explanations for the January effect
Schultz (1985) examined whether there is a January effect in the returns of small firms in the
period from 1900 to 1929 and 1963 to 1980, and if the tax-loss selling hypothesis could
7
sell their stocks which declined in value during the year to realize losses and therefore reduce
tax to be paid at the year-end. In January investors buy back these shares resulting in upward
price pressure. Schultz (1985) compared the difference between small firm returns and the
Dow Jones Industrials Index. The particular periods of interest are the periods from 1900
through 1917 against 1918 trough 1929 and 1963 through 1980. The difference between this
periods is indicated by the War Revenue Act. This law passed the U.S. Congress in 1917 and
greatly increased the federal income tax to finance the First World War (Www.history.com).
Schultz found that in the years prior to 1918, the differences between small firm returns and
the Dow Jones return were negative for 10 of the 18 Januarys. According to Schultz, this
implies there is no January effect in this period. For the two periods after 1917, the
differences between small firm returns and the Dow Jones return were positive for all
Januaries. Therefore, Schultz argued that in these periods the January anomaly exists, and this
is consistent with Keim (1983). The tax-loss selling explanation could not be rejected.
However, Schultz stated that it's troubling to accept tax-loss selling as the explanation for the
January effect and leaves the question why small stocks in January are preferred over
December to investors. A suggestion to this question is provided next, by research of Roll
(1981)
In addition to his earlier research (Roll, 1981), in which difference in risk seemed to be the
explanation for the difference in small firm returns and large firm returns, Roll (1983)
provided another factor responsible for this difference. This more recent investigation
supports tax-loss selling being responsible for the abnormal returns in January. Roll argues
that the stock returns of small firms are more likely to incur losses in the preceding year than
the returns of large firms. The reason is that the small firm returns are more volatile because,
8
firm stocks are more likely to be sold at the yearend for tax purposes. Roll (1983) states that
stocks which incurred losses in the previous year show higher returns in January. He
interprets this as supporting the tax-loss selling hypothesis which is consistent with Schultz
(1985).
A more recent research of Bentzen (2009) investigated the January effect in the period from
1964 to 2008. He used data of stock traded at the NYSE, AMEX and national Association of
Securities Dealers Automated Quotations (NASDAQ). Monthly returns are calculated with
taking into account the stock splits and dividends and ten portfolios are created based on the
market capitalization. The portfolios are ranked from 1 to 10 with 1 representing the smallest
10% and 10 the largest 10% of the market capitalization. Bentzen splits up the period into two
sub-periods to investigate the influence of the signing of the Tax reform act in 1986. After the
tax reform act was signed, capital gains were taxed on the same level as ordinary income
(Investopedia, 2014). The pre-Tax Reform Act period is from 1964-1986 and the after-Tax
Reform Act is from 1987-2008. Regression are done for both the sub-periods and the
complete period.
For the latter mentioned period, Bentzen found that the January returns are significantly
different for all the portfolios, November returns are significant for portfolio 3-10 and
December shows significant returns for portfolio 2-10. This is not consistent with the earlier
findings of Roll (1981) that the difference in firm size is the explanation for the difference in
returns.
The first sub-period shows significant and positive January returns for portfolio 1-8.
However, none of the portfolios show significant returns for December. For the second
sub-period, Bentzen found that January is significantly for portfolio 1-3 and 9. The returns in
9
this implies that firm size is not necessarily an explanation for the January effect. Bentzen
draws the conclusion that the January effect has shifted to December after the Tax Reform
Act was signed in 1986.
Jones et al. (1987) examined the tax-loss selling hypothesis as well. Two periods are being
researched in the Dow Jones Index and the Cowles Commission. First, the period from
February 1871 through December 1938 is examined. This period is much earlier and longer
than the period of Schultz. Second, the same period as Schultz is used. Both periods are split
up into a pretax and a post tax period. The pretax period ends in December 1917 and thus the
post-tax period starts in January 1918. Contrary to prior research, Jones et al. (1987) used a
parametric model to account for statistical problems like autocorrelation. They found that the
January effect existed before the income-tax was introduced in 1917. Therefore, their results
contradict with the tax-loss selling hypothesis, whereas they are consistent with other studies
which provide evidence for the relation between the January effect and firm size.
The rejection of the tax-selling hypothesis is also supported by Constantinides (1984). He
claims that it predicts the January anomaly only when investors are irrational.
Like the previous described research of Jones, Pearce and Wilson, the investigation of Chan
(1986) on the period from 1962 to 1982 does not support the tax-loss selling hypothesis as an
explanation of the January effect. Although his result shows a relation between the January
effect and the decline of stock prices, it does not necessarily mean that losses in stock price
are caused by tax-loss selling. However, Chan stated that the relation between the January
10
Keim (1983) examined the relation between the returns and the size of stock, which is defined
as the total value of common equity, of the NYSE and the AMEX in the period from
1962-1979. He found a negative relation between the returns and firm size and he argued that nearly
half of the extent of the premium of small firms over large firms is due to abnormal returns in
the January months. Furthermore, Keim found that more than half of the January premium is
due to the first 5 days of the year.
Lakoniskhok and Smidt (1984) investigated the relation between firm size and
turn-of-the-year behavior as well. The data used contained the daily returns on the companies of the
NYSE and AMEX from 1970 through 1981. The daily returns were adjusted for stock splits,
stock dividends and cash dividends. Whereas most of previous research created portfolios of
companies based on market capitalization, Lakoniskhok and Smidt used on the value of
trading during October and November prior to December and January. They found the
existence of a seasonal pattern in the returns of small companies. This is consistent with the
results of Keim (1983) and Roll (1983).
Ogden( 1990) examined the turn-of-the-month liquidity hypothesis in the period from 1969 to
1986. This hypothesis claims that the standardization of payments at the end of the month
causes stock prices to rise around the turn of the month. Ogden compared the means of the
turn-of-the-month trading days and remaining trading days. He found that the returns in the
days around the turn of the month are significantly higher than the remaining days of the
month. He argued that the liquidity of investors is the highest at the end of each month and
especially at the end of December due to profits from privately owned businesses and
year-end bonuses. Therefore, investors are willing to invest their money which results in higher
11
2.4
Summary and conclusion
Seasonalities in stock returns is a frequently researched topic in the past decades. Much of
these investigations are focused on the U.S. stock markets, although some more recent were
concerned about this phenomenon outside the U.S. And indeed, there is evidence for
seasonality is stock returns, especially for the January effect. However, this proof is outdated.
Explanations for the existence of the January effect are tax-loss selling, firm size and personal
liquidity.
3
Data and Methodology
This section describes the data and method used in this research. In this research, the data will
be tested for existence of the January effect for the most recent period. Therefore, the returns
for the period from 1985 to 2014 are taken into account. The particular European countries of
interest are The Netherlands, Germany, United Kingdom, France, Spain, Greece and Norway.
These countries and their indices differ among them, for example, in market capitalization,
number of companies, geography. In this way, a fair reflection of the European stock markets
is attained.
3.1
Method
The model that will be used to estimate the monthly returns for each portfolio, is the same as
Keim (1983) Jones et al. (1987) and Bentzen (2009) used in their research. The following
regression, which is called model 1 throughout the remainder of this research, is estimated:
12
Where
-
= stock return in month t
-
are dummy variables with
if month is February...December, 0 otherwise
-
represents the month January and is the return of the first month, whereas
through
measure the difference between January and the other 11 months. The dummy variable JAN
is not in the model to prevent for perfect multicollinearity
- ε is the error term
Because the estimated coefficients
reflect the difference between January and the other 11
months, the following hypothesis is constructed:
This hypothesis is in favor of the January effect because it states that the difference between
January and all the other 11 months are negative, meaning that the returns of the other 11
months are smaller than January returns. The January effect will be present if all β
i(i>1) are
significantly negative.
To investigate whether there is a trend in the existence of the January effect, the following
model, which is referred to as model 2, will be tested for each country:
(2)
Where
-
stock return in month t
-
a constant
13
- is a dummy variable with if year is 1985, if year is 1986...
.... if year is 2014
- is an interaction term
- is the error term
Previous literature like Haugen and Jorion (1996) states that the magnitude of the January
effect decreased over time, but not significantly. To test this, the following hypothesis is
constructed:
This hypothesis supports the findings in previous literature that the magnitude of the January
effect decreases over time.
3.2
Data
From the Compustat Global Securities Daily database, like Keim (1983), the daily closing
prices for all the listed companies per country are obtained as well as the shares outstanding, a
dividend adjustment factor, a stock split adjustment factor and a total return adjustment factor
for the stock prices. The advantage of using daily prices per company over monthly closing
prices per index is that more observations are used which increases the accuracy of the
estimation. Moreover, the specific indices are not investigated on their own. They will be split
up into 10 equal size portfolios to take into account the firm size effect which is described in
section 2.3.
The data in Compustat are the daily prices To get the daily returns adjusted for stock
splits and dividends the following formula is used:
(Wrds, 2014)
14
Where:
-
is the return on time t
-
is the stocks' daily closing price on time t
-
is an adjustment factor for dividends on time t
-
is an adjustment factor including cash distributions and reinvestment of dividends
Like the daily returns, the size of each company has to be computed as well. Size is defined as
the numbers of shares outstanding times the corresponding share price. This is referred to as
the market capitalization.
10 portfolios are created. Each portfolio refers to a decile of the market capitalization in
which portfolio 1 represents the smallest 10% of the firms and portfolio 10 represents the
largest 10%. This is done because in the previously discussed literature, Keim (1983) showed
that the January effect is chiefly a size effect. To investigate the influence of the last financial
crisis, the sample period (1985-2014) will be divided into three sub-periods. Namely period 1
(1985-2006), period 2 (2007-2010) and period 3 (2011-2014).
4
Empirical results and analyses
This section will describe and analyse the empirical results. It is structured as follows. In
section 4.1 some key figures are given for the data, as well as a comparison of the mean
returns. In section 4.2, for model 1, the results for the complete period (1985-2014) will be
shown first. Thereafter, the results for the three sub-periods will be shown and discussed in
section 4.3. In the fourth section, the results of model 2 will be analyzed. Note that the tables
of model 1 for Norway can be found in the next section, because the results for Norway differ
the most from the other countries. The tables of the other countries are in the appendix. At
15
4.1
Key figures
Table 1 shows the descriptive key figures for all the countries for the complete sample, the
first size quartile and the third size quartile. It draws attention that the January mean return in
the first quartile is larger than the mean return in the residual months for each country.
However, in the third quartile the mean return in January is lower than in the residual months
in all countries except Norway. This seems to be consistent with findings of Keim (1983) that
the January effect is particularly present in small firm stocks.
Table 1 – Key figures for all Countries
Note: Returns in %. Market capital in millions of dollars
In table 2, the mean return of January and the residual month are tested with a one sample
t-test. The null hypothesis states that the difference between January and the other months is
zero which implies that January doesn't have significantly higher returns. The alternative
hypothesis states that the difference is larger than zero. This implies that the mean return in
January is higher than in the other 11 months. Looking at the p-values in table 2, one can see
that the difference for Norway and Germany is significant.
Complete Sample Size Quartile 1 Size Quartile 3
Avg Market Cap. Mean return Jan Mean return other months Avg Market Cap. Mean return Jan Mean return other months Avg Market cap. Mean return Jan Mean return other months Norway 5960 .3014141 .051734 126 .6046779 .0487273 1710 .2106341 .0516759 Germany 19900 .2066223 .1641343 5.47 .6498167 .1677049 218 .0592563 .1583779 France 5750 .0733222 .0667646 14.8 .1924636 .0601596 625 .0599507 .0622595 Greece 24500 -.0808671 .0214293 7.89 .0527912 .0184682 178 -.3589185 .0243069 The Netherlands 5380 .0502305 .0651181 29.5 .1338728 .0631824 775 .0437753 .062595
16
Table 2 – Mean comparison for all countriesMean Jan Mean Other P-value t-test
Norway .3014141 .0539136 0.0000***
Germany .2066223 .1641343 0.0013***
France .0733222 .0667646 0.2262
Greece -.0829216 .0201121 1.0000
The Netherlands .0502305 .0651181 0.8430
Note: *** Significant at 1%. Returns in %
4.2
Complete period model 1
In table 3, the Stata output for regression model 1 for the complete period of Norway is
shown. Mostly all of the Januaries show significant positive returns. Only in the largest
portfolio, the January return is not significant. There is a decreasing value of the January
returns from portfolio 1 to 10, with the exception of portfolio 5 and 7. This is consistent with
findings in Reinganum (1983). In portfolios 1-3, all other months show significant lower
returns compared to January at a 1% significance level. This strongly implies the existence of
the January anomaly in the three smallest portfolio’s. In portfolio 4,5 and 7 the returns of all
the other months are negative as well, although not all of them are significant. Therefore, with
great reticence, these portfolios show signs of the January effect. The rest of the portfolios
show positive returns in some of the other months. This implies that there is no January effect
17
Table 3 - Model 1Norway, 10 size portfolios, 1985-2014Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
Table 4 in the appendix shows the complete period estimates for Germany. All the January
returns are significant at a 10% significance level and positive, except for portfolio 10. The
January returns show a decreasing value. Contrary to the results of Norway, there is no clear
evidence of the January effect in one or more of the portfolio. Only portfolio 2 shows a
positive return in January and lower returns in the other 11 months. However, only the returns
in November and December are significant at a 10% level. This is a very weak evidence of
existence of the January effect in the second size decile.
1 2 3 4 5 6 7 8 9 10 Jan 0,7848*** (0,1191) 0,5325*** (0,0797) 0,3803*** (0,0586) 0,2713*** (0,0508) 0,2854*** (0,0517) 0,2116*** (0,0434) 0,2229*** (0,0342) 0,1676*** (0,0341) 0,1482*** (0,0323) 0,0444 (0,0306) Feb -0,9624*** (0,1744) -0,5069*** (0,1149) -0,2453*** (0,0845) -0,1741** (0,0724) -0,1086 (0,0744) -0,0880 (0,0623) -0,1424*** (0,0496) 0,0207 (0,0486) -0,0241 (0,0467) 0,1168 (0,0437) Mar -0,7517*** (0,1705) -0,3767*** (0,1128) -0,2735*** (0,0837) -0,1631** (0,0711) -0,1849** (0,0728) -0,1201* (0,0618) -0,0949* (0,0487) -0,0484 (0,0481) -0,0113 (0,0456) 0,1194 (0,0428) Apr -0,8403*** (0,1743) -0,4957*** (0,1164) -0,2658*** (0,0866) -0,0381 (0,0723) -0,0954 (0,0752) -0,0044 (0,0636) -0,0374 (0,0501) 0,0415 (0,0494) 0,0373 (0,0467) 0,1042 (0,0441) May -0,7836*** (0,1685) -0,4437*** (0,1142) -0,2327*** (0,0861) -0,2027*** (0,0718) -0,1437* (0,0746) -0,1315** (0,0630) -0,1343*** (0,0502) -0,0393 (0,0492) -0,0686 (0,0467) 0,0494 (0,0444) Jun -0,9388*** (0,1697) -0,4900*** (0,1142) -0,2955*** (0,0861) -0,3353*** (0,0714) -0,2061*** (0,0738) -0,2070*** (0,0617) -0,2778*** (0,0500) -0,1935*** (0,0485) -0,1109** (0,0460) -0,0389 (0,0444) Jul -0,6005*** (0,1694) -0,3311*** (0,1147) -0,2849*** (0,0851) -0,1096 (0,0720) (0,0741) -0,1647** -0,1170* (0,0618) -0,0834* (0,0497) -0,0506 (0,0482) 0,0165 (0,0455) 0,0677 (0,0438) Aug -0,9331*** (0,1687) -0,5216*** (0,1126) -0,4350*** (0,0838) -0,2791*** (0,0716) -0,2392*** (0,0736) -0,3386*** (0,0603) -0,2462*** (0,0497) -0,1920*** (0,0479) -0,1142** (0,0451) -0,0176 (0,0438) Sep -1,1485*** (0,1695) -0,6290*** (0,1124) -0,4721*** (0,0845) -0,4864*** (0,0720) -0,3793*** (0,0738) -0,3744*** (0,0604) -0,3446*** (0,0502) -0,3036*** (0,0486) -0,2286*** (0,0456) -0,0702 (0,0445) Oct -0,9117*** (0,1663) -0,5784*** (0,1115) -0,3770*** (0,0830) -0,4241*** (0,0714) -0,1589** (0,0731) -0,1344** (0,0602) -0,1484*** (0,0492) -0,0982** (0,0489) -0,0292 (0,0454) 0,1059 (0,0441) Nov -1,2014*** (0,1672) -0,4840*** (0,1118) -0,3106*** (0,0830) -0,3093*** (0,0723) -0,1919*** (0,0734) -0,1192* (0,0610) -0,1235** (0,0491) -0,1128** (0,0491) -0,0814* (0,0458) 0,0526 (0,0444) Dec -1,1878*** (0,1643) -0,3851*** (0,1122) -0,2952*** (0,0832) -0,1077 (0,0726) -0,0854 (0,0744) 0,0631 (0,0621) -0,0095 (0,0496) 0,0838* (0,0498) 0,0891* (0,0466) 0,2291 (0,0448) #Obs 86698 86687 86697 86916 86467 86688 86722 86675 86694 86677 Adj R2 0.0008 0.0004 0.0004 0.0010 0.0003 0.0010 0.0010 0.0012 0.0007 0.0007 F-statistic 7.47 4.09 3.93 8.76 3.17 8.92 8.74 10.29 6.57 6.70
18
Table 5 in the appendix contains the model 1 regression result for the complete period
of France. Only in portfolio 1,2 and 5-9 the results are significant at a 10% level. There is no
pattern in the January returns between the portfolios. Only in the smallest portfolio there is
evidence of the January effect at a 5% significance level.
The complete period estimations for Greece are summarized in Table 6. Portfolios 3
and 5-8 show negative returns of which 4 of them are significant at a 1% level. Portfolio 1, 2,
9 and 10 contain positive January returns of which only the return of portfolio 2 is not
significant. The returns for the other 11 months are all significantly lower in portfolio 1,
except for July. This return is higher than in January, but not significant. In portfolio 2-10
there is no pattern in the returns. Therefore the conclusion is that there is no January effect.
Lastly, in table 7 the estimation results are shown for the Netherlands. In none of the
portfolios there appears to be a pattern in the returns. Thus, there is no evidence of the
January anomaly.
4.3
Robustness
To investigate if the existence of the January anomaly changes due to the last financial crisis,
first the period before 2007 is estimated. In table 8 the results for 1985-2006 for Norway are
shown. The January returns are drawing attention. These are positive for all portfolios and
significant for portfolio 1-8 at a 1% level. The January returns decrease when firm size
increases, with two exceptions. This is again consistent with Reinganum (1983). The January
effect exists in the smallest portfolio. Portfolios 2-5 and 7 contain lower returns for all months
different from January. However, they are not all significant. Thus, for portfolio 2-5 and 7
there are signs of the January effect.
After the first sub-period, the second sub-period (2007-2010) ,which is the financial crisis, is
19
regression of model 1 results in positive and significant returns for January. From portfolio 2
to 10, the return decreases as firm size increases. Only in portfolio two the other 11 months
are significantly lower than January. This provides evidence for the January effect in portfolio
2. In the other 9 portfolio, there is no evidence for this seasonality.
The results for the last sub-period regression (2011-2013) are shown in table 18. All
the January returns are positive and significant, except for the largest portfolio. For portfolio
1, 2, 4-6 and 8, the returns of the other 11 months are all lower compared to January, although
they are not all significant. Therefore, these portfolios show signs of the January effect, but
this is stated with prudence.
Table 8 - Model 1Norway, 10 size portfolios, 1985-2006
Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
1 2 3 4 5 6 7 8 9 10 Jan 1,2285*** (0,1565) 0,4308*** (0,0887) 0,3795*** (0,0688) 0,3761*** (0,0746) 0,4000*** (0,0669) 0,2533*** (0,0553) 0,2764*** (0,0452) 0,2551*** (0,0476) 0,2683 (0,0470) 0,1583 (0,0475) Feb -1,3361*** (0,2300) -0,1947 (0,1272) -0,2014** (0,0998) -0,1488 (0,1054) -0,1652* (0,0974) -0,0807 (0,0785) -0,2193*** (0,0653) -0,0404 (0,0669) -0,1594 (0,0681) 0,0431 (0,0680) Mar -0,8773*** (0,2278) -0,3587*** (0,1253) -0,1879* (0,0973) -0,2515** (0,1046) -0,2854*** (0,0942) -0,1410* (0,0776) -0,1641** (0,0645) -0,1159* (0,0664) -0,1106 (0,0656) 0,0646 (0,0662) Apr -1,0566*** (0,2349) -0,2556** (0,1286) -0,3321*** (0,1007) -0,1963* (0,1082) -0,1990** (0,0958) -0,1467* (0,0800) -0,0652 (0,0669) -0,0771 (0,0680) -0,0389 (0,0681) 0,0141 (0,0676) May -1,1531*** (0,2292) -0,2897** (0,1259) -0,2973*** (0,1001) -0,3291*** (0,1084) -0,1406 (0,0941) -0,1423* (0,0804) -0,1746*** (0,0663) -0,1386** (0,0682) -0,1189 (0,0662) -0,0571 (0,0676) Jun -1,2855*** (0,2284) -0,5232*** (0,1238) -0,3404*** (0,0994) -0,4888*** (0,1066) -0,3431*** (0,0937) -0,2433*** (0,0779) -0,3336*** (0,0658) -0,2662*** (0,0669) -0,2070 (0,0649) -0,0409 (0,0670) Jul -0,9356*** (0,2308) -0,3342*** (0,1246) -0,1176 (0,1005) -0,2456** (0,1064) -0,1830* (0,0952) -0,1013 (0,0776) -0,1540** (0,0661) -0,0488 (0,0659) -0,0353 (0,0649) -0,0272 (0,0657) Aug -1,2557*** (0,2222) -0,3195*** (0,1217) -0,3431*** (0,0975) -0,3605*** (0,1043) -0,3421*** (0,0924) -0,3032*** (0,0769) -0,3133*** (0,0654) -0,2374*** (0,0651) -0,2060 (0,0639) -0,0770 (0,0652) Sep -1,6941*** (0,2216) -0,5763*** (0,1211) -0,6358*** (0,0986) -0,5005*** (0,1059) -0,5800*** (0,0917) -0,3722*** (0,0775) -0,4042*** (0,0653) -0,3832*** (0,0668) -0,3274 (0,0641) -0,1682 (0,0660) Oct -1,2131*** (0,2186) -0,4125*** (0,1201) -0,3440*** (0,0967) -0,3537*** (0,1044) -0,2870*** (0,0908) -0,1697** (0,0775) -0,1839*** (0,0644) -0,1800*** (0,0669) -0,1672 (0,0645) -0,0003 (0,0656) Nov -1,7173*** (0,2185) -0,3183*** (0,1217) -0,2187** (0,0958) -0,2817*** (0,1044) -0,2481*** (0,0913) -0,0543 (0,0773) -0,1907*** (0,0645) -0,1360** (0,0675) -0,1416 (0,0653) -0,0181 (0,0653) Dec -1,7115*** (0,2152) -0,3575*** (0,1226) -0,1751* (0,0965) -0,2212* (0,1050) -0,1964** (0,0931) 0,1160 (0,0792) -0,0315 (0,0648) 0,0609 (0,0688) -0,0190 (0,0662) 0,1706 (0,0665) #Obs 45611 45611 45615 45604 45611 45610 45658 45567 45602 45609 Adj R2 0.0020 0.0005 0.0010 0.0006 0.0009 0.0011 0.0014 0.0014 0.0008 0.0005 F-statistic 9.11 2.91 5.05 3.46 4.79 5.68 6.68 6.81 4.47 3.18
20
Table 13 - Model 1Norway, 10 size portfolios, 2007-2010Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
1 2 3 4 5 6 7 8 9 10 Jan 0,2440 (0,3193) 0,4911*** (0,1707) 0,3606* (0,2072) 0,2027** (0,1308) 0,1446 (0,1099) -0,1608 (0,1264) -0,0725 (0,0848) -0,1084 (0,0860) -0,1003 (0,0676) -0,1655** (0,0727) Feb -0,6255 (0,4513) -1,0759*** (0,2453) -0,4592 (0,2966) -0,3858 (0,1839) -0,2349 (0,1575) 0,0344 (0,1808) 0,1169 (0,1227) 0,2018* (0,1228) 0,1810* (0,0981) 0,2214** (0,1036) Mar -0,3313 (0,4372) -0,6599*** (0,2408) -0,3324 (0,2890) -0,1613 (0,1805) -0,2163 (0,1586) 0,3478* (0,1825) 0,2230* (0,1219) 0,2813** (0,1224) 0,2686*** (0,0972) 0,4690*** (0,1028) Apr -0,1564 (0,4448) -0,6108** (0,2447) -0,2556 (0,2939) 0,1959 (0,1815) 0,2547 (0,1589) 0,6417*** (0,1840) 0,5201*** (0,1238) 0,6080*** (0,1235) 0,3596*** (0,0964) 0,4572*** (0,1027) May -0,0311 (0,4158) -0,4208* (0,2275) -0,0303 (0,2858) -0,1106 (0,1770) 0,0415 (0,1501) 0,3206* (0,1817) 0,2236* (0,1224) 0,3338*** (0,1192) 0,2302** (0,0979) 0,3589*** (0,1017) Jun -0,5280 (0,4155) -0,6662*** (0,2256) 0,2127 (0,2825) -0,2683 (0,1737) -0,1712 (0,1493) 0,5244*** (0,1771) -0,0405 (0,1219) 0,0446 (0,1157) 0,0562 (0,0961) 0,0866 (0,1006) Jul -0,2241 (0,4173) -0,5096*** (0,2239) -0,0415 (0,2802) -0,0368 (0,1737) -0,1047 (0,1501) 0,0968 (0,1750) 0,1839 (0,1211) 0,1339 (0,1149) 0,1887** (0,0957) 0,2272** (0,1003) Aug -0,7569* (0,4232) -0,7563*** (0,2272) -0,4279 (0,2816) -0,3482** (0,1771) -0,0520 (0,1512) -0,0030 (0,1751) 0,0033 (0,1220) -0,0465 (0,1165) 0,1524 (0,0959) 0,1590 (0,1015) Sep -0,7870* (0,4176) -0,6851*** (0,2256) -0,5250* (0,2799) -0,3690** (0,1772) -0,3014** (0,1482) -0,2296 (0,1756) -0,0864 (0,1238) -0,0404 (0,1160) -0,0936 (0,0980) 0,1420 (0,1025) Oct -0,5678 (0,4147) -0,5872*** (0,2230) -0,6151** (0,2771) -0,5013*** (0,1763) -0,2915** (0,1464) 0,1674 (0,1741) 0,0666 (0,1224) 0,1655 (0,1156) 0,1580 (0,0974) 0,1490 (0,1030) Nov -0,6279 (0,4195) -0,6644*** (0,2210) -0,3908 (0,2757) -0,3629** (0,1826) -0,3621** (0,1502) 0,2620 (0,1776) 0,0070 (0,1215) -0,0282 (0,1181) 0,0433 (0,0976) 0,1858* (0,1038) Dec -0,5400 (0,4063) -0,3912* (0,2191) -0,2718 (0,2817) -0,0573 (0,1847) 0,0096 (0,1524) 0,3253* (0,1764) 0,2393* (0,1243) 0,2171* (0,1210) 0,4315*** (0,0990) 0,5392*** (0,1048) #Obs 20841 20898 20785 20849 20832 20843 20840 20847 20837 20838 Adj R2 0.0000 0.0007 0.0004 0.0009 0.0009 0.0015 0.0013 0.0022 0.0019 0.0023
21
Tabel 18 - Model 1Norway, 10 size portfolios, 2011-2013Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
Table 9 contains the estimates for the first sub-period for Germany. Eight of the January
returns are significantly positive, and it's again decreasing from portfolio 1 to 10. However, in
none of the 10 portfolios all the returns are (significantly) lower than the corresponding
January returns. This is also the case for the first sub-period estimates for France, Greece and
the Netherlands. Therefore, there is no evidence for the January anomaly in this period. These
results are summarized in table 10, 11 and 12 respectively.
For the second sub-period, table 14, 15, 16 and 17 show the results for Germany, France,
Greece and the Netherlands respectively. In none of these countries there appears a pattern
which supports existence of the January effect during the financial crisis.
1 2 3 4 5 6 7 8 9 10 Jan 0,3979* (0,2162) 0,6809*** (0,1518) 0,3800*** (0,1334) 0,3702*** (0,0934) 0,1661** (0,0786) 0,2930*** (0,0714) 0,2166*** (0,0756) 0,2856*** (0,0559) 0,1442*** (0,0495) 0,0472 (0,0410) Feb -0,5368* (0,3184) -0,5409** (0,2245) -0,1289 (0,1904) -0,2041 (0,1357) -0,0846 (0,1142) -0,2429** (0,1022) 0,1127 (0,1116) -0,0768 (0,0800) 0,1017 (0,0714) 0,1117* (0,0587) Mar -0,5157* (0,3082) -0,4925** (0,2211) -0,2428 (0,1878) -0,3106** (0,1330) -0,0994 (0,1105) -0,1248 (0,1010) -0,1548 (0,1099) -0,2643*** (0,0787) -0,1340* (0,0701) -0,0206 (0,0576) Apr -0,5968* (0,3156) -0,7786*** (0,2236) -0,4553** (0,2014) -0,2494* (0,1391) (0,1128) -0,1342 -0,2977*** (0,1063) (0,1133) -0,1328 -0,3204*** (0,0827) -0,1707** (0,0731) (0,0610) -0,0339 May -0,5941* (0,3189) -0,6371*** (0,2251) -0,4673** (0,2047) -0,4430*** (0,1440) -0,1432 (0,1149) -0,3975*** (0,1085) -0,3211*** (0,1127) -0,3369*** (0,0861) -0,1436* (0,0756) -0,1163* (0,0639) Jun -0,4781 (0,3288) -0,5274** (0,2253) -0,4048* (0,2141) -0,5364*** (0,1471) -0,1128 (0,1174) -0,2827** (0,1097) -0,2979*** (0,1133) -0,3149*** (0,0897) -0,1526** (0,0775) -0,0906 (0,0659) Jul (0,3196) -0,3097 (0,2294) -0,3329 (0,2104) -0,3224 -0,3568** (0,1448) (0,1151) -0,1119 (0,1092) -0,0732 (0,1118) -0,0854 -0,3320*** (0,0894) (0,0763) -0,1050 0,1535** (0,0646) Aug -0,7689** (0,3185) -0,2121 (0,2247) -0,5230** (0,2095) -0,4802*** (0,1428) -0,2870** (0,1161) -0,3419*** (0,1079) -0,3341*** (0,1110) -0,3804*** (0,0878) -0,1148 (0,0772) -0,1135* (0,0640) Sep -0,5973* (0,3267) -0,5050** (0,2283) -0,0621 (0,2123) -0,3473** (0,1485) -0,2916** (0,1191) -0,1856* (0,1097) -0,1576 (0,1127) -0,4152*** (0,0915) -0,2107*** (0,0783) -0,0848 (0,0654) Oct -0,5457* (0,3215) -0,8882*** (0,2245) (0,2069) -0,2339 -0,2900** (0,1449) (0,1177) -0,1458 (0,1082) -0,1976 (0,1130) -0,0675 (0,0888) -0,1000 (0,0771) 0,0657 0,2410*** (0,0642) Nov -0,6320** (0,3169) -0,8196*** (0,2311) 0,0903 (0,2085) -0,4667*** (0,1417) -0,1974 (0,1229) -0,2732** (0,1107) -0,1397 (0,1120) -0,2021** (0,0896) -0,0973 (0,0779) 0,0925 (0,0647) Dec -0,5375* (0,3189) -0,5412** (0,2305) (0,2064) -0,2894 (0,1439) -0,1790 (0,1205) -0,0412 -0,2485** (0,1141) (0,1121) 0,0576 (0,0897) -0,1044 (0,0772) -0,0673 (0,0644) 0,0325 #Obs 20244 20240 20244 20239 20240 20253 20231 20243 20238 20241 Adj R2 -0.0001 0.0008 0.0003 0.0007 0.0000 0.0006 0.0011 0.0020 0.0011 0.0024 F-statistic 0.75 2.48 1.63 2.21 1.04 2.16 3.03 4.66 3.02 5.49
22
Lastly, the third sub-period is analyzed. In table 19, the third sub-period results for
Germany are tabulated. All the January returns are significantly positive. In portfolio 2, 4, 6
and 8, the returns for the other 11 months are all lower than January, but not all of these 11
months are significant. Therefore in these portfolio, there are some signs of the January effect.
In the other portfolios, there is at least one month with a higher return compared to January,
which doesn't support the existence of the January effect is these portfolios.
In contrast to Germany, the January effect is unveiled in three portfolios in the third
sub-period for France. Table 20 contains the results for this sub-period. All the January returns are
significantly positive. The January effect is present in portfolio 2, 4 and 5 because the returns
in the other 11 months are all significantly lower compared to January. Portfolios 1, 3, and
6-8 contain lower returns for the remaining months, but with at least one month that is not
significant. Therefore these portfolios show signs of the January effect. Portfolios 9 and 10
both have a higher return in February. Thus, in the two largest portfolios there is no January
effect.
Table 21 contains the results for the most recent period in Greece. Again, all January returns
are positive and significant for portfolio 2-9 and insignificant for portfolio 1. It is remarkable
that the January return is the lowest in de smallest portfolio and the highest in the largest
portfolio. In none of the portfolio's the January effect exists, but there are some signs of the
anomaly in portfolios 2 and 5-9. The returns in the other months of these years are all lower
than January, but at least one is not significant.
The results for the Netherlands are tabulated in table 22. The January returns are all positive
and significant, except for portfolio 2 and 10. Only portfolio 3 and 7-9 contain signs of the
23
4.4
Model 2
In table 23 the regression results of model 2 are shown for all countries. Only for Norway, the
return in January is significantly positive. The January returns in the other countries are
significantly negative. Thus, in these countries the January effect is not supported.
Although there seems to be a January effect in Norway, the interaction term is
significantly negative. This implies that the magnitude of the January effect decreases over
time in Norway. This is consistent in a moderate way with prior research of Haugen & Jorion
(1996).
However, in the other countries the interaction term is significantly positive. This implies that
the returns in January are increasing over time, but not the magnitude of the January effect
because the returns in January are significantly negative.
Table 22 - Model 2, all countries, 1985-2014
Constant Jan Trend Jan x Trend Adj. R2 #obs
Norway .0641474*** (.0164353) .3866214*** (.0568403) -.0006469 (.000803) -.0069004** (.0027061) 0.0002 867561 Germany -.1113665*** (.0131429) -.1342989*** (.0448567) .0142878*** (.0006466) .0085193*** (.0021518) 0.0001 4350420 France .0470527*** (.0071307) -.2420462*** (.0244442) .0010782*** (.0003639) .0131461*** (.0012138) 0.0000 3842796 Greece .1306804*** (.0155913) -.9006137*** (.0543177) -.0055184*** (.000753) .038733*** (.0025495) 0.0001 1164071 Netherlands .0134434 (.0114103) -.1235303*** (.0393597) .0030392*** (.0006209) .0060988*** (.0020847) 0.0000 1267447 Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
24
4.5
Conclusion
Concluding, the regression results of model 1 show no evidence of the January effect for
Germany, Greece and the Netherlands in both the complete period and the sub-periods.
However, the January effect is supported in the smallest portfolios in Norway for the
complete period, and in portfolio 2 during the financial crisis. In France, the January effect is
only supported in the smallest portfolio in the complete period and in portfolio 2, 4 and 5 after
de financial crisis. This could be explained by an increasing demand for small firm stocks
which declined during the financial crisis. The findings in Norway and France are consistent
with previous research of Roll (1981) that the January effect exists in small firm stocks.
The results of model 2 show that if the January effect exists, its magnitude decreases
over time. This is the case for Norway. According to the results of model 2, in the other
countries there is no January effect. However, for model 2 no distinction is made in firm size.
According to Roll (1981), the January effect could exist for small firm stocks if firm size
would be taken into account.
5
Summary and conclusion
The January anomaly is one of the most notorious stock return patterns. It refers to the
seasonality in which the stock returns in January are higher than in the other 11 months of the
year (Gultekin & Gultekin, 1983). There is a lot of evidence for the existence of the January
effect in the U.S. stock markets, for example Rozeff and Kinney (1976) and Gultekin and
Gultekin (1983).
However, the evidence for Europe is scarce and outdated. Therefore, this research examines
whether the January effect exists in the European stock markets. The main question in this
research is: Does the January effect show up in Europe, and will its magnitude decrease over
25
The empirical results of the regression of model 1 provide support for the existence of the
January anomaly in Norway during the financial crisis in the second smallest portfolio. This
seasonality is also present in the complete period for Norway in the smallest 30% of the firms.
In France, the January effect exists in the smallest 10% of the firms in the complete period, as
well as in portfolio 2, 4 and 5 in the years after the financials crisis. This is consistent with
findings of Roll (1981) that the January effect is especially a size effect. For The Netherlands,
Germany and Greece, there is no strong support for the existence of the January effect in any
of the investigated periods.
The regression results of model 2 show that the magnitude of the January effect
decreases in Norway over time. In the other countries, there is no January effect according to
model 2. So the conclusion for these countries is that the January returns increase over time.
The main conclusion of this research is that the January effect only exists for small
firm stocks in Norway during the financial crisis and in France after the financial crisis. The
magnitude of the January effect decreases over time.
Since the period after the financial crisis contains only a few years, future research has
to determine if the conclusions regarding this period are consistent. Furthermore, the
magnitude of the January effect should be investigated individually for small firm stocks and
26
6
References
Bentzen, E. (2009). Seasonality in stock returns. Applied Financial Economics, 19(20), 1605–1610.
doi:10.1080/09603100902984368
Bodie, Z., Kane, A., & Marcus, A. (2011). Investments and portfolio management (p. 161).
Bullock, C., Williams, J., & Tucker, R. (1919). The Review of Economic statistics (p. 147). Harvard
University Press.
Chan, K. C. (1986). Can Tax-Loss Selling Explain the January Seasonal in Stock Returns ? The
Journal of Finance, 41(5), 1115–1128.
Cohen, R. (2014, June 3). Technische analyse lijkt houvast te geven, maar is een wassen neus.
Financieel Dagblad, p. 21.
Constantinides, G. M. (1984). Optimal stock trading with personal taxes: Implications for Prices and
the Abnormal January Returns. Journal of Financial Economics, 13, 65–89.
Fama, E. F. (1970). Efficient Capital Markets : A Review of Theory and Empirical Work, 25(2), 28–
30.
Fama, E. F. (1995). Random Walks in Stock Market Prices. Financial Analysts Journal, 51(1), 75–80.
Gultekin, N. B., & Gultekin, M. N. (1983). Stock market seasonality: International Evidence. Journal
of Financial Economics, 12, 469–481.
Haugen, R. A., & Jorion, P. (1996). January All Effect : These Still Year There. Financial Analysts
Journal, 52(1), 27–31.
Investopedia. (2014). Tax Reform act. Retrieved June 28, 2014, from
http://www.investopedia.com/terms/t/taxreformact1986.asp
Jones, C. P., Pearce, D. K., & Wilson, J. W. (1987). Can Tax-Loss Selling Explain the January Effect ?
A Note. Journal of Finance, 42(2), 453–461.
Keim, B. (1983). Size-related anomalies and stock return seasonality. Journal of Financial Economics,
12, 13–32.
Kendall, M. G. (1953). The Analysis of Economic Time-Series-Part I : Prices. Journal of the Royal
Statistical Society, 116(1), 11–34.
Lakoniskhok, J., & Smidt, S. (1984). VOLUME AND TURN-OF-THE-YEAR BEHAVIOR*. Journal
of Financial Economics, 13(June), 435–455.
Ogden, P. J. (1990). Turn-of-Month Evaluations of Liquid Profits and Stock Returns : A Common
Explanation for the Monthly and January Effects. The Journal of Finance, 45(4), 1259–1272.
Owens, R. N., & Hardy, C. O. (1925). Interest rates and stock speculation: a study of the influence of
27
Reinganum, M. R. (1983). The anomalous stock market behavior of small firms in January. Journal of
Financial Economics, 12(1), 89–104. doi:10.1016/0304-405X(83)90029-6
Roll, R. (1981). A Possible Explanation of the Small Firm Effect. Journal of Finance, 36(4), 879–888.
Roll, R. (1983). Vas ist das? The turn-of-the-year effect and the return premia of small firms. The
Journal of Portfolio Management, 9(2), 18–21.
Rozeff, S., & Kinney, R. (1976). Capital market seasonality: the case of stock returns. Journal of
Financial Economics, 3, 379–402.
Schultz, P. (1985). Personal Income Taxes and the January Effect : Small Firm Stock Returns Before
the War Revenue Act of 1917 : A Note. Journal of Finance, 40(1), 333–343.
Wachtel, S. B. (1942). Certain oberservations on seasonal movements in stock prices. The Journal of
Business of the University of Chicago, 15(2), 184–193.
Wrds. (2014). Compustat Manual. Retrieved June 09, 2014, from
http://wrds-web.wharton.upenn.edu//wrds/support/Additional Support/WRDS Knowledge Base with
FAQs.cfm?article_id=5327&folder_id=660
Www.history.com. (n.d.). War Revenue Act passed in U.S. Retrieved from
http://www.history.com/this-day-in-history/war-revenue-act-passed-in-us
28
7
Appendix
Table 4
Model 1 -Germany, 10 size portfolios, 1985-2014
Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
1 2 3 4 5 6 7 8 9 10 Jan 1.2866*** (0.1141) 0.2800*** (0.0479) 0.1406*** (0.0280) 0.1240*** (0.0221) 0.0753*** (0.0225) 0.0482** (0.0189) 0.0696*** (0.0171) 0.0491*** (0.0156) 0.0250* (0.0143) -0.0189 (0.0124) Feb -0.2920* (0.1640) -0.0147 (0.0689) -0.0320 (0.0406) -0.0381 (0.0317) -0.0232 (0.0324) 0.0605** (0.0272) 0.0754*** (0.0246) 0.1030*** (0.0224) 0.1255*** (0.0206) 0.1341*** (0.0177) Mar -0.2998* (0.1609) -0.0201 (0.0674) -0.0994** (0.0398) -0.0843*** (0.0311) -0.0073 (0.0319) 0.0450* (0.0268) 0.0098 (0.0239) 0.0245 (0.0220) 0.0646*** (0.0202) 0.1212*** (0.0174) Apr -0.4651*** (0.1651) -0.0627 (0.0688) 0.0161 (0.0409) 0.0323 (0.0317) 0.0708** (0.0327) 0.1301*** (0.0276) 0.0891*** (0.0244) 0.1338*** (0.0226) 0.1349*** (0.0206) 0.1596*** (0.0178) May -0.3016* (0.1646) -0.0193 (0.0688) -0.0100 (0.0405) -0.0368 (0.0316) -0.0032 (0.0323) 0.0142 (0.0275) -0.0098 (0.0242) 0.0156 (0.0224) 0.0486** (0.0205) 0.0853*** (0.0178) Jun -0.5002*** (0.1645) -0.0687 (0.0688) -0.0960** (0.0406) -0.1023*** (0.0316) -0.0758** (0.0323) -0.0366 (0.0274) -0.0774*** (0.0243) -0.0121 (0.0223) 0.0232 (0.0205) 0.0618*** (0.0178) Jul -0.3385** (0.1625) -0.0220 (0.0680) -0.1368*** (0.0401) -0.1422*** (0.0314) -0.0012 (0.0321) -0.0371 (0.0271) -0.0481** (0.0240) 0.0239 (0.0221) 0.0434** (0.0202) 0.1126*** (0.0175) Aug -0.1843 (0.1623) -0.0813 (0.0676) -0.0953** (0.0400) -0.1237*** (0.0313) -0.1071*** (0.0321) -0.0762*** (0.0270) -0.0933*** (0.0240) -0.0473** (0.0220) -0.0288 (0.0202) -0.0119 (0.0175) Sep 0.1391 (0.1631) -0.1403 (0.0681) -0.2251*** (0.0400) -0.1660*** (0.0316) -0.1402*** (0.0324) -0.1296*** (0.0273) -0.1572*** (0.0243) -0.1496*** (0.0223) -0.0789*** (0.0204) -0.0574*** (0.0177) Oct -0.2464 (0.1607) -0.0205 (0.0672) -0.0468 (0.0396) -0.0107 (0.0313) -0.0058 (0.0321) -0.0276 (0.0269) -0.0230 (0.0242) 0.0203 (0.0222) 0.0255 (0.0203) 0.1106*** (0.0176) Nov -0.2586 (0.1617) -0.1211* (0.0679) -0.0811** (0.0399) -0.1093*** (0.0314) -0.0835*** (0.0321) -0.0838*** (0.0271) -0.0600** (0.0243) -0.0254 (0.0223) 0.0455** (0.0205) 0.1145*** (0.0177) Dec -0.2914* (0.1631) -0.2414*** (0.0688) -0.1068*** (0.0404) -0.0931*** (0.0319) -0.0359 (0.0327) 0.0206 (0.0274) 0.0794*** (0.0247) 0.0823*** (0.0227) 0.1011*** (0.0208) 0.1463*** (0.0179) #Obs 434237 434221 434229 434230 434227 434232 434231 434228 434225 434228 Adj R2 0.0000 0.0000 0.0001 0.0002 0.0001 0.0003 0.0004 0.0005 0.0004 0.0007 F-statistic 2.29*** 2.09** 5.43*** 7.45*** 6.18*** 12.93*** 18.60*** 21.26*** 17.77*** 30.15***
29
Table 5Model 1 -France, 10 size portfolios, 1985-2014
Note: * Significant at 10%, **Significant at 5%, *** Significant at 1%. Returns in %. Standard errors in parentheses
1 2 3 4 5 6 7 8 9 10 Jan 0.3413*** (0.0522) 0.1132*** (0.0324) 0.0321 (0.0247) -0.0238 (0.0241) -0.0328* (0.0194) 0.0670*** (0.0242) 0.0544*** (0.0176) 0.0594*** (0.0160) 0.0510*** (0.0159) 0.0288 (0.0133) Feb -0.2888*** (0.0752) -0.0209 (0.0467) 0.0847** (0.0355) 0.1570*** (0.0345) 0.1546*** (0.0280) 0.0908*** (0.0346) 0.1027*** (0.0252) 0.1550*** (0.0229) 0.1624*** (0.0228) 0.1171 (0.0190) Mar -0.2643*** (0.0739) 0.0256 (0.0459) 0.0949*** (0.0350) 0.1226*** (0.0338) 0.1362*** (0.0277) 0.0879*** (0.0340) 0.0839*** (0.0247) 0.0609*** (0.0225) 0.0883*** (0.0224) 0.1014 (0.0186) Apr -0.1527** (0.0761) 0.1261*** (0.0471) 0.1539*** (0.0359) 0.1665*** (0.0346) 0.1850*** (0.0283) 0.1429*** (0.0348) 0.0763 (0.0254) 0.1242*** (0.0230) 0.0985*** (0.0228) 0.1100 (0.0190) May -0.2511*** (0.0754) -0.0288 (0.0469) 0.0700** (0.0354) 0.1601*** (0.0345) 0.1130*** (0.0281) 0.0216 (0.0347) 0.0334** (0.0253) 0.0326 (0.0230) 0.0152 (0.0229) 0.0361 (0.0191) Jun -0.3501*** (0.0755) -0.0343 (0.0469) 0.0183 (0.0356) 0.0426 (0.0344) 0.0268 (0.0281) -0.0546 (0.0348) -0.0565 (0.0251) -0.0838*** (0.0227) -0.0703*** (0.0227) -0.0372 (0.0189) Jul -0.3663*** (0.0745) -0.0688 (0.0465) -0.0381 (0.0353) 0.0243 (0.0341) 0.0421 (0.0279) -0.0581* (0.0346) -0.0208 (0.0250) -0.0063 (0.0227) 0.0349 (0.0226) 0.0587 (0.0189) Aug -0.1866** (0.0748) -0.0066 (0.0462) 0.0722** (0.0351) 0.0983*** (0.0339) 0.1243*** (0.0278) 0.0030 (0.0345) -0.0055*** (0.0250) -0.0352 (0.0227) -0.0329 (0.0226) -0.0489 (0.0188) Sep -0.3933*** (0.0748) -0.1798*** (0.0465) -0.1174*** (0.0354) -0.0391 (0.0343) -0.0258 (0.0278) -0.1275*** (0.0346) -0.1247* (0.0250) -0.1413*** (0.0227) -0.1257*** (0.0226) -0.0788 (0.0189) Oct -0.3225*** (0.0735) -0.0191 (0.0460) 0.0026 (0.0350) 0.0636* (0.0342) 0.0744*** (0.0275) -0.0319 (0.0343) -0.0430 (0.0249) -0.0670*** (0.0225) -0.0488** (0.0226) 0.0529 (0.0188) Nov -0.2151*** (0.0740) -0.0121 (0.0463) -0.0184 (0.0355) 0.0631* (0.0345) 0.0674** (0.0277) -0.0540 (0.0346) -0.0319 (0.0251) -0.0055 (0.0229) 0.0190 (0.0230) 0.0577 (0.0191) Dec -0.4063*** (0.0734) -0.0277 (0.0460) 0.1008*** (0.0354) 0.1103*** (0.0344) 0.0877*** (0.0277) 0.0525 (0.0345) 0.0332 (0.0249) 0.0175 (0.0227) 0.0845*** (0.0228) 0.1012 (0.0189) #Obs 381291 381304 381278 381294 381289 381307 381273 381291 381291 381290 Adj R2 0.0001 0.0001 0.0002 0.0002 0.0003 0.0003 0.0004 0.0007 0.0007 0.0007 F-statistic 4.88*** 4.20*** 8.55*** 7.38*** 10.13*** 9.75*** 13.62*** 26.60*** 25.42*** 24.64***
