Dutch Individual Investor Performance*
Trading, Gender, and Overconfidence
Master thesis: Erik van Aalzum**June 26, 2014
ABSTRACT
This paper examines the impact of trading on individual investor performance using a random sample of Dutch retail investors. The results show that the average individual investor pays a performance penalty for active trading. Overconfidence can explain the high trading levels and the resulting poor performance. Psychological research shows that in finance, men are more overconfident than women and therefore trade more excessively. This research shows that men execute 46.2% more trades than woman and the turnover ratio of men is 52.9% higher of that of women. The only risk adjusted underperformance is for the investment accounts opened by men. Therefore men trade more than women and by trading more, hurt their performance more.
JEL Classification: G11, G12, G14, G24
Keywords: Individual investor performance, overconfidence, brokerage
*I would like to thank the brokerage firm for providing the data for this thesis. I would especially like to thank Bert Hodes for his efforts and support during the beginning of my thesis. Thijs Wille, thanks for providing me with the data by writing al the queries. Also many thanks to my supervisor Marc Kramer for numerous discussions and helpful comments during the writing of my thesis.
2
1. Introduction
The total equity investments for the Dutch individual investor were €32.1 billion in 20131. More than two thirds (68.3%) of these investments consisted of securities
issued by Dutch companies. The average Dutch individual equity investor trades his equity frequently. The turnover ratio2 was 70.8% in 2012 (WorldBank). The Dutch
Central Bank (DNB) documents that between March 2009 and March 2010 the total gross transactions (all sales and purchases) in listed stocks by the Dutch individual investor had a value of €30 billion. Despite having such large ownership in equities and transactions, there is little research on the return performance of equities held by Dutch individual investors. One of the few papers concerning this subject is from Bauer, Cosemans, and Eichholtz (2009) who show that equity and option trading hurts the average Dutch individual investor, however this research does not take into account any differences in trading activity between individual investors. Barber and Odean (2000) show that trading hurts the performance of the average U.S. individual investor. They show that the individual investors who trade frequently hurt their net performance significantly more than those who trade infrequently. The authors state that overconfidence can explain the high trading levels and the resulting poor performance. Odean (1999) states that U.S. individual investors with discount brokerage accounts on average reduce their returns by trading, even when trading costs are ignored. Psychological research shows that men are more prone to overconfidence than women (Lundeberg, Fox, and Punocachar 1994). Barber and Odean (2001) also analysed the individual activities and performance and sorted the U.S. individual investor on gender. They state that men trade 45% more and that trading hurts their net return significantly more than women.
This paper attempts to shed light on the common stock portfolio performance held by Dutch individual investors. Two competing theories of trading are tested. The first theory is from Grossman and Stiglitz (1980) who use a rational expectation framework. They state that the rational investor only executes a trade when the
1 http://www.statistics.dnb.nl/huishoudens/index.jsp
2 The WorldBank calculates the turnover ratio as the total value of shares traded during the period divided
3
potential benefit is equal or higher than the cost of the trade. Contrary, the theoretical models in the field of behavioural finance suggest that the individual investor suffers from overconfidence when trading. This implies that the individual investor overestimates the precisions of his or her knowledge and abilities and that trading will adversely affect portfolio returns for overconfident investors.
The main question in this thesis is: Does the Dutch individual equity investor suffers from overconfidence? To answer this question four sub questions are answered. (1) How does the average Dutch individual equity investor perform? (2) What is the impact of trading for the performance of the Dutch individual equity investor? (3) Do men trade more than women? (4) Do men hurt their performance more than women?
To answer the first en second question an approach identical to Barber and Odean (2000) is followed. I use monthly position statements and trading records of 1,726 individual investors for the years 2012 and 2013 and combine this with common stocks gross returns obtained from Reuters Datastream. I calculate gross and monthly returns for every individual investor and classify on turnover ratio and number of trades. Question 3 and 4 are answered by classifying the individual investor returns on the gender of the person who opened the account.
4
This thesis differs from current research in two ways. First, the research is not based on U.S. individual investors but on a unique data set of Dutch individual investors. Cross-cultural differences in overconfidence are present. Respondents in China exhibit higher degrees of overconfidence than respondents in the United States or Japan (Yates, Lee, Shinotskuka, Patalano and Sieck 1998). Acker and Duck (1998) show that, by using a stock market game, Asian students are consistently more overconfident than British students. It remains unknown what the differences in overconfidence between U.S. and Dutch individual investors are since no research is found with respect to these cultures. Bauer, Cosemans, and Eichholtz (2009) do examine Dutch individual investor performance but do not sort on trading activity, put more focus on the option investor and use a completely different methodology then this thesis.
Second, the research from Barber and Odean (2000) and Bauer, Cosemans, and Eichholtz (2009) is performed on data from 1991 until 1997 and 2000 until 2006 respectively whereas this thesis uses a more recent data sample (2012-2013). In 2008 the global financial crisis started and research shows that individual investor behavioural is influenced by this event. Kuo, Huang and Jane (2013) state that after the financial crisis the average Taiwanese individual investor became more risk-averse, traded less, and were more inclined to realize losses. Dorn and Weber (2013) state that the financial crisis of 2008 affects the composition of the German investors’ equity portfolio. They shift their equity towards individual stocks instead of actively managed funds. For the investors who make this shift end up with portfolios that are 30% riskier in terms of volatility compared to their pre-crisis positions.
5
2. Literature review
This section introduces the theoretical background and describes the outcome of important articles. Section 2.1 discusses the empirical findings of overconfidence. In section 2.2, I discuss the efficient market hypothesis whereas section 2.3 discusses the rational expectation framework view from Grossman and Stiglitz (1980), which has a strong link with the efficient market hypothesis. Section 2.4 describes the main empirical findings of individual investor characteristics and performance.
2.1 The overconfidence view
6
think that they are above average in terms of stock picking or performance (but did not have better than average past performance) trade more. Research also shows that individual investors who trade excessively, share the same features. Anderson (2013) shows that individual investors with lower income, wealth, age, and education trade excessively indicating that they lack investment expertise.
With respect to gender, men are generally more overconfident than women (Lundeberg, Fox, and Punocochar 1994). Lewellen, Lease, and Schlarbaum (1997) show that women rely more on their brokers, believe that returns are less highly predictable and make fewer transactions. Barber and Odean (2001) show that men trade more than women and by trading more, men hurt their performance more than women do.
Other studies conclude that overconfidence is not the only possible reason of excessive trading. Kumar (2009) concludes that individual investors invest disproportionately more in stocks with higher idiosyncratic volatility, higher skewness, and lower prices even though these stocks have lower mean returns. Dorn and Sengmueller (2009) state that investors who report that they enjoy gambling turn over their portfolios twice the rate of their peers. Grinblatt and Keloharju (2008) show that investors most prone to sensation seeking trade more frequently.
2.2 The efficient market hypothesis
7
The semi-strong-form of the EMH implies that when (economic) news arises, this news spreads quickly and the information is immediately incorporated in the stock prices. This spreading of information is so fast that no excess returns can be earned by trading on that information and therefore fundamental analysis will not be able to generate excess returns. The last and most inclusive form is the strong-form of the EMH. In this form of efficiency, share prices reflect all information, public and private and no one can earn excess returns. Assuming that stock prices are fully efficient in reflecting information, trading will not generate positive or negative excess gross returns for the average individual investor since the price is a perfect reflection of all information public and private. Therefore active traders will earn on average negative excess net returns when trading costs and bid ask spreads are taken into account.
2.3 A rational expectations framework
Grossman and Stiglitz (1980) argue that the strong-form of the EMH is impossible because there would be no incentive to trade since no excess gross returns are generated. If no incentive is present for trading, the authors state that markets would eventually disappear. Since this is not the case they argue that the gross return for active traders should be higher than the gross return for passive traders in order to outweigh the cost of the trade. This will eventually lead to, on average, equal net returns for the passive and active investor and therefore they argue that an investor only trades when the benefit of the trade is equal or larger than the cost of the trade.
2.4 The performance and characteristics of the average individual investor
8
of international diversification. French and Poterba (1991) show that 94% of the equity portfolio of American investors is invested domestically. United Kingdom (82%) and Japan (98%) show the same results. Bauer, Cosemans, and Eichholtz (2009) show that for the investment portfolio of Dutch investors, 95% (in terms of volume) of their trades are transactions in Dutch securities.
Odean (1998) shows that by holding losers to long and selling winners to short, average individual investors earn a poor performance relative to the market. This effect is known as the disposition effect. One of the major costs of the disposition effect is the tax paid over capital gain. However, in the Netherlands, the tax authorities do not distinguish between realized or paper losses and gain. Dhar and Zhu (2006) conducted an analysis of the disposition effect at an individual level. They show that more financial sophisticated individual investors were less prone to the disposition effect. In the study of Dhar and Zhu (2006), income and professional occupation were used as proxies for sophistication. By studying trading records, Boolell-Gunesh, Broihanne, and Merli (2012) show that the disposition effect decreases over time and that this decrease is partly caused by sophistication variables. Anderson (2012) shows that lower income, wealth, age, and education increases the level of excessive trading.
9
3. Methodology
3.1 Measuring return performance
The approach that is used to calculate the gross and net returns is based on Barber & Odean (2000). The monthly gross return for investor ℎ in month 𝑡 is given as:
𝑅ℎ𝑡𝑔𝑟𝑜𝑠𝑠 = ∑ 𝑝𝑖𝑡∗
𝑠ℎ𝑡
𝑖=1
𝑅𝑖𝑡𝑔𝑟𝑜𝑠𝑠 (1) where 𝑝𝑖𝑡 is the end of the month market value for stock 𝑖 held by investor ℎ in month 𝑡 divided by the end of the month market value for all stocks held by investor ℎ in month 𝑡, 𝑅𝑖𝑡𝑔𝑟𝑜𝑠𝑠 is the total gross monthly return3 for stock 𝑖 derived from
Reuters Datastream, and 𝑠ℎ𝑡 is the number of stocks held by investor ℎ in month 𝑡. For stock 𝑖 in month 𝑡, the monthly net return including transaction costs is given as:
(1 + 𝑅𝑖𝑡𝑛𝑒𝑡) = (1 + 𝑅𝑖𝑡𝑔𝑟𝑜𝑠𝑠) ∗ (1 − 𝑐𝑖𝑡
𝑠)
(1 + 𝑐𝑖,𝑡−1𝑏 ) (2)
Where 𝑐𝑖𝑡𝑠 is the cost of sales for stock 𝑖 divided by the sales price for stock 𝑖 in
month 𝑡 and 𝑐𝑖,𝑡−1𝑏 is the cost of purchase for stock 𝑖 divided by the purchase price
for stock 𝑖 in month 𝑡 − 1. The cost of sales and purchases consists of two variables: the commission which need to be paid to the broker and the bid-ask spread.
The market impact of trading is not taken into account since the trades of most individual investors are relatively small, so their impact is likely to be limited. This is in line with Bauer, Cosemans, and Eichholtz (2009) who make the same assumption. The commission cost is calculated as the price paid on commissions divided by the sale/purchase price. The best measure to calculate the bid-ask spread is to take the absolute difference between the actual price paid/received for stock 𝑖
3 The return index for every stock is used assuming that dividends are re-invested to purchase additional
10
at time 𝑡 with the then last known stock price for stock 𝑖 at time 𝑡 and divide this value by the actual price paid/received for stock 𝑖 at time 𝑡. Since this data is not available the methodology used by Barber and Odean (2000) seems like a good estimate. They estimate the bid-ask spread component of transaction costs for purchase and sales as:
𝑠𝑝𝑑𝑟𝑑𝑠 = (𝑃𝑑𝑠 𝑐𝑙 𝑃𝑑𝑠𝑠− 1) and 𝑠𝑝𝑑𝑟𝑑𝑏= − ( 𝑃𝑑𝑐𝑙𝑏 𝑃𝑑𝑏𝑏− 1) (4) Where 𝑃𝑑𝑐𝑙𝑠 and 𝑃 𝑑𝑏
𝑐𝑙 are the reported closing prices (instead of last know stock price
just before the trade) from the Center for Research in Security Prices (CRSP) daily stock return files on the day of a sale and purchase respectively, and 𝑃𝑑𝑠𝑠 and 𝑃
𝑑𝑏
𝑏 are
11
by open-end mutual funds form 1966 to 1993. The bid-ask spread estimate in this thesis will use the average values of the paper of Barber and Odean (2000) and Carhart (1997) and use these values as the basis for the stocks that are listed on the midcap index of the Netherlands. The large cap index stocks are given a reduction in bid-ask spread by a factor of 0.75 where the not large cap and midcap stocks are given a raise of bid-ask spread by a factor of 1.25.
Barber and Odean (2000) also state that this bid-ask spread estimate includes any market impact that might result from a trade but they do not further elaborate on that aspect and I find it therefore trivial how this bid-ask spread estimate does take into account the market impact of a trade. But in line with Bauer, Cosemans, and Eichholtz (2009) the market impact of a trade is neglected since the impact is assumed not present and/or insignificant.
When the net return of stock 𝑖 in month 𝑡 is calculated by including the direct and indirect costs when a trade occurred for stock 𝑖 in month 𝑡, the stock returns for investor ℎ can be summed up for month 𝑡 to come up with a monthly net common stock portfolio return for investor ℎ in month 𝑡 as:
𝑅ℎ𝑡𝑛𝑒𝑡= ∑ 𝑝𝑖𝑡∗ 𝑠ℎ𝑡 𝑖=1
𝑅𝑖𝑡𝑛𝑒𝑡 (5)
12
The average gross and net monthly returns in month 𝑡 are then calculated as:
𝑅𝐴𝑣𝑡𝑔𝑟𝑜𝑠𝑠 = 1 𝑛ℎ𝑡 ∑ 𝑅ℎ𝑡𝑔𝑟𝑜𝑠𝑠 𝑛ℎ𝑡 ℎ=1 and 𝑅𝐴𝑣𝑡𝑛𝑒𝑡 = 1 𝑛ℎ𝑡 ∑ 𝑅ℎ𝑡𝑛𝑒𝑡 (6) 𝑛ℎ𝑡 ℎ=1
Where 𝑛ℎ𝑡 is the number of accounts with stocks invested in month 𝑡.
3.2 Risk adjusted return performance
The monthly gross and net returns are attributed to different risk and style factors to obtain the abnormal performance. This thesis only examines common stocks that are listed in the Netherlands and therefore the risk and style factors are constructed for the Dutch market.
First, Jensen’s alpha is estimated by regressing the monthly gross and net return on the market return given by the capital asset pricing model:
𝑅𝐴𝑣𝑡− 𝑅𝑓𝑡 = 𝛼𝑖 + 𝛽𝑖(𝑅𝑚𝑡− 𝑅𝑓𝑡) + 𝜖𝑖𝑡 (7)
Here 𝑅𝑓𝑡 is the monthly return on the 10 year yield to maturity Dutch government bond, 𝛼𝑖 is the investors excess return, 𝛽𝑖 is the beta, 𝑅𝑚𝑡 is the monthly return of the MSCI Netherlands equity index, 𝜖𝑖𝑡 is the error term.
Second, the Fama and French (1993) three factor model is used to adjust the average investor return not only to the market but also to size and book to market style tilts:
13
The size factor is characterized by the return of the MSCI Netherlands large cap index4 minus the return of the MSCI Netherlands small cap index5. The book to
market factor is characterized by the return of the MSCI Netherlands value index6
minus the return of the MSCI Netherlands growth7 index.
4. Data
The data for this thesis is provided by a bank in the Netherlands. The bank provided me with a random sample of 1,726 investors including the end of the month stock positions and trading records of their Dutch listed stocks starting at January 2012 until December 2013. The only restriction in the sample selection is that the account is not closed during the sample period. Due to a recent migration of a new system, older data could not be obtained.
Only looking at Dutch listed stocks means that not the entire portfolio of common stocks is taken into account however the bank shows that 91% of all investments are domestic stocks. Bauer, Cosemans, and Echholtz (2009) also show that the majority of transactions in their sample of Dutch investors are transactions in Dutch securities. In terms of volume this was 95%.
The data from the bank consists of two different files. The first file is an overview of end of the month common stock positions for every account. The second file consists of an overview of all trades that were executed by the 1,726 individual investors including the commissions. A trade can be seen as buying or selling a stock and this is possible in two different ways. An individual investor can buy or sell a stock with money from the brokerage account or the individual investor can reinvest the received dividend. Dividend can be paid out in three different ways namely stock, cash or optional (the investor can choose if it wants to receive cash or
4 The MSCI Netherlands large cap index is given by the first 70% of stocks based on their full market
capitalization.
5 The MSCI Netherlands small cap index is given by the last 15% of stocks based on their full market
capitalization.
6 The MSCI Netherlands Value Index depends on the book value to price ratio, the 12 month forward
earnings to price ratio and the dividend yield.
7 The MSCI Netherlands Growth Index depends on the long-term forward earnings per share growth rate,
14
stocks) dividend. Receiving cash or optional dividend and reinvesting the dividend can be seen as a choice of the investor to reinvest and therefore this action is seen as a trade. When dividend is paid out in stocks there was no real choice of willing to trade by the individual investor since receiving stock dividend is the only option. Therefore this reinvestment is not seen as a real trade.
For every individual investor the zip-code, gender, date of birth, and client group is present. The different client groups are linked with the total amount of wealth (cash or securities) an individual investor has at the bank. The bank separates its clients into four groups. Mass clients have a total wealth of no more than €75,000, whereas Personal Banking clients have a total wealth of more than €75,000. Private Banking clients have a total invested wealth (only securities) of more than €500,000 or a total wealth of more than €1,000,000, whereas Wealth Management clients have a total wealth of more than €25,000,000. In total there are 1120 Mass clients, 580 Personal Banking clients, 24 Private Banking clients, and 2 Wealth Management clients in the data sample.
15
Table 1
Descriptive statistics on investor accounts and trades
Table 1 presents the descriptive statistics for a sample of 1,726 individual accounts at a Dutch bank. The sample period is from January 2012 to December 2013. The variables are defined as follows: Gender and Age are the gender and age of the primary account holder. Trades is the total number of trades per account during the sample period. Turnover is total value of all purchases and sales divided by the sum of the portfolio values. Income is the average yearly disposable household income assigned to investors based on their four digits zip-code. The table shows for each variable the mean,
median, and standard deviation, as well as the 5th, 25th, 75th, and 95th percentile values.
Mean Std dev 5th 25th Median 75th 95th
Male Gender (%) 67.94 Age (years) 56.28 14.33 33 47 56 66 80 Trades (#) 5.14 32.07 0 0 0 2 20 Turnover (%) 13.88 87.35 0 0 0 2.95 44.75 Income (€) 36,591 7,048 26,500 31,800 36,300 40,500 47,700 Stocks (#) 2.2 1.81 1 1 1 3 6 Cost of trade (%) 4.26 78.0 0.1 0.2 0.4 0.8 50.0 Trade value (€) 17,443 43,076 3 1,080 2,900 9,340 107,440
16
indicating a less skewed distribution. Bauer, Cosemans, and Echholtz (2009) however show a mean turnover ratio of 23.68% and a median turnover ratio of 0% indicating a more skewed distribution but also shows that the at least 50% of the individual investors do not execute any trade. Only domestic stocks held by the individual investor are taken into account leading to the fact that the turnover ratio of the total portfolio is higher, however the bank shows that 91% all investments are domestic stocks.
The Central Agency for Statistics (CBS) calculated the average yearly disposable household income per zip-code area. I assign this yearly disposable household income to the first four digits of the investors’ zip-code. For 29 individual investors no income could be assigned since their zip-codes were not in the CBS data file. I calculate the average stocks in a portfolio as the average of monthly stocks held in the investors’ portfolio. With an average of 2.2 stocks and a median of 1 it can be concluded that a lot of idiosyncratic risk is present in the portfolio of the average individual investor. This is in line with the findings of Blume and Friend (1975) who find that individual investors based in the U.S. hold only one or two stocks. Kelly (1995), Polkovnichenko (2005) and Goetzmann and Kumar (2008) also confirm the poor diversification of U.S. individual portfolios.
The cost of a trade is given by the commission of a trade divided by the gross value of that trade. Where the average cost of a trade is 4.26%, the median value is only 0.4% and the 75th percentile is only 0.8% meaning that there are a few trades who
17
telephone, the individual investor pays €8 + 0.24% with a maximum of €150 for a trade executed through the internet.
A third data file is obtained from Reuters Datastream. This file includes the total return per month for 122 individual stocks that are listed in the Netherlands. Taking the total return includes all dividends and corrects for (reverse) stock splits. The monthly gross return for month 𝑡 is then calculated as the total return at month 𝑡 minus the total return at month 𝑡 − 1 divided by the total return at month 𝑡 − 1. Also the MSCI Netherlands equity return index, the MSCI Netherlands large cap index, the MSCI Netherlands small cap index, the MSCI Netherlands value index and the MSCI Netherlands growth index are retrieved from Reuters Datastream. From the last four indexes the Fama & French HML and SMB factors are created as described in the former section.
5. Results
5.1 Average investor performance
18
Figure 1
Net returns vs. Number of trades
Figure 1 shows a scatterplot of my sample of individual investors by looking at the number of trades and the annualized geometric mean gross return. The number of trades is given on a logarithmic scale. The black horizontal line shows the annualized geometric mean return of the MSCI Netherlands equity index.
Where the MSCI Netherlands equity index generated an annualized geometric mean return of 24.88%, the annualized geometric mean gross return for the average individual investor was only 13.75%. When commission costs and bid-ask spreads are taken into account this average return lowers to 12.67%.
36.6% of the individual investors outperform the MSCI Netherlands equity index but when transaction costs are taken into account, 34.1% outperforms the market. 11.9% of the individual investors beat the market after costs with at least 10% annually, however 49.1% underperforms the market after costs with at least 10% annually.
19
Table 2 shows the risk adjusted gross and net returns for the average individual investor. I conclude that there is a negative excess return present for the market adjusted return, CAPM and Fama French three factor model intercepts, although there is only a significant negative excess return for the net return on the Fama French three factor model. The average investor underperforms the market significantly on the 10% level with 67 basis points (8.3% annually) adjusted for beta, book to market and size.
The average individual investor tilt towards value and small stocks since the HML and SMB factors are positive and reliably different from zero. So in line with the existing literature, the average investors underperforms the market however there are individual investors that outperforms the market.
Table 2
Percentage monthly abnormal return measures for the average individual investor
Table 2 shows the returns which are based on month ultimo position statements of 1726 individual investors with a time span from January 2012 to December 2013. The upper table presents the gross percentage monthly returns and the lower table presents the net percentage monthly returns. The CAPM is a time-series regression of the average individual excess return on the market excess return. Fama French is a time-series regression of the average individual excess return on the market excess return, the high minus low factor and the small minus big factor. The p-values are given in the brackets where *, ** and *** are the significance levels at the 10, 5 and 1% level respectively.
Alpha Rmt - Rft HMLt SMBt adj. R2
Gross Percentage Monthly Returns for the Average Investor
Market adjusted -0.657 Return (0.595) CAPM -0.644 0.993 64.9 (0.324) (0.965) Fama French -0.391 0.807* 0.454*** 0.300*** 90.1 (0.289) (0.061) (0.000) (0.000)
Net Percentage Monthly Returns for the Average Investor
20
5.2 Sorting on trading activity
In the first part of this thesis two contradictory models are described, the overconfidence model and the rational expectation framework. In this part, I provide the empirical evidence of the testing of these competing models. To test these models the sample of individual investors is sorted on the number of trades and their turnover ratio, leading to an absolute (amount of total trades) and a relative (turnover ratio) measurement as a proxy for trading activity. Taking the amount of trades as a classification criterion can lead to a portfolio size bias since one can assume that large portfolio consists of more stocks and therefore individual investors with larger portfolios perform more trades. Since there are a few very active trades who also have a large portfolio size, there is indeed a portfolio size bias when looking at the mean amount of trades. However using the median amount of trades no correlation between portfolio value and number of trades is present. Also there is no strong positive linear relation between the average amount of stocks in the individual investor portfolio and the total number of trades (r = 0.07).
5.2.1 Sorting on the number of trades
Since only a small amount of individual investors trade very often, the classified groups differ in size. The criteria of being an active trader is that the individual investor on average trades ones a month implying a minimum of 24 executed trades for the sample period. 75 individual investors meet this criterion.
The group of individual investors that do trade but no more than ones per month on average is significantly larger than the active traders group. No distinction is made in this group since the maximum difference in the amount of trades is only 22 and it is therefore hard to state that a subgroup of this group of individual investors is significantly more active. This group contains the passive traders.
21
groups. Group 0 consists of 959 individual investors who did not execute any trade and earn an average gross and net annualized geometric mean return of 17.0%. Since the bank is a listed firm on the AEX index the individual investor can invest in it. 187 (10.8%) only invest in the stock of the bank and do not execute any trade. When these individual investors are left out, an average gross and net annualized geometric mean return of 13.8% is earned. Group 1 are all individual investors who traded and therefore are active in the stock market but have not traded more than one trade per month on average. These are the passive traders. This group contains 692 individual investors. Their average gross and net annualized geometric mean return is 9.7% and 7.6% respectively. Group 2 contains all investors who trade ones per month or more and consists of 75 individual investors. I see this group as the active traders. Their average gross and net annualized geometric mean return is 9.7% and 4.0% respectively. This group consist of 75 individual investors. Even though the group size of the active traders is only 75, the difference in net return for the passive and active trader is significantly different on the 10% level.
Figure 2
Individual investor return classified on the number of trades
Figure 2 shows the average gross annualized geometric mean return and the average net annualized geometric mean return distributed over the amount of trades the individual investor has executed. Group 0 contains all individual investors who did not trade during the sample period. Group 0* excludes the investors who only hold the stock of the bank and do not trade. Group 1 contains all individual investors who traded, but no more than one trade per month on average. Group 2 contains all individual investors who at least traded ones per month on average.
17.0% 13.8% 9.7% 9.7% 17.0% 13.8% 7.6% 4.0% 0 0* 1 2
22
5.2.2 Sorting on mean turnover ratio
Taking the absolute amount of trades as a proxy for the amount of trading activity, one does not take the impact of the trade compared to the total portfolio value into account. Second, no equal group size of passive and active traders can be used in this case since only a very small amount of individual investors do trade very often. Therefore I also use the turnover ratio as a proxy for trading. Although the turnover ratio also has its drawbacks as described in section 4, this ratio takes the impact of a trade relative to the total portfolio value into account. Also a better distinction can be made between the individual investors in terms of group size. Figure 3 shows the average gross and net annualized mean returns classified on turnover ratio. The turnover ratio is calculated as the total value of all purchases and sales during the sample period divided by the sum of the monthly portfolio values of the individual investor. Where group 0 and 0* are identical to figure 2 since no trading means zero turnover, the other groups are different in terms of classification and size. Group 1, 2 and 3 represent individual investor tertiles classified on monthly turnover where group 1 turns over their portfolio the least.
Figure 3
Individual investor return classified on the average turnover ratio
Figure 3 shows the average gross annualized geometric mean return and the average net annualized geometric mean return based on average turnover. Group 0 is the group of investors who did not during the sample period. Group 0* excludes the investors who only trade in the stock of the bank. The other groups represent individual investor tertiles based on monthly turnover where group 1 turns over their portfolio the least and group 3 turns over their portfolio the most.
17.0% 13.8% 10.0% 12.6% 6.5% 17.0% 13.8% 7.3% 10.5% 3.9% 0 0* 1 2 3
23
Where Barber and Odean (2000), who only use the turnover ratio as a proxy for trading, show that the gross raw return for the most passive and the most active investor does not significantly differ from each other, this cannot be concluded for this research.
Barber and Odean (2000) show that the 20% of households that trade most often earn the worst returns. This is in line with Figure 2 and 3 who show indeed a significant lower net return for the most active group based on amount of trades and turnover ratio. However a decreasing net return with an increasing turnover ratio is not found.
Table 2 provides the risk adjusted gross and net returns on a monthly basis based on the same classifications as is done in figure 2 and 3. The market adjusted return, CAPM and Fama & French three-factor model are used as risk adjusted models. The coefficient estimates are based on the Fama & French three-factor model where the dependent variable is the gross monthly return. No statistical significant difference is found between coefficient estimates of the passive and active individual investors or between the high and low turnover tertiles.
24
based on the mean turnover ratio. When the turnover ratio increases, the risk adjusted gross underperformance decreases. Where the prediction of the Grossman and Stiglitz model suggests a higher risk adjusted gross return for the active trader this thesis does not support this. Barber and Odean (2000) also show no consistency with the Grossman and Stiglitz model. Looking at the gross returns for both classifications one can conclude that by increasing the amount of trades or turnover ratio the gross returns are worse. This is in line with Odean (1999) who also concludes that investors lower their returns trough trading even when trading costs are ignored but not in line with Barber and Odean (2000) who show equal gross returns for every quintile classified on turnover ratio.
By including trading costs the differences in the risk adjusted underperformance of the active and the higher turnover investors becomes larger and more significant although no statistical difference is found between the long short portfolio of being short in the active portfolio and long in the passive portfolio or being short in the high turnover tertile and long in the low turnover tertile.
Where the passive individual investor underperforms the market with 76 basis points adjusted for beta, book to market and size although not statistical significantly different from zero, the active individual investor underperforms the market with 131 basis points statistical significant on the 1% level. On an annual basis this will lead to an underperformance of 9.5% for the passive investor and no less than 16.9% for the active investor. However the long short portfolio for the Fama & French three factor model shows no significant difference.
25
Table 3
Descriptive statistics, gross returns, and net returns of common stock investments for the individual investor classified on the number of trades and
the mean turnover ratio
Table 3 shows the returns which are based on month ultimo position statements of 1726 individual investors with a time span from January 2012 to December 2013 classified on the amount of actual trades and turnover. The passive portfolio contains the individual investors who have traded one but no more than 23 times, the active portfolio contains the individual investors who have traded more than 23 times. The 1726 individual investors are also sorted into tertiles based on their mean turnover ratio. Tertile 1 contains the individual investors with the lowest turnover, whereas tertile 3 contains the individual investors with the highest turnover ratio. The coefficient estimates are those from a time-series regression of the gross average household excess return on the market excess return, the high minus low factor and the small minus big factor. The raw return is the average monthly return for the average individual investor. The market adjusted return is the monthly return on the individual investor minus the monthly return on the MSCI Netherlands Equity Index. The CAPM is a time-series regression of the average individual excess return on the market excess return. Fama French is a time-series regression of the average individual excess return on the market excess return, the high minus low factor and the small minus big factor. The p-values are given in the brackets where *, ** and *** are the significance levels at the 10, 5 and 1% level respectively.
Zero Trades Number Of Trades Average Turnover Ratio
Passive Active Active-Passive 1 2 3 3-1
Descriptive Statistics and Coefficient Estimates
Mean monthly 0 11.66 211.77 N.A. 0.17 4.66 89.4 N.A.
turnover (%)
Mean total trades 0 3.57 85.29 81.72 2.00 4.15 28.58 26.58
Median total trades 0 2 46 44 2 2 7 5
Number of 959 692 75 N.A. 256 256 255 N.A.
investors Rmt - Rft 0.819** 0.777 0.927 0.150 0.693 0.838* 0.843 0.150 (0.042) (0.152) (0.521) (0.340) (0.347) (0.097) (0.210) (0.673) HMLt 0.535*** 0.337** 0.499*** 0.162 0.242 0.419*** 0.404*** 0.163 (0.000) (0.019) (0.000) (0.302) (0.401) (0.000) (0.001) (0.605 SMBt 0.164* 0.458*** 0.575*** 0.117 0.718** 0.285*** 0.407*** -0.311 (0.078) (0.009) (0.000) (0.531) (0.046) (0.009) (0.005) (0.413) adjusted R2 92.2 73.2 89.3 0.16 89.7 90.6 85.7 6.7
Gross Percentage Monthly Returns
Raw Return 1.151 0.994 0.971 -0.023 1.054 1.169 0.752 -0.302
Market adj. Return -0.536 -0.744 -1.182 -0.438 -0.397 -0.788 -1.118 -0.783
(0.407) (0.343) (0.177) (0.507) (0.766) (0.211) (0.113) (0.545)
CAPM -0.524 -0.755 -1.162 -0.406 -0.428) -0.783 -1.118 -0.750
(0.410) (0.327) (0.177) (0.531) (0.744) (0.206) (0.107) (0.555)
Fama French -0.227 -0.566 -0.882** -0.316 -0.290 -0.549 -0.952** -0.663
(0.469) (0.320) (0.046) (0.628) (0.809) (0.127) (0.046) (0.615)
Net Percentage Monthly Returns
Raw Return 1.151 0.838 0.535 0.303 0.864 1.012 0.548 -0.316
Market adj. Return -0.536 -0.953 -1.611* -0.659 -0.711 -0.961 -1.386* -0.675
(0.407) (0.238) (0.070) (0.326) (0.605) (0.133) (0.065) (0.604)
CAPM -0.524 -0.960 -1.592* -0.632 -0.732 -0.955 -1.384* -0.652
(0.410) (0.227) (0.068) (0.337) (0.588) (0.129) (0.061) (0.610)
Fama French -0.227 -0.760 -1.314*** -0.554 -0.570 -0.719* -1.157** -0.588
26
5.3 Sorting on gender
Table 4 presents the raw and risk adjusted gross and net returns held by men and women. The literature shows that, in areas such as finance, men are generally more overconfident than women. To be in line with this psychological research, I state that men should trade more than women and by trading more, men hurt their net performance more than women. Table 4 shows that men indeed trade significantly more than women. For all accounts, men trade 46.2% more than woman and the turnover ratio of men is 52.9% higher. Taking only the single registered accounts (one accountholder) into account, men trade even more (52.8%) than women and have an even higher (71.0%) turnover ratio. Barber and Odean (2001) see that as a confirmation that gender serves as a proxy for overconfidence. The reason for this is that the authors state that for the double registered accounts one accountholder can make or influence decisions made by the other account holder.
27
Table 4
Number of trades, average turnover ratio, and return performance of common stock investments for the individual investor classified on gender
Table 4 shows the returns which are based on month ultimo position statements of 1726 individual investors with a time span from January 2012 to December 2013 classified on gender. The accounts are classified as female or male based on the gender who opened the account. A single registered accounts implies that there is only one person who owns the account whereas a double registered account implies an account which belongs to two persons. The coefficient estimates and returns are identical calculated as is described in table 3.
All accounts Single registered account Double registered account
Women Men Difference Women Men Difference Women Men Difference
Descriptive statistics and Coefficient Estimates
Number of trades 3.9 5.7 -1.8* 3.6 5.5 -1.9 4.6 5.9 -1.3
(0.098) (0.115) (0.282)
Mean monthly 10.239 15.601 -5.362* 9.374 15.910 -6.536* 11.903 15.178 -3.278
turnover (%) (0.081) (0.066) (0.325)
Number of 559 1167 N.A. 358 674 N.A. 201 493 N.A.
investors Rmt - Rft 0.817* 0.802* 0.015 0.831* 0.799** 0.032 0.793 0.806 -0.014 (0.086) (0.054) (0.683) (0.080) (0.032) (0.406) (0.107) (0.069) (0.654) HMLt 0.451*** 0.456*** -0.004 0.440*** 0.430*** 0.010 0.472*** 0.490*** -0.018 (0.000) (0.000) (0.870) (0.000) (0.000) (0.778) (0.000) (0.000) (0.498) SMBt 0.326*** 0.287** 0.038 0.352*** 0.302*** 0.050 0.280** 0.268** 0.011 (0.006) (0.011) (0.221) (0.002) (0.004) (0.230) (0.044) (0.034) (0.723) adjusted R2 89.6 90.1 8.2 91.4 91.5 12.8 85.0 87.9 5.9
Gross Percentage Monthly Returns Classified On Gender
Raw Return 1.323 1.259 0.064 1.370 1.227 0.100 1.238 1.303 -0.065
Market adj. Return -0.617 -0.661 0.044 -0.582 -0.662 0.080 -0.681 -0.660 -0.021
(0.365) (0.316) (0.682) (0.383) (0.293) (0.579) (0.346) (0.349) (0.849)
CAPM -0.613 -0.660 0.046 -0.577 -0.663 0.086 -0.678) -0.0654 -0.024
(0.360) (0.310) (0.664) (0.379) (0.285) (0.546) (0.340) (0.346) (0.823)
Fama French -0.362 -0.406 0.044) -0.332 -0.423 0.092 -0.416 -0.382) -0.034
(0.345) (0.270) (0.683) (0.340) (0.204) (0.524) (0.372) (0.364) (0.762)
Net Percentage Monthly Returns Classified On Gender
Raw Return 1.121 1.056 0.065 1.175 1.023 0.152 1.023 1.103 -0.081
Market adj. Retun -0.939 -0.962 0.023 -0.899 -0.956 0.058 -1.01 -0.971 -0.041
(0.203) (0.174) (0.837) (0.216 (0.156) (0.710) (0.191) (0.201) (0.712)
CAPM -0.925 -0.952 0.027 -0.884) -0.949 0.066 -0.100 -0.956 -0.044
(0.202) (0.171) (0.809) (0.216) (0.152) (0.666) (0.189) (0.200) (0.688)
Fama French -0.651 -0.679* 0.027 -0.616 -0.693* 0.0764 -0.714 -0.660 -0.054
28
5.4 Investor sophistication
Research shows that lower income, wealth, age, and education increases the level of excessive trading (Anderson 2012). This thesis uses beginning portfolio size, and income as proxies for sophistication. To test if sophisticated individual investors are less prone to excessive trading, portfolios of individual investors classified on portfolio size and income are created.
5.4.1 Sorting on portfolio size
The individual investors are classified into quintiles based on their beginning portfolio value on January 2012. Every quintile consists of 345 individual investors except the fifth quintile which consists of 346 individual investors. The descriptive statistics show that the largest mean portfolio value is €34,555 whereas the smallest mean portfolio value is only €139. When the portfolio value increases the mean monthly turnover ratio and the mean total trades also increase. However the median values do not show this relation. The coefficient estimates reveal that small portfolios have a tilt towards low beta, small, value stocks. Barber and Odean (2000) also find that small portfolios tilt towards small value stocks however they find a decreasing beta when portfolio value increases.
29
Table 5
Descriptive statistics, gross returns and net returns of common stock investments for the individual investor classified on beginning portfolio value
Table 5 shows the returns which are based on month ultimo position statements of 1726 individual investors with a time span from January 2012 to December 2013 classified into quintiles based on the portfolio value on January 2012. Quintile 1 contains the individual investors with the smallest portfolio value, whereas quintile 5 contains the individual investors with the largest portfolio value. The coefficient estimates are those from a time-series regression of the gross average household excess return on the market excess return, the high minus low factor and the small minus big factor. The raw return is the average monthly return for the average individual investor. Fama French is a time-series regression of the average individual excess return on the market excess return, the high minus low factor and the small minus big factor. The p-values are given in the brackets where *, ** and *** are the significance levels at the 10, 5 and 1% level respectively.
Portfolio Size
1 2 3 4 5 5-1
(small) (large)
Descriptive Statistics and Coefficient Estimates
Mean portfolio value 139 816 2,471 6,403 34,355 N.A.
Mean monthly turnover (%) 9.1 5.1 15.1 18.21 22.0 12.9**
(0.037)
Median monthly turnover (%) 0.000 0.000 0.005 0.000 0.294 0.294
Mean total trades 0.9 1.4 5.1 7.5 10.9 10***
(0.000)
Median total trades 0 0 1 0 1 N.A.
Rmt - Rft 0.698*** 0.739* 0.815 0.883 0.899 0.201*** (0.003) (0.087) (0.237) (0.264) (0.137) (0.006) HMLt 0.467*** 0.609*** 0.500*** 0.414*** 0.281*** -0.187*** (0.000) (0.000) (0.001) (0.000) (0.000) (0.004) SMBt 0.351*** 0.283*** 0.311* 0.276** 0.279*** -0.072 (0.001) (0.000) (0.068) (0.019) (0.000) (0.311) adjusted R2 89.6 82.5 80.1 89.6 94.7 4.8
Gross Percentage Monthly Returns Classified On Portfolio Value
Raw Return 1.321 1.119 1.075 1.450 1.433 0.112
Fama French -0.153 -0.477 -0.621 -0.350 -0.357 -0.203
(0.640) (0.384) (0.283) (0.364) (0.155) (0.413)
Net Percentage Monthly Returns Classified on Portfolio Value
Raw Return 1.270 1.048 0.971 1.358 1.344 0.074
Fama French -0.212 -0.564 -0.747 -0.457 -0.461 -0.249
30
5.4.2 Sorting on income
The individual investors are classified into quintiles based on their income which is retrieved from the CBS and linked to the investor’s first four digits of their zip-code. The quintiles consist of 339 or 340 individual investors. Since 29 individual investors cannot be assigned an income due to a missing zip-code, these quintiles are a fraction smaller than for the classification on beginning portfolio value.
31
Table 6
Descriptive statistics, gross returns and net returns of common stock investments for the individual investor classified on income
Table 6 shows the returns which are based on month ultimo position statements of 1726 individual investors with a time span from January 2012 to December 2013 classified into quintiles based on the income retrieved from the CBS and linked to the investor’s first four digits of their zip-code. Quintile 1 contains the individual investors with the lowest income, whereas quintile 5 contains the individual investors with the highest income. The coefficient estimates are those from a time-series regression of the gross average household excess return on the market excess return, the high minus low factor and the small minus big factor. The raw return is the average monthly return for the average individual investor. Fama French is a time-series regression of the average individual excess return on the market excess return, the high minus low factor and the small minus big factor. The p-values are given in the brackets where *, ** and *** are the significance levels at the 10, 5 and 1% level respectively.
Income
1 2 3 4 5 5-1
(Low) (High)
Descriptive Statistics and Coefficient Estimates
Mean income 27,832 32,868 36,232 39,676 46,480 N.A.
Mean monthly turnover (%) 14.2 13.9 9.3 8.6 22.4 8.2
(0.158)
Median monthly turnover
(%) 0.000 0.000 0.000 0.000 0.000 0.000
Mean total trades 4.3 5.9 5.1 3.4 6.8 2.5
(0.200)
Median total trades 0 0 0 0 0 N.A.
Rmt - Rft 0.774**. 0.816 0.823 0.8073** 0.814* 0.040 (0.024) (0.100) (0.109) (0.047) (0.075) (0.286) HMLt 0.494*** 0.392*** 0.496*** 0.416*** 0.474*** -0.020 (0.000) (0.001) (0.000) (0.000) (0.000) (0.546) SMBt 0.305*** 0.350*** 0.263** 0.298*** 0.276*** -0.030 (0.000) (0.006) (0.029) (0.000) (0.000) (0.457) adjusted R2 91.0 87.6 89.3 90.7 90.0 10.8
Gross Monthly Returns Classified On Income
Raw Return 1.221 1.187 1.167 1.167 1.205 -0.016
Fama French -0.299 -0.366 -0.401 -0.355 -0.325 -0.026
(0.394) (0.365) (0.316) (0.306) (0.367) (0.412)
Net Monthly Returns Classified On Income
Raw Return 1.127 1.092 1.074 1.080 1.121 -0.006
Fama French -0.400 -0.470 -0.497 -0.448 -0.410 -0.010
32
5.5 Cross-sectional analysis of investor performance and characteristics
I estimate a cross-sectional analysis between investor performance and their characteristics. I perform this analysis to check if the results from the time series analysis are robust in the cross-section. This cross-sectional analysis uses the annualized geometric mean gross and net returns, and the gross and net Sharpe ratio8 for every individual investor as the dependent variables. The gross and net
Sharpe ratios are the risk adjusted performance measurements. This sample consists of 1682 individual investors instead of 1726. All 17 accounts that have an account holder which is younger than 18 are left out. Another 27 accounts are left out since no income could be assigned to these individual investors. I include the following investor characteristics as independent variables: Turnover ratio, which is the total value of all purchases and sales during the sample period divided by the sum of the monthly portfolio values of the individual investor; active, a dummy variable equal to one if at least one trade is executed; single woman, a dummy variable equal to one if the account is opened by a woman and held only by a woman; gender, a dummy variable equal to one if the account is opened by a woman; number of trades, which is the total number of trades the individual investor has executed during the sample period; age, which is the age of the individual investor who opened the account; income, which is the yearly income assigned to investors based on their first four digits of their zip-code; and portfolio
value which is the investors’ portfolio market value of January 2012.
The results of this analysis confirm the earlier findings. I find a negative relation between the individual investors’ turnover ratio and their net performance. The estimated dummy active is highly significant and indicates that the individual investor portfolio return for inactive investors is between 7.9 and 11.4 percent higher per year than for the individual investor who executed at least one trade. The estimated dummies gender and single female show a positive relation with
8 The Sharpe ratio 𝑆
𝑖 is calculated as: 𝑆𝑖=
𝑅𝑖−𝑅𝑓
𝜎(𝑅𝑖), where 𝑅𝑖 is the average monthly return of investor 𝑖, 𝑅𝑓 is
the average monthly risk free rate, and 𝜎(𝑅𝑖) is the standard deviation of the average monthly return of
33
individual investor performance however insignificant. I do not find any relation between portfolio value and individual investor performance and between income and performance.
Table 7
Investor performance and their demographic characteristics
The left hand side uses raw returns as dependent variables and the right hand sight refers to the Sharpe ratio. Both gross and net returns are used in the cross-sectional regression. Each regression is estimated using data from 1682 individual investors. I estimate the cross-sectional analysis for the period between January 2012 and December 2013. The independent variables are investor characteristics. Age is the age of the individual investor who opened the investment account. Active is a dummy variable equal to one if an investor has executed at least one trade. Gender is a dummy variable equal to one if the account is opened by a woman. Single female is a dummy variable equal to one if the account is opened by a woman and it is a single registered account. Income is the yearly income assigned to investors based on their first four digits of their zip-code. Portfolio value is the investors’ portfolio market value of January 2012. Number of trades is the number of trades the individual investor has executed between January 2012 and December 2013. Turnover ratio is the the total value of all purchases and sales during the sample period divided by the sum of the monthly portfolio values of the individual investor.
Raw Return Sharpe Ratio
Gross Net Gross Net
34
6. Summary and conclusion
This research analyses the monthly returns earned on equity investments by 1726 Dutch individual investors at a Dutch bank for 2 years ending in December 2013. I sort the individual investors on trading activity and gender to test two competing theories. The first theory from Grossman and Stiglitz (1980) suggests that an individual investor only executes a trade when the benefit of the trade is equal or larger than the cost of the trade. Contrary, the theoretical models in the field of behavioural finance suggest that the individual investor suffers from overconfidence when trading. This implies that the individual investor will adversely affect their portfolio returns by trading.
Where the MSCI Netherlands equity index generated an annualized geometric mean return of 24.88%, the annualized geometric mean gross return for the average individual investor was 13.75%. When commission costs and bid-ask spreads are taken into account this average return lowers to 12.67%. The average individual investor net return, adjusted for beta, value and size show a negative excess net return of 67 basis points per month. This result is in line with Bauer, Cosemans and Eichholtz (2009) who also show a negative excess risk adjusted net return for the average Dutch equity investor.
35
By taking the mean turnover ratio as a proxy for trading activity and dividing every active (performed at least one trade) individual investor into tertiles, the lowest turnover tertile earns a risk adjusted negative excess gross return of 29 basis points and a risk adjusted negative excess net return of 57 basis points per month however both not significantly different from zero. The highest turnover tertile earns a risk adjusted negative excess gross return of 95 basis points and a risk adjusted negative excess net return of 116 basis points per month statistically significantly different from zero on the 5% level. Conclusive, the individual investor who trades most (in terms of total trades or turnover ratio) is hurt the most. This is in line with the theoretical models of overconfidence that state that trading hurts the average individual investor. Barber and Odean (2000) also show that the households that trade the most, hurt their net returns the most.
36
Overall, it can be concluded that the results that are presented in this thesis do support the behavioral finance literature on overconfidence. However, the data does have its limitations since the number of individual investors and the time span is small compared to the dataset as used by Barber and Odean (2000) or Bauer, Cosemans and Eichholtz (2009). It must also be stated that the conversion from gross returns towards net returns is prone to possible errors since the bid-ask spread is an approximation based on average values from previous research. The best measure to calculate the bid-ask spread is to take the absolute difference between the actual price paid/received for stock 𝑖 at time 𝑡 with the then last known stock price for stock 𝑖 at time 𝑡 and divide this value by the actual price paid/received for stock 𝑖 at time 𝑡.
37
Bibliography
Acker, D., Duck, N., 2008. Cross-cultural overconfidence and biased self-attribution. Journal of Socio-Economics 37, 1815-1824.
Anderson, A., 2013. Trading and under-diversification. Review of Finance, 1-43. Barber, B., Odean, T., 2000. Trading is hazardous for your wealth: the common stock investment performance of individual investors. Journal of Finance 55, 773-806. Barber, B. Odean, T., 2001. Boys will be boys: Gender, overconfidence, and common stock investment. Quarterly Journal of Economics 116, 261-292.
Bauer, R., Cosemans, M., Eichholtz, P., 2009. Option trading and individual investor performance. Journal of Banking & Finance 33, 731-746.
Bessembinder, H., Venkataraman, K., 2010. Bid-ask spreads: measuring trade excecution costs in financial markets. In: Cont, R., Encyclopedia of Quantitative Finance.
Blume, M., Friend, I., 1975. The asset structure of individual portfolios and some implications for utility functions. Journal of Finance 30, 585-603.
Boolell-Gunesh, S., Broihanne, M., Merli, M., 2012. Sophistication of individual investors and disposition effect dynamics. Finance 33, 9-37
Cabellé, J., Sákovics, J., 2000. Speculating against an overconfident market. Journal of Financial Markets 6, 199-225.
38
Cooper, A., Carolyn, W., Dunkelberg, W., 1988. Entrepreneurs perceived chances for success. Journal of Business Venturing 3, 97-108.
Coval, J., Hirshleifer, D., Shumway, T., 2005. Can individual investors beat the market? Unpublished working paper. Harvard University, Boston.
Dhar, R., Zhu, N., 2006. Up close and personal: investor sophistication and the disposition effect. Management Science 52, 726-740.
Dorn, D., Sengmueller, P., 2009. Trading as entertainment? Management Science 55, 591-603.
Dorn, D., Weber, M., 2013. Individual investors’ trading in times of crisis: going it alone or giving up? Unpublished working paper. Drexel University, Philadelphia. Fisher, K., Statman, M., 2002. Bubble expectations. Journal of Wealth Management 5, 17-22.
Glaser, M., Weber, M., 2007. Overconfidence and trading volume. Geneva Risk and Insurance Review 32, 1-36.
Goetzman, W., Kumar, A. 2008. Equity portfolio diversification. Review of Finance 12, 433-463.
Grinblat, M., Keloharju, M., 2009. Sensation seeking, overconfidence and trading activity. Journal of Finance 64, 549-578.
39
Kelly, M., 1995. All their eggs in one basket: portfolio diversification of US households. Journal of Economic Behaviour & Organization 27, 87-96.
Kidd, J., 1970. The utilization of subjective probabilities in production planning. Acta Psychologica 34, 338-347.
Kumar, A., 2009. Who gambles in the stock market? Journal of Finance 64, 1889-1933.
Kuo, W., Lin, T., 2013. Overconfident individual day traders: evidence from the Taiwan futures market. Journal of Banking & Finance 37, 3548-3561.
Kuo, M., Huang, P., Jane, W., 2013. The impact of the financial tsunami on investment behavior: the case of individual investors in Taiwan. Investment Management and Financial Innovations 1, 207-217.
Lundeberg, M., Fox, P., Punccohar, J. 1994. Highly confident but wrong: gender differences and similarities in confidence judgements. Journal of Educational Psychology 86, 114-121.
Odean, T., 1999. Do investors trade to much? American Economic Review 89, 1279-1298.
Oskamp, S., 1965. Overconfidence in case-study judgments. Journal of Consulting Psychology 29, 261-265.
40
Wagenaar, W., Keren, G. 1986. Does the expert know? The reliability of predictions and confidence ratings of experts. In: Hollnagel, E., Mancini, G., Woods, D., Intelligent Decision Support in Process Environments. Berlin, Germany, pp. 87-103.
Weinstein, N., 1982. Unrealistic optimism about susceptibility to health problems. Journal of Behavioral Medicine 5, 441-460.
Lewellen, W., Lease, R., Schlarbaum, G. 1977. Patterns of investment strategy and behavior among individual investors. Journal of Business 50, 296-333.
