# Stockholdings of first-time and more experienced investors

Kent Baker (Department of Finance and Real Estate, Kogod School of Business, American University, Washington, District of Columbia, USA)
Annalien De Vries (Department of Business Management, Stellenbosch University, Stellenbosch, South Africa)

ISSN: 1940-5979

Publication date: 11 June 2018

## Abstract

### Purpose

The purpose of this paper is to examine whether socio-economic factors influence portfolio composition of individual investors investing in stocks for the first time and how these factors relate to stock portfolio performance.

### Design/methodology/approach

The study uses cross-sectional time-series analysis to examine a unique and detailed data set of Swedish stockholdings.

### Findings

The results show that first-time investors do not hold diversified portfolios. They experience high market risk and, on average, underperform more experienced investors. Males have higher unsystematic risk in their portfolios than females and older investors have more diversified portfolios compared to younger investors.

### Research limitations/implications

The results show that individual investors should improve their insights by incorporating risk when investing in stocks.

### Practical implications

Given the results of this paper, the movement from defined benefit to defined contribution pension schemes in many countries raises the issue of the need to better understand and monitor the risks in stock portfolios.

### Originality/value

This study provides insights into whether socio-economic factors influence portfolio composition, the extent to which socio-economic factors and portfolio characteristics relate to portfolio returns, and whether portfolio performance between first-time and more experienced investors differs.

## Keywords

#### Citation

Baker, K., De Ridder, A. and De Vries, A. (2018), "Stockholdings of first-time and more experienced investors", Review of Behavioral Finance, Vol. 10 No. 2, pp. 146-162. https://doi.org/10.1108/RBF-11-2016-0077

### Publisher

:

Emerald Publishing Limited

## 1. Introduction

In his seminal paper, Markowitz (1952) shows that the portfolio problem represents a choice between a portfolio’s mean return and its variance. Following mean-variance optimization, investors should maximize their return for a given level of risk (i.e. variance) or minimize risk for a given level of return. This approach leads to the formulation of an efficient frontier from which investors can choose a preferred (optimal) portfolio given their risk-return preferences. In doing so, investors must consider how each stock in the portfolio co-moves with all other stocks rather than examining risks and returns on individual stocks. Consequently, investors should hold diversified portfolios rather than a single stock. Nevertheless, DeMiguel et al. (2009) report that simple investment strategies outperform more sophisticated models.

To determine whether individual investors hold diversified portfolios, the literature often relies on survey data or data from a sample of investors from a stock brokerage firm. As Campbell (2006, p. 1557) contends, “While these brokerage records are highly accurate reports of holdings and trades in individual stocks, they sample customers of the brokerage house rather than the entire population. Furthermore, these records do not necessarily represent the total wealth of even these customers, as they may have other accounts elsewhere.” Unlike the tenets of investor rationality in traditional finance, evidence from behavioral finance suggests that investors do not always act rationally. For example, Kourtidis et al. (2011) find evidence of irrational trading by individual investors suggesting that emotions, not firm fundamentals, drive decisions. Thus, some portray individual investors as being unsophisticated “noise” traders who are subject to fads and psychological biases (Kaniel et al., 2012). Apart from making less informed investment decisions, individual investors also hold undiversified portfolios and trade actively and speculatively, which occurs to their detriment (Barber and Odean, 2001). Hoffmann et al. (2010) report, using a survey of 5,500 individual investors, differences in their objectives (speculation or long-term saving) and also that investors relying on fundamental analysis outperform those who rely on technical analysis.

Our study has three main objectives involving individual investors: to determine whether socio-economic factors influence portfolio composition; to identify the extent socio-economic factors and portfolio characteristics relate to portfolio returns, and to examine the stock performance of first-time and more experienced investors’ portfolios. This study uses detailed data of stockholdings from the Central Securities Registrar in Sweden (Euroclear Sweden). It also relies on data from the Swedish Tax Agency (Skatteverket) on the annual income for all individuals who are registered as shareholders and stock prices from the economic research department at the Stockholm Stock Exchange (SSE). Our sample consists of 179,128 individuals with stockholdings in 344 different firms between 2004 and 2010.

Our results support prior studies that individual investors do not hold diversified stock portfolios. We find that males have significantly higher systematic risk in their portfolios than females and older investors have more diversified portfolios compared to younger investors. We also document that investors with high portfolio values outperform investors with low portfolio values and first-time investors underperform more experienced investors.

The remainder of the study proceeds as follows. Section 2 provides background information followed by our hypotheses. Section 3 provides a discussion of our sample data and methodology. Section 4 contains the results and discussion and Section 5 concludes.

## 2. Literature review and hypotheses

Prior research documents that individual investors make investment mistakes as a result of both cognitive and emotional biases. For example, evidence shows that these investors often do not hold well-diversified portfolios as suggested by Markowitz (1952) and are attracted to low-priced stocks (Gompers and Metrick, 2001; Dyl and Elliott, 2006; Kumar and Lee, 2006; Kumar, 2009). As discussed below, socio-economic factors also influence portfolio composition of individual investors.

### 2.1 Gender

Prior studies find that decision-making strategies differ between males and females primarily because of differences in risk-taking behavior. Women are less prone to overconfidence than are males and thus tend to be more risk averse. As a result of having less risk tolerance, women are generally more detail orientated, invest more conservatively, and trade less often than men (Arano et al., 2010). Barber and Odean (2001) support this argument by reporting that male investors, especially young, single men, tend to trade frequently because they are more confident and thus willing to take greater risks.

Investor attitudes toward risk are likely to affect their diversification decisions. According to Goetzmann and Kumar (2008), investors with high (low) risk tolerance might hold a less (more) diversified portfolio. Moreover, investors who are overconfident in their ability to interpret information intentionally choose to hold-focused and under-diversified portfolios. Based on the findings of previous studies on gender and risk-taking behavior, we propose the following hypothesis:

H1.

Males have a higher propensity to hold fewer stocks in their portfolios than females, and hence hold portfolios containing more unsystematic risk.

### 2.2 Age

Theory suggests that age should affect portfolio choice because older investors have shorter investment horizons and generally greater wealth than younger investors (Campbell and Viceira, 2002). Thus, we expect a positive relation between age and the level of portfolio diversification. Prior studies suggest that risk aversion (risk tolerance) increases (decreases) with age (Blume and Friend, 1975; Morin and Suarez, 1983; Grable and Lytton, 1998; Yao et al., 2011). According to Kumar (2009), older and more experienced investors hold less risky portfolios and trade less frequently. Goetzmann and Kumar (2008) support this view by finding that older and more experienced investors are more diversified than younger investors. Based on previous research evidence, we formulate our second hypothesis as follows:

H2.

A positive relation exists between the age of individual investors and the level of portfolio diversification.

### 2.3 Wealth

A final factor to consider in the decision-making process of individual investors is their wealth. Prior studies show that individual investors (institutional investors) dominate the holdings of low-priced (high-priced) stocks (Muscarella and Vetsuypens, 1996; Dhar et al., 2003; Dyl and Elliott, 2006). Goetzmann and Kumar (2008) report that high-income investors are better diversified than low-income investors. Studies suggest that risk aversion increases with both age and wealth (Blume and Friend, 1975; Morin and Suarez, 1983). Based on the literature, we propose the following hypothesis:

H3.

A positive relation exists between the portfolio value (wealth) of individual investors and portfolio diversification.

## 3. Data

### 3.1 Sample description

We obtain data on stockholdings from the Central Securities Registrar in Sweden (Euroclear Sweden), which monitors the ultimate stockholdings in public firms in Sweden. From this source, we gather detailed data on stock ownership for all firms listed on the SSE. The initial sample includes about 2.2 million individual investors in 375 different firms, who are recorded as a shareholder between 2002 and 2010. We chose this period because data on annual income are only available through 2010. Having access to social security numbers provides information on the birth year and gender for each individual.

We define first-time investors as those who are not recorded as a shareholder in the two-year period before we begin our analysis. For our sample in 2004, we require that the individual is not recorded as a shareholder at the end of 2003 and 2002. We repeat this screening for all years in our study. In total, this process provides an initial sample of 243,866 individual investors. We exclude investors who are younger than 18 and older than 99 years of age and also investors where the number of shares in a firm is limited to just one.

We also exclude individuals where our data, obtained from the Swedish Tax Agency, do not show any income. Finally, we require that the stocks held by investors must exist in the intersection set of the data from Euroclear Sweden and stock price data obtained from the economic research department at the SSE and Datastream. After implementing these filters, our final sample consists of holdings by 179,128 individuals in 344 firms between 2004 and 2010.

### 3.2 Portfolio value

Besides analyzing each individual’s portfolio value, we also compute the weighted average price in the portfolio to examine if the nominal stock price (or its square) is related to stock ownership. We define portfolio value of investors as the sum of total holdings in firms multiplied by its stock price at the end of each calendar year. If the investor has holdings in a firm with a different series of stocks, we compute the portfolio value accordingly. The following equation shows the calculation of portfolio value:

(1) P O R T F O L I O V A L U E = i = 1 j n i , t × P R I C E i , t ,
where ni,t is the number of stock in firm i at time t and PRICEi,t is the closing price at the end of December for each calendar year. We compute the weighted average stock price in the portfolio in a similar way by first multiplying the weight for each stock by its stock price, which we then sum to obtain the portfolio’s average stock price.

To further investigate the composition of an investor’s portfolio, we compute and examine the Herfindahl index, which is a traditional ownership concentration measure (Demsetz and Lehn, 1985; Fuertes et al., 2014). This index is constructed as follows. For each individual we first compute the portfolio value as in Equation (1). For each stock in the portfolio, we then compute the total value of the stock. Next, we divide this value by the portfolio value and then square this fraction. We repeat this procedure for each different stock in an individual’s portfolio. We compute the Herfindahl index as the sum of all these squares. For example, an equally weighted portfolio containing only three stocks has a Herfindahl index of 0.33 whereas an equally weighted portfolio of four different stocks has an index of 0.25. A portfolio with a single stock has an index of 1.

### 3.3 Portfolio performance

To investigate how stock returns are related to investor portfolios, we begin the analysis by grouping our sample of investors in several different classes such as gender, investor age, level of diversification, and portfolio value and then use calendar-time methodology. We identify the composition of each investor’s stock portfolio at the end of each calendar year where the composition at the end of 2003 is the first observation. Beginning with January 2004 and for each subsequent calendar month over the sample period, we compute the portfolio return with reinvested dividends. We then compute the equally weighted mean return across investors for each month until December 2010. The portfolios are re-balanced at the end of December each calendar year as investors enter and exit. Thus, all trades by investors and their portfolios take place in the secondary market. We estimate abnormal returns using a four-factor model, consisting of the three Fama and French (1993) factors augmented by a momentum factor (Carhart, 1997). For each investor portfolio, we regress the monthly return on the following regression model:

(2) R p , t R f , t = α p + β p [ M K T t ] + s p [ S M B t ] + h p [ H M L t ] + m p [ M O M t ] + ε p , t ,
where Rp,t−Rf,t is the monthly return on the equally weighted calendar-time portfolio of investors less the return on the three-month Treasury bill, Rf, in calendar month t; MKT is the return on the value weighted market portfolio (proxied by the OMXSPI index) minus the monthly return on the three-month Treasury bill; SMB is the difference in the return of a portfolio of small firms and a portfolio of large firms; HML is the difference in the return of a portfolio of high book-to-market (B/M) stocks and a portfolio of low B/M stocks; and MOM is the Carhart momentum factor based on ranking firms by their return over the previous 12-month period and using the return on a portfolio of high momentum stocks less the return on a portfolio of low momentum stocks[1]. The estimated intercept reflects the mean monthly abnormal return on an equally weighted portfolio for investors and corresponds to the well-known Jensen’s α in a CAPM context. We estimate the model beginning in January 2004 and ending in December 2010 reflecting 84 monthly observations.

With access to the composition of each investor’s stock portfolio and also their age and gender, the analysis allows a detailed examination of four subsamples and sorted by: gender, the age group 18-44, 45-64, and 65-99 years, level of diversification, and the value of the stock portfolio.

## 4. Results and discussion

### 4.1 Summary statistics

Table I presents descriptive statistics for our sample. As reported, the mean (median) age of the 179,128 first-time investors is 44 (42) years, consisting of 57 percent males and 43 percent females. The average annual income, expressed in 2010 prices, is SEK291,700 whereas the median annual income of SEK279,600 is slightly lower. The average portfolio value, expressed in 2010 prices, is SEK51,900 with a median value of SEK11,300[2]. The mean ratio between the portfolio value divided by income and expressed as a percentage is 24.56 percent indicating that the total market value of the average portfolio amounts to about a quarter of an individual’s income. However, a wide range exists for this measure with a value of 0.40 percent in the 10th percentile and 61.99 percent in the 90th percentile. The mean number of stocks in the portfolio for a first-time investor is 2, with a median value of 1 and a standard deviation of 2. However, only 10 percent of all investors have a portfolio consisting of at least five different stocks. These results suggest that first-time investors do not have highly diversified portfolios, which is consistent with the results by Kelly (1995).

### 4.2 Characteristics of first-time investors’ stock portfolios

Table II shows the mean, standard deviation, and median for two socio-economic variables (age and income) and three portfolio variables (portfolio value, portfolio value divided by income (P/I), and the weighted average stock price (price) grouped by the number of stocks in the portfolio). The average age of first-time investors is fairly constant but an increase in age occurs when the portfolio has six or more stocks. The median annual income is relatively stable across the groups although the highest mean income occurs for investors holding at least 11 stocks in their portfolios. The mean portfolio value ranges between SEK25,300 for the portfolio with one stock and SEK246,500 for the portfolios with at least 11 stocks. The average ratio of portfolio value to the annual income is 0.136 for the portfolio with a single stock and 1.028 for the portfolio with at least 11 stocks. Overall, the finding suggests that investors with higher incomes hold more stocks in a portfolio than those with lower incomes and also that the weighted average stock price in the portfolio decreases as the number of stocks increases.

### 4.3 Multivariate analysis

We next examine the relation between portfolio value and first-time investor characteristics as well as portfolio characteristics. To control for the possibility that the differences in portfolio composition as previously reported are due to changes in the characteristics related to age, we sort our sample into three groups based on age: 18-44, 45-64, and 65-99 years, respectively, and examine our baseline models. Table III presents the results. In model (1) the only independent variable is INCOME. As the table shows, the estimated coefficient is positive and statistically significant at the 1 percent level for all our age groups. The variable INCOME in the 65-99 age group is more sensitive to the portfolio value compared to the other age groups.

In model (2) we add the number of stocks in the portfolio (NUMBER), and in model (3) we add PRICE and (PRICE)2, AGE, and the dummy variable GENDER as additional control variables. As reported, the estimated coefficient on NUMBER is positive and statistically significant at the 1 percent level for all age groups suggesting that high portfolio values are associated with more stocks in the portfolios. The adjusted R2 increases from 0.4 to 11.1 percent for the first age group and even more in the two remaining groups after adding this control variable.

In model (3) the estimated coefficient on PRICE is positive whereas the estimated coefficient (PRICE)2 is negative and both estimates are statistically significant at the 1 percent level for all three groups. Thus, the evidence suggests that the relation between PORTFOLIO VALUE and PRICE represents an inverted U-parabola. The estimated coefficient on AGE is positive and suggests that higher age is associated with higher portfolio values. The estimated coefficient on the dummy variable reflects gender.

As Table III shows, for the first age group, the estimated coefficient is negative (−0.093) and statistically significant at the 1 percent level. For the two other groups, the estimated coefficients are positive (0.069 and 0.062, respectively) and statistically significant at the 1 percent level. To summarize, Table III confirms that portfolio value of first-time investors is related to socio-economic variables and reveals a quadratic relation between the value of the stock portfolio and the weighted average stock price in the portfolio.

### 4.4 Stock performance for portfolios

Table IV reports factor loadings and Jensen’s αs for equally weighted portfolios for first-time investors[3]. Panel A reports the results for all individual investors and also for male and female investors as well as the difference between males and females. Panel B presents the results after sorting investors into three age groups.

We first focus on the regressions that use only MKT as the independent variable. As Panel A shows, no statistically significant abnormal performance exists across investors and for our subgroups. First-time investors, on average, have stock portfolios with a β of 1.19 for females and 1.23 for males. Male investors have higher systematic risk in their portfolios compared to females, as reported in the next to last column, and the difference is statistically significant at the 10 percent level (p-value=0.076). The adjusted R2 is lower for males than females suggesting that the males have more unsystematic risk in their portfolios, a finding consistent with H1.

When using the four-factor model, as reported in the second column for all investors, equally weighted portfolio returns are associated with a statistically significant positive abnormal return of 50 basis points (p-value=0.098) a month using a heteroscedasticity consistent standard error relative to the benchmark model. The estimated monthly abnormal return corresponds to an annual abnormal return of about 6 percent. The return on the portfolio is negatively related to the SMB although the estimated coefficient is not statistically significant. The estimated coefficient on HML and MOM is negative suggesting that both the portfolio is sensitive to the B/M ratio and investors trade less than the momentum on the market. Furthermore, our results show that the portfolios for males outperform those for females as reported in the last column. Specifically, the difference, representing 20 basis points a month, is statistically significant at the 10 percent level (p-value=0.077). This portfolio is negatively related to the HML factor.

To investigate the age effect of first-time investors, we continue our analysis after grouping them into the three age groups. As Panel B somewhat surprisingly shows, across all three age groups, the market risk of their portfolios is higher than unity. In other words, the risk of the stock portfolios for first-time investors is, on average, higher than the market irrespective of the investor’s age. The estimated MKT coefficient for the portfolios in the different age groups is 1.14, 1.15, and 1.11, respectively, using the four-factor model. However, the difference in β is statistically insignificant between young and old first-time investors.

As the estimated intercepts in Panel B of Table IV using the four-factor model show, a statistically significant abnormal return exists only for investors in the age group 45-64 years. Specifically, the abnormal return for this group is 50 basis points a month (p-value=0.097) and corresponds to 6 percent a year relative to the model. Also, as reported in the last column, no statistically significant difference in the abnormal return exists between young and old investors using the four-factor model.

Finally, we observe a negative and statistically significant coefficient (p-value<0.000) for the HML risk factor between young and old investors indicating fewer holdings of value stocks for younger investors. H2 posits a positive relation between the age of individual investors and the level of portfolio diversification. As reported, the adjusted R2 increases by age and thus supports H2. Hence, older investors have more diversified portfolios compared to younger investors.

As Goetzmann and Kumar (2008) report, portfolios of most individual investors contain a maximum of five stocks. Statman (1987) finds that a well-diversified portfolio consists of at least 30 stocks. Hence, most individual investors are under-diversified potentially resulting in poor returns. More recent evidence by Barber and Odean (2013) finds that investors who hold poorly diversified portfolios not only have perverse stock selection ability incurring unnecessary investment costs but also trade frequently and generally to their detriment.

To investigate the effect of different attributes among first-time investors’ stock portfolios, we separately investigate the importance of portfolio diversification based on the degree of diversification and the value of the portfolio. Specifically, we first rank and sort first-time investors’ portfolios into quintiles based on the Herfindahl index at the end of December between 2004 and 2010. Then, within each quintile, we form equally weighted portfolios across years. In our analysis, we denote investors with a Herfindahl index in quintile 1 as investors with a high degree of diversification and those with a Herfindahl index in quintile 5 with low degree of diversification. As some first-time investors only hold a single stock, we treat these investors as a special group and separately examine this group. Similarly, we sort individuals based on their portfolio value and use the same sorting procedure stated previously where portfolios are sorted into quintiles based on value at the end of each calendar year. We denote portfolios in quintile 1 (5) as low (high) portfolio values and portfolios in quintile 2, 3, and 4 as portfolios with moderate values.

Table V presents the results sorted on the degree of diversification (Panel A) and on portfolio value (Panel B). Focusing first on the first two columns in Panel A, labeled low degree of diversification, we see a positive and statistically significant abnormal return of 60 basis points a month (p-value=0.098) using the four-factor model. For investors who have more diversified portfolios, as reported in columns 3 and 4, we find similar results. However, the monthly α on the portfolio reflecting the difference between low and high diversified portfolios is not statistically significant (p-value=0.844) whereas the difference on MKT is negative and statistically significant (p-value=0.003). Thus, we conclude that no significant difference exists in stock performance between portfolios with a low vs high degree of diversification and also that less diversified portfolios have higher market risk. The return on this difference is positively related to the SMB factor (p-value=0.097) and the HML factor (p-value=0.006) indicating presence of small capitalization risk and evidence of value stocks in the portfolio.

The last two columns in Panel A of Table V report estimated results for investors with no diversification. As reported, no evidence of abnormal return exists for these investors. Also, for these investors, the portfolio return is positively related to MKT and negatively related to the other factor loadings.

Panel B of Table V investigates the effect of stock portfolio value on both the αs and the difference between portfolios with a high and a low value. We find that the estimated abnormal return, using the four-factor model, increases with the overall portfolio value. For the group denoted as low, the monthly α is –30 basis points but is not statistically significant (p-value=0.408). The α for the second group (moderate) is 60 basis points and statistically significant at the 5 percent level. Finally, for high portfolio values, we see a monthly statistically significant abnormal return representing 100 basis points (p-value<0.000). The difference between portfolios with a high and low value, as reported in the last column, is a monthly α of 130 basis points a month, which is statistically significant at the 1 percent level.

The results reported in Table V show that first-time investors with high portfolio value outperform portfolios with a low portfolio value. Also, standard risk characteristics cannot explain the abnormal performance.

To provide further insight into the association between portfolio value and abnormal return, we re-examine our data but focus on the first-time investors who have a portfolio value in quintile 1 (low portfolio value) and 5 (high portfolio value) as reported in Table[4]. For these investors, we then group them into quintiles based on portfolio value.

Table VI shows the results for the four-factor model. Panel A reports estimations for first-time investors with a low value of their stock portfolio whereas Panel B reports results for investors with a high value of their portfolios. As reported, investors in the first quintile have a statistically significant monthly negative abnormal return at the 10 percent level of 80 basis points. For investors in the fifth quintile, the estimated abnormal return is negative but not statistically significant. However, the difference between investors with a high and a low portfolio value is positive and statistically significant at the 10 percent level. This finding indicates that investors with a high portfolio value outperform investors with a low portfolio value by about 7 percent annually. Panel B of Table VI reports estimation results for investors with high portfolio values. As reported in the last row, no statistically significant difference exists between investors with high vs low portfolio value (p-value=0.774).

Of particular interest in Table VI is the difference between investors with high vs low portfolio values. Panel A shows a negative abnormal return in all five regressions whereas Panel B shows four of five regressions with positive abnormal returns. The abnormal return for individuals with the highest portfolio value indicates an abnormal return of about 12 percent a year relative to the four-factor model. Also, as reported in the last column, the adjusted R2 is also higher for the portfolios representing investors with high portfolio values suggesting that the four-factor model better explains the returns for these investors and is also supportive of H3.

Overall, the results of the analysis of abnormal returns for portfolios of first-time investors show a strong difference between portfolios with a high vs low portfolio value. Specifically, we document that the difference between high vs low portfolio values yield a monthly statistically significant abnormal return at the 1 percent level, relative to the four-factor model of 1.3 percent. We also document that the market risk for portfolios held by first-time investors is greater than 1.0.

### 4.5 Further tests of stock performance for portfolios

Our evidence of the stock portfolio performance for first-time investors shows statistically significant differences in abnormal returns after controlling for gender and portfolio value. Based on the above results, we conduct further tests on the abnormal performance for first-time investors’ stock portfolios and more experienced investors by defining the latter group as the remaining investors (i.e. investors who are recorded as a shareholder for a period of at least two years). This latter sample represents about 1,756,300 individual investors compared with 179,128 individual investors. Specifically, we examine the portfolios between first-time and more experienced investors as well as using differences-in-differences regressions and repeat our previous analysis.

Table VII presents the results using the four-factor model. The first column indicates that first-time investors have a higher monthly abnormal return of 50 basis points (p-value=0.028) compared to more experienced investors. Also, first-time investors have higher market risk in their portfolios. The estimated coefficient on HML is negative indicating that the portfolio value of first-time investors is less than for more experienced investors. The estimated coefficient on MOM is negative and statistically significant suggesting no ability among first-time investors to pick winner stocks and sell losers. Column 2 of Table VII shows similar results for investors holding just one stock in their portfolio.

Additionally, we estimate and report the abnormal return using differences-in-difference methodology in columns 3 through 6. As reported, the difference between males and females is an abnormal return of 20 basis points a month, which is statistically significant at the 10 percent level. The difference between young and old investors is not statistically significant (p-value=0.504). Similarly, the results after controlling for the degree of diversification between the groups shows that first-time investors’ portfolios with a high diversification underperform portfolios with low diversification by ‒3.6 percent annually (p-value=0.081). The last column in Table VII shows the results between the two groups of investors after controlling for the portfolio value. As reported, the abnormal return is not statistically significant at conventional levels. These findings indicate that first-time investors outperform more experienced investors and also that first-timers holding just one stock outperform more experienced investors also holding just one stock[5].

Figure 1 plots the time-series of the mean portfolio concentration, using the Herfindahl index, for first-time and more experienced investors over the period December 2003 through December 2010. The figure shows semi-annual data between 2003 and December 2005 and quarterly data from 2006. As Figure 1 shows, more experienced investors have more diversified portfolios. Specifically, we find that the mean Herfindahl index for first-time investors for the full sample period is 0.80 and 0.73 for more experienced investors with a mean difference statistically significant at the 1 percent level (not reported). The figure also indicates that more experienced investors have a fairly stable composition of their portfolios.

## 5. Conclusions

This study provides new insights about stock investments by first-time and more experienced investors in Sweden. Specifically, we investigate whether socio-economic factors influence portfolio composition of individual investors, identify the extent socio-economic factors and portfolio characteristics relate to portfolio returns, and examine the stock portfolio performance of first-time and more experienced investors.

Our analysis of socio-economic factors including gender, age, and portfolio value (wealth) shows that males have significantly higher unsystematic risk in their portfolios than females; older investors have more diversified portfolios compared to younger investors; and investors with high portfolio values outperform those with low portfolio values. Not surprisingly, our evidence shows that first-time investors engage in little portfolio diversification. In fact, only 25 percent of all individuals hold two or more stocks in their portfolio. Multivariate analysis shows a positive relation between portfolio value and the number of stocks in the portfolio irrespective of the age groups used.

We also find a pattern in stock returns for individuals using a four-factor model together with our detailed ownership data that provides insights about investment behavior for first-time investors. Specifically, male investors not only outperform female investors but also assume more unsystematic risk in their portfolios. This latter finding supports H1. We document that the adjusted R2 is higher for the older age group 65-99 years (0.892) compared to the younger age group (18-44 years) (0.832). This evidence supports H2 that older investors have more diversified portfolios due to shorter investment horizons. We also find that investors who have high portfolio values have significantly higher monthly average returns relative to the four-factor model. This evidence helps to explain why first-time investors, specifically those with low income, make the investment mistake of failing to diversify their portfolios. Hence, the action of these investors suggests that they are either unaware of modern portfolio theory or fail to practice it. Our results also support H3 that first-time investors with a high portfolio value have more diversified portfolios.

Tests show that the stock portfolio performance of first-time investors outperforms more experienced investors. Using differences-in-differences methodology, we find that male first-time investors have superior returns. However, our evidence also shows that first-time investors underperform more experienced investors after controlling for the degree of diversification.

## Figures

#### Figure 1

Portfolio diversification for first-time and more experienced investors

## Table I

Summary statistics on first-time investors

Percentile
Variables Mean SD 10th 25th 50th 75th 90th
AGE 44 16.5 19 31 42 56 86
GENDER 0.57 0.50 0 0 1.00 0.45 0.55
INCOME SEK (000) 291.7 158.4 79.5 161.4 279.6 341.3 453.1
PORTFOLIO VALUE SEK (000) 51.9 108.2 1.1 3.4 11.3 36.6 112.3
Portfolio value divided by income (%) 24.56 55.51 0.40 1.43 5.11 18.35 61.99
Number of stocks in the portfolio 2 2 1 1 1 2 5

Notes: This table reports descriptive statistics on the sample of individual shareholders, who invest in Swedish stocks for the first time (first-time investors) between 2004 and 2010. The sample size for all variables is 179,128. The Central Security Registrar in Sweden (Euroclear Sweden) and the Swedish Tax Agency (Skatteverket) serve as the data sources. All cross-sectional statistics are calculated at the end of December for each calendar year. AGE is the age of the individual investor when investing in stocks for the first time. GENDER is a dummy variable with the value of 1 for male investors and 0 otherwise. INCOME is annual income. PORTFOLIO VALUE is the total value of the stock portfolio. Number of stocks in the portfolio is the total number of different stocks in the portfolio. Income and portfolio value are inflation adjusted and expressed in 2010 price level using the consumer price index from Statistics Sweden as a deflator. To reduce the impact of outliers, all variables are winsorized at the 1 percent level

## Table II

Diversification and characteristics of first-time investors

Characteristics of first-time investors and stock portfolio
Number of stocks in the portfolio AGE INCOME (SEK 000) PORTFOLIO VALUE (SEK 000) P/I PRICE (SEK)
1 44
(16.4)
42
267.8
(154.6)
258.1
25.3
(68.3)
7.2
0.136
(0.367)
0.031
84.66
(91.91)
58.80
2 44
(17.1)
42
259.1
(155.9)
248.1
44.8
(90.7)
15.9
0.249
(0.509)
0.073
70.76
(66.00)
53.75
3 43
(16.5)
41
267.3
(163.7)
255.4
61.6
(107.7)
24.9
0338
(0.614)
0.110
67.78
(58.80)
54.95
4 44
(16.3)
43
274.9
(167.4)
261.8
79.8
(128.5)
32.1
0.415
(0.698)
0.139
69.52
(55.93)
60.06
5 44
(16.4)
43
278.8
(171.8)
263.1
108.7
(157.5)
45.0
0.547
(0.831)
0.195
67.62
(54.06)
58.79
6 46
(16.3)
46
285.5
(175.2)
269.4
124.0
(166.7)
58.2
0.610
(0.873)
0.236
68.17
(50.91)
60.46
7 46
(16.5)
46
277.0
(170.5)
259.3
155.3
(192.5)
73.5
0.732
(0.944)
0.314
70.27
(50.22)
63.76
8 47
(16.6)
47
279.1
(172.2)
259.2
162.7
(197.6)
81.1
0.765
(0.970)
0.337
70.58
(50.12)
64.53
9 48
(16.9)
48
282.7
(179.2)
255.9
189.5
(214.4)
97.6
0.854
(1.020)
0.401
71.47
(48.50)
66.19
10 48
(16.1)
48
283.2
(173.4)
261.8
191.8
(219.7)
(95.9)
0.871
(1.072)
0.364
71.36
(44.82)
67.25
11+ 49
(15.7)
49
290.9
(174.3)
266.3
246.5
(256.4)
132.1
1.028
(1.141)
0.507
66.40
(41.70)
62.46

Notes: This table reports the number of stocks in a portfolio and characteristics for first-time investors as well as the portfolio characteristics. AGE is the age at the year when first-time investors invest in stocks for the first time. INCOME is the annual income. PORTFOLIO VALUE is the value of the portfolio as at the end of the calendar year when the investors entered the market. P/I is the ratio between portfolio value and annual income. PRICE is the weighted average price of all stocks in the portfolio. For each characteristic, the table contains the mean, standard deviation (in parentheses), and median. The total number of investors is 179,128. The data are retrieved from the Central Security Registrar in Sweden (Euroclear Sweden) and the Swedish Tax Agency (Skatteverket). All data are calculated as at the end of December for each calendar year and values (reported in SEK 000) are expressed in 2010 price levels using the consumer price index from Statistics Sweden as a deflator. To reduce the impact of outliers, all variables except age are winsorized at the 1 percent level

## Table III

Regression of portfolio value

Age group
18-44 45-64 65-99
(1) (2) (3) (1) (2) (3) (1) (2) (3)
INTERCEPT 7.340*** (88.03) 6.586*** (83.20) 6.860*** (85.39) 9.059*** (62.45) 8.4792*** (63.38) 6.992*** (48.36) 7.556*** (29.37) 8.501*** (37.61) 5.069*** (15.24)
INCOME 0.130*** (19.02) 0.143*** (22.03) 0.037*** (4.80) 0.042*** (3.62) 0.033*** (3.12) 0.038*** (3.51) 0.197*** (9.17) 0.049*** (2.57) 0.189*** (8.19)
NUMBER 0.301*** (99.57) 0.293*** (97.80) 0.307*** (114.90) 0.304*** (115.21) 0.376*** (80.40) 0.377*** (82.25)
PRICE 0.007*** (38.44) 0.008*** (43.57) 0.008*** (21.41)
(PRICE)2 −0.002*** (−37.83) −0.002*** (−32.14) −0.001*** (−11.75)
AGE 0.028*** (24.54) 0.018*** (20.84) 0.016*** (10.59)
GENDER −0.093*** (−7.46) 0.069*** (5.65) 0.062*** (2.48)
Adjusted R2 0.004 0.111 0.135 0.001 0.151 0.183 0.004 0.233 0.269
n 82,981 82,981 82,981 74,514 74,514 74,514 21,633 21,633 21,633

Notes: This table reports the OLS regressions estimating the portfolio value (PORTFOLIO VALUE) for first-time investors in Sweden between 2004 and 2010. The explanatory variables are as follows: the annual income for each investor (INCOME); the number of stocks in the portfolio (NUMBER); the weighted average price for a stock in the portfolio (PRICE) and its square (PRICE)2; age of the individual (AGE); and a dummy variable with the value of 1 if the investors is a male and 0 otherwise (GENDER). The variables PORTFOLIO VALUE and INCOME are expressed in natural logarithms. The data reflect a total of 179,128 observations. Heteroskedasticity-robust t-statistics are reported in parentheses. n is the number of observations. ***Indicate statistical significance at the 1 percent level

## Table IV

Abnormal returns for first-time investors sorted by gender and age

Stock performance
Panel A: individual investors sorted by gender
All investors Male investors Female investors Difference (male−female)
INTERCEPT 0.004 (0.224) 0.005* (0.098) 0.005 (0.206) 0.006* (0.089) 0.003 (0.316) 0.004 (0.150) 0.002 (0.125) 0.002* (0.077)
MKT 1.216*** (<0.000) 1.140*** (<0.000) 1.233*** (<0.000) 1.153*** (<0.000) 1.190*** (<0.000) 1.120*** (<0.000) 0.043* (0.076) 0.034 (0.127)
SMB –0.052 (0.587) –0.050 (0.629) –0.061 (0.452) 0.010 (0.765)
HML –0.147* (0.070) –0.189** (0.035) –0.099 (0.149) –0.090*** (0.003)
MOM –0.277*** (<0.000) –0.299*** (<0.000) –0.241*** (<0.000) 0.058** ( 0.015)
Adjusted R2 0.811 0.866 0.780 0.848 0.856 0.901 0.026 0.339
Panel B: individual investors sorted by age
Age group 18-44 (young) Age group 45-64 Age group 65-99 (old) Difference (young−old)
INTERCEPT 0.005 (0.243) 0.006 (0.121) 0.004 (0.237) 0.005* (0.097) 0.003 (0.272) 0.004 (0.102) 0.002 (0.471) 0.002 (0.506)
MKT 1.216*** (<0.000) 1.141*** (<0.000) 1.224*** (<0.000) 1.148*** (<0.000) 1.191*** (<0.000) 1.112*** (<0.000) 0.025 (0.622) 0.029 (0.454)
SMB –0.015 (0.889) –0.075 (0.388) –0.107 (0.185) 0.091 (0.139)
HML –0.230** (0.016) –0.112 (0.128) 0.009 (0.895) –0.239*** (<0.000)
MOM –0.306*** (<0.000) –0.261*** (<0.000) –0.231*** (<0.000) –0.075* (0.072)
Adjusted R2 0.742 0.832 0.846 0.886 0.871 0.892 0.091 0.518

Notes: This table presents regression results of stock portfolios for first-time investors in Sweden during the sample period 2004-2010 using monthly data where the portfolios are re-balanced at the end of each calendar year. The estimates are based on the calendar-time portfolio methodology from regressing calendar-time portfolio monthly returns constructed on the Fama and French (1993) three factors (MKT, SMB, and HML) and Carhart’s (1997) momentum factor (MOM). Results are presented in two panels: individuals sorted by gender (Panel A) and individuals sorted by age (Panel B). The p-values, which are adjusted for heteroskedasticity and autocorrelations, are in parentheses. *,**,***Indicate statistical significance at the 10, 5 and 1 percent levels, respectively

## Table V

Calendar-time regressions of monthly stock returns

Stock performance
Panel A: stock portfolios grouped by diversification (Herfindahl)
Low degree of diversification High degree of diversification Difference (low−high) No diversification
INTERCEPT 0.006* (0.098) 0.006*** (0.098) 0.005*** (0.005) 0.006*** (0.000) 0.001 (0.960) −0.001 (0.844) 0.004 (0.393) 0.004 (0.247)
MKT 1.178*** (<0.000) 1.109*** (<0.000) 0.997*** (<0.000) 0.953*** (<0.000) –0.182*** (<0.000) −0.157*** (0.003) 1.235*** (<0.000) 1.161*** (<0.000)
SMB 0.102 (0.289) 0.036 (0.319) 0.066* (0.097) –0.055 (0.625)
HML –0.134 (0.104) 0.065* (0.036) 0.198*** (0.006) –0.212** (0.028)
MOM –0.229*** (<0.001) –0.152*** (<0.000) –0.077 (0.170) –0.284*** (<0.000)
Adjusted R2 0.823 0.851 0.942 0.966 0.132 0.203 0.765 0.823
Panel B: individual investors sorted by portfolio value
Low Moderate High Difference (high−low)
INTERCEPT –0.003 (0.461) ‒0.003 (0.408) 0.005 (0.148) 0.006** (0.049) 0.009*** (0.004) 0.010*** (<0.000) 0.012*** (<0.000) 0.013*** (<0.000)
MKT 1.192*** (<0.000) 1.135*** (<0.000) 1.229*** (<0.000) 1.145*** (<0.000) 1.201*** (<0.000) 1.127*** (<0.000) 0.008 (0.869) ‒0.009 (0.859)
SMB 0.097 (0.405) –0.076 (0.429) –0.126 (0.142) –0.223*** (0.005)
HML –0.110 (0.269) –0.166** (0.046) –0.130* (0.076) –0.020 (0.763)
MOM –0.259*** (0.001) –0.296*** (<0.000) –0.236*** (<0.000) 0.024 (0.657)
Adjusted R2 0.738 0.808 0.807 0.864 0.853 0.883 0.115 0.183

Notes: This table presents regression results of stock portfolios for first-time investors in Sweden during the sample period 2004-2010 using monthly data where the portfolios are re-balanced at the end of each calendar year. The estimates are based on the calendar-time portfolio methodology from regressing calendar-time portfolio monthly returns constructed on the Fama and French (1993) three factors (MKT, SMB, and HML) and Carhart’s (1997) momentum factor (MOM). Results are presented in two panels: portfolios with respect to diversification levels (Panel A) and portfolios sorted by low, moderate, and high portfolio value (Panel B). The p-values, which are adjusted for heteroskedasticity and autocorrelations, are in parentheses. *,**,***Indicate statistical significance at the 10, 5 and 1 percent level respectively

## Table VI

Calendar-time regressions of monthly stock returns

INTERCEPT MKT SMB HML MOM n Adjusted R2
Panel A: equally weighted portfolios based on low portfolio value
Low –0.008* (0.097) 1.126*** (<0.000) 0.185 (0.186) –0.058 (0.625) –0.268*** (<0.000) 84 0.748
2 –0.002 (0.593) 1.155*** (<0.000) 0.074 (0.536) –0.135 (0.185) –0.265*** (<0.000) 84 0.805
3 –0.003 (0.469) 1.142*** (<0.000) –0.045 (0.737) –0.267** (0.02) –0.281*** (<0.000) 84 0.771
4 –0.002 (0.725) 1.143*** (<0.000) 0.178 (0.191) –0.120 (0.300) –0.209** (<0.025) 84 0.753
High –0.001 (0.695) 1.112*** (<0.000) 0.092 (0.411) 0.029 (0.758) –0.272*** (<0.000) 84 0.815
High−low 0.006* (0.054) –0.014 (0.814) –0.093 (0.339) 0.087 (0.293) –0.005 (<0.945) 84 0.092
Panel B: equally weighted portfolios based on high portfolio value
Low 0.010*** (0.001) 1.142*** (<0.000) –0.155 (0.108) –0.193** (0.020) –0.287*** (<0.000) 84 0.864
2 –0.002 (0.593) 1.155*** (<0.000) 0.074 (0.566) –0.135 (0.185) –0.262*** (<0.000) 84 0.815
3 0.010*** (0.001) 1.127*** (<0.000) –0.117 (0.187) –0.145* (0.055) –0.249*** (<0.000) 84 0.876
4 0.011*** (<0.000) 1.116*** (<0.000) –0.125 (0.166) –0.109 (0.156) –0.218*** (0.001) 84 0.868
High 0.010*** (<0.000) 1.101*** (<0.000) –0.120* (0.084) –0.061 (0.301) –0.169*** (<0.000) 84 0.913
High−low –0.001 (0.774) –0.041 (0.239) 0.035 (0.525) 0.132*** (0.006) 0.118*** (0.002) 84 0.285

Notes: This table presents regression results of stock portfolios for first-time investors in Sweden during the sample period 2004-2010 using monthly data where the portfolios are re-balanced at the end of each calendar year. The estimates are based on the calendar-time portfolio methodology from regressing calendar-time portfolio monthly returns constructed on the Fama and French (1993) three factors (MKT, SMB, and HML) and Carhart’s (1997) momentum factor (MOM). Results are presented in two panels: individuals sorted by gender (Panel A) and individuals sorted by age (Panel B). The p-values, which are adjusted for heteroskedasticity and autocorrelations, are in parentheses. *,**,***Indicate statistical significance at the 10, 5 and 1 percent levels, respectively

## Table VII

Abnormal returns for first-time and more experienced investors

Stock performance between first-time and more experienced investors
Difference-in-differences
All investors No diversification Male−female Young−old High−low diversification High−low portfolio value
INTERCEPT 0.005** (0.028) 0.006** (0.029) 0.002* (0.087) 0.001 (0.504) –0.003* (0.081) 0.004 (0.238)
MKT 0.091** (0.025) 0.123** (0.023) 0.016 (0.382) 0.087** (0.013) –0.087*** (0.009) 0.053 (0.384)
SMB –0.049 (0.437) –0.073 (0.390) –0.009 (0.751) –0.016 (0.769) 0.067 (0.201) –0.172* (0.078)
HML –0.132** (0.016) –0.251*** (0.001) –0.011 (0.657) 0.026 (0.572) 0.016 (0.712) 0.351*** (0.001)
MOM –0.154*** (0.001) –0.166*** (0.005) –0.033* (0.100) –0.146*** (<0.000) 0.059* (0.094) –0.214*** (0.001)
Adjusted R2 0.325 0.339 0.027 0.293 0.136 0.433

Notes: This table presents regression results of differences in stock portfolios between first-time and more experienced investors in Sweden during the sample period 2004-2010 using monthly data where the portfolios are re-balanced at the end of each calendar year. The estimates are based on the calendar-time portfolio methodology from regressing calendar-time portfolio monthly returns constructed on the Fama and French (1993) three factors (MKT, SMB, and HML) and Carhart’s (1997) momentum factor (MOM). Results are first presented between all investors and investors with no diversification (Herfindahl=1) in columns 1 and 2, respectively. Columns 3 through 6 show regression results using difference-in-differences methodology between first-time and more experienced investors for four groups: male and females; young and old; differences between portfolios with a high degree of diversification less portfolios with a low degree of diversification (measured by the Herfindahl index); and differences between portfolios with a high and low portfolio value. The p-values, which are adjusted for heteroskedasticity and autocorrelations, are in parentheses. *,**,***Indicate statistical significance at the 10, 5 and 1 percent levels, respectively

## Notes

1.

We compute the factors SMB, HML, and MOM using Swedish data from Datastream.

2.

In US dollars, the average portfolio value corresponds to a mean of $7,100 and a median of$1,500.

3.

The mean value of the factor loadings MKT, SMB, HML, and MOM over the sample period was 0.87, 0.34, ‒0.17, and 0.05 percent, respectively.

4.

For first-time investors with a portfolio value in group “low” in Table V, mean age is 39.9 with an average annual income of SEK252,000. The mean portfolio value for this group is SEK326,800 with a mean Herfindahl index of 0.87. For investors in the high group, mean age is 51.0 with an average annual income of SEK318,000. The mean value of the portfolio for these investors is SEK1,500,000 with a mean Herfindahl index of 0.59.

5.

We perform additional analyses to confirm our results. For instance, instead of using a two-year horizon in the definition of a first-time investor, we used three and four years. Although not reported, we find that these two definitions do not change our results.

## References

Arano, K., Parker, C. and Terry, R. (2010), “Gender-based risk aversion and retirement asset allocation”, Economic Inquiry, Vol. 48 No. 1, pp. 147-155.

Barber, B.M. and Odean, T. (2001), “Boys will be boys: gender, overconfidence, and common stock investment”, Quarterly Journal of Economics, Vol. 116 No. 1, pp. 261-292.

Barber, B.M. and Odean, T. (2013), “The behavior of individual investors”, in Constantinides, G.M., Harris, M. and Stulz, R.M. (Eds), Handbook of the Economics of Finance, Elsevier B.V., Amsterdam, pp. 1533-1570.

Blume, M.E. and Friend, I. (1975), “The asset structure of individual portfolios and some implications for utility functions”, Journal of Finance, Vol. 30 No. 2, pp. 585-603.

Campbell, J.Y. (2006), “Household finance”, Journal of Finance, Vol. 61 No. 4, pp. 1553-1604.

Campbell, J.Y. and Viceira, L.M. (2002), Strategic Asset Allocation: Portfolio Choice for Long-term Investors, Oxford University Press, New York, NY.

Carhart, M. (1997), “On persistence in mutual funds fund performance”, Journal of Finance, Vol. 52 No. 1, pp. 57-82.

DeMiguel, V., Garlappi, L. and Uppal, R. (2009), “Optimal versus naive diversification: how inefficient is the 1/N portfolio strategy?”, Review of Financial Studies, Vol. 22 No. 5, pp. 1915-1953.

Demsetz, H. and Lehn, K. (1985), “The structure of corporate ownership: causes and consequences”, Journal of Political Economy, Vol. 93 No. 6, pp. 1155-1177.

Dhar, R., Goetzmann, W. and Zhu, N. (2003), “The impact of clientele changes: evidence from stock splits”, working paper, Yale University, New Haven.

Dyl, E.A. and Elliott, W.B. (2006), “The share price puzzle”, Journal of Business, Vol. 79 No. 4, pp. 2045-2066.

Fama, E. and French, K. (1993), “Common risk factors in the returns of stocks and bonds”, Journal of Financial Economics, Vol. 33 No. 1, pp. 3-56.

Fuertes, A., Muradoglu, G. and Ozturkkal, B. (2014), “A behavioral analysis of investor diversification”, European Journal of Finance, Vol. 20 No. 6, pp. 499-523.

Goetzmann, W.N. and Kumar, A. (2008), “Equity portfolio diversification”, Review of Finance, Vol. 12 No. 3, pp. 433-463.

Gompers, P. and Metrick, A. (2001), “Institutional investors and equity prices”, Quarterly Journal of Economics, Vol. 116 No. 1, pp. 229-259.

Grable, J. and Lytton, R.H. (1998), “Investor risk tolerance: testing the efficacy of demographics as differentiating and classifying factors”, Financial Counseling and Planning, Vol. 9 No. 1, pp. 61-70.

Hoffmann, A., Shefrin, H. and Pennings, J. (2010), “Behavioral portfolio analysis of individual investors”, available at: http://ssrn.com/abstract=1629786 (accessed April 2016).

Kaniel, R., Liu, S., Saar, G. and Titman, S. (2012), “Individual investor trading and return patterns around earnings announcements”, Journal of Finance, Vol. 67 No. 2, pp. 639-680.

Kelly, M. (1995), “All their eggs in one basket: portfolio diversification of US households”, Journal of Economic Behavior & Organization, Vol. 27 No. 1, pp. 87-96.

Kourtidis, D., Sevic, Z. and Chatzoglou, P. (2011), “Investors’ trading activity: a behavioural perspective and empirical results”, Journal of Socio-Economics, Vol. 4 No. 5, pp. 548-557.

Kumar, A. (2009), “Who gambles in the stock market?”, Journal of Finance, Vol. 64 No. 4, pp. 1889-1933.

Kumar, A. and Lee, C. (2006), “Retail investor sentiment and return comovements”, Journal of Finance, Vol. 61 No. 5, pp. 2451-2486.

Markowitz, H. (1952), “Portfolio selection”, Journal of Finance, Vol. 7 No. 1, pp. 77-91.

Morin, R.A. and Suarez, A.F. (1983), “Risk aversion revisited”, Journal of Finance, Vol. 38 No. 4, pp. 1201-1216.

Muscarella, C.J. and Vetsuypens, M.R. (1996), “Stock splits: signaling or liquidity? The case of ADR’s solo-splits”, Journal of Financial Economics, Vol. 42 No. 1, pp. 3-26.

Statman, M. (1987), “How many stocks make a diversified portfolio?”, Journal of Financial and Quantitative Analysis, Vol. 22 No. 3, pp. 353-363.

Yao, R., Sharpe, D.L. and Wang, F. (2011), “Decomposing the age effect on risk tolerance”, Journal of Socio-Economics, Vol. 40 No. 6, pp. 879-887.

## Acknowledgements

The authors thank the participants at the Midwest Finance Association Conference and Southern Finance Association Conference both in 2016 for their helpful comments and suggestions. The authors also appreciate the constructive feedback from Jay Ritter, Ki Han, and Tao Shu, computer support from Daniel Hallgren, and Krister Modin at Euroclear Sweden for providing the authors with data.