The January effect on the European market

(1)

The January Effect on the European Market

Bachelor Thesis Finance

Name: Sebastiaan de Vries Student number: 10666168 Study: Economics & Business Track: Finance & Organization Supervisor: Evgenia Zhivotova Date: January 2, 2017

Abstract

In this thesis highly significant evidence for the post millennium January Effect on the European market is empirically analyzed with an ordinary least squares regression for firms with small and high market value. Moreover, the exposure to risk of these stocks is reviewed by measuring Sharpe Ratios for all the monthly periods. Prior to this, presumable causes of the calendar effect such as tax-loss selling, window dressing, and equity home bias are discussed and substantiated by referring to earlier research.

(2)

2

This document is written by Student Sebastiaan de Vries, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

1. Introduction

Since the day in 1942 when the January Effect was noticed by the American investment banker Sidney B. Wachtel1, financial economists have been investigating this seasonal stock price movements for many reasons. The January Effect states that stock returns, especially for small market cap stocks, tend to be higher in the first month of the year compared to others. But what causes this price fluctuation during the beginning of the year? According to Reinganum (1982) the main factor for this phenomenon is tax-loss selling at the end of a year. Some researchers suggest that bonus payments at the end of the year will drive up stock prices (Ogden, 1990). While others believe it is caused by a renewed sense of optimism among investors (Ciccone, 2011).

As aforementioned there has been several researches on these topics. However most of these are aged and generally done for the same environment, namely the American. Would all of these empirical tests have the same outcome for the European environment? There are a lot of differences among those two; tax regulations, currencies, industries and so on. Besides that, did the global financial crisis, starting in 2008, had an impact on the January Effect? During this crisis volatility of stocks could increase, several stocks might have negative returns and therefore have an impact on the seasonality.

In this thesis there will be investigated if the January Effect does affect returns of European firms with a small sized market capitalization from the year 2000. To test whether the financial crisis had any influence on the seasonal movement, there will be a divide of time series pre and post 2008. The prediction is that there is a perceptible effect in Europe,

however this may vary from the amount encountered to the United States due to different economic circumstances. Furthermore, it is expected that the effect will be less than before because stock market information is widespread nowadays and therefore price anomalies will be reduced.

The following research will create new information about the contemporary January Effect in the European market. Investors might be able to attain an alpha taking advantage of a consistent European January Effect. Additionally it will be interesting to observe if the home bias and trade restrictions will play a part in trading in European stocks instead of U.S. stocks. In short, to investigate the seasonal price movement stock price data is collected of the Stoxx 600 index, which represents the 600 largest listed companies of Europe. Afterwards

(4)

4 a selection is made of the highest and lowest 10% market capitalized firms on this index. Monthly average returns are calculated through the year. Ultimately, a dummy regression with control variables is set up to discover whether January has a significant higher return per risk unit over the given time series.

The further framework of this thesis is as follows. In the next chapter existing

literature will be reviewed to make relevant assumptions and hypotheses for the research and improve upon earlier probes. Subsequently, the research method and data description will be shown in the methodology to define how the research question is investigated and where the data is obtained from. Afterwards the empirical results are presented and analyzed with corresponding tables and figures. Finally, the research results are discussed and concluded about how they relate to previous probes. Also, potential limitations of the research method are debated to make a suggestion for further research.

(5)

5

2. Literature Review

In this section related literature is reviewed to show insight and break down the January Effect. Furthermore awareness of different theories and approaches used in earlier research are debated. The knowledge retrieved out of these publications forms an overview where the research is built upon.

2.1 Price Anomalies

According to Fama’s (1969) theory about the efficient market hypothesis security market prices should always fully reflect available information of firms, making it impossible to beat the market and purchase under- or overvalued stocks. The efficient market hypothesis states that stock prices fluctuate randomly without serial correlation. This implies that stock prices are unpredictable on previous behavior. Jensen (1978) extended this theory with the

following way to explain the efficient market hypothesis:

‘A market is efficient with respect to information set θt if it is impossible to make

economic profits by trading on the basis of information set θt’

Adding to this Praetz (1973, p. 77) suggested that if a stock market is inefficient and contains seasonal movement, speculators would make use of this to make abnormal returns. Which effectively leads to the removal of the seasonality.

However, a few years later economists Rozeff and Kinney (1976) analyzed market data of the New York Stock Exchange from the beginning of the 20th century and stated that there was a significant evidence for seasonality. Particularly January had relatively high returns in comparison with other months. The investigators ultimately compared their results with a research based on the data of the Australian capital market (Officer, 1974, p. 30). They concluded that the seasonality measured on the New York Stock Exchange was weaker than the price movements in Australia for the same time series. Furthermore the empirical studies of Officer show the highest seasonal effect in July instead of January. In other words,

empirical studies has shown that there are opportunities to realize abnormal returns in contrary to the efficient market hypothesis. This would suggest that not every broker has access to the same information set, otherwise profits cannot be made.

(6)

6 Regarding to this research, there is proof of existence of inefficient markets in

different countries which contribute to share price seasonality. Following the theory of Jensen and Fama, when information is not widely available this could be a source for abnormal returns in combination with seasonal price fluctuation.

2.2 January Effect

For a long time since the first datasets of the U.S. stock markets came publicly available calendar effects were not considered to be present. As written before, Wachtel (1942) was the first to discover a significant seasonal effect. He did a brief examination of the Dow-Jones Index from 1927 until 1942 showing that there is a frequent stock price appreciation in 11 of these 15 years. These bullish tendencies all happened in a timeframe starting from December until January. Ever since these kind of tests were performed multiple times on different datasets, with most of them resulting in significant evidence of the January Effect.

More interesting are the possible causes of such a calendar effect. How is it possible that stock prices constantly bullish in certain time periods? And which stocks are most affected by these causes? Wachtel and other economists formulated several theories to link likely causes to the January Effect. A greater exposure to a certain cause could lead to a more drastic price fluctuation during December, where the effects starts, until the end of January. In order to give an accurate prediction for the effect on the European market the most probable causes will be discussed.

2.2.1 Tax-loss Selling

Branch (1977) investigated if tax-loss selling at the end of the year has an impact on the stock prices. His dataset consisted of percentages change in the NYSE composite index during the first four Fridays in January from 1965 until 1974. He concluded that on average the tax-loss selling had little to no impact on the general stock price level during the year. However, when using a different set of stocks the results clearly changed. Depressed issues, reaching a yearly low at the end of December, tend to rise in the early days of the next year.

(7)

7

Table 1

Percentage Change in NYSE Stocks Making Lows at Year End

Number of Weeks after

Year _{Stocks in} Ending Sample 1 2 3 4 1974 306 13.2 27.28 _{28.55 32.78} 1973 231 10.3 7.97 _{11.61 14.78} 1972 70 3.62 5.2 -1.71 -4.35 1971 7 3.54 4.47 5.35 3.92 1970 13 4.52 8.63 _{11.43 12.89} 1969 151 4.84 6.12 2.3 0.97 1968 39 3.97 -1.3 -0.43 0.13 1967 39 2.28 4.35 4.78 3.04 1966 77 5.81 9.58 _{13.39 14.68} 1965 29 1.46 3 2.22 1.2 Average - 5.35* 7.53* 7.50** 8.00**

* Significant at 99 percent level ** Significant at 95 percent level *** Significant at 90 percent level

Neglecting commissions, the average gain which could be made is between 5 and 8 percent. This means that depressed shares are a significant driver of the January Effect, due to

investors who wants to decrease their capital gain tax at the end of the year. Thus, while more pressure to sell issues lowers prices at the year-end, prices rebound in the beginning of next year resulting in high returns.

In addition to this information, Starks et al. (2006) more recently published evidence of an increasing year-end trading volume when municipal bonds experience severe price loss. Moreover, the researchers did not understand why the January Effect, what is commonly known nowadays, is not arbitraged away. This statement implies that, even though almost every broker knows about the effect, the market does not take advantage of these high returns. This could be a reason that till this date the January Effect is present in Europe as long as there are depressing stocks on the market.

Small capitalized firms tend to have higher returns, in particular in the first few days in January. Reinganum (1982) examined whether there is a correlation between those large returns and tax-loss selling in the last stages of previous year. He concluded that stocks that were underperforming at the end of a year had abnormally high returns in January due to tax-loss selling. However, small capitalized stocks who were least likely to be sold for tax reasons in the previous year (winning stocks), also exhibit returns in January.

(8)

8

2.2.1.1 Capital Gain Tax Environments

In order to get a clear view of tax-loss selling incentives in Europe, statutory capital gain taxes has to be compared with the U.S. tax environment. The following figure includes both corporate and individuals levels of taxation in the member countries of the Organisation for Economic Co-operation and Development. This is a useful benchmark because all the big players of the European market are listed here as well as the country where most of the prior literature is based on, namely the U.S.

Figure 1

Combined corporate and private statutory tax rates on capital gains on shares as at 1 July 20121

On average the European countries have a statutory capital gain tax of 42%. This is

calculated by combining the corporate tax and personal tax levels of each country. Compared to the U.S. tax regulation (52%) that is about 10% points less. But we have to take into account that countries like France, The Netherlands, Germany, Great Britain, and Sweden all have reasonably high capital tax rates and own a great part on the European stock indexes. This could imply that if most of the shareholders have the same domestic background as the firms, tax-loss selling incentives are about the same as in the U.S.

2.2.2 Window Dressing

Another possible cause is the use of window dressing by portfolio managers of mutual funds. This theory hypothesize that those managers sell stocks which are losing or bear high risk and

1_{Figure 1 shows the combined statutory tax rates based on the data of OECD taxation working papers No. 19.}

ITAex shows the combined tax rate on existing equity; ITAnew shows the combined tax rate on new equity. Table is provided in the appendix.

(9)

9 buy stocks of which prices have been rising all year and have promising expectations. So at the end of the year, the time of disclosure, the portfolio composition only contains of well performing stocks. The reason managers do this is to improve appearance of the financial statements and to impress sponsors (Haugen & Lakonishok, 1997). With the increasing amount of mutual funds, the January Effect could be even higher when portfolio managers continue to window dress. Because when the share of stocks in a market held by mutual funds increases, window dressing will have a larger impact because more losing stocks will be sold at the end of the year and therefore create an incentive to repurchase those stock at the beginning of the year.

2.2.3 Investor Optimism

A behavioral cause was given by Ciccone (2001) as he emphasizes that at a start of a new year investors have a renewed sense of optimism. He indicates that firms with high dispersion outperform firms with a low dispersion in January. This seems in line with the theory that was earlier mentioned in terms of tax-loss selling. The researcher also states that February returns are lower because optimists are subsequently disappointed. But why don’t

disappointed brokers learn from previous times and cause the January Effect to disappear? Furthermore, he theorizes why the January Effect is not arbitraged away by unsentimental arbitrageurs. These arbitrageurs could take long positions in high dispersion and small stocks and short positions in low dispersion stocks. He thinks that short-selling constraints and high costs of trading those high dispersion stocks are key in this price anomaly.

2.3 Equity Home Bias

The equity home bias refers to a natural tendency of investors to focus and invest mostly on the domestic markets. However, internationally diversifying your stock portfolio hedges systematic risk. French and Poterba (1991) investigated this investment strategy and collected data from the U.S., Japan and Britain. The results showed that investors had a strong

preference for domestic stock markets since respectively 98%, 94% and 82% of their portfolio were issued by firms of their own country.

Coval and Moskowitz (1999) searched for reasons why investors choose positions that bear more risk than diversified international portfolios. They found a couple of barriers such as governmental restrictions on foreign and home capital flows, foreign taxes, exchange rate

(10)

10 risk and high transaction costs. Even though those obstacles became almost negligible in the last decades, investors still prefer investing in domestic stocks. Besides these trade barriers, a lack of international market and firm information is argued as one of the main reasons of restraint to stock traders. The latter statement is questionned by Van Nieuwenburgh and Veldkamp (2009) because they believe that investors are free to learn about foreign firms. They found out that there is little to no information advantage for domestic investors, but that traders are willing to obtain different information sets than other traders to ultimately benefit from this information asymmetry.

A few years earlier Cooper and Kaplanis (1994) published a paper with the percentages of domestic shares held in some of the biggest participants on the European market equity market.

Table 2

The home bias in equity portfolios, December 1987

Percentage of Country portfolio in domestic equities France 64.4 Italy 91.0 Spain 94.2 Sweden 100.0 UK 78.5 Germany 75.4

The percentages differ a lot from one country to another. Besides that almost every investor in each country, except from Sweden, has a lower percentage of domestic stock in their total portfolio compared to the American investors.

This could have an impact on the January Effect as well. The equity home bias insulates stock markets from each other. Because if every investor would globally diversify his stock portfolio the January Effect on each market would equalize to a certain point. The previous figures show that the rational choice to create a perfectly diversified portfolio from different stock markets is hardly noticeable. Taking this into account, the January Effect could positively be affected by the European home equity bias but should be less compared to the U.S. bias under the assumption that there are markets bearing less seasonality in their stock returns.

(11)

11

2.4 Risk Factors of Stock Returns

Fama and French (1993) created a model to predict stock returns differently than the Capital Asset Pricing Model. Their Three-Factor Model suggests that stock returns are driven by the market exposure of a certain stock, the size of the firm and the value of the firm. This model is also called an extension of CAPM, since the first factor is also implemented in the latter model. The researchers analyzed the fact that investors not only are concerned about the market risk. The market cap size, often referred to as the small minus big variable, measures the abnormal return traders previously received by investing in shares with a relatively small market capitalization. The last factor measures the value premium where investors are exposed to when investing in firms with a high book-to-market ratio. Companies with a high B/M ratio are typically undervalued because their book value exceeds the market value.

To make an unbiased prediction of the January Effect in Europe these risk factors of stock returns can be used as variables to control the seasonality effect in the following research method.

2.5 Hypothesis

Given the information about all papers discussed above there could be enough evidence to also find a January Effect on the European market. However, based on the probable causes that affect the seasonality it could less present compared to the U.S. As mentioned before, the incentive for tax-loss selling in Europe presumably is lower which decreases the effect. But taking into accountant that, with a higher amount of mutual funds, window dressing of portfolio managers could play a role in the effect, maybe stimulates the rising stock prices in January. Surprisingly, the advanced access to firm information for brokers would not per se mean that portfolios get more diversified as seen in the equity home bias puzzle. With this lack of diversification the market will not correct for price anomalies which leads to a possible existence of the effect. Finally it is interesting to see whether the returns are mostly driven by the monthly periods or by the control variables in the regression.

(12)

12

3. Methodology

In this section the method used to research the January Effect in Europe is explained. First, the data used in this research is presented and explained. Afterwards the estimation method is given about how to measure such a calendar effect with the corresponding formula’s and regressions. At last the hypothesis will be stated with a test to calculate the significance of the hypothesis.

3.1 Data

To give an answer to the research question, an equity index is needed that represents the European stock market. The aim is to investigate this in a market where illiquidity and transaction costs are minor because these risk factors may affect the returns. But there also has to be a clear divide between firms with a low and high market value to test whether there is a difference in the seasonality between these types of companies. Given these conditions the Stoxx 600 is the most suitable. This index tracks the 600 most valuable firms based in Europe. The composition of the index with respect to the sectors and countries is listed in table 3. All the index data is retrieved from DataStream by Thomas Reuters.

Table 3

Weights for the top 10 sectors and countries in Stoxx 6002

Sector Country

Health Care 12.3% _{Great Britain} 28.7%

Banks 11.8% _France 15.8%

Industrial Goods & Services 11.3% _Germany 14.4% Personal & Household Goods 8.7% _Switzerland 14.1%

Food & Beverage 7.0% _Spain 5.1%

Oil & Gas 6.1% _Netherlands 4.7%

Insurance 5.9% _Sweden 4.6% Chemicals 5.1% _Italy 3.4% Telecommunications 4.3% _Denmark 2.7% Utilities 4.1% _Belgium 2.1% Total 76.6% _Total 95.6%

2_{Table 3 data is retrieved from STOXX Limited’s current factsheet (2016) about the STOXX Europe 600 Index}

(13)

13 The daily stock prices and the market values are retrieved from DataStream for the time period from the 1st of January 2000 until the 31st of December 2015. These time series are used to check if there is a January Effect in the current millennium. In this time frame 3 periods will be tested; the first eight years, the last eight years and the whole time period. It could be that the financial crisis, starting in 2007, had an impact on the trading and returns of stocks. Investor’s optimism could be impaired by this event resulting in overall lower returns. To test for these time frames, two groups of firms are created based on their market

capitalization. The first group consists of 60 companies having the smallest market capitalization and the second of 60 companies having the largest market capitalization.

Those groups are based on the fact that small cap firms tend to have a higher price fluctuation during January according to reviewed literature (Reinganum, 1982).

Apart from using data from the Stoxx 600 index variables are used that are believed to influence stock prices. The Fama and French (1993) three-factor model is used to control the ordinary least squares regression since it explains 90% of diversified portfolio returns in comparison to only 70% when using the capital asset pricing model.

The first control variable is the excess market return given by the European market return minus the risk free rate. The second control variable is the small minus big, which measures the historic excess returns of small cap firms over big cap firms. The last one is the high minus low variable and is calculated on the basis of the market ratio. Low book-to-market stocks are also referred to as growth stocks, while high book-to-book-to-market stocks are commonly named as value stocks. The dataset is retrieved from the official data library3 of Kenneth French and corresponds with the daily returns used in the empirical research. The 3 control variables used are based on the daily European market factors.

3_{The data consists of the daily developed market factors and returns on the European market retrieved from:}

(14)

14

3.2 Research Method

First of all the two firm groups are established by tracking the market value data of all 600 constituents listed on the Stoxx 600. To investigate whether there are differences regarding the January Effect between high and small cap, only the top and low 10% market valued companies are considered. The in total 120 companies are selected by taking the first and last decile of the arithmetic mean market values over the entire time period. All the market values are denoted in euros calculated by the daily conversion rates.

𝜇𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 =

𝛴𝐷𝑎𝑖𝑙𝑦 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 𝑁

µmarket value = Average market value

Daily market value = Share price multiplied by shares outstanding N = Number of trading days

Afterwards adjusted closed daily stock prices, denoted in euros, are collected of every single company in the two groups. Adjusted closed prices are used to correct for any

distributions to the shareholders in terms of dividends or rights offerings because these are the most accurate prices. Based on these stock prices, daily returns for all companies are calculated according to the following formula.

𝑅_𝑡 =𝑃𝑡− 𝑃𝑡−1 𝑃_𝑡−1

Rt = Daily return

Pt = Price at time t

Pt-1 = Price at time t-1

To test for seasonality in the stock market returns for the small and high cap firms an ordinary least squares regression is used. The regression is set up with 12 dummy variables and equals 1 if the stock return was obtained in a particular month of the year. They will equal 0 if the stock return was not obtained in a particular month. Because the monthly periods are not the only factors driving the stock prices, 4 control variables are added from the Fama and French factor model.

(15)

15 The regression formula is as follows:

𝑅_𝑡 = 𝛽₀+ 𝛽₁𝐽𝐴𝑁 + 𝛽₂𝐹𝐸𝐵 + 𝛽₃𝑀𝐴𝑅 + 𝛽₄𝐴𝑃𝑅 + 𝛽₅𝑀𝐴𝑌 + 𝛽₆𝐽𝑈𝑁 + 𝛽₇𝐽𝑈𝐿 + 𝛽₈𝐴𝑈𝐺 + 𝛽9𝑆𝐸𝑃 + 𝛽10𝑂𝐶𝑇 + 𝛽11𝑁𝑂𝑉 + 𝛽12𝐷𝐸𝐶 + 𝛽13𝑀𝑅𝐾𝑅𝐹 + 𝛽14𝑆𝑀𝐵 + 𝛽₁₅𝐻𝑀𝐿 + 𝛽₁₅𝑅𝐹

Rt = Daily return

JAN, FEB, MAR = Dummy variables of months MRKRF = Daily market return minus risk-free rate

SMB = Small cap minus big cap HML = High minus low book-to-market ratio

RF = Risk-free rate

This regression is tested for different time frames to analyze the effect on the January Effect possibly caused by the financial crisis. These time series will consist of the years 2000 – 2015, 2000 – 2008 and 2008 – 2015.

To conclude if a seasonal effect is present an F-test is used according to the following hypotheses:

𝐻_𝑜: 𝛽₁ = 𝛽₂ = ⋯ = 𝛽₁₂ 𝐻1: 𝛽1 ≠ 𝛽2≠ 𝛽3 ≠⋯ ≠ 𝛽12

So the null hypothesis states that returns should be indifferent between all months, where the alternative hypothesis states that the returns in January are significantly higher compared to the rest of the year.

The volatility of the stocks during different months is analyzed using the Share Ratio. Because if a stock bears more risk, investors wants to be compensated for this in terms of returns. The Sharpe Ratio is calculated by the following equation:

𝑆ℎ𝑎𝑟𝑝𝑒 𝑅𝑎𝑡𝑖𝑜 =𝑅𝑡− 𝑅𝑓 𝜎_𝑡

Rp = Daily return

Rf = Risk-free rate

(16)

16

4. Results

In this section the results are presented testing the existence of the January Effect on the European market. First of all the regression results of the different time frames are briefly analyzed prior to the according F-tests, to test whether the beta of the January variable is significantly higher than the other variables corresponding to the rest of the year. Ultimately, the Sharpe Ratios of the high and low cap will be reviewed and tested to check for seasonal volatility diversities. All regressions contain a robustness check to correct in the event of outliers or heteroscedasticity.

4.1 January Effect

The regression results of the 60 small cap companies in the different time frames is listed below and thereafter the regression table of the 60 high cap companies:

Table 4

Regression table small capitalized companies

Daily Return 2000 - 2015 2000 - 2008 2008 - 2015

Regression _Robust Regression _Robust Regression _Robust

Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

January 0.0017583*** 0.000545 0.0009159 0.000711 0.0007063 0.000786 February 0.0012332* 0.000542 (omitted) 0.0004463 0.00077 March 0.0005803 0.000525 -0.0003809 0.000694 -0.0005008 0.000768 April 0.000659 0.000522 -0.0003572 0.000684 -0.0002541 0.000764 May 0.0003758 0.000573 -0.0013001 0.000758 0.0001705 0.000841 June -0.0003332 0.000533 -0.0014781* 0.000717 -0.0010772 0.000769 July 0.0004783 0.00052 -0.0010702 0.000685 -0.0000391 0.000756 August 0.0009757 0.000562 -0.0003254 0.000701 0.0003595 0.000844 September (omitted) -0.0018829* 0.000768 -0.0002391 0.000751 October 0.0007451 0.000569 -0.0005496 0.000726 0.0001241 0.000857 November 0.0005649 0.00056 -0.0007764 0.0008 (omitted) December 0.0002208 0.000521 -0.0012166 0.000682 -0.000275 0.000762 Market premium 0.0070849*** 0.00017 0.0055945*** 0.000291 0.0078353*** 0.000255

Small minus big 0.0022808*** 0.000328 0.0016064*** 0.000416 0.0025907*** 0.000602

High minus low -0.0007185* 0.000291 -0.0018138*** 0.000428 -0.0015431*** 0.000464

Risk-free rate -0.0277065* 0.013601 -0.0188745 0.018334 -0.0953214 0.06343

Constant 0.0002127 0.000389 0.0015074 0.000562 0.0009209 0.000541

R2 _0.5664 _0.4288 _0.6482

Observations 4174 2086 2088

* p < 0.05, ** p <0.01, *** p < 0.001

(17)

17 The first column of regression table 4 shows that there is evidence for the January Effect over the complete time frame. The regression coefficient is highly significant different from zero and the seasonality seems to continue, although with less magnitude, during February. There is not enough evidence for any other calendar effects but investing in June is the least

profitable since it is the only month bearing a negative return. Along with the expectations the control variables are a large influence on the daily returns, especially the market premium and small minus big variables. The latter presumes that small capitalized firms outperform the big capitalized firms, because the coefficient is highly significant different from zero and positive. The significant negative coefficient of the risk-free rate can be explained by the fact that when interest rates increases, investors have more incentives to purchase bonds instead of stocks and will cause a negative effect on the share prices. However, it seems that there are some factors ignored in the regression because only 57% of the regression is explained by the independent variables.

The second column reveals that returns in the run-up to the financial crisis are almost entirely negative. The exemption from this is the month January and thus could imply that the January Effect holds even when the stock market is falling. The discrepancy between the returns in December and January could be an explanation for the seasonality to start from the end of the year due to tax-loss selling. Repeatedly, the control variables are highly significant different from zero and thus the main factors influencing the returns instead of the calendar effects. In this period even a lower percentage (43%) is explained by the independent variables. The reason for this may be that there are more externalities having an impact on returns such as bankruptcies and investor’s confidence.

In the last phase, the post-crisis period, the stocks partly recover and generate positive effects on returns in some months. In the most recent years there is no significant evidence for calendar effect, neither negative nor positive return effects. Beside this, January’s positive return impact compared to the previous month December is again much higher. Moreover, the last model seems to predict the outcomes of the daily returns better than the prior time frame because the determination coefficient is 65%.

(18)

18

Table 5

Regression table high capitalized companies

Daily Return 2000 - 2015 2000 - 2008 2008 - 2015

Regression Robust Regression Robust Regression Robust

Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

January .0017423*** 0.0004212 .0006836 0.0005108 .0015062** 0.0006074 February .0010728** 0.0003974 (omitted) .000894 0.0005514 March .0004578 0.0004023 -.0005054 0.0005172 .0000914 0.0005704 April .0007488 0.0003967 -.0005757 0.0005129 .0008179 0.0005491 May .0005228 0.0004146 -.0006725 0.0005522 .000553 0.000569 June .0001442 0.0003921 -.0007831 0.0005213 -.0001347 0.0005393 July .0005891 0.0003875 -.0006726 0.0005058 .0005632 0.0005334 August .0007126 0.0004224 -.0006295 0.0004902 .0007468 0.000645 September (omitted) -.0017772** 0.0005691 .0003646 0.0005439 October .0005612 0.0004836 -.000285 0.0005122 .0001272 0.0007914 November .0002259 0.000386 -.0007418 0.0004951 (omitted) December .0003525 0.0003816 -.0008833 0.0004417 .0002798 0.0005707 Market premium .0056498*** 0.0001517 .0047671*** 0.0002347 .0058675*** 0.0003036

Small minus big -.0098683*** 0.0002785 -.0104066*** 0.000323 -.0095315*** 0.0006048

High minus low .0006359 0.0003898 -.0008958** 0.0003202 .0011951 0.0008178

Risk-free rate -.0064602 0.0105912 -.0080815 0.0138143 -.0949072** 0.0431127

Constant -.0004141 0.0002955 .0009473 0.0003933 -.0002779 0.0003831

R2 _0.8483 _0.8491 _0.852

Observations 4174 2086 2088

* p < 0.05, ** p <0.01, *** p < 0.001

Some values are omitted due to multicollinearity

While observing the regression table for the high cap companies, there are a lot of similarities compared to table 4. First of all, the complete time frame represents a significant January Effect with an extension into February. Secondly, the monthly effects on the daily return for the pre-crisis period are negative except for January. Above all, the January coefficient is the largest compared to the other months.

But there are some differences compared to the small cap results. The post-crisis period only has one month with a negative influence on the daily return, while in the same period on small cap stocks there are 6 months. This could be an effect of investors investing in bigger capitalized stocks after the crisis because they are less volatile compared to small capitalized stocks and thus resulting in higher returns. Furthermore, the goodness of fit of the model for the high capitalized stocks is better with an overall coefficient of 85%. This could mean that there are less externalities in returns compared to the small cap stocks.

(19)

19 To check for both regression models if the January variable is significantly different from the other dummy variables an F-test is performed on the complete time frame as follows: 𝐹 = 𝑆𝑆𝑅𝑟− 𝑆𝑆𝑅𝑢𝑟 𝑞 𝑆𝑆𝑅_𝑢𝑟 𝑛 − 𝑘 − 1

SSRr = Sum squared of residuals unrestricted model

SSRur = Sum squared of residuals restricted model

q = Amount of restrictions n = Number of observations

k = Number of independent variables in the unrestricted model

Here the unrestricted model represents the regression stated in the methodology section, and the unrestricted model omits all monthly dummy variables. If the outcome of the F-test is not significant this means that there is too little evidence to reject het null hypothesis, and thus the January Effect is not significantly different from the other monthly effects.

Table 6

Test for seasonal effects on daily returns

Small cap High cap

firms firms

F-statistic 2.04 2.56

p-value 0.0211 0.0032

It seems that, against expectations, there is more difference in seasonality effect among high market value stocks compared to the low market value stocks. Although both tests reject the null hypothesis at a 5% significance level, the difference in seasonal effect is more plausible when investing in high cap stock.

4.2 Sharpe Ratio

As mentioned before, the research would be biased if results are only tested on daily returns. Because bearing risk gives a level of uncertain income, investors want to be compensated for this exposure. For both the high and low cap stocks there is an observable January Effect by

(20)

20 looking at table 4 and 5. It could be the case that not only daily returns are noticeably higher, but also the volatility of the stocks when trading in January.

The Sharpe Ratio is a measure which scales these returns taking into account the amount of risk of a portfolio and the outside option of investing in risk-free investments such as bonds-, and government securities. The outcome of the Sharpe Ratio reflects the return an investor earns by being exposed to one unit of risk, where a higher Sharpe Ratio is preferable. In the table below Sharpe Ratios of both the small and high cap companies are listed per month.

Table 7

Average monthly Sharpe Ratios from 2000 - 2015

January February March April May June July August September October November December

High cap firms 0.0281 0.0814 0.0516 0.0971 0.0088 -0.0444 0.0739 0.0146 0.0292 0.1393 0.0328 0.0874

Small cap firms 0.2707 0.2824 0.1176 0.1705 0.0696 -0.0293 0.0824 0.1406 0.0580 0.1778 0.1558 0.2130

There is a notable difference between the Sharpe Ratios of both firm groups. The average yearly Sharpe Ratio of the high cap firms equals 0.05 against a 0.14 average of the small cap firms. Secondly, it is expected, under the assumption that volatility is equal throughout the year, that the results from table 4 and 5 hold. However, the month January is underperforming in the Sharpe Ratio test for high cap stocks. In contrast with the regression, October earns the most return per unit of risk under the high cap firms. This means that the stocks that underperform compared to the Sharpe Ratios are exposed to more risk. The risk-free rate has been reasonably stable and low, and thus have little influence in these results.

The reason that the Sharpe Ratios of the small cap firms is almost three times higher could either mean that returns of these stocks exceed high cap share returns or that small cap stocks are less exposed to risk. The latter could imply that the biggest firms were exposed the most to the financial crisis and therefore had more volatile returns compared to the smaller companies of the Stoxx 600. Furthermore, for both groups it is meritorious to invest in

February instead of January by looking at the Sharpe Ratios. Due to intensive stock trading at the beginning of the year as a result of tax-loss selling, volatility could be higher in January compared to February and therefore lead to a lower Sharpe Ratio.

(21)

21

5. Conclusion

In this research study the aim was to investigate whether there is a perceptible January Effect on the European market, where returns tend to be regularly higher in the first month of the year. The grandiosity of this analysis is to search for a seasonal pattern to attain an alpha for investors. In extension of this, literature research has been done to determine by what such seasonality is caused.

In advance it was predicted that, although observable firm information is almost widely available and tax-loss selling incentives are lower compared to the American environment, there still is a noticeable January Effect on the European market. Further

reasons for a possible seasonality could be window dressing and increased investor optimism. Finally a lack of diversification due to the home equity bias of traders could be a source for not correcting the seasonal effect.

The empirical results show with high significance that there is a present January Effect in both firm groups. The ordinary linear regression displays additionally that the effect of the January dummy on the daily returns is mostly larger than any other month. However, for the small cap firm group there seems to be more externalities due to a low goodness of fit. For any further research, it is advisable to draft a regression including more control variables that influence daily returns and have a better representing model. When comparing the model for both firms groups it stands out that determination coefficient of the high cap group is almost 30 percent points larger. In line with the hypothesis, the January Effect starts affecting daily returns in December, where presumably losing stocks are sold to generate tax losses. In both firm groups the January dummy is remarkably larger than the December coefficient.

Now that it is known that there is evidence for a January Effect on the European market, it would be interesting to for further research to see which factors of the reviewed literature have a big influence on the seasonality. This could be done by adding data on tax-loss selling, window dressing and investor optimism to the model. Additionally, a larger firm composition could give a better view on the January Effect because small firm groups have exposure to idiosyncratic risk. Finally, a comparison of similar time frames in different market environments on the January Effect would filter common factors affecting seasonality and therefore reveal different causes.

(22)

22

References

Branch, B. (1977) A Tax Loss Trading Rule. The Journal of Business, Vol. 52, No. 2, pp. 198-207

Ciccone, S. J. (2011) Investor Optimism, False Hopes and the January Effect. Journal of

Behavioral Finance, Vol. 12, pp. 158-168

Cooper, I. and Kaplanis, E. (1994) Home Bias in Equity Portfolios, Inflation Hedging, and International Capital Market Equilibrium. The Review of Financial Studies, Vol. 7, No. 1, pp. 45-60

Coval, J. D. and Moskowitz, T. J. (1999) Home Bias at Home: Local Equity Preference in Domestic Portfolios. The Journal of Finance, Vol. 54, No. 6, pp. 2045-2073

Fama, E. F. (1970) Efficient Capital Markets: A Review of Theory and Empirical Work. The

Journal of Finance, Vol. 25, No. 2, pp. 383-417

Fama, E. F. and French, K. R. (1992) Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, Vol. 33, pp. 3-56

Fama, E. F. and French, K. R. (1992) The Cross-Section of Expected Stock Returns. The

Journal of Finance, Vol. 47, No. 2, pp 427-465

French, K. R and Poterba, J. M. (1991) Investor Diversification and International Equity Markets. National Bureau of Economic Research, No. 3609

Harding, M. (2013) Taxation of Dividend, Interest, and Capital Gain Income. OECD

Taxation Working Papers, No. 19

Haugen, R. A. and Lakonishok, J. (1988) The Incredible January Effect: The Stock Market’s Unsolved Mystery. Homewood, Illinois: Dow Jones-Irwin.

Jensen, M. C. (1978) Some Anomalous Evidence Regarding Market Efficiency. Journal of

Financial Economics, Vol. 6, pp. 95-101

Musto, D. K. (1997) Portfolio Disclosures and Year-End Price Shifts. The Journal of

Finance, Vol. 52, No. 4, pp. 1563-1588

Officer, R. R. (1975) Seasonality in Australian Capital Markets. Journal of Financial

Economics, Vol. 2, pp. 29-51

Ogden, J. P. (1990) Turn-of-Month Evaluations of Liquid Profits and Stock Returns: A Common Explanation for the Monthly and January Effects. The Journal of Finance, Vol. 45, No. 4, pp. 1259-1272

(23)

23 Praetz, P. D. (1973) A Spectral Analysis of Australian Share Prices. Australian Economic

Papers, Vol. 12, Issue 20, pp. 70-78

Reinganum, M. R. (1983) The Anomalous Stock Market Behavior of Small Firms in January.

Journal of Financial Economics, Vol. 12, pp. 89-104

Rozeff, M. S. and Kinney Jr., W. R. (1976) Capital Market Seasonality: The Case of Stock Returns. Journal of Financial Economics, Vol. 3, pp. 379-402.

Starks, L. T., Yong, L., Zheng, L. (2006) Tax-Loss Selling and the January Effect: Evidence from Municipal Bond Closed-End Funds. Journal of Finance, Vol. 61, No. 6, pp. 3049-3067

Wachtel, S. B. (1942) Certain Observations on Seasonal Movements in Stock Prices. Journal

(24)

24

Appendix A: Combined Statutory Tax Rate

Figure 1 table Firm compositions Regressions Sharpe ratios

(25)

25

Appendix B: Firm Group Compositions

Firm group compositions ranked from high to low market value:

Rank Small Market Capitalization High Market Capitalization

1 Rheinmetall (Xet) Bp

2 Howden Joinery Gp. Hsbc Hdg. 3 Spirax-Sarco Engr. Nestle 'R'

4 Rotork Novartis 'R'

5 Castellum Vodafone Group

6 Huhtamaki Total

7 Britvic Glaxosmithkline

8 Aalberts Industries Royal Dutch Shell A 9 Grafton Group Uts. Roche Holding

10 Essentra Plc Sanofi

11 Wh Smith Eni

12 Sydbank Telefonica

13 Georg Fischer 'R' Siemens (Xet)

14 Cofinimmo Edf

15 Berendsen Nokia

16 Carillion Banco Santander

17 Wirecard (Xet) Deutsche Telekom (Xet) 18 Gerresheimer (Xet) Astrazeneca

19 Shaftesbury Ubs Group

20 Hexpol 'B' Anheuser-Busch Inbev

21 Micro Focus Intl. Bnp Paribas 22 Swedish Orphan Biovitrum L'Oreal

23 Amer Sports Allianz (Xet)

24 Asm International Royal Bank Of Sctl.Gp. 25 Aareal Bank (Xet) Orange

26 Btg Daimler (Xet)

27 Fabege Sap (Xet)

28 Genmab British American Tobacco

29 Atkins (Ws) Barclays

30 Deutsche Euroshop (Xet) Lloyds Banking Group

31 Jm Engie

32 Eurofins Scientific Statoil 33 Plastic Omnium E On (Xet) 34 Ultra Electronics Hdg. Ing Groep

35 Ubisoft Entm. Lvmh

36 Bovis Homes Group Bbv.Argentaria 37 Moneysupermarket Com Gp. Rio Tinto

(26)

26

39 Victrex Basf (Xet)

40 Viscofan Axa

41 Billerud Korsnas Glencore

42 Dorma Kaba Hold Diageo

43 Temenos Group Credit Suisse Group N 44 Fuchs Petrolub Pf. (Xet) Enel

45 Beazley Deutsche Bank (Xet)

46 Intrum Justitia Unilever Dr

47 Rubis Ericsson 'B'

48 Austriamicrosystems Unicredit 49 Nibe Industrier 'B' Inditex

50 Duerr (Xet) Bhp Billiton

51 Dialog Semicon. (Xet) Bt Group 52 Unibet Group Sdb Tesco

53 Centamin Anglo American

54 Domino'S Pizza Group Societe Generale

55 Rpc Group Novo Nordisk 'B'

56 Paysafe Group Vivendi

57 Fastighets Balder 'B' Assicurazioni Generali 58 Sartorius Pref. (Xet) Bmw (Xet)

59 Gvc Holdings Unilever (Uk)

(27)

27

Appendix C: Regression Tables

Regression table of small capitalized firms 2000 – 2015:

Linear regression Number of obs 4174

F( 15, 4158) 211.78 Prob > F 0 R-squared 0.5664 Root MSE 0.00717 Robust

Daily return Coefficient Std. Err. t P>t [95% Conf. Interval]

January 0.0017583 0.0005454 3.22 0.0010 0.000689 0.0028275 February 0.0012332 0.0005418 2.28 0.0230 0.000171 0.0022954 March 0.0005803 0.0005251 1.11 0.2690 -0.0004492 0.0016099 April 0.000659 0.0005221 1.26 0.2070 -0.0003646 0.0016827 May 0.0003758 0.0005726 0.66 0.5120 -0.0007467 0.0014983 June -0.0003332 0.0005325 -0.63 0.5320 -0.0013772 0.0007108 July 0.0004783 0.0005204 0.92 0.3580 -0.000542 0.0014987 August 0.0009757 0.0005619 1.74 0.0830 -0.0001258 0.0020773 September (omitted) Oktober 0.0007451 0.0005692 1.31 0.1910 -0.0003709 0.0018611 November 0.0005649 0.0005596 1.01 0.3130 -0.0005322 0.001662 December 0.0002208 0.000521 0.42 0.6720 -0.0008006 0.0012422 Market premium 0.0070849 0.0001703 41.61 0.0000 0.006751 0.0074187 Small minus big 0.0022808 0.0003281 6.95 0.0000 0.0016375 0.002924 High minus low -0.0007185 0.0002906 -2.47 0.0130 -0.0012882 -0.0001488 Risk-free rate -0.0277065 0.0136013 -2.04 0.0420 -0.0543723 -0.0010406 Constant 0.0002127 0.0003892 0.55 0.5850 -0.0005503 0.0009757

(28)

28 Regression table of small capitalized firms 2000 – 2008:

January 0.0009159 0.0007105 1.29 0.198 -0.0004775 0.0023093 February (omitted) March -0.0003809 0.0006938 -0.55 0.583 -0.0017416 0.0009798 April -0.0003572 0.0006842 -0.52 0.602 -0.0016989 0.0009846 May -0.0013001 0.0007579 -1.72 0.086 -0.0027865 0.0001862 June -0.0014781 0.0007171 -2.06 0.039 -0.0028845 -0.0000718 July -0.0010702 0.0006848 -1.56 0.118 -0.0024132 0.0002728 August -0.0003254 0.0007008 -0.46 0.643 -0.0016998 0.0010491 September -0.0018829 0.0007678 -2.45 0.014 -0.0033886 -0.0003772 Oktober -0.0005496 0.0007259 -0.76 0.449 -0.0019732 0.0008739 November -0.0007764 0.0008002 -0.97 0.332 -0.0023456 0.0007929 December -0.0012166 0.0006817 -1.78 0.074 -0.0025534 0.0001202 Market premium 0.0055945 0.0002906 19.25 0 0.0050247 0.0061643 Small minus big 0.0016064 0.0004163 3.86 0 0.0007899 0.0024229 High minus low -0.0018138 0.0004277 -4.24 0 -0.0026526 -0.0009751 Risk-free rate -0.0188745 0.0183344 -1.03 0.303 -0.0548303 0.0170812 Constant 0.0015074 0.000562 2.68 0.007 0.0004053 0.0026095

(29)

29 Regression table of small capitalized firms 2008 – 2015:

January 0.0007063 0.0007859 0.9 0.369 -0.0008349 0.0022475 February 0.0004463 0.0007695 0.58 0.562 -0.0010628 0.0019554 March -0.0005008 0.0007675 -0.65 0.514 -0.0020058 0.0010043 April -0.0002541 0.000764 -0.33 0.739 -0.0017524 0.0012442 May 0.0001705 0.0008411 0.2 0.839 -0.001479 0.0018199 June -0.0010772 0.000769 -1.4 0.161 -0.0025853 0.0004308 July -0.0000391 0.0007564 -0.05 0.959 -0.0015224 0.0014442 August 0.0003595 0.0008437 0.43 0.67 -0.0012951 0.0020142 September -0.0002391 0.0007512 -0.32 0.75 -0.0017122 0.0012341 Oktober 0.0001241 0.0008567 0.14 0.885 -0.001556 0.0018043 November (omitted) December -0.000275 0.0007615 -0.36 0.718 -0.0017683 0.0012184 Market premium 0.0078353 0.0002553 30.69 0 0.0073345 0.008336 Small minus big 0.0025907 0.0006018 4.31 0 0.0014105 0.0037708 High minus low -0.0015431 0.0004639 -3.33 0.001 -0.0024529 -0.0006334 Risk-free rate -0.0953214 0.0634303 -1.5 0.133 -0.2197152 0.0290725 Constant 0.0009209 0.0005412 1.7 0.089 -0.0001404 0.0019822

(30)

30 Regression table of high capitalized firms 2000 – 2015:

January .0017423 0.0004212 4.14 0 0.0009165 0.002568 February .0010728 0.0003974 2.7 0.007 0.0002935 0.001852 March .0004578 0.0004023 1.14 0.255 -0.0003309 0.0012466 April .0007488 0.0003967 1.89 0.059 -0.0000289 0.0015264 May .0005228 0.0004146 1.26 0.207 -0.0002901 0.0013356 June .0001442 0.0003921 0.37 0.713 -0.0006244 0.0009129 July .0005891 0.0003875 1.52 0.129 -0.0001707 0.0013489 August .0007126 0.0004224 1.69 0.092 -0.0001154 0.0015407 September (omitted) Oktober .0005612 0.0004836 1.16 0.246 -0.0003869 0.0015094 November .0002259 0.000386 0.59 0.558 -0.0005309 0.0009827 December .0003525 0.0003816 0.92 0.356 -0.0003956 0.0011007 Market premium .0056498 0.0001517 37.24 0 0.0053523 0.0059472 Small minus big -.0098683 0.0002785 -35.4 0 -0.0104144 -0.0093223 High minus low .0006359 0.0003898 1.63 0.103 -0.0001283 0.0014001 Risk-free rate -.0064602 0.0105912 -0.61 0.542 -0.0272246 0.0143043 Constant -.0004141 0.0002955 -1.4 0.161 -0.0009933 0.0001652

(31)

January .0006836 0.0005108 1.34 0.181 -0.0003182 0.0016854 February (omitted) March -.0005054 0.0005172 -0.98 0.329 -0.0015197 0.0005089 April -.0005757 0.0005129 -1.12 0.262 -0.0015816 0.0004302 May -.0006725 0.0005522 -1.22 0.223 -0.0017555 0.0004105 June -.0007831 0.0005213 -1.5 0.133 -0.0018056 0.0002393 July -.0006726 0.0005058 -1.33 0.184 -0.0016645 0.0003192 August -.0006295 0.0004902 -1.28 0.199 -0.0015908 0.0003318 September -.0017772 0.0005691 -3.12 0.002 -0.0028933 -0.000661 Oktober -.000285 0.0005122 -0.56 0.578 -0.0012896 0.0007195 November -.0007418 0.0004951 -1.5 0.134 -0.0017127 0.0002291 December -.0008833 0.0004417 -2 0.046 -0.0017496 -0.000017 Market premium .0047671 0.0002347 20.31 0 0.0043069 0.0052274 Small minus big -.0104066 0.000323 -32.2 0 -0.01104 -0.0097733 High minus low -.0008958 0.0003202 -2.8 0.005 -0.0015237 -0.0002679 Risk-free rate -.0080815 0.0138143 -0.59 0.559 -0.035173 0.0190099

(32)

January .0015062 0.0006074 2.48 0.013 0.0003151 0.0026974 February .000894 0.0005514 1.62 0.105 -0.0001873 0.0019752 March .0000914 0.0005704 0.16 0.873 -0.0010271 0.00121 April .0008179 0.0005491 1.49 0.137 -0.000259 0.0018948 May .000553 0.000569 0.97 0.331 -0.0005629 0.0016689 June -.0001347 0.0005393 -0.25 0.803 -0.0011924 0.0009231 July .0005632 0.0005334 1.06 0.291 -0.0004829 0.0016094 August .0007468 0.000645 1.16 0.247 -0.0005181 0.0020118 September .0003646 0.0005439 0.67 0.503 -0.0007021 0.0014313 Oktober .0001272 0.0007914 0.16 0.872 -0.0014248 0.0016792 November (omitted) December .0002798 0.0005707 0.49 0.624 -0.0008393 0.0013989 Market premium .0058675 0.0003036 19.33 0 0.005272 0.0064629 Small minus big -.0095315 0.0006048 -15.8 0 -0.0107176

-0.0083454 High minus low .0011951 0.0008178 1.46 0.144 -0.0004087 0.002799 Risk-free rate -.0949072 0.0431127 -2.2 0.028 -0.1794559

-0.0103585 Constant -.0002779 0.0003831 -0.73 0.468 -0.0010291 0.0004733

The January effect on the European market