1 s2.0 S0378426614001253 Main PDF

Journal of Banking & Finance 46 (2014) 372386
Contents lists available at ScienceDirect
Journal of Banking & Finance

journal homepage: www.elsevier.com/locate/jbf
Do investors put their money where their mouth is? Stock market
expectations and investing behavior
Christoph Merkle , Martin Weber
Chair of Finance and Banking, University of Mannheim, Germany
a r t i c l e
i n f o
Article history:
Received 21 February 2013
Accepted 30 March 2014
Available online 24 April 2014
JEL-Classication Codes:
D81
G02
G11
Keywords:
Expectations
Beliefs
Risk
Return
Trading behavior
Portfolio choice
a b s t r a c t
To understand how real investors use their beliefs and preferences in investing decisions, we examine a
panel survey of self-directed online investors at a UK bank. The survey asks for return expectations, risk
expectations, and risk tolerance of these investors in three-month intervals between 2008 and 2010. We
combine the survey data with investors actual trading data and portfolio holdings. We nd that investor
beliefs have little predictive power for immediate trading behavior. The exception is a positive effect of
increases in return expectation on buying activity. Portfolio risk levels and changes are more systematically related to return and risk expectations. In line with nancial theory, risk taking increases with
return expectations and decreases with risk expectations. In response to their expectations, investors also
adjust the riskiness of assets they trade.
2014 Elsevier B.V. All rights reserved.
1. Introduction
There is a large gap between what nance models predict for
individual investor behavior and what can be observed in their
actual behavior. Portfolio theory assumes that investors form
expectations about return and risk of securities and select portfolios according to their expectations and risk preferences
(Markowitz, 1952). As a consequence, they should hold broadly
diversied portfolios and trade very little. But instead, private
investors have been shown to hold underdiversied portfolios
(Goetzmann and Kumar, 2008), to trade frequently (Odean, 1999;
Barber and Odean, 2000), to take high idiosyncratic risk (Calvet
et al., 2007), and to gamble in the stock market (Kumar, 2009).
There is also evidence that they use various investment strategies
different from pure meanvariance optimization (Lewellen et al.,
1977; Grinblatt and Keloharju, 2000). Often these deviations have
been explained by specic psychological biases, e.g., excessive
trading by overcondence (Odean, 1998; Glaser and Weber, 2007).
Corresponding author. Address: Lehrstuhl fr ABWL und Finanzwirtschaft,
Universitt Mannheim, 68131 Mannheim, Germany. Tel.: +49 6211811531; fax:
+49 6211811534.
E-mail address: [email protected] (C. Merkle).
http://dx.doi.org/10.1016/j.jbankn.2014.03.042
0378-4266/ 2014 Elsevier B.V. All rights reserved.
However, this way one learns very little about the actual decision making process people go through when they invest. How
do investors use their beliefs and preferences in this process?
Empirically, there is only scarce evidence on this question as the
input parameters are hard to obtain. The economic paradigm of
revealed preferences states that beliefs and preferences can be
inferred from observed actions (Samuelson, 1938). But this already
implies that they are perfectly converted into actions. In order to
reveal whether and where this transfer might fail, direct information on beliefs and preferences is needed.
To this end, we collect return and risk expectations in a
repeated panel survey of self-directed private investors at a large
UK online brokerage provider. These investors are not representative for the overall investor population including institutions,
which imposes some limits on the generality of the results. However, our focus is on individual investors for which our sample is
rather typical. Participants are well informed about nancial markets as, e.g., their responses in a nancial literacy questionnaire
show. They also have on average many years of investment experience and invest non-trivial amounts of money. In three-month
intervals, survey participants are queried for numerical and qualitative expectations and their risk tolerance. We then match expectations of investors to their actual transactions in their online
C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386
brokerage accounts. We observe volume, timing, and direction of

all trades within the survey period, and are able to calculate portfolio holdings of participants.
We develop different measures of nancial risk taking based on
trading behavior and portfolio holdings of investors. In a rst step,
we consider the direction of stock trading and calculate the ratio of
buys over total trades, referred to as buysell ratio. This corresponds to an increase or decrease in investors total equity position. We nd that the absolute levels of expectations for market
return and risk do not predict buying and selling behavior. An
explanation could be that previous expectations are already
reected in investors portfolios and there is no need for investors
to engage in further transactions. We therefore also test whether
changes in expectations explain buying and selling behavior corresponding to trades reecting changes in portfolios. Indeed, improving return expectations have a positive impact on buysell ratios.
Thus, quite intuitively, positive return expectations foster buying
activity, but there is no effect of changes in risk expectations or risk
attitude on buysell ratios.
While immediate trading behavior and direction of trade is a
means to alter ones risky position, we also directly investigate
portfolio risk. We calculate portfolio volatility and beta for investors in our panel as standard risk measures. This is complemented
by additional measures such as relative volatility and average
component volatility (Dorn and Huberman, 2005). We consider
both, levels of portfolio risk at the point in time of survey rounds
and changes in portfolio risk between survey rounds. Levels of
risk taking of investors can be well explained by their beliefs,
preferences and demographics. All portfolio risk measures are
positively related to return expectations and risk tolerance, and
negatively related to risk expectations, age, and wealth of investors. These results are consistent with nancial theory and previous literature.
An advantage of our dataset is that it allows studying the
dynamics of this relationship between expectations and risk taking, i.e., whether investors react to changes in expectations by
changing their portfolio composition and thus alter risk exposure.
For the volatility measures this is the case, as we nd a positive
change in volatility when return expectations improve and a negative change if investors expect increasing stock market risk. The
relationship is weakest for short-term volatility and portfolio beta,
indicating that investors manage their portfolios rather based on
long-term volatility as a proxy for risk taking. Our results are
robust to several alternative specications including the use of
lagged values to address endogeneity concerns. Risk tolerance
remains insignicant in most of our regressions (both levels and
changes), which sheds some light on the debate, whether investors
can translate their level of risk aversion into an adequate portfolio
choice (Ehm et al., 2014).
Finally, we combine the perspectives of trades and portfolio risk
and analyze the volatility of transactions by investors. This allows
us to gain a deeper understanding of how investors regulate their
portfolio risk. The analysis reveals that more optimistic investors
shift part of their investments to more volatile securities. In addition to expanding their total equity position by purchases in excess
of sales, they also buy riskier assets. This is consistent with the
nding that portfolio volatility not just passively moves with market volatility, but also relative portfolio volatility increases for optimistic investors.
We continue with a theoretical motivation and an overview of
related literature in Section 2, followed by a description of the data
set, which contains two main sources, the survey and the trading
data. In Section 4 we present results about the relationship
between investor expectations and trading behavior, which we
then discuss in Section 5. A nal section concludes.
373
2. Theory and literature

People acting on their beliefs and preferences are such a basic
assumption in economic theory that it has seldom been contested.
Exemplarily, portfolio theory as the canonical nance model posits
that investors form expectations about return and risk of securities
and then choose an optimal portfolio according to their risk preferences (Markowitz, 1952). We will now in a more formal but simple
way derive directional predictions for the inuence of return
expectations, risk expectations, and risk tolerance on nancial risk
taking behavior.
We assume an investor to have power utility dened over
wealth W of the form UW W 1h 1=1 h. Power utility
has the desirable property of declining absolute risk aversion and
constant relative risk aversion, which is most consistent with real
world observations. The investor in a simple two-period economy
faces the budget constraint W 1 W 0 1 r0;1 , implying that the
only source of wealth at time t = 1 is wealth in t = 0 plus the return
earned on wealth. The corresponding maximization problem thus
is:
max E0 W 0 1 r 0;1 1h =1 h:
Under the additional assumption that future wealth W 1 is lognormally distributed, expression (1) simplies to (for a detailed derivation cp. Campbell and Viceira, 2002):
1
max ln E0 1 r 0;1 hr20 ;
2
where r20 is the conditional variance of the log return,

1
r20 Var0 ln W
Var0 ln1 r0;1 . In expression (2) the ingredients
W0
of the maximization problem are visible: the investor trades off

expected return against expected risk (variance of returns). The
parameter h of the utility function describes the investors relative
risk aversion.
With only two assets, a risky asset s and a riskless asset f, return
on wealth is r 0;1 rf ;0;1 ws;0 rs;0;1 r f ;0;1 , where ws;0 represents
the weight an investor puts on the risky asset. However, while
r0;1 is a linear combination of the two asset returns, the log return
on wealth cannot be expressed as a linear combination of the log
returns. Instead, Campbell and Viceira (2002) suggest a Taylor
approximation to rewrite (2) in the form
1
1
max ws;0 E0 r s;0;1 r f ;0;1 ws;0 1 ws;0 r20 1 hw2s;0 r20 ;
2
2
which can be solved by
ws;0
E0 r s;0;1 r f ;0;1 r20 =2

:
hr20
The equation implies that the share of risky investment should

increase with expected returns for the risky asset, and decrease
with risk expectations and risk aversion. This result can be generalized to a multi-asset or multi-period framework and is fairly robust
to the relaxation of several of the chosen assumptions. A simple
meanvariance optimization comes to the same conclusions, as
doesfrom a slightly different anglerisk-value theory (Sarin and
Weber, 1993). We take the results of this model as a prediction
for the role of expectations and risk preferences in investing
behavior.
Empirically, risk taking behavior of individual investors has
been studied using different approaches and datasets. VissingJorgensen (2003) analyzes a US individual investor survey by
UBS/Gallup and nds a strong positive effect of expected return
on equity share in self-reported investor portfolios. Dorn and
Huberman (2005) report portfolio volatilities for a sample of
German brokerage clients and identify risk aversion as most
374
predictive for portfolio volatility. Moreover, younger, selfemployed, less sophisticated, and poorer investors tend to hold
more risky portfolios. Calvet et al. (2007) examine disaggregated
wealth data covering the entire Swedish population and show a
positive impact of wealth, income, and education on risk taking
measured by portfolio volatility.1 They also break down portfolio
risk in its various components and reveal interesting patterns of risk
taking. In a follow-up study, Calvet et al. (2009) present evidence on
rebalancing, suggesting that investors actively control their share of
risky investments and offset changes brought about by passive market variations.
While this literature addresses risk taking behavior of private
investor, it lacks a systematic study of the input variables we are
interested in: individual investor beliefs in form of return and risk
expectations, and investor risk preferences. Closest related to our
study is the work by Amromin and Sharpe (2009), Weber et al.
(2013), Hoffmann et al. (2013), and Guiso et al. (2011). Similar to
us Amromin and Sharpe (2009) use panel data, in their case coming
from the Michigan Survey of Consumer Attitudes. However, they
analyze self-reported portfolio shares of survey participants and
do not have access to their transactions or actual portfolios. They
concentrate on the interrelation of return expectations and risk
expectations, but also provide some evidence of the inuence of
these variables on portfolio composition. Consistent with nancial
theory, higher return expectations and lower risk expectations
increase the share of equity in portfolios of investors. Hoffmann
et al. (2013) study an investor survey in the Netherlands, which
is matched to brokerage account data. Their data spans a time
period from April 2008 to March 2009 and survey rounds are
administered monthly. By eliciting expectations and portfolio
characteristics, Hoffmann et al. (2013) establish a link between
the beliefs of investors and their investing behavior. They nd a
positive impact of risk tolerance, risk perception, and return expectations on trading activity, while risk tolerance is identied as a
main driver for risk taking behavior.
Guiso et al. (2011) concentrate in their analysis on risk aversion
measured by a qualitative and a quantitative approach. They report
a substantial increase of risk aversion in the nancial crisis compared to pre-crisis levels. Ownership of risky assets is negatively
related to risk aversion. Guiso et al. (2011) suggest psychological
factors as drivers of risk aversion, as they are able to rule out alternative explanations such as wealth or background risk.
In a previous analysis of our dataset, Weber et al. (2013) report
a relationship between expectations and investing decisions. They
analyze a survey question which asks participants to split a hypothetical amount of 100,000 between an investment in the UK
stock market and a riskless asset. With this investment task they
are able to show a strong inuence of changes in expectations
and risk attitude on changes in the proportion of risky investment.
This inuence is in the expected direction: increases in expected
returns or risk tolerance lead to an increase in risky investment,
while higher risk expectations render investors more cautious.
We extend this research by relating return and risk expectations
to the actual trades and portfolios of investors. By analyzing various aspects of investing behavior, we present a more complete portrayal of the underlying relationships. We also exploit the full time
series of the survey which was not available to the earlier study by
Weber et al. (2013).
1
The seemingly contradictory results might be explained by the different
composition of the datasets. While Dorn and Huberman (2005) analyze stock
portfolios, where wealth and nancial sophistication usually lead to a better
diversication (and thus less risk), Calvet et al. (2007) use total wealth portfolios
for which wealth and sophistication typically lead to a greater equity share (and thus
more risk).
3. Data
We obtain survey responses and transaction data for a sample
of clients at Barclays Stockbrokers, a UK direct brokerage provider.
Barclays is one of the largest brokers in the UK and attracts a wide
variety of customers (for demographic characteristics of its clients
see below). The accounts are self-directed in the sense that customers can inform themselves on special webpages provided by
the bank but receive no direct investment advice. Most transactions are processed online.
3.1. Survey data
In collaboration with Barclays Wealth, we conduct a repeated
survey taking place every three months, beginning in September
2008 and ending in September 2010. Fig. 1 shows the development
of the UK stock market represented by the FTSE all-share index and
the timing of survey rounds. Our panel consists of nine rounds covering a time period of highly volatile market environment. We thus
expect participants to express changing beliefs about market prospects; in the standard model of Eq. (4) this would in turn lead to
changes in their portfolios.
In the initial survey a stratied sample of the banks client base
was invited via e-mail to participate in the online questionnaire
(for details on the sampling procedure see Weber et al., 2013). In
total 617 clients of the bank participated in the survey, 394 of which
participated multiple times. 189 participants have completed at
least ve rounds, and 52 have participated in all nine rounds. We
have a minimum of 130 observations for each of the nine rounds.
We will discuss potential selection effects in Section 4.4.
Table 1 shows some demographic characteristics of survey participants. Investors are predominantly male, and they are older and
more afuent than the general population (for an explicit
Fig. 1. FTSE all-share index and survey rounds. Development of the FTSE all-share
index (covers 98% of UK market capitalization) between June 2008 and December
2010. Vertical lines represent the timing of the nine survey rounds.
Table 1
Demographics of participants.
Age (in years)

Gender (male = 1)
Financial literacy
(% correct)
Wealth (in categories)
Income (in categories)
Mean
Median
Std.dev.
Min
Max
613
617
614
51.4
0.93
3.49
53
1
4
12.9
0.25
0.68
21
0
0
84
1
4
502
494
4.80
3.88
5
4
2.39
1.80
1
1
9
8
Notes: The table shows descriptive statistics about demographics of participants.

Age is reported in years, gender as a dummy variable taking a value of 1 for male
participants. Financial literacy is the number of correct responses in a 4-item
nancial literacy test (see Appendix A). Wealth and income are self-reported and
measured in categories (see Appendix A). Number of observations varies due to
refusals.
comparison see again Weber et al., 2013). However, they closely

resemble typical investor populations in other studies (e.g.,
Barber and Odean, 2001). The nancial literacy of survey participants is relatively high with on average 3.5 correct responses out
of four questions. This exceeds values usually found for these questions in household surveys (van Rooij et al., 2011).
We elicit beliefs about return and risk expectations in two ways,
by a numerical question asking for return expectations in percentage
terms and a more subjective evaluation of risk and return
on a bipolar scale. The wording of the numerical question is as
follows:
We would like you to make three estimates of the return of the UK
stock market (FTSE all-share) by the end of the next three month.
Your best estimate should be your best guess.
Your high estimate should very rarely be lower than the actual outcome of the FTSE all-share (about once in 20 occasions).
Your low estimate should very rarely be higher than the actual outcome of the FTSE all-share (about once in 20 occasions).
Please enter your response as a percentage change.
The question asks participants to predict the three-month
return of the UK stock market. We use this time horizon to avoid
overlapping observations as the distance between survey rounds
is three month as well. One might argue that these short-term
expectations will be irrelevant, if investors have a longer investment horizon. However, we nd them to be highly correlated with
one year expectations which were elicited twice during the survey.
We suspect that three-month expectations express an investors
current optimism or pessimism about the market not limited to
the particular time interval. In addition, high portfolio turnover
reported below implies that short-term expectations should
certainly matter.
In a design similar to Glaser and Weber (2005), participants have
to submit a best estimate as well as a high and a low estimate, which
together yield a 90%-condence interval. We take the best estimate to
represent an investors return expectation about the UK stock market.
The high and low estimates allow calculating implicit expected volatility of investors which we use as numerical risk estimate applying
the method of Keefer and Bodily (1983). We use this indirect way as
it has been shown that people often have difculties with numeric
risk estimates (Windschitl and Wells, 1996; Dave et al., 2010).
Furthermore, numeric estimates may not cover all aspects of
expected risks and benets which are partly emotional. The riskas-feelings hypothesis maintains that subjective risk perceptions
will often differ from cognitive assessment of risk (Loewenstein
et al., 2001). It is unclear, whether investors primarily act on their
numerical expected volatility or an affective impression of risk.
The nance literature uses many different ways to measure risk
expectations, and it is still debated which best explains investor
behavior (Hoffmann et al., 2013; Weber et al., 2013). We therefore
include qualitative questions, which ask people to evaluate return
and risk on a seven-point scale.
375
Risk tolerance of investors is measured as agreement to the

statement It is likely I would invest a signicant sum in a high risk
investment (on a seven-point scale). The statement is part of a
more complete assessment of risk attitude (eight items) in the
entry questionnaire to our survey. Factor analysis and Cronbachs
alpha show high consistency between the items, and for brevity
the set of statements was reduced to one for the panel survey.
The selected statement was chosen not for wording, but for statistical properties such as to capture maximal information from the
multi-dimensional measure (for the construction of psychometric
risk tolerance scores cp. also Egan et al., 2010; Kapteyn and
Teppa, 2011). The correlation between the eight-item risk
tolerance score and single-item risk tolerance is 0.77 for the entry
round when both were elicited. Weber et al. (2013) nd high predictive power of the single-item risk tolerance measure for hypothetical investment decisions. Besides the core variables of
beliefs and preferences, the survey asks for demographics, psychological dispositions, and investment objectives. All variables used
in our analysis are described in Appendix A.
3.2. Survey responses

Average numeric return expectations are relatively low before
the peak of the nancial crisis, then rise during the crisis and fall
again, when the UK stock market recovers. Fig. 2 shows the pattern
in detail. In general investors tend to be more optimistic about
their own portfolios: the average return expectations are consistently higher and the difference is non-trivial (24%-points). In
contrast to market expectations, average portfolio expectations
remain high throughout 2009 and only decline afterwards. While
market expectations are in a reasonable range adding up to an
annual return of 812% (compared to a FTSE all-share historical
return of about 8%), the absolute level of portfolio expectations
seems unrealistically high (probably explained by overcondence
cp. Merkle, 2012).
Investors in our panel (numerically) underestimate stock market risk (cp. Glaser et al., 2013). The implied volatilities calculated
from the condence intervals of investors return expectations are
much lower than volatility expectations of sophisticated market
participants (represented by implied option volatilities, see
Fig. 3). While condence intervals are too narrow in the initial survey round, investors seem to learn from observed outcomes that
extreme realizations are possible and enlarge their condence
intervals. Expected volatility thus increases, but is still below
implied option volatility. Furthermore after the initial adjustment,
the condence intervals remain insensitive to subsequent market
developments.
How would you rate the returns you expect from an investment in
the UK stock market (FTSE all-share) over the next 3 months?
Over the next 3-months, how risky do you think the UK stock market (FTSE all-share) is?
In the rst question answer alternatives range from extremely
bad to extremely good, in the second question from not risky
at all to extremely risky. We ask equivalent questions for investors own portfolios held with Barclays. In total we thus collect eight
belief items per investor per round.
Fig. 2. Numerical return expectations of investors. Average quarterly return

expectations of investors for their own portfolio and the UK stock market (FTSE
all-share).
376
Fig. 3. Risk expectations of investors. Qualitative risk expectations for market and
own portfolio (scale 17, right axis), and numerical risk expectations as implied by
condence intervals (volatilities, left axis). For comparison implied option volatility
(FTSE 100 VIX, left axis).
Compared to the quantitative measure, qualitative risk expectations elicited on a seven-point scale reect more closely implied
market risk expectations represented by the FTSE 100 VIX. While
it is not possible to compare the absolute magnitudes, we nd a
correlation of 0.78 (p < 0:02) between average qualitative risk
expectations and implied option volatilities. Quite intuitively, risk
expectations rise with the peak of the nancial crisis and then fall
afterwards. However, there are two further increases in panelists
risk expectations: one without a corresponding rise in option market expectations (September to December 2009), and another,
which falls together with the onset of the European debt crisis
(June 2010). In general, expectations for own portfolio risk follow
this trend but are on average slightly lower and more stable than
market expectations. It is noteworthy that investors appear to
believe they can earn higher returns bearing less risk (cp. Kempf
et al., 2014).
For investigating trading behavior over time, changes in expectations are particularly important. Table 2 shows average changes
for all expectation variables. We observe a signicant increase in
average return and risk expectations between round one and three
followed by a very mixed pattern from round three to four (further
increase of qualitative return and numerical risk expectations, but
sharp drop of qualitative risk expectations). Changes in expectations are less pronounced for the time after the immediate crisis.
An exception is the very last survey round for which we observe
strongly increasing return expectations and decreasing risk expectations. Similar to Weber et al. (2013), we nd that the correlations
between changes of numeric and qualitative expectations are often
low (return) or insignicant (risk). Stronger correlations exist
between market and portfolio expectations. Average risk tolerance
remains fairly stable over the whole survey period.
3.3. Trading data
Our data also include the trading records of all investors active
in the panel survey. We include three months prior to our rst survey round and three months after our last survey round. In the
resulting period between June 2008 and December 2010 we
observe 49,372 trades with a total trading volume of
258,940,694. Of these trades 37,022 or 75% are in stocks (63% of
trading volume). In some parts of the analysis, we will concentrate
on these equity transactions as they are closest related to the
expectations we elicit among investors. The remaining trades
include bonds, derivatives, mutual funds and ETFs. The average trader in the panel trades 84.1 times within the 2.5 year period (about
three times per month), with an average trading volume of
441,126. However, the distribution is strongly skewed; the median trader trades only 33 times (about once a month), the median
trading volume is 72,805. We observe most pronounced trading

activity in the initial phase of the nancial crisis; investors seem
to feel a need to react to the turbulent times on asset markets.
Combining trading data with a snapshot of investors portfolios,
we are able to calculate portfolio statistics for our survey period.
The median portfolio is worth 41,687 (average 314,663) and
median portfolio turnover on a per round basis (three months) is
19% (mean 77%), which means that the median investor turns over
his portfolio about twice in the survey period of 2.5 years, and
some turn over their portfolio ten times or more.2
We use the transaction records to develop several measures of
risk taking behavior. As we cannot directly observe the share of
risky assets as described in Eq. (4), we dene two alternatives that
cover different aspects of risk taking. First, we consider the balance
of purchases and sales of stocks in the trading records of investors,
as in most cases, extending ones equity position corresponds to an
increase in nancial risk taking, while a reduction of ones equity
position corresponds to a decrease in risk taking. We form two
ratios of buys divided by total trades, based on the number and
volume of investors equity transactions, respectively. The ratios
thus attain values between 0 and 1. Similar buysell ratios have
been used by Ritter (1988), Grinblatt and Keloharju (2000), and
Bhattacharya et al. (2012).
We expect buysell imbalance to be related to investors stock
market expectations: with high return expectations for the stock
market, the propensity to buy should rise relative to the propensity
to sell, while the opposite effect is predicted for high risk expectations and high risk aversion. More precisely, only changes in expectations and preferences should be relevant for changes in portfolios
(cp. Weber et al., 2013). However, as this is a stark theoretical
assumption, we analyze both levels and changes of expectations.
A second strategy to assess nancial risk taking of investors is
by measures of portfolio risk such as volatility and beta (cp. Dorn
and Huberman, 2005; Calvet et al., 2007; McInish, 1982). Financial
theory posits that the composition of the risky portfolio should not
change, but risk is entirely adjusted via the share of the risky portfolio (fund separation, Tobin, 1958). However, in practice there are
large differences in composition and risk of portfolios suggesting
that investors manage their overall risk taking at least in part by
portfolio risk. Therefore, we apply the theoretical predictions in
Eq. (4) also to portfolio risk measures, and expect higher portfolio
risk in response to a positive change in return expectations or a
negative change in risk expectations.
We calculate volatility of portfolios over one-year and over
three-month horizons. We calculate portfolio beta over a one-year
horizon using the FTSE all-share index as corresponding market
index (this choice seems justied as survey participants hold most
of their investments (>90%) in the UK stock market). Taking into
account that within a volatile market environment a large part of
the changes in portfolio volatility will be passively caused by
changes in market volatility, we also measure relative volatility
as the ratio of portfolio volatility divided by market volatility.
Dorn and Huberman (2010) argue that portfolio volatility is not
the correct measure of risk, if investors disregard correlations
between securities. They propose a value-weighted average of
the return volatilities of portfolio components (ACV), which
reects risk taking if investors mainly orient themselves at the volatility of individual securities rather than portfolio volatility. Again,
we consider levels and changes of these variables. Exact denitions
of all variables can be found in Appendix A.
2
Compared to similar studies, portfolio value is high. Glaser and Weber (2007)
report a median portfolio value of 15,630, Barber and Odean (2000) of $16,210,
and Dorn and Huberman (2005) of DM55,000 (about 23,000). On a monthly basis,
median turnover is in the same range as in Barber and Odean (2000) and Dorn and
Huberman (2005) with 6% and 9% respectively.
377

Table 2
Changes in expectations of investors.
Market
Own portfolio
Round
D risk tolerance
D num. return
D qual. return
D num. risk
D qual. risk
D num. return
D qual. return
D num. risk
D qual. risk
2
3
4
5
6
7
8
9
0.23
0.10
0.07
0.15
0.14
0.03
0.21
0.22
0.020
0.014
0.010
0.008
0.016
0.004
0.008
0.009
0.12
0.20
0.30
0.01
0.03
0.05
0.27
0.45
0.023
0.001
0.014
0.008
0.011
0.004
0.001
0.015
0.43
0.03
0.77
0.38
0.07
0.22
0.17
0.29
0.026
0.030
0.003
0.019
0.014
0.047
0.010
0.045
0.09
0.20
0.33
0.10
0.17
0.11
0.09
0.35
0.023
0.007
0.018
0.008
0.008
0.017
0.010
0.011
0.28
0.13
0.24
0.02
0.06
0.12
0.13
0.29
(December 08)
(March 09)
(June 09)
(September 09)
(December 09)
(March 10)
(June 10)
(December 10)
Notes: The table states changes in risk tolerance and changes in numerical and qualitative expectations of investors (compared to the previous survey round).
Changes are signicantly different from zero at 10%-level (one-sided t-test).
3.4. Descriptive statistics of investor risk-taking

For all rounds, average buysell ratios exceed 50%, which
implies that investors are net buyers. There is almost no difference
between ratios based on number of trades and volume, correlation
is 0.94 (p < 0:01). We observe the highest buysell imbalance for
late 2008, at the peak of the nancial crisis, when the ratios reach
about 0.66. This suggests that investors in our sample view the crisis as an opportunity to buy at low prices. There is also large crosssectional variation in buysell ratios between investors, which is
crucial for our analysis of the differential inuence of expectations
and preferences.
Fig. 4 displays portfolio volatilities of the median investor, the
rst-quartile investor, and third-quartile investor in our panel at
the time of each survey round. The volatility of the FTSE all-share
index serves for comparison. Median portfolio volatility in our
panel rises from 0.26 in June 2008 to about 0.40 during the crisis,
before falling to values around 0.18 for the last year of the survey.
It remains constantly above market volatility, which indicates that
a majority of investors hold portfolios that are riskier than the UK
market portfolio. The difference between median portfolio volatility and market volatility is strongly signicant for all rounds
(p < 0:01, Wilcoxon signed-rank test). The third quartile shows
that many investors hold very volatile portfolios compared to the
market index, while the rst quartile is still close to that index.
The average component volatility (ACV, not displayed) exceeds
these portfolio volatilities by about 40% as it does not account for
diversication effects.
0.7
0.6
0.5
3rd Qrt Investor

Median Investor
Market
1st Qrt Investor
0.4
0.3
0.2
0.1
0
Fig. 4. Portfolio volatility of investors and UK stock market volatility. Portfolio

volatility is the one-year standard deviation of daily portfolio returns at point in
time of survey rounds. Displayed are the median investor, the rst-quartile and
third-quartile investor. UK stock market volatility uses the FTSE all-share index.
High portfolio volatility of investors is not due to high levels of

systematic risk, as the median beta is around 0.8 over the whole
sample period and most investors hold portfolios with a beta smaller than one. Instead, high volatility is driven by idiosyncratic risk
as a result of a low degree of diversication. Relative volatilities
suggest that investors in the immediate phase of the nancial crisis
try to reduce their risk exposure relative to the market, while they
increase it again afterwards. Changes in beta conrm a reduction in
systematic risk for the rst phase of the crisis, while for later
rounds the results remain inconclusive.
4. Results
4.1. Investor trading behavior
We rst investigate whether market expectations drive the
decision of investors to increase or decrease their stock market
exposure, which is measured by buysell ratios. We estimate a
panel Tobit model with random effects as the buysell ratios are
limited on the interval between 0 and 1, and values on the boundaries occur frequently. We consider two specications, one in
which the absolute levels of expectations are relevant for investors,
and another in which investors are supposed to react on changes in
expectations.
Columns 1 and 5 of Table 3 show the results of the buysell
ratios regressed on expectation levels. More precisely, we measure
expectations at the time of the survey and then observe buysell
ratios in the three month afterwards until the next survey takes
place. Levels of expectations seem to have little effect on subsequent buying and selling behavior. Among the few marginally signicant effects is a negative coefcient for risk tolerance. An
explanation might be that risk tolerant investors already hold high
equity positions and tend to reduce their exposure during the
nancial crisis. However, this effect is not robust to the inclusion
of additional explanatory variables.
Changes in expectations are dened over the same time horizon
(between surveys), for which buysell ratios are calculated. The
lower number of observations in the changes regressions is due to
the fact that for changes in expectations, we need investors to participate in the survey for two consecutive rounds. Among the
changes variables, changes in numeric return expectations exert a
signicant effect on buysell behavior (columns 2 and 6). If return
expectations improve, investors tend to move to the buying side of
the market, which is consistent with the theoretic prediction. For
additional equity purchases thus not the absolute level of return
expectations is relevant, but instead changes in these expectations.
This result is robust to the inclusion of the levels variables (columns
3 and 7) and of demographic variables: age, gender, wealth, income,
and nancial literacy (columns 4 and 8). Income quite intuitively
378
Table 3
Buying and selling behavior.
Buysell ratio
(1)
Num. return
Num. risk
Qual. return
Qual. risk
Risk tolerance
D
D
D
D
D
Buysell volume ratio

(2)
0.068
0.059
0.014
0.007
0.010
0.150
0.058
0.010
0.007
0.012
num. return
num. risk
qual. return
qual. risk
risk tolerance
(3)
(4)
(5)
0.275
0.043
0.009
0.018
0.010
0.257
0.029
0.013
0.022
0.010
0.040
0.040
0.014
0.008
0.012
0.293
0.074
0.014
0.016
0.007
0.288
0.070
0.017
0.018
0.007
Age
Gender (male = 1)
Wealth
Income
Fin. literacy
Pseudo-R2
n
(6)
0.186
0.063
0.012
0.010
0.011
(7)
(8)
0.251
0.124
0.011
0.020
0.011
0.234
0.106
0.015
0.024
0.009
0.321
0.115
0.016
0.020
0.005
0.316
0.111
0.019
0.022
0.006
0.002
0.089
0.013
0.034
0.034
0.002
0.108
0.010
0.028
0.034
0.039
0.035
0.049
0.067
0.035
0.035
0.049
0.064
1376
769
769
767
1376
769
769
767
Notes: The table shows results of a panel Tobit regression with random effects and round dummies. Dependent variable is buysell ratio dened over number of trades (# of
buys/# of total trades) for columns 14 and buysell volume ratio dened over trading volume (buying volume/total trading volume) for columns 58. Columns 1 and 5
include levels of expectations and columns 2 and 6 changes of expectations as explanatory variables. Columns 3 and 7 show regressions on both, levels and changes, in
columns 4 and 8 additionally controlled for demographics. Demographic variables include age, gender, wealth, and nancial literacy. The table displays marginal effects,
which are the coefcients of the uncensored dependent variable. The pseudo-R2 is McKelvey and Zavoinas R2 .
They are signicant at 10%-level.
has a positive effect on buysell ratios as it is a proxy for additional

liquidity investors might want to invest. For the remaining demographic variables we nd no signicant effect. Overall the low
pseudo-R2 in the regressions suggests that the predictive power
of beliefs for immediate trading behavior is rather low.
The coefcients in Table 3 represent marginal effects (the coefcients for the latent variable), which directly allow an interpretation in terms of economic signicance. A 10%-point increase in
return expectations will raise buysell ratios by about three percent. For comparison, moving upward one category in income
has about the same effect. In unreported results, we exclude heavy
traders (the top 10% in number of trades and trading volume), as
these investors might be engaged in trading activity independent
of their current beliefs or other situational factors. When investors,
who trade less frequently, place an order, this order might be more
closely related to personal return and risk expectations. However,
there is almost no change in the results under this restriction.
For robustness, as the presented panel Tobit model cannot account
for potential heteroscedasticity, we test several alternatives: A linear panel regression with clustered standard errors by individual, a
xed effects regression, and a regression with least absolute deviation (LAD) estimators. The results are reported in Table 4.
Clustered standard errors take into account the non-independence of observations within our sample. Columns 1 and 4 conrm
the strongly positive impact of changes in return expectations. In a
xed effects model, results are less pronounced and only marginally
signicant as much of the cross-sectional variation is eliminated.
Part of the effect is picked up by changes in qualitative expectations.3 Finally, the LAD regression (columns 3 and 6) has favorable
small sample properties in reducing the importance of outliers. The
effect of changes in return expectations is robust to this specication.
4.2. Investor portfolio risk

We now turn to investor portfolio risk, which might be a more
stable measure of investor risk taking. In our analysis, we interpret
the volatility levels of investors portfolios when the survey takes
place as the level of risk an investor is taking at this point in time.
Consequently, changes in volatility correspond to changes in risk
taking.4 Similarly, we use levels and changes of other portfolio risk
measures (beta, relative volatility, average component volatility).
Panel A of Table 5 shows correlations between the levels of these
measures; all correlations are positive as they share a common concept of risk, but the variables also capture different aspects of risk as
correlations are not perfect. In particular, portfolio beta shows the
weakest relation to other risk measures with coefcients between
0.23 and 0.43. When considering changes (Panel B) the picture
becomes even more mixed. All but one correlation are still positive,
but especially for beta and three-month volatility (which is the only
measure calculated over a shorter time horizon) coefcients are
low. As portfolio risk measures differ, we consider most of them
in our regression analysis (except relative volatility which is redundant in the levels analysis). We take the natural logarithm of the
volatility variables, as volatilities are skewed within our sample.
We use market expectations as explanatory variables to avoid
reverse causality inherent with portfolio expectations, as current
portfolio volatility will determine expectations for future portfolio
returns and volatility. Table 6 shows the results of a panel GLS
regression with random effects and clustered standard errors (columns 14) and a xed effects regression (columns 58). We nd
that the risk level investors take on in their portfolios depends
on their expectations. In all regressions, a positive impact of
numerical return expectations on volatilities and a negative impact
3
The correlation between changes in numerical and qualitative expectations is
positive but low (0.26), suggesting that multicollinearity is not an issue. An
explanation for the emergence of the qualitative rating effect is that the used scale
lacks inter-subject comparability, but is a good predictor within subjects (xedeffects model).
4
This is a deliberate analogy to levels and changes in the hypothetical risk taking
task analyzed by Weber et al. (2013). In this task investors had to divide 100,000
between the FTSE all-share and a riskless asset. If we assume a volatility of 0 for the
riskless asset, the volatility of the chosen portfolio is monotonically increasing with
the fraction invested in the FTSE.
379

Table 4
Robustness tests: Buying and selling behavior
Buysell ratio
Buysell volume ratio
(1)
clus. SE
(2)
FE
(3)
LAD
(4)
clus. SE
(5)
FE
(6)
LAD
0.199
0.025
0.014
0.017
0.007
0.134
0.090
0.038
0.025
0.023
0.343
0.151
0.008
0.019
0.006
0.173
0.130
0.018
0.021
0.006
0.132
0.106
0.034
0.031
0.022
0.174
0.162
0.011
0.016
0.009
num. return
num. risk
qual. return
qual. risk
risk tolerance
0.230
0.034
0.019
0.013
0.006
0.179
0.006
0.033
0.015
0.005
0.252
0.025
0.026
0.004
0.006
0.276
0.091
0.024
0.019
0.004
0.239
0.127
0.035
0.023
0.006
0.255
0.020
0.028
0.010
0.002
Age
Gender (male = 1)
Wealth
Income
Fin. literacy
0.002
0.088
0.013
0.029
0.029
0.001
0.146
0.006
0.032
0.010
0.002
0.118
0.009
0.022
0.030
R2
n
0.055
0.073
0.047
0.053
0.075
0.043
767
769
767
767
769
767
Num. return
Num. risk
Qual. return
Qual. risk
Risk tolerance
D
D
D
D
D
0.002
0.136
0.012
0.030
0.023
Notes: The table shows results of a panel GLS regression with random effects and standard errors clustered by participant (columns 1 and 4), a panel regression with xed
effects (columns 2 and 5), and a regression using least absolute deviation and bootstrapped standard errors (columns 3 and 6). Dependent variable is buysell ratio dened
over number of trades (# of buys/# of total trades) for columns 13 and buysell volume ratio dened over trading volume (buying volume/total trading volume) for columns
46. Independent variables are as specied in Table 3. For random effects regressions overall R2 is reported, for xed effects regressions within R2 , for LAD-regressions
pseudo-R2 .
Coefcients are signicant at 10%-level.
Table 5
Correlation of portfolio risk measures.
Levels of portfolio risk
Panel A
Volatility 1y
Volatility 3m
Rel. volatility
Portfolio beta
ACV
Vol 1y
Vol 3m
Rel. Vol
Beta
ACV
1.00
0.76
0.89
0.42
0.64
1.00
0.59
0.28
0.54
1.00
0.43
0.50
1.00
0.23
1.00
Changes of portfolio risk
Panel B
D Volatility 1y
D Volatility 3m
D Rel. volatility
D Portfolio beta
D ACV
D Vol 1y
D Vol 3m
D Rel. Vol
D Beta
D ACV
1.00
0.60
0.39
0.13
0.60
1.00
0.11
0.05
0.32
1.00
0.40
0.06
1.00
0.05
1.00
Notes: The table shows pairwise Pearson correlations of levels (Panel A) and changes
(Panel B) of portfolio risk measures. All correlations are signicant at 1%-level.
of numerical risk expectations can be observed. Both effects are

signicant in most specications, the effects are weakest for portfolio beta (also conrmed by low R2 ). Risk tolerance and qualitative
expectations mostly have no predictive power for portfolio risk.
Among the demographic variables, we nd signicant effects for
age, wealth, and nancial literacy. Younger investors hold more
volatile portfolios, while wealthier investors tend to own less risky
portfolios.5 This result is consistent with the ndings of Dorn and
Huberman (2005).
Even though using market expectations addresses the most
obvious endogeneity problem, there might still be concerns that
5
We do not nd signicant results for portfolio value / wealth as a measure of
relative importance of investors portfolios for their overall wealth.
own portfolio risk determines also market expectations. Therefore,

we repeat the previous analysis using lagged expectations and
lagged preferences. The timing now is such that we use the expectations of each survey date to explain portfolio risk three month
later. Results in Table 7 conrm the impact of numerical return
and risk expectations on portfolio risk. The most notable difference
is that in the lagged regression risk tolerance has a more consistent
positive effect on risk taking, suggesting that it takes some time for
investors to implement their risk preferences.6
The interpretation in terms of economic signicance is straightforward, as the dependent variable is log transformed. 10%-points
higher return expectation will induce investors to hold a portfolio
with 1.65% higher volatility (1.27% for lagged expectations). Analogously, a 10%-points higher expected volatility relates to a 1.31%
decrease in portfolio volatility (1.16% for lagged expectations).
Up to this point, we dealt with state variables that give us some
information which portfolio risk investors choose depending on
their expectations, risk tolerance, and demographics. The panel
structure of our data allows us to investigate in more detail the
dynamics of these relationships. We now analyze changes of portfolio risk in response to contemporaneous changes in investor
expectations and preferences. The assumption is that investors in
addition to adjusting their risky share as suggested by Eq. (4) also
change portfolio composition. We adopt a parallel approach to the
levels regression and again estimate a random effects and a xed
effects model.
Table 8 shows the results of these regressions. With changes in
one-year portfolio volatility (columns 1 and 6) we observe the
same patterns as in the levels regression. Positive changes in
6
As a further test we instrument contemporaneous expectations by lagged
expectations. While the results are consistent in direction, signicance is weak.
However, instrumentation is costly in terms of statistical power, as it requires
consecutive observations. Additionally, there are concerns about weak instruments as
correlations between expectations and lagged expectations are only around 0.3
(common tests for weak instruments attain borderline results).
380
Table 6
Portfolio risk and expectations.
Random effects model
Fixed effects model
ln(Vol 1y)
(1)
ln(Vol 3m)
(2)
Beta
(3)
ln(ACV)
(4)
ln(Vol 1y)
(5)
ln(Vol 3m)
(6)
Beta
(7)
ln(ACV)
(8)
Num. return
Num. risk
Qual. return
Qual. risk
Risk tolerance
0.165
0.131
0.001
0.000
0.001
0.165
0.187
0.000
0.013
0.007
0.070
0.030
0.002
0.003
0.004
0.135
0.086
0.002
0.010
0.011
0.129
0.106
0.003
0.000
0.002
0.102
0.154
0.003
0.012
0.002
0.042
0.006
0.004
0.003
0.003
0.098
0.048
0.001
0.010
0.006
Age
Gender (male = 1)
Income
Wealth
Fin. literacy
0.006
0.085
0.016
0.031
0.066
0.006
0.115
0.014
0.030
0.066
0.002
0.104
0.004
0.016
0.054
0.003
0.059
0.018
0.022
0.059
R2
n
0.368
0.495
0.036
0.290
0.746
0.753
0.089
0.608
1924
1911
1926
1817
1924
1911
1926
1817
Notes: The table shows results of a GLS panel regression with random effects and clustered standard errors (columns 14) and xed effects (columns 58). All regressions
contain round dummies. Dependent variables are portfolio risk measures: the natural logarithm of portfolio volatility calculated over a 1-year and 3-month horizon, portfolio
beta and the log of average component volatility (ACV). Expectation variables and demographics are as dened before. For random effects regressions overall R2 is reported,
for xed effects regressions within R2 .
Table 7
Portfolio risk and lagged expectations.
Fixed effects model
ln(Vol 1y)
(1)
ln(Vol 3m)
(2)
Beta
(3)
ln(ACV)
(4)
ln(Vol 1y)
(5)
ln(Vol 3m)
(6)
Beta
(7)
ln(ACV)
(8)
0.127
0.116
0.005
0.007
0.006
0.101
0.136
0.006
0.022
0.013
0.060
0.048
0.008
0.001
0.003
0.188
0.147
0.002
0.008
0.014
0.092
0.095
0.007
0.006
0.005
0.012
0.080
0.008
0.020
0.009
0.036
0.034
0.010
0.000
0.002
0.144
0.119
0.002
0.006
0.011
Age
Gender (male = 1)
Income
Wealth
Fin. literacy
0.006
0.080
0.019
0.030
0.058
0.006
0.075
0.025
0.025
0.036
0.002
0.069
0.007
0.016
0.047
0.003
0.078
0.017
0.018
0.059
R2
n
0.400
0.542
0.027
0.297
0.794
0.772
0.134
0.641
1923
1908
1925
1808
1923
1908
1925
1808
Lagged
Lagged
Lagged
Lagged
Lagged
num. return
num. risk
qual. return
qual. risk
risk tolerance
contain round dummies. Dependent variables are portfolio risk measures: the natural logarithm of portfolio volatility calculated over a 1-year and 3-month horizon, portfolio
beta and the log of average component volatility (ACV). Expectation variables are included as lagged variables. For random effects regressions overall R2 is reported, for xed
effects regressions within R2 .
numerical return expectations are accompanied by increased risk

taking, while higher numerical risk expectations result in
decreased risk taking. In the regressions of changes in three-month
volatilities on changes in expectations (see Table 8), the coefcients for numerical expectations maintain their direction but no
longer reach statistical signicance. This may be due to the diminished statistical power of the changes regressions, as we can only
consider investors who participate in two consecutive survey
rounds. However, another interpretation is that investors have
rather long-term objectives and do not manage their portfolios
according to three-month volatilities. In our questionnaire, most
investors state an investment horizon of 35 years.
For relative volatility and average component volatility similar
patterns as for volatility emerge. In particular numerical return
expectations positively inuence risk taking. Changes in relative
volatility most closely reect investors active interventions to
alter portfolio risk, as raw portfolio volatility is in large part driven

by changes in market volatility. As already documented for levels,
beta is the risk measure least related to expectations. It is likely
that beta has little relevance to participants in managing the risk
of their portfolios. Many private investors may not even know
about this concept.
In unreported results, we substitute round dummies by market
volatility, which is constant across participants and will thus capture the part of changes in portfolio volatility caused by a passive
change in overall market volatility. In portfolio volatility regressions, the coefcient of market volatility is about 0.7, which means
that about 70% of changes in portfolio volatilities are driven by
changes in market volatility. Interestingly, changes in market volatility have a negative impact on relative volatility, suggesting
investors attempt to counteract rising market volatility by reducing their portfolio risk relative to the market.
381

Table 8
Changes in portfolio risk.
Fixed effects model
DVol 1y
(1)
DVol 3m
(2)
DRel. Vol
(3)
DBeta
(4)
DACV
(5)
DVol 1y
(6)
DVol 3m
(7)
DRel. Vol
(8)
DBeta
(9)
DACV
(10)
num. return
num. risk
qual. return
qual. risk
risk tolerance
0.062
0.079
0.003
0.001
0.005
0.114
0.128
0.005
0.008
0.008
0.059
0.073
0.000
0.000
0.004
0.000
0.017
0.005
0.002
0.000
0.080
0.020
0.006
0.003
0.004
0.051
0.072
0.007
0.000
0.004
0.107
0.149
0.014
0.010
0.009
0.047
0.069
0.004
0.000
0.002
0.009
0.005
0.001
0.000
0.000
0.083
0.014
0.008
0.005
0.004
Age
Gender (male = 1)
Income
Wealth
Fin. literacy
0.001
0.005
0.003
0.001
0.008
0.001
0.018
0.012
0.008
0.012
0.000
0.012
0.005
0.002
0.003
0.001
0.001
0.002
0.003
0.009
0.000
0.018
0.004
0.001
0.005
R2
n
0.682
0.652
0.246
0.141
0.417
0.733
0.682
0.275
0.170
0.394
1038
1031
1038
1009
1018
1038
1031
1038
1009
1018
D
D
D
D
D
contain round dummies. Dependent variables are changes in portfolio risk measures: changes in portfolio volatility calculated over a 1-year and 3-month horizon, changes in
relative volatility and portfolio beta, and changes in average component volatility (ACV). Independent variables are demographics and changes in expectations, both as
dened before. For random effects regressions overall R2 is reported, for xed effects regressions within R2 .
4.3. Volatility of trades

We combine the two approaches of measuring nancial risk
taking and examine the volatility of securities investors are trading. For this purpose, all securities traded by survey participants
(and for which a sufcient time series of returns is available) are
sorted by return volatility throughout the survey period. We form
ten volatility deciles and hereby establish a ranking of securities by
their relative riskiness. We then calculate the value-weighted average of volatility decile each investor trades in. We also compute the
volatility of purchases and the volatility differential between purchases and sales. The latter two measures we interpret as indicators of nancial risk taking as investors shift money to volatile
securities.
Table 9 shows population averages of volatility of trades, of volatility of purchases, and the average buysell volatility differential.
We observe that investors trade securities that are slightly more
volatile than the total sample of securities (which of course has
an average decile rank of 5.5). This is due to the fact that mutual
funds and ETFs are less frequently traded than more volatile securities such as stocks and options. Volatility of trades and purchases
is highest in the rst two rounds of the survey; these are also the
Table 9
Volatility of securities traded.
only rounds where the buysell volatility differential is positive

which conrms the earlier nding that private investors in our
sample seem to view the crisis as an opportunity to buy risky securities. This behavior then turns around, in particular for a period of
high stock market gains in mid-2009 (cp. also Fig. 1). Investors
move back into safer securities, a behavior that repeats itself for
the nal survey rounds, for which the average volatility of trades
is lowest.
When we regress the three measures dened above on the levels of investors expectations and risk tolerance (Table 10), we nd
no effect on overall trade volatility, a slight effect on the volatility
of purchases and a pronounced impact on the buysell volatility
differential. This means that investors shift capital towards riskier
securities when they have high return expectations. This extends
the results of the previous section, as we now learn how investors
adjust their portfolio volatility in response to positive expectations: they buy high volatility securities and sell low volatility
securities. We also nd that less risk-averse investors buy securities with higher volatility, in line with risk habitat theory which
states that investors select securities of which volatilities are commensurate with their risk aversion (Dorn and Huberman, 2010).7
Again older, wealthier, and more sophisticated investors trade less
volatile securities. We do not report results for a regression on
changes in expectations in this case, as we nd no signicant results.
4.4. Selection effects
Round
Trade
volatility
Buy
volatility
Buysell
vol. diff.
Pre-survey (June 08September 08)

Round 1 (September 08December 08)
Round 2 (December 08March 09)
Round 3 (March 09June 09)
Round 4 (June 09September 09)
Round 6 (December 09March 10)
Round 7 (March 10June 10)
Round 8 (June 10September 10)
6.20
6.23
6.02
5.99
6.00
5.96
6.03
5.75
5.86
5.82
6.42
6.37
6.05
5.79
5.88
6.07
6.04
5.60
5.84
5.76
0.33
0.29
0.24
0.39
0.32
0.02
0.06
0.36
0.29
0.23
Notes: The table shows for all survey rounds the average volatility decile of trades
and purchases, and the average volatility differential between purchases and sales.
This difference is signicant by a Wilcoxon signed-rank test at 10%-level.
Our sample is clearly not representative, neither for the total UK

population, nor for UK stock market investors, maybe not even for
Barclays online brokerage clients. We make no claim in this
regard. However, we do believe that our data are meaningful and
allow to draw some inferences about investing behavior in
response to personal expectations and preferences. While one
has to be careful not to overgeneralize our ndings, we have no
evidence of systematic selection in our sample, which would invalidate our results. In this section we analyze selection issues in a
formal way.
7
There is no such effect for buysell volatility differentials, but this is not
surprising as both risk-averse and risk tolerant investors will sometimes augment and
sometimes reduce risk (though on different levels).
382
Table 10
Volatility of trades explained by expectations
Fixed effects model
Trade volatility
(1)
Buy volatility
(2)
Buysell vol. diff.

(3)
Trade volatility
(4)
Buy volatility
(5)
Buysell vol. diff.

(6)
Num. return
Num. risk
Qual. return
Qual. risk
Risk tolerance
0.467
0.042
0.003
0.038
0.045
0.840
0.431
0.002
0.031
0.082
1.374
0.377
0.001
0.049
0.050
0.127
0.201
0.027
0.035
0.015
0.514
0.199
0.038
0.037
0.001
2.367
0.421
0.051
0.089
0.065
Age
Gender (male = 1)
Income
Wealth
Fin. literacy
0.017
0.704
0.047
0.102
0.391
0.017
0.604
0.048
0.116
0.376
0.010
0.077
0.023
0.031
0.005
R2
n
0.085
0.108
0.041
0.010
0.021
0.049
1467
1343
890
1467
1343
890
Notes: The table shows results of panel regression with random effects with clustered standard errors (columns 13) or xed effects (columns 46), all regressions contain
round dummies. Dependent variables are the volatility of trades, the volatility of purchases and the difference between volatility of purchases and sales. For random effects
regressions overall R2 is reported, for xed effects regressions within R2 .
Given the relatively low (but not uncommon8) response rate and
the presence of attrition in our panel, there are two potential channels of selection. Specic investors might be more attracted to participate in the survey, or they leave and rejoin the sample in a
non-random way, both potentially biasing our results. We have only
limited data on non-participants, including age and gender, as well
as some portfolio information (portfolio value, number of positions,
number of transactions).9 We use these items as explanatory variables in a participation regression, results are reported in column 1
of Table 11. We nd that male investors and investors with a higher
number of holdings and transactions are more likely to participate in
the survey. The latter are potentially more active and interested in
nancial markets, which would explain this result.
While this supports the presence of selection on observables in
our sample, it may remain inconsequential for our results. We run
a two-stage Heckmann selection model to test for this possibility.
In columns 2a and 2b, we reproduce the regression of portfolio
value on expectations including the inverse Mills ratio of the rst
stage. The inverse Mills ratio is highly signicant, again suggesting
a selection effect. However, our main result regarding the inuence
of expected return and expected risk on risk taking remains intact.
It is also robust to an inclusion of the set of variables from the participation regression (column 2b). Not surprisingly, portfolio value
and number of positions are strongly negatively related to portfolio
volatility, as they come along with a diversication effect. In contrast, number of transactions has a positive effect on volatility. In
this specication, the signicance of the inverse Mills ratio is much
reduced, as the additional variables capture part of the selection
effect.
We nd similar results for the other levels specications, meaning that despite selection is present in our sample, our results are
mainly unaffected by it. The changes regressions, by making use
of the in-sample variation over time, are per se less vulnerable
against this type of selection.
Next, we analyze the participation in the panel over time to
detect any signs of systematic attrition. To make sure that this type
8
In similar survey studies Graham and Harvey (2001) report a response rate of 9%,
Glaser and Weber (2007) of 7%, Dorn and Sengmueller (2009) of 6%, compared to our
3% for a repeated survey.
9
The remaining demographic variables such as income and wealth were selfreported survey items.
of selection does not bias our main results we again use a Heckman
selection model. We follow Wooldridge (1995) in estimating the
participation equation separately for each round of the panel,
including demographics and lagged survey variables. Instead of
displaying these roundwise rst stage regressions, Table 11 shows
a panel probit version of the participation regression (column 3). It
demonstrates that wealthier investors are more likely to participate, while higher income investors are less likely to participate.
Intuitively, those with higher income might be more time-constraint. More importantly, lagged expectations do not explain subsequent participation, which means that it is not the case that e.g.,
optimistic or more risk tolerant investors are more likely to continue the survey.
We then re-estimate in the second stage the panel regression as
before, including now inverse Mills ratios from the roundwise participation regressions. This time, we nd no signicance for the Mills
ratio, suggesting no strong evidence for selection effects in the sense
of systematic panel attrition. Our main results are unchanged in
both specications, whether using random effects (4a) or xed
effects (4b). We also nd no evidence that the changes regression
of Table 8 is affected by selection. We thus conclude that while selection is present in our sample, it seems to have little inuence on the
effect of expectations and preferences on risk taking behavior.
5. Discussion
A main problem any research in beliefs and expectations
encounters is whether responses in a survey are valid representations of the internal beliefs of participants. The challenge is twofold, questions need to be stated in a way that participants are
able to answer them in a sensible way, and participants need to
be motivated to do so. For the latter we rely on the intrinsic motivation of participants as they completed the survey voluntarily,
and many found it interesting enough to take part multiple times.
As in most large-scale surveys, monetary incentives were not feasible, but we are in this case not aware of any obvious reason to
conceal or distort beliefs in their absence.10 Additionally, we build
10
For a discussion about when monetary incentives are useful see Camerer and
Hogarth (1999). Other surveys that do not incentivize participants include the
Michigan Survey of Consumer Finances, the German Socioeconomic Panel and most
surveys on investing behavior.
383

Table 11
Sample selection.
Participation
Part.
ln(Vol 1y)
Part.
ln(Vol 1y)
(1)
(2a)
(2b)
(3)
(4a)
(4b)
0.131
0.128
0.005
0.004
0.002
0.136
0.136
0.007
0.002
0.001
0.065
0.315
0.006
0.010
0.002
0.129
0.152
0.003
0.001
0.005
0.106
0.121
0.004
0.000
0.003
0.006
0.219
0.014
0.031
0.040
0.004
0.117
0.015
0.016
0.033
0.004
0.192
0.122
0.115
0.043
0.004
0.098
0.007
0.042
0.022
0.001
0.008
1033
1033
Num. return
Num. risk
Qual. return
Qual. risk
Risk tolerance
Age
Gender (male = 1)
Income
Wealth
Fin. literacy
0.000
0.168
Portfolio value
Portfolio positions
Transactions
0.015
0.062
0.064
Inv. Mills ratio

n
Panel attrition
19,609
0.046
0.109
0.044
0.594
0.354
1536
1518
1825
Notes: The table shows two-stage Heckman selection models for participation in the survey and panel attrition. Column 1 displays a probit regression of participation
including age and gender, and portfolio value, portfolio positions, and transactions (all logarithmized). Columns 2a and 2b reproduce results of Table 6 including the inversed
Mills ratio of the rst stage. Column 3 shows a probit regression for participation within survey, columns 4a and 4b the associated second stage estimated with random effects
(4a) and xed effects (4b).
on the nding of Weber et al. (2013)who use the same survey

that the elicited expectations are effective and consistent predictors
of decisions, which should attenuate concerns about their validity.
The other concern that participants might not be able to express
their beliefs in the question format provided to them is taken into
account by the use of both, numerical and qualitative elicitation of
expectations. While the numerical estimates are more demanding,
in particular with respect to condence intervals, they have the
advantage of being comparable across participants. On the other
hand, qualitative estimates may capture aspects of value and risk
not comprised in the rst two moments of a distribution. Interestingly, we nd with rare exceptions that only numerical expectations are relevant for actual nancial risk taking decisions, which
is in contrast to the results of Weber et al. (2013) who establish
a strong inuence of qualitative expectations on allocations in
the hypothetical investment task. We test whether the explanatory
power of numerical expectations changes over to qualitative
expectations if we drop numerical expectations from the regressions. In general, this is not the case and the impact of qualitative
expectations remains weak. When in turn qualitative expectations
are excluded, our results are unchanged.
An explanation for this nding has to consider the decision process in the hypothetical investment task compared to actual
investing. First of all, our measures of nancial risk taking are only
weakly correlated with the proportion of risky investment in the
survey task, which already hints at the two being different. In particular, the changes of risk taking in the task and investors portfolios are unrelated. We conjecture that the qualitative expectations
are affective evaluations of the market situation, while the numerical estimates draw more on cognitive resources (cp. Kuhnen and
Knutson, 2011). We would then expect these evaluations to be predictive for decisions that are made in the same mode of
thinking.11 If the actual investment decisions of investors are preceded by a more deliberate thought process than the allocations in
the hypothetical task, this would at least partly explain the greater
11
Support for this dual-process theories of information processing and decision
making can be found, e.g., in Kahneman (2003).
predictive power of numerical expectations for these decisions. As

we cannot fully explore the underlying mechanisms, this might be
an interesting avenue for future research.
We also consider the time structure of expectations and trading,
and throughout the paper we opted for an approach that tries to
explain changes in investing behavior by contemporaneous
changes in expectations. Another possibility would be that investors need some time to react on changes in expectations, for example because of inertia. When we use lagged level variables many of
the described relationships between expectations and investment
behavior can still be observed (cp. Table 7).12 However, the effects
are in general about equal or weaker than for contemporaneous
expectations. We thus conclude that investors tend to implement
their beliefs in a timely manner.
Instead of studying return and risk expectations separately, Eq.
(4) can also be interpreted in terms of Sharpe ratios. A higher
expected Sharpe ratio then implies a higher share in the risky
investment. We calculate expected Sharpe ratios using investors
market return expectations, the three-month risk-free rate (represented by the LIBOR), and the expected market volatility from
investors condence intervals.13 We nd that higher levels of
Sharpe ratios are related to higher portfolio risk, supporting our
results from Tables 6 and 7. However, there are in general no significant results for changes in Sharpe ratios. We attribute this to the fact
that Sharpe ratios are a combination of several survey items, each
subject to noise and measurement error. In particular for changes,
such constructs may become unreliable.
As a complement to our research, the investor survey of
Hoffmann et al. (2013) has an overlap of seven month with our
data. Elicited expectations and portfolio characteristics show some
similarities: For instance return expectations of Dutch investors
also rise from September to December 2008 and further to March
2009, and trading and buying activity initially increases in
response to the crisis. Similarly to us, Hoffmann et al. (2013) nd
12
A similar analysis for changes is precluded by the fact, that a change Dt1;t is
mechanically (negatively) correlated with Dt;t1 over the shared observation in t.
13
As negative Sharpe ratios are not well-dened, we have to exclude observations
where Er s < r f .
384
that median portfolio volatility is higher than market volatility and

closely tracks the market index. However, there are some differences as well, e.g., risk perceptions fall gradually after a peak in
September 2008, while in our data they rise and then stay on a high
level until March 2009. This might be due to the different wording
of the question, which in Hoffmann et al. (2013) refers to current
risk perception, while our approach is more forward looking. Nevertheless, taken together the ndings suggest that there exist some
more general properties in expectations of private investors that
are not limited to a particular dataset.
In a regression of buysell ratios on beliefs and preferences,
Hoffmann et al. (2013) use qualitative measures of expectations as
explanatory variables (which in our case remained insignicant).
They demonstrate a signicantly positive inuence of risk tolerance
on buying activity, but nd little effect of return and risk expectations (levels and changes). This contributes to our impression that
immediate trading behavior is hard to predict from elicited beliefs.
For portfolio volatility, both datasets share the intuitive positive
result for risk tolerance and the insignicant result for qualitative
return expectations. However, Hoffmann et al. (2013) identify a positive effect of risk perception on portfolio volatility. They explain this
result by investors being aware of the risk of their investment portfolio, which suggests a reverse causality from portfolio risk to risk
perception. Our ndings for numerical risk expectations strongly
point in an opposite direction, i.e., investors taking less risk when
they perceive risk to be higher. This discrepancy might again be a
result of different measurement, as we strictly distinguish between
portfolio and market expectations, and use the latter in our regressions to avoid reverse causality. Risk perception in Hoffmann et al.
(2013) refers more general to riskiness of investing.
6. Conclusion
We investigate the functional relationship between beliefs and
preferences of investors and their trading behavior. While we are
still far from suggesting a denite functional form in the spirit of
Eq. (4), our ndings are a rst step to improve the understanding
of this complicated but fundamental relationship. We provide evidence that expectations are relevant for risk taking of investors,
and that they are used in a predominantly rational and intuitive way.
Higher return expectations lead to increased risk taking in
terms of portfolio volatility among investors, while higher risk
expectations have the opposite effect. Even more, changes in portfolio risk are predicted by contemporaneous changes in return and
risk expectations. We nd evidence that investors counteract
changes in market volatility by adjusting their portfolio volatility
relative to the market. In general, the best t of our model is

achieved for long-term portfolio volatility. Changes in short-term
portfolio volatility and changes in portfolio beta are less well or
not at all predicted by changes in expectations. This relates directly
to the question how private investors manage their portfolio risk
and which risk measure is closest to their subjective experience
of risk. As long-term volatility measures react strongest to investor
expectations, we take this as tentative evidence that they are a
good proxy for experienced risk.
Expectations have less predictive power for immediate trading
activity of investors. We nd a positive effect of return expectations on equity buying activity, which proxies for an adjustment
of the (unobserved) risky share. However, trading is often noisy,
inuenced by liquidity and other exogenous trading motives,
which might be a reason why we nd no inuence of risk expectations and preferences. Investors also engage in risk shifting within
their portfolio, replacing less volatile securities by more volatile
ones. We infer that contrary to two-fund separation investors use
several channels to adjust their risky position. They not only
increase or decrease a xed risky portfolio, but also change this
risky portfolio according to their expectations.
Taken together, our results suggest that nancial theory in general correctly predicts the role of return and risk expectations for
actual trading behavior. Private investors take their expectations
into account to determine whether to buy or sell and whether to
increase or decrease portfolio risk. But at the same time investors
reaction to expectations and preferences is more nuanced and
more ambiguous than in the theoretical model. Not only do individual investors use different ways to alter their investment risk,
but also some nancial risk measures such as equity beta seem
to bear little relevance for them. Instead, we conjecture that a multitude of other factors, which to describe and identify is beyond the
scope of this paper, play a role in investment decisions.
Acknowledgements
We are grateful to Barclays Stockbrokers for providing access to
their online investor client base, and to Barclays Behavioural
Finance team for joint design and execution of the survey. We
thank Daniel Egan, Christian Ehm, Greg Davies, Victor Fleisher,
Alen Nosic, participants of the 2011 Boulder Summer Conference
on Consumer Financial Decision Making and the 2011 SPUDM Conference, and seminar participants in Mannheim and Luxemburg.
For research assistance we thank Robin Cindori. Research reported
in this article was supported by the Observatoire de lEpargne
Europenne (OEE) and Deutsche Forschungsgemeinschaft (DFG,
Grant We993).
Appendix A
Description of variables
Variable
Origin
Description
Num. return
Survey
Num. risk
Survey
Qual. return
Survey
Return in % in response to survey question We would like you to make three estimates of the return of the
UK stock market (FTSE all-share) by the end of the next three month. Your best estimate should be your best
guess
Volatility calculated from condence intervals using the methodology of Keefer and Bodily (1983)
using responses to survey question We would like you to make three estimates of the return of the UK
stock market (FTSE all-share) by the end of the next 3 month. Your high estimate should very rarely be lower
than the actual outcome of the FTSE all-share (about once in 20 occasions). Your low estimate should very
rarely be higher than the actual outcome of the FTSE all-share (about once in 20 occasions)
Rating on scale 17 in response to question How would you rate the returns you expect from an
investment in the UK stock market (FTSE all-share) over the next 3 months?
385
Appendix A (continued)
Description of variables
Variable
Origin
Description
Qual. risk
Survey
Risk tolerance
Survey
D num. return
D num. risk
D qual. return
D qual. risk
D risk tolerance
Age
Gender
Wealth
Survey
Survey
Survey
Survey
Survey
Bank data
Bank data
Survey
Income
Survey
Fin. literacy
Survey
Buysell ratio
Buysell
volume ratio
Volatility 1y
Volatility 3m
Rel. Volatility
Portfolio beta
Bank data
Bank data
Rating on scale 17 in response to question Over the next 3-months, how risky do you think the UK stock
market (FTSE all-share) is?
Agreement on Likert scale 17 to statement It is likely I would invest a signicant sum in a high risk
investment
Num. return (t) num. return (t 1)
Num. risk (t) num. risk (t 1)
Qual. return (t) qual. return (t 1)
Qual. risk (t) qual. risk (t 1)
Risk tolerance (t) risk tolerance (t 1)
Age of participants in years
Gender of participants, dummy variable 1 if male, 0 if female
Self-reported wealth using nine categories provided in the survey: 010,000; 10,00150,000;
50,001100,000; 100,001150,000; 150,001250,000; 250,001400,000; 400,001600,000;
600,0011,000,000; >1,000,000. Missing values were imputed
Self-reported income using eight categories provided in the survey: 020,000; 20,00130,000;
30,00150,000; 50,00175,000; 75,001100,000; 100,001150,000; 150,001200,000;
>200,000. Missing values were imputed
Number of correct responses in a 4-item nancial literacy test using questions by van Rooij et al.
(2011)
Number of purchases divided by number of total trades (range 01)
Volume of purchases divided by total trading volume (range 01)
ACV
Bank data
Trade volatility
Bank data
Buy volatility
Bank data
Buysell vol.
diff.
Portfolio value
Bank data
Portfolio
positions
Transactions
Bank
Bank
Bank
Bank
data
data
data
data
Bank data
Bank data
Bank data
One-year historical portfolio volatility at time t

Three-month historical portfolio volatility at time t
One-year historical portfolio volatility divided by one-year historical market volatility at time t
One-year historical portfolio beta from a one factor model using the FTSE all-share index as
corresponding market index and the LIBOR as riskfree rate
Average component volatility calculated using a weighted average of one-year historical volatility of
portfolio components owned at time t
Weighted average of volatility deciles for all securities traded between t an t + 1. Volatility is calculated
over total survey period and sorted into deciles
Weighted average of volatility deciles for all securities purchased between t an t + 1. Volatility is
calculated over total survey period and sorted into deciles
Difference of volatility for securities purchased and volatility for securities sold between t and t + 1.
Volatility is calculated over total survey period and sorted into deciles
Portfolio value before the start of the survey (participants and non-participants), and at each survey
round (only participants)
Number of holdings before the start of the survey (participants and non-participants), and at each
survey round (only participants)
Transactions in the year before the survey start (participants and non-participants), and between
survey round (only participants)
References
Amromin, G., Sharpe, S.A., 2009. Expectations of Risk and Return among Household
Investors: Are their Sharpe Ratios Countercyclical? Working Paper.
Barber, B.M., Odean, T., 2000. Trading is hazardous to your wealth: the common
stock investment performance of individual investors. The Journal of Finance 55
(2), 773806.
Barber, B.M., Odean, T., 2001. Boys will be boys: gender, overcondence, and
common stock investment. Quarterly Journal of Economics 116 (1), 261292.
Bhattacharya, U., Holden, C.W., Jacobsen, S., 2012. Penny wise, dollar foolish:
buysell imbalances on and around round numbers. Management Science 58 (2),
413431.
Calvet, L.E., Campbell, J.Y., Sodini, P., 2007. Down or out: assessing the welfare costs
of household investment mistakes. Journal of Political Economy 115 (5), 707
747.
Calvet, L.E., Campbell, J.Y., Sodini, P., 2009. Fight or ight? Portfolio rebalancing by
individual investors. The Quarterly Journal of Economics 124 (1), 301348.
Camerer, C.F., Hogarth, R.M., 1999. The effects of nancial incentives in
experiments: a review and capitallaborproduction framework. Journal of
Risk and Uncertainty 19 (13), 742.
Campbell, J.Y., Viceira, L.M., 2002. Portfolio Choice for Long-Term Investors. Oxford
University Press, New York, NY.
Dave, C., Eckel, C.C., Johnson, C.A., Rojas, C., 2010. Eliciting risk preferences: when is
simple better? Journal of Risk and Uncertainty 41 (3), 219243.
Dorn, D., Huberman, G., 2005. Talk and action: what individual investors say and
what they do. Review of Finance 9 (4), 437481.
Dorn, D., Huberman, G., 2010. Preferred risk habitat of individual investors. Journal
of Financial Economics 97 (1), 155173.
Dorn, D., Sengmueller, P., 2009. Trading as entertainment? Management Science 55
(4), 591603.
Egan, D., Davies, G.B., Brooks, P., 2010. Comparisons of risk attitudes across
individuals. In: Cochran, J. (Ed.), Wiley Encyclopedia of Operations Research and
Management Science. John Wiley and Sons, Hoboken, NJ.
Ehm, C., Kaufmann, C., Weber, M., 2014. Volatility inadaptability: investors care
about risk, but cant cope with volatility. Review of Finance (in press).
Glaser, M., Langer, T., Weber, M., 2013. True overcondence in interval estimates:
evidence based on a new measure of miscalibration. Journal of Behavioral
Decision Making 26 (5), 405417.
Glaser, M., Weber, M., 2005. September 11 and stock return expectations of
individual investors. Review of Finance 9 (2), 243279.
Glaser, M., Weber, M., 2007. Overcondence and trading volume. The GENEVA Risk
and Insurance Review 32 (1), 136.
Goetzmann, W.N., Kumar, A., 2008. Equity portfolio diversication. Review of
Finance 12 (3), 433463.
386
Graham, J.R., Harvey, C.R., 2001. The theory and practice of corporate nance:
evidence from the eld. Journal of Financial Economics 60 (23), 187243.
Grinblatt, M., Keloharju, M., 2000. The investment behavior and performance of
various investor types: a study of Finlands unique data set. Journal of Financial
Economics 55 (1), 4367.
Guiso, L., Sapienza, P., Zingales, L., 2011. Time Varying Risk Aversion. Working
Paper.
Hoffmann, A.O.I., Post, T., Pennings, J.M.E., 2013. Individual investor perceptions and
behavior during the nancial crisis. Journal of Banking and Finance 37 (1), 60
74.
Kahneman, D., 2003. Maps of bounded rationality: a perspective on intuitive
judgement and choice. The American Economic Review 93 (5), 14491475.
Kapteyn, A., Teppa, F., 2011. Subjective measures of risk aversion, xed costs, and
portfolio choice. Journal of Economic Psychology 32 (4), 564580.
Keefer, D.L., Bodily, S.E., 1983. Three-point approximations for continuous random
variables. Management Science 29 (5), 595609.
Kempf, A., Merkle, C., Niessen, A., 2014. Low risk and high return affective
attitudes and stock market expectations. European Financial Management (in
press).
Kuhnen, C.M., Knutson, B., 2011. The inuence of affect on beliefs, preferences, and
nancial decisions. Journal of Financial and Quantitative Analysis 46 (3), 605
626.
Kumar, A., 2009. Who gambles in the stock market? The Journal of Finance 64 (4),
18891933.
Lewellen, W.G., Lease, R.C., Schlarbaum, G.G., 1977. Patterns of investment strategy and
behavior among individual investors. The Journal of Business 50 (3), 296333.
Loewenstein, G.F., Hsee, C.K., Weber, E.U., Welch, N., 2001. Risk as feelings.
Psychological Bulletin 127 (2), 267286.
Markowitz, H., 1952. Portfolio selection. The Journal of Finance 7 (1), 7791.
McInish, T.H., 1982. Individual investors and risk-taking. The Journal of Economic
Psychology 2 (2), 125136.
Merkle, C., 2012. Financial Overcondence Over Time Foresight, Hindsight, and
Insight of Investors. Working Paper.
Odean, T., 1998. Volume, volatility, price, and prot when all traders are above
average. The Journal of Finance 53 (6), 18871934.
Odean, T., 1999. Do investors trade too much? The American Economic Review 89
(5), 12791298.
Ritter, J.R., 1988. The buying and selling behavior of individual investors at the turn
of the year. The Journal of Finance 43 (3), 701717.
Samuelson, P.A., 1938. A note on the pure theory of consumers behaviour.
Economica 5 (17), 6171.
Sarin, R.K., Weber, M., 1993. Risk-value models. European Journal of Operational
Research 70 (2), 135149.
Tobin, J., 1958. Liquidity preference as behavior towards risk. The Review of
Economic Studies 25 (2), 6586.
van Rooij, M., Lusardi, A., Alessie, R., 2011. Financial literacy and stock market
participation. Journal of Financial Economics 101 (2), 449472.
Vissing-Jorgensen, A., 2003. Perspectives on behavioral nance: does irrationality
disappear with wealth? Evidence from expectations and actions. In: Gertler,
Rogoff (Eds.), NBER Macroeconomics Annual 2003, vol. 18. The MIT Press,
Cambridge, MA, pp. 139208.
Weber, M., Weber, E.U., Nosic, A., 2013. Who takes risk and why: determinants of
changes in investor risk taking. Review of Finance 17 (3), 847883.
Windschitl, P.D., Wells, G.L., 1996. Measuring psychological uncertainty: verbal
versus numeric methods. Journal of Experimental Psychology 2 (4), 343364.
Wooldridge, J.M., 1995. Selection corrections for panel data models under
conditional mean independence assumptions. Journal of Econometrics 98 (1),
115132.

1 s2.0 S0378426614001253 Main PDF

Uploaded by

Copyright:

Available Formats

1 s2.0 S0378426614001253 Main PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0378426614001253 Main PDF

Uploaded by

Copyright:

Available Formats

Journal of Banking & Finance 46 (2014) 372386

Contents lists available at ScienceDirect

Journal of Banking & Finance

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

brokerage accounts. We observe volume, timing, and direction of

2. Theory and literature

max E0 W 0 1 r 0;1 1h =1 h:

where r20 is the conditional variance of the log return,

of the maximization problem are visible: the investor trades off

which can be solved by

E0 r s;0;1 r f ;0;1 r20 =2

The equation implies that the share of risky investment should

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Age (in years)

Notes: The table shows descriptive statistics about demographics of participants.

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

comparison see again Weber et al., 2013). However, they closely

Risk tolerance of investors is measured as agreement to the

3.2. Survey responses

Fig. 2. Numerical return expectations of investors. Average quarterly return

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

trading volume is 72,805. We observe most pronounced trading

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Changes are signicantly different from zero at 10%-level (one-sided t-test).

Changes are signicantly different from zero at 5%-level (one-sided t-test).

Changes are signicantly different from zero at 1%-level (one-sided t-test).

3.4. Descriptive statistics of investor risk-taking

3rd Qrt Investor

Fig. 4. Portfolio volatility of investors and UK stock market volatility. Portfolio

High portfolio volatility of investors is not due to high levels of

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Buysell volume ratio

They are signicant at 10%-level.

They are signicant at 5%-level.

They are signicant at 1%-level.

has a positive effect on buysell ratios as it is a proxy for additional

4.2. Investor portfolio risk

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Buysell volume ratio

Coefcients are signicant at 10%-level.

Coefcients are signicant at 5%-level.

Coefcients are signicant at 1%-level.

Changes of portfolio risk

of numerical risk expectations can be observed. Both effects are

own portfolio risk determines also market expectations. Therefore,

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Fixed effects model

Coefcients are signicant at 10%-level.

Coefcients are signicant at 5%-level.

Coefcients are signicant at 1%-level.

Fixed effects model

Coefcients are signicant at 10%-level.

Coefcients are signicant at 5%-level.

Coefcients are signicant at 1%-level.

numerical return expectations are accompanied by increased risk

alter portfolio risk, as raw portfolio volatility is in large part driven

C. Merkle, M. Weber / Journal of Banking & Finance 46 (2014) 372386

Fixed effects model

Coefcients are signicant at 10%-level.