Stock Market Volatility and Changes in Financial Risk

Tolerance During the Great Recession

Abed G. Rabbani,a John E. Grable,b Wookjae Heo,c Liana Nobre,d and Stephen Kuzniake

This study investigated the degree to which the financial risk tolerance of individuals was influenced by volatility
in the U.S. equities market during the period of the Great Recession. Based on data from a valid and reliable
risk tolerance scale and return information for the Standard and Poor’s (S&P) 500 index, there does appear
to be some associations between daily market volatility and changes in risk tolerance scores. Changes in risk
tolerance scores were also calculated using short- and intermediate-term volatility measures. The relationships
do vary, however, with evidence supporting the relationship only 64% of the time. Overall, changes in financial
risk tolerance scores were found to be modest. Although not following hypothesized directions at all times, risk
tolerance was not influenced by the length of volatility measurements.

Keywords: Great Recession, risk taking, risk tolerance, volatility

T
here is a general consensus among research- Many economists mathematically measure risk tolerance as
ers that risk tolerance involves the willingness, the reciprocal of risk aversion (Barsky, Juster, Kimball, &
rather than preference or capacity, to trade off Shapiro, 1997; Gron & Winton, 2001; Walls & Dyer, 1996),
the possibility of incurring an almost certain small gain which describes an investor as risk averse if she always
with the potential of making a larger gain with an equally prefers a sure wealth level to a lottery, as risk seeking if
high potential of losing wealth (Davies & Brooks, 2013; opposite, and as risk neutral if indifferent. Within modern
Grable & Joo, 2004; Pan & Statman, 2012). Given that portfolio theory and the capital asset pricing model, when
wealth is associated with income and expenditures, it is the level of risk aversion is identified, investors can select
a person’s willingness to accept variability in lifetime in- their preferred allocation from several efficient portfolios.
come (Guillemette, Finke, & Gilliam, 2012; Hanna, Fan, In practice, self-developed or commercially available ques-
& Chang, 1995). tionnaires are used widely to frame financial management
recommendations that provide the highest level of benefit
Financial risk tolerance has emerged as a very impor- for a given level of risk.
tant variable needed to match the suitability of finan-
cial management recommendations to the clients’ needs It was speculated in the public media that the Great Reces-
(Hung et al., 2008). Risk tolerance evaluation became a sion (from late 2007 to early 2010) prompted households
requirement in Australia since 2001 (Gilliam, Chatterjee, to become less risk tolerant in reaction to volatility in the
& Grable, 2010) and in United States since at least 2006. securities markets. The Standard and Poor’s (S&P) 500 fell
However, Bhattacharya, Hackethal, Kaesler, Loos, and 51.7% from its high point during this period (Guillemette &
Meyer (2012) and Roth and Voskort (2014) highlighted Finke, 2014). Some financial pundits question the validity
the danger of using a measurement tool that is not valid of risk tolerance questionnaires as a tool for use within the
or reliable. financial counseling and planning process (e.g., Roth, 2013),

a
Assistant Professor, Department of Personal Financial Planning, 239E Stanley Hall, University of Missouri, Columbia, MO 65203. E-mail: rabbania@
missouri.edu
b
Professor, Department of Financial Planning, Housing, and Consumer Economics, University of Georgia, Athens, GA 30602. E-mail: [email protected]
c
Assistant Professor, Department of Consumer Science, South Dakota State University, Brookings, SD. E-mail: [email protected]
d
Adjunct Professor, Universidade Federal Rural do Semi-Árido, Mossoro, RN, Brazil. E-mail: [email protected]
e
Financial Planner, Cannon Financial Strategists, Inc. 649-8 South Milledge Avenue, Athens, GA 30605. E-mail: [email protected]

140 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017, 140–154
© 2017 Association for Financial Counseling and Planning Education®
http://dx.doi.org/10.1891/1052-3073.28.1.140
and they argue that the risk tolerance questionnaires do not Although the aforementioned relationships may represent
account for the emotional tendencies of investors. Critics normative predictions, researchers such as Yao and Curl
also point out that investor behavior appears to be driven (2011) have noted attitudes and behavior that diverge from
by the market environment changes (Guiso, Sapienza, & predictions in practice. They found that exhibiting either
Zingales, 2013; Hoffmann, Post, & Pennings, 2013; Sahm, very low or very high risk tolerance can hinder wealth ac-
2007; Yao & Curl, 2011). If true, this would seriously under- cumulation over the lifespan. Extraordinarily high risk-
mine the assumption that risk attitudes are relatively stable tolerance often leads to excessive trading (Barber & Ode-
(Yao & Curl, 2011). Making recommendations based on an, 2001) and problematic asset allocation frameworks
such an unstable tool could be detrimental especially during for investors nearing retirement (Hariharan, Chapman, &
high volatility period, such as the Great Recession. A signifi- Domian, 2000). On the other hand, very low risk toler-
cant instability in risk tolerance may cause corresponding ance can create a myopic investing outlook that limits an
changes in portfolio holdings, hence may decrease a house- investor’s willingness to take necessary risks (Barsky et al.,
hold’s potential to accumulate wealth. 1997; Hariharan et al., 2000). In addition, switching from a
low- to high-risk tolerance, or vice versa, can cause prob-
The purpose of this study was to test the degree to which lems if an investor’s goal is wealth accumulation (Friesen
risk tolerance attitudes of individuals are influenced by vol- & Sapp, 2007). This phenomenon of risk tolerance shift can
atility in the equities markets. It may be that risk attitudes sometimes be confused with the use of tactical asset alloca-
are most influenced by periods when volatility increases tion strategies. For example, Glumov (2013) noted that at
rather than periods when volatility decreases or through the moment of transition from bear to bull market in 2010,
normal patterns of price changes. Data from a unique mul- the portfolio allocation to risky assets declined to its lowest
tiyear survey that included information covering the Great level in the history of the Survey of Consumer Finances
Recession and recovery period were analyzed to address the (SCF) for people younger than the age of 35 years. Was this
research purpose. It was determined that market volatility because investor risk tolerance fell or was this a result of
does, as questionnaire skeptics claim, influence risk atti- tactical reallocations unrelated to risk attitudes? It is pos-
tudes, but counter to such claims, the overall effect of such sible that risk tolerance did fall, which, in turn, generated
changes was found to be relatively small. Investor behavior a need to reallocate portfolio assets. Whether or not such
that appears to be disjointed from assessed risk attitudes a change was (or is) prudent is beyond the scope of this
(e.g., selling portfolio holdings after a steep decline in eq- review. What is of relevance is the extent to which inves-
uity prices) may be related to other investor attitudes, needs, tor risk tolerance changes in practice. This is the reason it
and capacities rather than significant shifts in risk tolerance. is important to have a better understanding of the stability
assumption underlying most financial risk tolerance attitu-
Literature Review and Hypotheses dinal assessments.
A noticeable increase in interest about investor risk tolerance
attitudes occurred during and shortly after the Great Reces- Several studies conducted prior to the Great Recession were
sion. This interest was driven, in part, by the apparent as- able to document that investor risk attitudes appear to vary
sociation between investor risk attitudes and market volatil- based on environmental factors (Grable & Joo, 2004; Sung
ity (Yao & Curl, 2011). Oscillating equity prices signaled a & Hanna, 1996). Clarke and Statman (1998) concluded that
change in market participants’ feelings regarding the stabil- risk tolerance tends to fall after financial newsletter writers’
ity of prices. For some, price instability can and did result in sentiment declines. Shefrin (2000) pointed to a positive as-
portfolio losses. For others, the same volatility provided an sociation between shifts in the markets and the risk-taking
opportunity for future gains. Conceptually, investors with a preferences of professional investors. More recently, others
long-term outlook should not be influenced by unpredictable have confirmed that risk tolerance is less stable than previ-
short-term instability. Those with a higher tolerance for risk ously thought. For example, Yao, Hanna, and Lindamood
should be willing to accept more volatility in pursuit of high- (2004) tracked changes in risk attitudes measured in the
er returns, regardless of short-term trends in prices. Investors SCF. The SCF uses a single-item risk tolerance measure.
with a lower willingness to engage in risky financial behav- Although the SCF risk measure suffers from validity and
ior should likewise be less willing to hold volatile assets. reliability complications (Grable & Lytton, 2001), general

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 141
risk attitudes varied based on patterns in the stock market (Hoffmann et al., 2013), this study did not explicitly re-
(Yao et al., 2004). Grable, Lytton, and O’Neill (2004) and port the extent to which volatility was associated with risk
Grable, Lytton, O’Neill, Joo, and Klock (2006) related tolerance.
changes in risk tolerance attitudes to investors engaging in
projection bias by forecasting recent price changes in the The most relevant study conducted on the topic of risk tol-
stock market into the future. They concluded that as prices erance attitude stability was published by Guillemette and
increase, risk attitudes increase. On the other hand, when Finke (2014). They set about attempting to determine if risk
prices in a previous period fall, investors tend to project this tolerance attitudes vary over time and, if so, whether this
into the future and reduce their willingness to take risk. variability is large enough to influence portfolio choices.
Their secondary purpose was to test whether changes in
Data published after the Great Recession generally supports risk tolerance correlate with the market environment. In
the hypothesis that risk tolerance is somewhat variable in this study, the authors used 5 years of data from FinaMet-
the short run (Guillemette & Finke, 2014), although there rica (i.e., a risk-assessment questionnaire used by financial
is also some evidence to suggest that risk tolerance is not planning professionals) to test their research questions.
affected by shifts in the markets (Weber, Weber, & Nosic, They noted an overall significant positive (r 5 .70) correla-
2013). Yao and Curl (2011) noted that the association be- tion between risk tolerance scores and changes in the S&P
tween risk tolerance attitudes and market returns was posi- 500 over the period. However, this correlation varied during
tive during the Great Recession. Van de Venter, Michayluk, the bear market period and the recovery period. During the
and Davey (2012) examined this issue in greater detail using bear market timeframe, the correlation was .90; on the other
risk tolerance data collected over a 5-year period. The aim hand, during the recovery period, there was little correla-
of their study was to test the proposition that risk tolerance tion (r 5 .01). Their conclusions highlighted the tendency
was a stable personality trait. They noted that risk attitudes of investor risk attitudes to move in step with the markets.
did appear to shift over time; however, unlike previous When stock prices were bullish during the 5-year period,
reports, annual changes in risk tolerance were subtle and investor risk attitudes hardly changed. Alternatively, when
likely not significant (Van de Venter et al., 2012). Rather prices were bearish, risk tolerance fell. However, the po-
than being linked to market expectations, they suggested tential drawback with this scenario is that investors might
that changes in risk attitudes might be more closely aligned have been selling winning stocks too early and riding losing
with household size change and help-seeking behavior. It is stocks too long, which Shefrin and Statman (1985) called
worth noting, however, that this study did not track varia- the “disposition effect,” which is, of course, the opposite of
tions in risk tolerance in relation to either change in market what a rational investor ought to do in practice.
conditions or market volatility.
There is evidence in the FinaMetrica dataset that investor
Hoffmann et al. (2013) looked specifically at investor per- risk tolerance was somewhat time-varying (Guillemette &
ceptions and behavior during the period of September and Finke, 2014). There is also support for this claim from a
October 2008, which represented a time in which stock study conducted by Barberis, Huang, and Santos (2001).
market experienced significant losses. They found inves- They developed a model based on prospect theory and used
tors’ expectations declined during these months. Of more varying levels of weights for losses. They found that inves-
importance in the context of this study, they noted that risk tors are loss averse over fluctuations in the value of their
tolerance decreased. Although investor risk tolerance does financial wealth. They also reported that the degree of loss
appear to vary with market fluctuations, the changes tend aversion depends on prior investment performance. This
to be temporary (Hoffmann et al., 2013). Rather than creat- finding is in line with Thaler and Johnson’s (1990) “house
ing a permanent shift in attitude, changes in market prices money effect,” which states that the degree of loss aversion
(either up or down) do not appear to last long. Investor increases after a prior gain and decreases after a prior loss.
attitudes tend to revert to their mean starting point. Their
findings were supported by Grable and Rabbani (2014) that However, Guillemette and Finke (2014) reported an es-
showed that individuals held risk tolerance as a generalized sential caveat associated with the findings of Barberis
attitude. Although there were changes in trading volatility et al. (2001); namely, the change in magnitude of market

142 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017
changes was significantly larger than changes in risk toler- As noted earlier, the online survey was open to anyone
ance scores. In other words, there was no evidence showing who had Internet access during the survey period. Thirteen
that if the markets fell by, for example, 5% risk tolerance risk tolerance questions, as originally compiled by Grable
scores would also drop by 5%. Although the correlation was and Lytton (1999), were asked of each survey respon-
evident, particularly in the bear market period, the level of dent. In addition, basic demographic inquiries were made,
comparative change in risk tolerance scores, based on mar- including questions about the gender, income, and marital
ket movements, was relatively modest. status of respondents. A separate question asked each re-
spondent to indicate whether they made their own invest-
Three observations emerge from the literature. First, it ment decisions, relied on someone else, or had no investable
does appear that risk tolerance is variable. The key ques- assets at the time of survey completion. In total, 152,766
tion is to what extent the variability is meaningful. Second, useable responses were available for analysis. Respondent
early reports of variability were based on nominal changes characteristics for these factors are shown in Table 1. The
in market prices, not the volatility in prices. It may be that following scoring guidelines (Grable & Lytton, 1999) were
risk attitudes are most influenced by highly volatile peri- provided to those who completed the risk questionnaire
ods rather than through normal patterns of price changes. online: (a) 13–18 5 low tolerance for risk, (b) 19–22 5
Third, and most important, only a handful of studies have below-average tolerance for risk, (c) 23–28 5 average/
explicitly documented the level of change in risk attitudes moderate tolerance for risk, (d) 29–32 5 above-average
associated with shifts in the markets. Sahm (2007) inves- tolerance for risk, and (e) 33–47 5 high tolerance for risk.
tigated risk tolerance and asset allocation of investors and Readers are encouraged to review their original article for
found that heterogeneity in risk preferences is an impor- further details.
tant source of the heterogeneity in financial portfolios. As
noted by Guillemette and Finke (2014) and Van de Ven- Analysis Method
ter et al. (2012), risk tolerance attitudes may, in fact, be Risk tolerance scores were calculated for each respondent
variable, but it may also be true that the variability is not by summing answers to the 13 risk tolerance questions, in
significant enough to impact most financial counseling the manner as described by Grable and Lytton (1999). The
and planning decisions. Hence, it was hypothesized in this closing value of the S&P 500 was then matched to each re-
article that spondent’s survey completion day. For example, if some-
one completed the online survey on March 14, the closing
1. Risk tolerance scores would fall during periods of value of the S&P 500 would be recorded for that person as
increased volatility of the 14th of March. In situations where someone com-
2. Risk tolerance scores would increase during pleted the survey over a weekend, the closing value of the
periods of decreased volatility last previous trading day was matched to the respondent’s
3. Changes in risk tolerance scores would be large risk tolerance score. The study used the nominal value
and meaningful of the S&P 500. This choice of measurement instead of
inflation-adjusted measurement was based on the assump-
Method tion that investors evaluate markets in nominal terms, not
Data inflation-adjusted terms, primarily because nominal returns
Data for this study were obtained from a proprietary mul- are reported by those in the media. Market data were then
tiyear data gathering survey hosted online by Rutgers New used to calculate volatility estimates. Rather than correlate
Jersey Agricultural Experiment Station (Cooperative Exten- changes in risk tolerance scores to nominal changes in S&P
sion, http://njaes.rutgers.edu:8080/money/riskquiz/). Daily values, it was hypothesized that what investors may actu-
data were collected from users who visited the website to ally be responding to volatility in the markets. For example,
obtain a risk tolerance assessment. Access to the survey was when viewed over an extended period, there may be little
free and open to anyone with Internet access. The number change in the nominal value of a market index. For instance,
of survey respondents ranged from a handful to several hun- a Monday opening value may be very close to that week’s
dred per day. This study’s analyses were based on data col- closing Friday value. If viewed this way, the change in value
lected from 2008 through 2013. would be close to zero. However, during the week (or, say,

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 143
TABLE 1. Descriptive Statistics for Respondents
Variable Frequency (n) % M Risk Score SD of Risk Score
Gender
 Female 64,492 42.22 25.94 4.94
 Male 88,274 57.78 28.72 5.54
Age (years)
  Younger than 25 81,926 53.63 27.37 5.53
  25–34 27,865 18.24 27.93 5.38
  35–44 13,934  9.12 28.26 5.39
  45–54 13,176  8.62 27.77 5.29
  55–64 11,143  7.29 27.02 5.10
  65–74   3,596   2.35 26.60 5.19
  75 and older  1,126  0.74 27.57 8.42
Marital status
  Never married 89,544 58.62 27.51 5.51
  Living with significant other  9,474  6.20 27.54 5.41
 Married 43,576 28.52 27.70 5.30
  Separated or divorced  6,747  4.42 27.12 5.47
 Widowed  1,516  0.99 26.96 6.66
  Shared living arrangement  1,909  1.25 27.80 6.39
Education
  Some high school or less 34,698 22.71 27.31 5.87
  High school diploma 21,032 13.77 27.16 5.51
  Some college 26,485 17.34 26.88 5.21
  Associate’s degree 10,527  6.89 26.81 5.26
  Bachelor’s degree 35,308 23.11 28.11 5.20
  Graduate or professional degree 24,716 16.18 28.43 5.39
Household income
  Less than $25,000 34,949 22.88 27.08 5.58
  $25,000–$49,999 29,924 19.59 26.84 5.34
  $50,000–$74,999 29,642 19.40 27.27 5.31
  $75,000–$99,999 20,316 13.30 27.72 5.27
  $100,000 or more 37,935 24.83 28.65 5.52
Decision making
  Make own investment decisions 90,197 59.04 27.89 5.53
  Rely on the advice of professional 17,869 11.70 27.78 5.13
  Do not have investment assets 44,700 29.26 26.76 5.40

a 20-day period) the level of volatility could be quite large. for volatility, an absolute value change variable was cre-
In fact, throughout 2008 this happened several times. Dur- ated for three periods: (a) daily, (b) weekly, and (c) over
ing some weeks, the intra-week market moved up or down a 20-day cycle. These periods were chosen based on com-
on a given day by more than 10%; however, this volatility monly used technical trading standards (Weinstein, 1988;
was disguised when viewed on an average basis. To account Wilder, 1978). Volatility, as defined in this study, was a

144 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017
measure of the magnitude of market fluctuations. Volatility at least were distributed somewhat evenly around the mean
was estimated as a local average of absolute price changes and more than the total range.
│G(t)│ over a suitable time interval (see Wilder, 1978).
Large values of │G(t)│ represent crashes and big rallies. Table 2 provides a snapshot of yearly risk tolerance score
In this study, Z(t) represented the closing value of the S&P and volatility data over the three evaluation periods. The first
500 index. As such, the absolute price change was measured number shown represents the average risk tolerance score
as: G(t)│ 5 │[Z(t 1 Dt) 2 Z(t)]/ Z(t)│. and volatility estimate for the year, respectively. The second
figure is the matching standard deviation. The third number
Full-year data for 2008, 2009, 2010, 2011, 2012, and 2013 represents the percentage increase or decrease in the average
were compiled based on respondent risk tolerance scores risk tolerance score from the previous year (December 2007
and market volatility. Changes in risk tolerance scores by data were used as the baseline for the 2008 changes). The
year were then calculated and compared against volatility final figure represents Cronbach’s alpha, which is a measure
estimates. Correlation coefficients were used to measure the of scale reliability for the risk tolerance scale scores. In gen-
level of association between the different measures of vola- eral, risk tolerance scores changed very little, whereas there
tility and risk tolerance scores. In addition, periods of vola- was more movement in year-over-year changes in volatility.
tility were categorized as either increasing or decreasing During the period of greatest market stress—corresponding
over a monthly cycle. These categorized volatility estimates to the Great Recession—daily, weekly, and 20-day volatility
were then compared to changes in risk tolerance scores dur- was large. Even so, the standard deviation associated with
ing these periods to determine if risk tolerance scores were volatility was relatively modest. This is somewhat surpris-
affected by shifts in volatility. ing given the previous literature. One would have expected
to see greater variability in risk tolerance scores from 1 year
Results to the next. However, others might argue that this stability
The 152,766 respondents were quite diverse. Approximately of risk tolerance score is indicative of the true nature of a
58% of respondents were male. The age of respondents person’s willingness to engage in risky financial behavior.
skewed younger, with nearly 70% reporting being 35 years In addition, the stability in average risk tolerance scores
of age or younger. Approximately 10% of respondents were points to the overall reliability of the risk-assessment scale.
older than age of 55 years. Given the age of respondents,
it is not surprising that the majority were also single. Ap- Figure 1 provides a visual interpretation of the data over the
proximately 28% of respondents were married. In terms periods of study. As shown, risk tolerance scores remained
of educational attainment, the sample was well educated. relatively stable year-over-year, as measured by average
Approximately 45% of respondents reported having an as- scores and standard deviations, especially when compared
sociate’s degree or higher level of education. About 23% to the volatility measures.
of respondents earned less than $25,000 per year, whereas
about 24% reported earnings of more than $100,000 per Table 3 provides data that help address the key question of
year. In general, respondents were more likely to report interest in this study. The second column reports the cor-
making their own investment and financial decisions (59%), relation coefficient between risk tolerance scores and each
compared to those who relied on others for advice (11.7%) of the three measures of volatility. The third column re-
and those who did not make meaningful financial decisions ports the actual percentage change in risk tolerance scores
because of a lack of assets (29%). when volatility increased. The fourth column shows the
percentage change in risk tolerance scores when volatility
As a reference point, it is important to note statistics about decreased. It is hypothesized that when volatility increased,
the S&P 500 during the periods of analysis. The mean level in any of the three evaluation periods, risk tolerance scores
of the S&P 500 was 1,348.86 (SD 5 228.41), which was would fall. Conversely, during periods of reduced volatility,
close to the median (1,338.31). The range in value during risk tolerance scores were hypothesized to increase.
the periods of analysis was relatively large at 1,171.83, with
a minimum of 676.53 and high of 1,848.36. This indicates All of the correlation coefficients reported in Table 3 were
that all of the data points were fairly close to each other or significant at the p , .01 level. Figure 2 shows that risk

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 145
TABLE 2. Risk Tolerance Score and Volatility Descriptive Data
Risk Tolerance
Score Daily Volatility Weekly Volatility 20-Day Volatility
SD SD SD SD
% Changea % Changea % Changea % Changea
Year N Alpha
2008 13,077 27.86 15.33 95.41 408.22
0.39 7.96 46.73 214.99
20.15 19.82 13.34 38.17
.78
2009 14,178 27.67 8.95 52.94 240.70
0.34 3.52 18.41 69.30
0.03 25.04 29.17 29.23
.76
2010 18,592 27.56 8.04 45.49 190.52
0.53 4.84 21.19 65.45
0.03 19.74 11.31 8.30
.77
2011 33,692 27.65 10.21 60.19 264.92
0.54 5.11 29.03 131.78
0.17 17.65 19.45 13.45
.75
2012 33,897 27.35 6.38 38.45 167.43
0.33 1.62 10.52 36.80
20.13 1.42  2.09 21.71
.76
2013 39,330 27.76 7.13 42.26 186.47
0.28 2.43 12.83 34.95
0.14 10.63  9.52 1.87
.75
Values are in average.
a

tolerance scores, although not following the hypothesized (i.e., risk tolerance score of 33–47) to a moderate category
directions at all times, were not influenced by the time (i.e., risk tolerance score of 29–32) with regularity. In other
boundaries associated with volatility measurement. words, although scores did vary, the variability within cat-
egories of risk tolerance was very restrained.
The percentage change in risk tolerance score data in Table 2
is of particular importance. Although it is true that risk at- It is important to note, however, that the observed re-
titudes and volatility were correlated, the actual change in lationships did not always match what was originally
risk tolerance scores was very modest. The greatest shift hypothesized. Figure 3 provides a visual summary of the
in scores was, at most, close to 2%. This means changes in findings. As illustrated, the hypotheses were confirmed in
risk tolerance scores, on average, were not large enough to 2008, which happened to be the most volatile of the six mar-
move a respondent from, say, a high-risk tolerance category ket periods. As market volatility increased, risk tolerance

146 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017
Figure 1. Variations in risk tolerance scores compared to market volatility.

1000 29
900 28.5
800 28
700

Risk Tolerance
27.5
600
Volatility

27
500
26.5
400
26
300
200 25.5

100 25
0 24.5
June, 08

June, 09

June, 10

June, 11

June, 12

June, 13
Dec, 07

Dec, 08

Dec, 09

Dec, 11

Dec, 12

Dec, 13
Dec, 10

Mar, 13
Mar, 08

Mar, 09

Sep, 09

Mar, 10

Sep, 10

Mar, 11

Sep, 11

Mar, 12

Sep, 13
Sep, 08

Sep, 12
Daily Vol Weekly Vol 20 Day Vol Risk Tolerance

scores dropped. When volatility decreased, risk tolerance periods of decreased volatility, and (c) changes in risk tol-
scores increased. However, the percentage change in scores erance scores would be large and meaningful. Data from
was not that great. The strength of the hypothesized asso- a unique multiyear survey that included data covering the
ciations weakened in 2009, 2010, 2011, and 2012, only to Great Recession and subsequent recovery period showed
come back into line with the hypotheses in 2013. that risk tolerance scores do, in fact, vary based on envi-
ronmental factors. There does appear to be some evidence
Consider the market in 2010 and 2011. Counter to expecta- of association between market volatility and changes in risk
tions, risk tolerance scores actually tracked positively with tolerance; however, the relationship was not consistent.
daily and weekly market volatility during this period. As The relationship matched the hypothesis for daily volatil-
volatility increase, so did risk tolerance scores in two of the ity 64% of the time. The relationship matched the hypoth-
daily and weekly scenarios. When volatility decreased, so esis for weekly volatility 50% of the time. It is important
did risk tolerance scores. These results suggest that although to note that the overall effect of these changes was rela-
there does appear to be some association between market tively small. Over the period from January 2008 through
volatility and risk tolerance scores, the relationship did December 2013, risk tolerance scores ranged from 13 to 47,
not always conform to expectations. Overall, the observed with average scores falling between 26 and 28. According
relationships matched the hypotheses approximately 64% to Grable and Lytton (1999), the average score is indica-
of the time. This can be seen in Figure 4. The lines repre- tive of an average/moderate willingness to engage in a risky
sent the average percentage change in daily volatility from financial behavior.
Table 2 and the risk tolerance and daily volatility correla-
tion from Table 3. Had the relationship been more robust, Results from this study indicate that although risk toler-
the two lines should have moved in diverging directions. ance scores are likely to vary based on market volatility, the
actual magnitude of such changes will generally be quite
Discussion modest. Even during periods of extreme market volatility,
It was hypothesized at the outset of this study that (a) risk for example, risk tolerance scores rarely moved by more
tolerance scores would fall during periods of increased than 2% points. It would take a much larger percentage
volatility, (b) risk tolerance scores would increase during change in a risk tolerance score to move someone from

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 147
TABLE 3. Correlation Estimates and Changes in Risk Tolerance Score Data
Correlation With Changes in Risk Tolerance Changes in Risk Tolerance
Changes in Risk Score During Period of Score During Period of
Volatility Tolerance Score p Value Increased Volatility Decreased Volatility
2008
 Daily 2.72 .009* 21.08% 1.13%
 Weekly 2.59 .042* 20.41% 1.04%
  20 day 2.47 .127 20.07% 1.06%
2009
 Daily 2.60 .041* 21.51% 1.74%
 Weekly 2.57 .053* 21.51% 1.39%
  20 day 2.32 .317 0.11% 20.02%
2010
 Daily .52 .084* 0.54% 21.07%
 Weekly .51 .089* 1.08% 21.07%
  20 day 2.05 .888 20.43% 20.01%
2011
 Daily .06 .845 0.80% 20.24%
 Weekly .21 .513 0.80% 20.24%
  20 day .11 .742 20.37% 0.42%
2012
 Daily 2.24 .450 20.10% 0.52%
 Weekly 2.22 .483 0.00% 0.77%
  20 day 2.36 .252 0.24% 0.10%
2013
 Daily .03 .937 20.12% 0.65%
 Weekly .01 .984 20.12% 0.65%
  20 day .04 .910 20.06% 0.08%

*p 5.10.

one category of risk tolerance (e.g., low, below-average, market volatility by reducing their portfolio risk relative to
high) to another. the market. The change in risk tolerance may also be re-
lated to the change in wealth during the Great Recession.
If fluctuations in risk tolerance scores do not necessitate It is likely that households who reported a decrease in
changes in most people’s household financial plans, then risk tolerance may have experienced a decrease in wealth,
why do financial counselors and planners witness some- whereas households who reported an increase in risk toler-
times odd behavioral changes in clients who are experienc- ance may have experienced an increase in wealth. Necker
ing dramatic market volatility? Merkle and Weber (2014) and Ziegelmeyer (2014) found that among German house-
reported that higher risk expectations lead to decreases in holds who attributed losses in wealth to the Great Reces-
risk tolerance in terms of volatility among investors, where- sion, many reported a decreased risk tolerance, whereas
as lower risk expectations have the opposite effect. They those who did not attribute losses to the global financial
also found evidence that investors counteract changes in crisis reported no change in risk tolerance. Shin and Hanna

148 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017
Figure 2. Risk tolerance and yearly volatility correlation coefficients.

0.60

Correlation Coefficient
0.40
0.20
0.00
0.20
0.40
0.60
0.80
2008 2009 2010 2011 2012 2013

Daily Weekly 20 Day

(2015) reported that minority investors seem to react more of a change in willingness to take risk, overconfident inves-
strongly to Great Depression than White investors because tors may be falling prey to a recency bias. This bias causes
the minority investors are more risk averse (Hanna, Wang, investors to ignore the actual risks inherent in a situation and
& Yuh, 2010). instead conclude that the recent past will continue into the
future. Yao and Curl (2011) reported that the fluctuation of
Another potential source of behavior change is the notion risk tolerance in relation to market returns may be because of
of loss aversion. Benartzi and Thaler (1995) found that projection bias and the recency effect. Significant volatility
loss aversion amplifies risk aversion in standard utility may create a psychological marker that prompts behavioral
models leading to inconsistent preferences. For example, change even though the underlying willingness of the inves-
the level of investor loss aversion fluctuates depending on tor to take risk remains unchanged.
prior investment performance (Barberis et al., 2001). Find-
ings reported by Barberis et al. (2001) suggest that after Limitations and Future Research
prior gains, an investor may become more risk tolerant and, Although the findings from this study are noteworthy, it is
conversely, after prior losses, the same investor may be- important to evaluate the results within the context of the
come more risk averse. Therefore, financial counselors and sample. The data were obtained from an open-access Inter-
planners may find it useful to monitor market performance net survey. It is possible that a response bias was present in
as a precursor to client discussions about risk and risk tak- the data and that respondents did not represent the typical
ing. Negative performance may cause a client to become investor. It is also possible that a self-selection bias exist-
increasingly loss averse, which can lead to a reduced will- ed in relation to those who completed the survey. Readers
ingness to hold risky asset (e.g., equities). need to remember that the measure of risk tolerance used in
this study was designed to measure a person’s willingness
Shifts in risk tolerance witnessed during periods of market to take financial risk. It is possible that had a measure of
volatility may also be because of several behavioral finance loss aversion been used, a different result might have been
biases that cause investors to misjudge environmental cues. found. In addition, it should be noted that the measure of
For example, Chatterjee (2015) reported that false confi- risk tolerance used in this study was limited in its ability
dence is a predictor of volatility in stock holdings within to suggest any particular portfolio allocation for a given
one’s portfolio. False confidence has been found to increase level of risk tolerance score. Finally, results may have dif-
investments following years of strong equity performance fered had other measures of volatility been used or had the
followed by decreased investments during periods following volatility measurement periods been different. Regardless
low returns. Rather than changing portfolio holdings because of these potential limitations, however, this research does

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 149
150
Figure 3. Variation in the observed relationships compared to the hypothesized associations.

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017

Figure 4. Comparison of risk tolerance scores and daily volatility correlations and percentage changes.

0.60 25.00%

0.40 20.00%

Correlation Coefficient
0.20 15.00%

% Change
0.00 10.00%
2008 2009 2010 2011 2012 2013
0.20 5.00%

0.40 0.00%

0.60 5.00%

0.80 10.00%
Correlation with Changes in Risk Tolerance Score
Average % Change in Daily Volatility

add to the existing literature by both confirming that risk is unlikely to be large enough to warrant a significant modi-
attitudes do appear to shift in relation to the market envi- fication in someone’s household financial plan. It is more
ronment and that such changes are most often very modest. likely that other behavioral biases are at play. As such, it is
important for financial counselors and planners to develop
Future studies should be conducted to verify the results techniques to evaluate potential behavioral biases and in-
presented in this article. Specifically, it would be useful to tervention techniques to ensure that biases do not unduly
compare shifts in risk tolerance based on the socioeconomic influence client decisions.
profile of respondents. In this study, for example, respon-
dents included the very young and very old and those who Financial counselors and planners may wish to incorpo-
owned and did not own securities. Based on the baseline rate aspects of behavioral portfolio theory (BPT; Shefrin &
data reported here, follow-up studies can be conducted to Statman, 2000) into their practice. A working assumption
compare results of targeted samples, including pre- and within BPT is that portfolio allocations may be based on a
postretirees, those who hold significant equity positions, combination of “insurance” (protection against losses) and
and those just beginning their careers. It may be that risk “lotteries” (small odds of a large gain). According to this
tolerance varies not only by market conditions but also by theory, investors consider their portfolios as pyramids of
the way those sharing certain socioeconomic characteristics goal-based mental accounts and have as many risk attitudes
view the markets. as mental accounts (Shefrin & Statman, 2000). Griesdorn,
Lown, DeVaney, Cho, and Evans (2014), in their investiga-
Implications for Financial Counselors and Planners tion of how self-control, mental accounting, and framing
The findings from this research have several important im- are related to risk tolerance, found a significant positive
plications for financial counselors and planners. First, dur- association between mental accounting and risk tolerance.
ing a period of market volatility, financial counselors and They reported that the use of mental accounts is common
planners should assume that the willingness of their clients among low- to moderate-income households, but their us-
to invest aggressively will change. In general, increased vol- age does not appear to influence household risk tolerance
atility appears to dampen risk tolerance, whereas decreased unless the household owns a brokerage account. If true, this
volatility increases risk tolerance. Second, it is important means that the impact of market fluctuations may influence
for financial counselors and planners to remember that mul- a client in multiple ways. When thinking about retirement
tiple factors may be at play when clients alter their financial assets, one client may move to protect a portfolio from
behavior in the context of market volatility. Although risk losses. At the same time, the client may make unreasonably
tolerance may shift, the actual magnitude of such a change large bets with more speculative money. Behavioral finance

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 151
technique, such as framing, can be used to successfully pre- Barberis, N., Huang, M., & Santos, T. (2001). Prospect the-
dict risk tolerance (Griesdorn et al., 2014). Financial coun- ory and asset prices. Quarterly Journal of Economics,
selors and planners should counsel clients to evaluate their 116, 1–53.
financial assets with the widest “frame” possible and avoid Barsky, R. B., Juster, F. T., Kimball, M. S., & Shapiro, M.
focusing on individual securities, instruments, or accounts. D. (1997). Preference parameters and behavioral het-
Financial management and portfolio discussions should be erogeneity: An experimental approach in the health and
framed in terms of long-term goals and each client’s total retirement study. Quarterly Journal of Economics, 112,
wealth picture. 537–579.
Benartzi, S., & Thaler, R. (1995). Myopic loss aversion
Another useful tool may be to implement precommitment and the equity premium puzzle. Quarterly Journal of
strategies to help clients formalize their household finan- Economics, 110, 73–92.
cial management objectives and requirements in a written Bhattacharya, U., Hackethal, A., Kaesler, S., Loos, B., &
investment policy statement (IPS) to help them make op- Meyer, S. (2012). Is unbiased financial advice to retail
timal long-term decisions (Winchester, Huston, & Finke, investors sufficient? Answers from a large field study.
2011). Laibson (1997) noted that individuals often forsake Review of Financial Studies, 25, 975–1032.
their long-term goals for instant gratification. Precommit- Chatterjee, S. (2015). False confidence, stock market par-
ment in this context can be thought of as a strategy where ticipation, and wealth accumulation of households:
an individual with a conflict between present and future An examination. Retrieved from http://ssrn.com/
objectives can strengthen their position by cutting off op- abstract=2566622
tions for instant gratification. The “Save More Tomorrow” Clarke, R. G., & Statman, M. (1998). Bullish or bearish?
strategy suggested by Thaler and Benartzi (2004) is an ex- Financial Analysts Journal, 54, 63–72.
ample of a precommitment strategy where individuals com- Davies, G. B., & Brooks, P. (2013). Risk tolerance: Essen-
mit in advance to allocating a portion of their future salary tial, behavioural and misunderstood. Journal of Risk
increase toward retirement savings. As markets move and Management in Financial Institutions, 7, 110–113.
emotions take hold, the IPS, coupled with some precommit- Friesen, G., & Sapp, T. (2007). Mutual fund flows and in-
ment strategies, can help prevent people from making snap vestor returns: An empirical examination of fund in-
judgments. vestor timing ability. Journal of Banking and Finance,
31, 2796–2816.
Financial counselors and planners should also visit with Gilliam, J., Chatterjee, S., & Grable, J. (2010). Measur-
their clients about the concept of risk tolerance, how risk ing the perception of financial risk tolerance: A tale of
tolerance is measured, and why risk tolerance is impor- two measures. Journal of Financial Counseling and
tant when making household financial management de- Planning, 21(2), 30–43.
cisions. This is especially important in light of the call Glumov, D. (2013). How did the great financial crisis af-
for increased diligence because of market volatility. Fi- fect portfolio allocations and attitudes towards risk?
nancial professionals can assist clients in maintaining a Evidence from the “Survey of Consumer Finances.”
sense of perspective during market volatility and keep a Retrieved from http://papers.ssrn.com/sol3/papers
long-term focus by reminding the client of the written IPS .cfm?abstract_id=2293189
that was signed to help them accomplish their long-term Grable, J. E., & Joo, S.-H. (2004). Environmental and bio-
goals. A long-run strategy that reduces the temptation to psychosocial factors associated with financial risk tol-
buy high and sell low can be a significant source of value erance. Journal of Financial Counseling and Planning,
provided by a financial counselor or planner (Guillemette 15(1), 73–82.
& Finke, 2014). Grable, J. E., & Lytton, R. H. (1999). Financial risk toler-
ance revisited: The development of a risk assessment
References instrument. Financial Services Review, 8, 163–181.
Barber, B., & Odean, T. (2001). Boys will be boys: Gender, Grable, J. E., & Lytton, R. H. (2001). Assessing the concur-
overconfidence, and common stock investment. Quar- rent validity of the SCF risk tolerance question. Journal
terly Journal of Economics, 116, 261–292. of Financial Counseling and Planning, 12(2), 43–53.

152 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017
Grable, J. E., & Rabbani, A. G. (2014). Describing risk Laibson, D. (1997). Golden eggs and hyperbolic discount-
tolerance: Does a risk characterization portrait exist? ing. Quarterly Journal of Economics, 112, 443–477.
Journal of Financial Counseling and Planning, 25(2), Merkle, C., & Weber, M. (2014). Do investors put their
174–183. money where their mouth is? Stock market expecta-
Grable, J. E., Lytton, R. H., & O’Neill, B. (2004). Projec- tions and investing behavior. Journal of Banking &
tion bias and financial risk tolerance. The Journal of Finance, 46, 372–386.
Behavioral Finance, 5, 142–147. Necker, S., & Ziegelmeyer, M. (2014). Household risk tak-
Grable, J. E., Lytton, R. H., O’Neill, B., Joo, S-H., & ing after the financial crisis (MEA Discussion Paper
Klock, D. (2006). Risk tolerance, projection bias, viv- Series No. 14279). Munich, Germany: Munich Center
idness and equity prices. Journal of Investing, 15(2), for the Economics of Aging at the Max Planck Institute
68–74. for Social Law and Social Policy.
Griesdorn, T., Lown, J., DeVaney, S., Cho, S., & Evans, D. Pan, C. H., & Statman, M. (2012). Questionnaires of risk
(2014). Association between behavioral life-cycle con- tolerance, regret, overconfidence, and other investor
structs and financial risk tolerance of low-to-moderate propensities. Investment Consulting, 13(1), 54–63.
-income households. Journal of Financial Counseling Roth, A. S. (2013). Why risk tolerance questionnaires don’t
and Planning, 25(1), 27–40. work? Retrieved from https://www.financial-planning
Gron, A., & Winton, A. (2001). Risk overhang and market .com/news/why-risk-tolerance-questionnaires-dont-work
behavior. The Journal of Business, 74, 591–612. Roth, B., & Voskort, A. (2014). Stereotypes and false con-
Guillemette, M. A., & Finke, M. (2014). Do large swings sensus: How financial professional predict risk prefer-
in equity values change risk tolerance. Journal of ences. Journal of Economic Behavior & Organization,
Financial Planning, 27(6), 44–50. 107, 553–565.
Guillemette, M. A., Finke, M., & Gilliam, J. (2012). Risk Sahm, C. R. (2007). How much does risk tolerance change?
tolerance questions to best determine client portfolio Washington, DC: Federal Reserve Board.
allocation preferences. Journal of Financial Planning, Shefrin, H. (2000). Beyond greed and fear: Understanding
25(5), 36–44. behavioral finance and the psychology of investing.
Guiso, L., Sapienza, P., & Zingales, L. (2013). Time vary- Cambridge, MA: Harvard University Press.
ing risk aversion (NBER Working Paper No. 19284). Shefrin, H., & Statman, M. (1985). The disposition to sell
Retrieved from http://www.nber.org/papers/w19284 winners too early and ride losers too long: Theory and
Hanna, S. D., Fan, J. X., & Chang, R. (1995). Optimal life evidence. Journal of Finance, 40, 777–790.
cycle savings. Journal of Financial Counseling and Shefrin, H., & Statman, M. (2000). Behavioral portfolio
Planning, 6(1), 1–15. theory. Journal of Financial and Quantitative Analysis,
Hanna, S. D., Wang, C., & Yuh, Y. (2010). Racial/ethnic 35, 127–151.
differences in high return investment ownership: A de- Shin, S. H., & Hanna, S. D. (2015). Decomposition analyses
composition analysis. Journal of Financial Counseling of racial/ethnic differences in high return investment
and Planning, 21(2), 44–59. ownership after the Great Recession. Journal of Finan-
Hariharan, G., Chapman, K. S., & Domian, D. L. (2000). cial Counseling and Planning, 26(1), 43–62.
Risk tolerance and asset allocation for investors near- Sung, J., & Hanna, S. (1996). Factors related to risk toler-
ing retirement. Financial Services Review, 9, 159–170. ance. Journal of Financial Counseling and Planning,
Hoffmann, A. O. I., Post, T., & Pennings, J. M. E. (2013). 7(1), 11–19.
Individual investor perceptions and behavior during Thaler, R. H., & Benartzi, S. (2004). Save more tomorrow:
the financial crisis. Journal of Banking & Finance, 37, Using behavioral economics to increase employee
60–74. saving. Journal of Political Economy, 112(Suppl. 1),
Hung, A. A., Clancy, N., Dominitz, J., Talley, E., Berrebi, S164–S187.
C., & Suvankulov, F. (2008). Investor and industry Thaler, R. H., & Johnson, E. J. (1990). Gambling with the
perspective on investment advisers and broker-dealers: house money and trying to break even: The effects of
Technical report. Santa Monica, CA: RAND Institute prior outcomes on risky choice. Management Science,
for Civil Justice. 36, 643–660.

Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017 153
Van de Venter, G., Michayluk, D., & Davey, G. (2012). A Wilder, J. W. (1978). New concepts in technical trading
longitudinal study of financial risk tolerance. Journal systems. Greensboro, NC: Trend Research.
of Economic Psychology, 33, 794–800. Winchester, D. D., Huston, S. J., & Finke, M. S. (2011).
Walls, M. R., & Dyer, J. S. (1996). Risk propensity and firm Investor prudence and the role of financial advice. Jour-
performance: A study of the petroleum exploration nal of Financial Service Professionals, 65(4), 43–51.
industry. Management Science, 42, 1004–1021. Yao, R., & Curl, A. L. (2011). Do market returns influence
Weber, M., Weber, E. U., & Nosic, A. (2013). Who takes risk tolerance? Evidence from panel data. Journal of
risks when and why: Determinants of changes in inves- Family Economic Issues, 32, 532–544.
tor risk taking. Review of Finance, 17, 847–883. Yao, R., Hanna, S. D., & Lindamood, S. (2004). Changes in
Weinstein, S. (1988). Secrets for profiting in bull and bear financial risk tolerance, 1983–2001. Financial Services
markets. Burr Ridge, IL: Irwin. Review, 13, 249–266.

154 Journal of Financial Counseling and Planning, Volume 28, Number 1, 2017