Perceived and Realized Risk Tolerance: Changes During the 2008 Financial Crisis

Boise State University

ScholarWorks

Marketing and Finance Faculty Publications and

Presentations

Department of Marketing and Finance

1-1-2016

Perceived and Realized Risk Tolerance: Changes

During the 2008 Financial Crisis

Diane K. Schooley

Boise State University

Debra Drecnik Worden

George Fox University

+is document was originally published by the Association for Financial Counseling and Planning Education in the Journal of Financial Counseling &

Planning. Copyright restrictions may apply. doi: 10.1891/1052-3073.27.2.265

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016, 265–276

http://dx.doi.org/10.1891/1052-3073.27.2.265

265

Perceived and Realized Risk Tolerance:

Changes During the 2008 Financial Crisis

Diane K. Schooley

and Debra Drecnik Worden

Using the 2007–2009 Survey of Consumer Finances panel data, this study examined changes in perceived

and realized risk tolerance after the nancial crisis. Households who perceived less risk tolerance were more

likely to have reduced their portfolio risk and vice versa. Furthermore, households whose wealth decreased

were more likely to perceive less risk tolerance and vice versa. Regression analysis revealed that change in risk

tolerance as measured by the change in nancial portfolio risk is related to perceived risk tolerance, education,

life cycle stage, and employment status. Single households, or those households whose head is less educated, or

self-employed or unemployed, may need nancial advice to prevent them from reducing their portfolio risk in

reaction to a nancial crisis.

Keywords: nancial crisis, risk tolerance, portfolio composition, behavioral nance

Associate Dean and Professor of Finance, College of Business and Economics, Boise State University, Boise, ID 83725. E-mail: [email protected]

Professor of Business and Economics, College of Business, George Fox University, 414 N. Meridian Street, 6263, Newberg, OR 97132.

E-mail: dworden@georgefox.edu

education to ensure that they do not overreact. Otherwise,

they may respond by reducing the risk of their portfolios

and foregoing potential wealth accumulation.

Much research has been devoted to identifying factors that

impact nancial risk tolerance. However, more insight into

the changes in risk tolerance that resulted from the 2008

nancial crisis will help nancial advisors to better coun-

sel their clients. The Survey of Consumer Finances (SCF)

interviewed the same sample of households before and after

the crisis. The responses provide a unique opportunity to

examine how perceived risk tolerance and portfolio risk

changed for the sample over this time period.

This article examines changes in household risk tolerance,

both perceived and measured by investment behavior pre-

ceding the 2008 nancial crisis and at its end. After a lit-

erature review and data and sample descriptions, univariate

analysis considers changes in portfolio risk across changes

in perceived risk tolerance. Because these changes could

depend on how the nancial crisis impacted household

wealth, the analysis distinguishes between households who

experienced a decrease in wealth during the nancial crisis

and those who experienced an increase in wealth. Multivar-

iate analysis then explores the factors that explain changes

any speculate that the recent nancial crisis has

inuenced households to become less tolerant of

nancial risk, that is, more averse to the vari-

ability in returns on investments. Between 2007 and 2009,

many households observed substantial declines in the values

of their stock portfolio and housing assets. Risk tolerance

is modeled, theoretically, to be a function of wealth; these

changes in wealth may have been met with changes in risk

tolerance. Lower risk tolerance, and thus less portfolio

risk, is especially concerning after a downturn. If investors

decrease the risk of their portfolios (hold less stock) in reac-

tion to a decline in the stock market, they may be selling

stock when the market is relatively low, decreasing their

actual wealth, and leading them to miss out on gains as the

market recovers.

Financial advisors make recommendations that can inuence

the nancial risk borne by their clients. According to Finke

and Huston (2003), willingness to take nancial risk is as-

sociated with a signicantly higher net worth and is a strong

predictor of higher net worth for those older than 65 years.

Thus, understanding a client’s risk tolerance is an impor-

tant responsibility for nancial advisors. If indeed individu-

als become temporarily more risk averse during economic

downturns, advisors may need to provide interventional

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016266

in portfolio risk between 2007 and 2009. Finally, results and

a discussion of the implications are provided. This study

extends the literature by identifying characteristics of those

who are most likely to reduce their portfolio risk in reaction

to a downturn in nancial markets.

Literature Review on Risk Tolerance

Insights From Behavioral Finance

Classic theory (Arrow, 1965) proposed that relative risk aver-

sion (RRA) is a function of wealth. Depending on assumptions

about the form of an individual’s utility function, RRA may

be increasing, constant, or decreasing with respect to wealth.

Such conventional theory assumes that individuals are ratio-

nal wealth maximizers. However, individuals do not always

behave according to rational economic model assumptions.

The relatively new eld of behavioral nance combines

psychology with nance to explain why actual behavior

may deviate from that of rational economic agents and can

be inuenced by emotion. Key concepts identied include

availability bias and overreaction. The availability heuristic,

introduced by Tversky and Kahneman (1973) is the notion

that people tend to weight more heavily the information and

events that come to mind more easily. Generally, that leads

to decisions that are based on the latest news or dramatic

and unexpected events. De Bondt and Thaler (1985) found

that investors overreact to new information. In their study,

investors overreacted to bad news, driving stock prices down

disproportionately, and overreacted to good news, moving

stock prices in the opposite direction. Eventually, the prices

rebounded as investors realized that they had overreacted.

Loewenstein, Weber, Hsee, and Welch (2001) proposed a

model of risky choice whereby strong anticipatory emotions

(e.g., fear and anxiety) felt at the moment of decision mak-

ing play a role in the outcome. This risk-as-feelings hypoth-

esis was tested in an experimental setting by Kuhnen and

Knutson (2011); events associated with positive emotions

were found to lead to riskier behavior, whereas those associ-

ated with negative emotions led to more risk-averse behavior.

Empirical studies have shown that traumatic events such as

the 2008 nancial crisis can affect willingness to take risk.

The crisis provided a convenient setting for applying the risk-

as-feelings hypothesis. Using results of a survey of a large

Italian bank’s customers in 2007 and repeated in 2009, Guiso,

Sapienza, and Zingales (2013) found that risk aversion

increased after the 2008 nancial crisis. They found sup-

port for a fear-based model that predicts that investors will

sell stock after a sharp decline. Investors become less risk

tolerant not simply because their wealth declined but also

because of a traumatic event such as the sharp drop in stock

prices. Availability bias impacted their risk tolerance.

In a study of German households, Necker and Ziegelmeyer

(2014) found that those who attributed losses in wealth

between 2007 and 2009 to the nancial crisis experienced

decreased risk tolerance, whereas those who did not attri-

bute losses to the crisis saw no change in risk attitude. They

attributed the change in risk tolerance to a psychological

reaction to the nancial crisis.

Risk tolerance has been found to be relatively stable over

time, but individuals may exhibit time-varying risk aver-

sion in the short run. Guillemette and Finke (2014) found

that although risk tolerance scores were relatively stable in

the long run, they were strongly correlated with recent stock

market movements in the short run. Risk tolerance increased

as stock market valuations increased and decreased during

market downturns. Investors’ appetites for risk tended to

change in reaction to market returns.

Studies have shown that risk tolerance tends to increase

when market returns increase, and decrease when mar-

ket returns decrease (Yao & Curl, 2011; Yao, Hanna, &

Lindamood, 2004). This tendency can be explained by

availability bias or recency effect, a phenomenon whereby

most recent events have the most impact on perceptions.

Unfortunately, if investors act accordingly, they may under-

mine their returns by buying after a gain (when prices are

high) and selling after a loss (when prices are low).

The recency effect is of special concern for young investors

because younger individuals were found to be more sensi-

tive to recent returns than older individuals (Malmendier

& Nagel, 2011). Some nancial advisors have expressed

concern that young investors hold portfolios that are too

conservative, selecting safe investments over equities

(Dagher, 2011; Light, Pilon, & Silver-Greenbert, 2011;

Yousuf, Wang, & Derousseau, 2011). The experience of

the depressed job market and decline in stock prices during

the nancial crisis may explain their reluctance to take risk.

Regrettably, inadequate retirement funds are a likely result

of this low risk tolerance.

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016 267

Recent research links behavioral nance constructs to risk

tolerance. Griesdorn, Lown, DeVaney, Cho, and Evans

(2014) investigated how decision-making strategies sug-

gested by the behavioral life cycle hypothesis are related to

risk tolerance. They found a signicant positive relationship

between self-control and risk tolerance. Those who scored

higher in self-control scored signicantly higher in risk toler-

ance. They also found to a lesser extent that mental account-

ing and framing constructs were related to risk tolerance.

Demographic and Socioeconomic Characteristics

Many empirical studies have shown that demographic

and socioeconomic characteristics impact risk tolerance,

although the evidence is mixed in terms of the direction of

the impact. A preponderance of research has provided evi-

dence that risk tolerance increases with wealth (Griesdorn

et al., 2014; Hartog, Ferrer-i-Carbonell, & Jonker, 2002;

Schooley & Worden, 1996) and income (Gibson, Michayluk,

& Van de Venter, 2013; Hartog et al., 2002). Some evidence

supported the belief that risk tolerance decreases with age

(Gibson et al., 2013; Griesdorn et al., 2014; Hallahan, Faff,

& McKenzie, 2004; Yao & Curl, 2011; Yao et al., 2004);

other studies found the relationship with age to be nonlinear

(Hallahan et al., 2004; Riley & Chow, 1992). Being married

has been shown to have a signicant negative impact on risk

tolerance (Hallahan et al., 2004; Yao et al., 2004), although

McInish (1982) did not nd a signicant relationship. Num-

ber of dependents has also been shown to have a negative

impact on risk tolerance (Hallahan et al., 2004).

Research has provided evidence that education has a posi-

tive effect on risk tolerance (Griesdorn et al., 2014; Hallahan

et al., 2004; Hartog et al., 2002; Yao & Curl, 2011; Yao et al.,

2004). Those who consulted a nancial advisor exhibited

lower risk tolerance (Gibson et al., 2013). Self-employed

workers have been shown to be more risk tolerant than

those who work for others (Colombier, Boèmont, Loeac, &

Masclet, 2008; Hartog et al., 2002; Sung & Hanna, 1996;

Yao et al., 2004). A study of how personality characteris-

tics inuence decisions on entrepreneurial status in German

households found that the probability of entry into self-

employment is higher for those with higher risk tolerance

(Caliendo, Fossen, & Kritikos, 2014). Interestingly, the prob-

ability of exiting from self-employment is higher for those

with low or high risk tolerance but lower for those with me-

dium risk tolerance. They also found evidence that those who

are more risk tolerant are more likely to be self-employed.

Conversely, Halek and Eisenhauer (2001) found evidence

that the self-employed are more averse to downside risk than

those employed by others. The inclusion of socioeconomic

factors in an examination of risk tolerance is supported by

these many and varied results.

The literature supports the notion that investors may not

always behave rationally. They may overreact to bad news

and behave in a more risk-averse manner when they experi-

ence negative emotions. Investors’ reactions to a decrease

in stock prices caused by a traumatic event such as the

nancial crisis can be much more emotional than to a grad-

ual decrease in stock prices. Italian investors became less

risk tolerant after the 2008 nancial crisis. German house-

holds who attributed losses in wealth to the nancial crisis

experienced decreased risk tolerance. Risk tolerance has

also been linked to socioeconomic and demographic char-

acteristics such as the use of a nancial advisor, education

level, stage of life cycle (incorporating age, marital status,

and children), employment status, income, and wealth.

The objective of this study was to examine how changes

in U.S. households’ risk tolerance (realized and perceived)

is related to the 2008 nancial crisis. Contrary to popular

belief, not everyone became less risk tolerant as a result of the

crisis. Univariate analysis tested the hypotheses that a house-

hold became more risk tolerant when its wealth increased

after the crisis and became less risk tolerant when its wealth

decreased after the crisis. Regression analysis tested the hy-

pothesis that realized risk tolerance increased for those who

perceive they became more risk tolerant and decreased for

those who perceive they became less risk tolerant.

Methods

Data and Sample

The data for this study were from the 2007–2009 SCF panel.

The SCF is a triennial interview survey of U.S. households

sponsored by the Federal Reserve Board of Governors

(FRB). The purpose of the SCF is to provide a comprehen-

sive view of the nancial behavior of a cross section of U.S.

households. Information is gathered on assets and liabilities

of the household as well as demographic and socioeconomic

characteristics such as age, income, education, and makeup

of the household unit. Attitudes toward risk taking are also

surveyed. Because of the severity of the nancial crisis and

recession, in 2009, the FRB implemented a follow-up sur-

vey of households that had participated in the 2007 triennial

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016268

SCF. The result was a panel database of information from

the pairs of interviews that allowed for comparison between

2007 and 2009.

About 89% (3,862) of the eligible households completed the

panel interview in 2009. To protect the privacy of several

wealthy households, the public dataset was reduced to 3,857.

Furthermore, because of extreme values that indicated data

error, one observation was eliminated from this analysis,

reducing the sample size to 3,856 households. The panel data

are available at the Federal Reserve website (FRB, 2009),

and the surveys are described by Bricker, Bucks, Kennickell,

Mach, and Moore (2011) and Kennickell (2011).

The 2007 SCF was conducted from May through December,

with a few interviews in early 2008. During that time, it was

obvious that the economy was slowing, even though the

National Bureau of Economic Research (NBER) declared

that the recession began in December 2007. The 2009 rein-

terview was conducted from July through December, with a

few completed in early 2010. Although the economic recov-

ery was not particularly strong, according to the NBER, June

2009 was the ofcial ending of the recession (NBER, 2010).

Variables

The conventional denition of nancial risk is uncertainty

in returns. Because equity securities historically have ex-

hibited the highest standard deviation of returns, investors’

risk-taking behavior—their realized risk tolerance—was

measured by the percentage of nancial assets held in equity

securities. Financial assets included transaction accounts,

certicates of deposits, savings and other bonds, stock,

pooled investment funds, retirement accounts, the cash

value of life insurance, and other managed accounts such as

trusts or annuities. Equity securities included stock held di-

rectly or indirectly through mutual funds and trusts. If a fund

was reported as diversied, the SCF coding was employed

to determine the value ascribed to equity holdings.

A household’s perceived risk tolerance was measured by a

response to a survey question. For each year, respondents

were asked to identify their risk tolerance by selecting which

one of the following statements reects the amount of risk

they were willing to take when saving or making investments:

• Take substantial nancial risks expecting to earn

substantial returns

• Take above-average nancial risks expecting to

earn above-average returns

• Take average nancial risks expecting to earn

average returns

• Not willing to take any nancial risks

Respondents who selected the “substantial” statement were

viewed to be the most risk tolerant (least averse) and those

who selected the “not willing” statement were considered to

be the least risk tolerant (most averse).

A comparison of the responses in the two interviews pro-

vided a measure of change in perceived risk tolerance after

the nancial crisis. A household who indicated a lower risk-

taking category in 2009, as compared to the 2007 choice,

was deemed to perceive a lower risk tolerance level (more

risk aversion). Conversely, one who indicated a higher risk-

taking category was deemed to perceive a higher risk toler-

ance (less risk aversion).

Analysis

This study’s investigation of changes in risk tolerance used

both univariate and multivariate analyses. Univariate analy-

sis examined the relationship between wealth changes and

changes in risk tolerance—both perceived and realized

through changes in portfolio composition. The SCF use of

a dual-frame sampling design that oversamples the wealthy

required that an analysis weight be employed so that the

distributions were representative of the populations of U.S.

households. Because previous studies showed that wealth is

signicantly related to risk tolerance, changes in perceived

risk tolerance were examined separately for households who

experienced a decrease in wealth and those whose wealth

increased. Also, changes in perceived risk tolerance and

portfolio risk were examined to determine whether those

who perceived that their risk tolerance had changed adjusted

their portfolio risk (realized risk tolerance) accordingly. The

analysis was extended further to determine if the results

depended on whether wealth had increased or decreased.

Multivariate analysis was then used to examine factors that

impacted the direction and magnitude of the change in the

riskiness of a household’s nancial portfolio (realized risk

tolerance) between the time of the 2007 and 2009 surveys.

The dependent variable “PortRisk07-09” measured the dif-

ference in the percentage of nancial assets held in stock:

2007 minus 2009. The larger the PortRisk07-09, the greater

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016 269

was the reduction in the percentage of nancial assets held

as stock. A negative difference indicates that the percentage

of nancial assets held in stock was greater in 2009 than

it was in 2007. Variables for demographic and socioeco-

nomic factors that have been shown in the literature to im-

pact risk tolerance were included as explanatory variables.

These include the use of a nancial advisor, education level,

stage of life cycle (incorporating age, marital status, and

children), employment status, income, and wealth.

Results

Descriptive Statistics

Table 1 provides the distribution of households whose port-

folio risk declined, increased, or saw no change between

2007 and 2009. Interestingly, the distribution is fairly even.

As shown in Table 1, 35.3% of households decreased the

percentage of stock held in their portfolios, whereas 32.7%

increased the percentage of stock and 32% had no change.

The median decline was 21.1%, whereas the median

increase was 19.8%.

The distribution of households’ perceived nancial risk toler-

ance in 2007 versus 2009, as measured by these responses, is

presented in Table 2. In the aggregate, households appeared

to perceive less risk tolerance in 2009 as compared to 2007.

Six percent fewer households in 2009 reported that they were

willing to take “above-average risk,” as compared to 2007.

And, 6% more households in 2009 reported that they would

not tolerate any risk in investments, as compared to 2007.

Univariate Analyses

As shown in Table 3, overall just over a quarter (25.9%) of

households reported less tolerance for risk after the nancial

crisis. Less than 16% of households reported a higher toler-

ance for risk after the nancial crisis.

Table 3 also illustrates how the change in perceived risk tol-

erance was associated with the change in household wealth,

dened as total assets minus total liabilities. The nan-

cial crisis had a varied impact, with 62.5% of households

experiencing a decrease in wealth and 36.8% experiencing

an increase in wealth. Only 0.7% experienced no change in

wealth. Although the difference was not highly signicant

(p 5 .086), the percentage reporting less risk tolerance was

greater (at 27%) for those households whose wealth de-

creased than for those whose wealth increased (at 24.4%).

A more signicant difference (p 5 .014) appeared for the

percentage reporting more risk tolerance. Only 14.6% of

households whose wealth decreased during the crisis per-

ceived more tolerance for risk, compared to 18% of house-

holds whose wealth increased.

The nancial portfolio composition for households

revealed that the change in wealth and reported risk toler-

ance was associated with a change in investment behav-

ior. Table 4 considers the composition of the portfolio of

nancial assets of households who experienced a change

in wealth during the nancial crisis—across the reported

change in risk tolerance. As discussed earlier, a household’s

TABLE 1. Change in Asset Allocation—Percentage of Financial Assets Held in Stock 2007 Versus 2009

Change in Percentage Held

“PortRisk07-09” Distribution of Households (%) Median Change 2007–2009 (%)

Decline 2007% . 2009%

35.3 21.1

No change 2007% 5 2009%

32.0 0.0

Increase 2007% , 2009%

32.7

219.8

Note. n 5 3,621; excludes 235 households that had no nancial assets in 2007 and/or 2009.

TABLE 2. Distribution of Financial Risk Tolerance 2007 Versus 2009 “Willing to Take . . . ”

Survey Year

Substantial Risk for

Substantial Returns (%)

Above-Average Risk for

Above-Average Returns (%)

Average Risk for

Average Returns (%) No Risk (%)

2007 3.5 17.5 38.3 40.7

2009 3.3 11.4 39.1 46.2

Note. n 5 3,856.

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016270

risk-taking behavior, or realized risk tolerance, was mea-

sured by the percentage of nancial assets held in equity

securities. Furthermore, because of their low standard

deviation of returns, liquid nancial assets were dened

as “risk-free,” and the percentage of nancial assets held

in risk-free securities provided a measure of realized risk

“intolerance.”

More than 60% of households experienced a decrease in

wealth during the time between the two surveys. Although a

decline in stock valuation alone could cause the percentage

of nancial assets held in stock to decline, it is only for

those households who reported less risk tolerance that the

median percentage declined (23.7% in 2007 to 14.8% in

2009). For those who reported a higher tolerance for risk,

the median percentage of nancial assets held in stock actu-

ally increased from 11.1% in 2007 to 17.1% in 2009. On

the other hand, the median percentage of risk-free assets

increased (from 23.8% in 2007 to 33.2% in 2009) for those

households who reported less risk tolerance.

TABLE 3. Change in Financial Risk Tolerance by Change in Wealth: 2007 Versus 2009

Percentage Reporting a Change in Risk Tolerance

Change in Wealth Between 2007 and 2009 Less Risk Tolerance More Risk Tolerance

Decrease in wealth 27.0% 14.6%

Increase in wealth 24.4% 18.0%

2.96 6.30

p value

.086

†

.014*

Across all households 25.9% 15.8%

Note. n 5 3,836; excludes 20 households with no change in wealth.

*p , .05.

†

p , .10.

TABLE 4. Portfolio Composition Across Change in Wealth and Risk Tolerance Percentage of Stock to

Financial Assets and Risk-Free to Financial Assets

2007 Versus 2009 (Median Values)

Reported Change in Risk Tolerance

Change in Wealth Between 2007 and 2009 Less Risk Tolerance (%) More Risk Tolerance (%)

Decrease in wealth

Stock/nancial assets: 2007 23.7 11.1

Stock/nancial assets: 2009 14.8 17.1

Risk-free/nancial assets: 2007 23.8 30.0

Risk-free/nancial assets: 2009 33.2 29.3

Increase in wealth

Stock/nancial assets: 2007 18.7 7.1

Stock/nancial assets: 2009 17.1 17.0

Risk-free/nancial assets: 2007 38.2 50.2

Risk-free/nancial assets: 2009 32.6 35.5

Note. n 5 3,621; excludes 235 households that had no nancial assets in 2007 and/or 2009.

Financial assets include transaction accounts, certicates of deposit, savings and other bonds, stock, pooled investment

funds, retirement account holdings, the cash value of life insurance, and other managed accounts such as trusts and

annuities. Stock includes equity shares held directly or indirectly through mutual funds and trusts, both in retirement and

nonretirement accounts. Risk-free assets include transaction accounts, savings accounts, certicate of deposits, and the cash-

value of life insurance.

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016 271

About 37% of households experienced an increase in wealth

during the time between the two surveys. For those who

reported less risk tolerance in 2009, the median percentage

of nancial assets held in stock was relatively unchanged.

However, for those who reported higher tolerance for risk,

not only did the median percentage of nancial assets held

in stock increase from 7.1% in 2007 to 17.0% in 2009, but

the median percentage of nancial assets held in liquid

assets declined from 50.2% to 35.5%.

These univariate analyses indicated that the nancial crisis

affected individuals’ attitudes toward risk and the composi-

tion of their portfolio of nancial assets in different ways

depending on how their wealth changed.

Multivariate Analyses

Variable descriptions and statistics for the multivariate

analysis are presented in Table 5.

The change in perceived risk tolerance was included in the anal-

ysis by two indicator variables. The reduction in risk tolerance

was captured by the variable “less risk tolerance,” which took

the value 1 when the household reported lower risk tolerance

in 2009 as compared to 2007. An increase in risk tolerance was

captured similarly by the variable “more risk tolerance.” No

change in perceived risk tolerance was included in the constant

term. Financial sophistication and knowledge was measured by

the variable “UseFinProfAdvisor,” which indicated whether the

household used advice from a nancial professional to make

decisions about savings and investments in 2007. A house-

hold’s behavior was expected to be inuenced during the nan-

cial crisis by an established relationship with an advisor.

The impact of the education level achieved by the household

head was measured by the indicator variables high school,

some college, and graduate degree, with college degree in-

cluded in the constant term. The household’s stage of life and

family structure was measured by a modied version of the

life cycle variable developed by Bojanic (1992). The life cy-

cle variables captured the household head’s age, the presence

of a spouse or partner, and the presence of children. Employ-

ment status variables indicated whether the household head

was self-employed or not employed, as compared to being

employed by others. These socioeconomic factors were mea-

sured at the time of the 2009 reinterview because the analysis

is interested in determining their impact on the change in the

portfolio’s composition between 2007 and 2009.

Only 2% of household heads exhibited a change in educa-

tion level between 2007 and 2009. Although 20% of house-

holds changed life cycle category during that period, the

percentage for each category was minute because of the

number possible. For employment status, 83% of household

heads maintained the same employment status at the time

of the surveys, with the remaining 17% again distributed in

small percentages among the various possibilities.

Table 6 presents results of the linear regression analy-

sis on the change in the ratio of stock to nancial assets

(PortRisk07-09).

The overall F statistic for the regression had a signicant

p value. Because there is evidence that a change in risk

tolerance is related to wealth, the percentage change in the

household’s wealth during the nancial crisis was included

as a control variable. In addition, the percentage change in

household income was held constant, so that the inuence

of education and employment status could be more clear-

ly identied. Because what is of interest for this analysis

is the change in the portfolio composition, the “starting

value,” that is, the percentage of nancial assets held in

stock in 2007 (PortRisk07), was included as an additional

control variable. A PortRisk07-09 value of 5% is more

noteworthy if the percentage held in 2007 was 7% than if

it was 30%.

Those households who perceived a higher level of risk tol-

erance in 2009 had a signicantly smaller reduction in the

percentage of nancial assets held as stock. In fact, they

could have experienced an increase in this ratio. As ex-

pected, households who reported that they became more

risk tolerant either increased the riskiness of their portfo-

lios or reduced the equity holdings by a signicantly smaller

amount. However, the change in the portfolio composition

was not signicantly different for those households who per-

ceived a lower level of risk tolerance. The other socioeco-

nomic factors must have been the contributing factors to the

change in their portfolios, not their perceived risk tolerance.

Whether the household used a nancial professional to

make savings and investment decisions at the beginning of

the nancial crisis was expected to have an impact on the

change in the portfolio allocation, but it proved to be insig-

nicant. The regression analysis was repeated using indica-

tor variables that measured changes in the use of a nancial

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016272

TABLE 5. Variable Denitions and Descriptive Statistics

Dependent Variable Denition M

PortRisk07-09

5 change in percentage of nancial assets invested in stock

between 2007 and 2009 (2007 minus 2009)

2.28%

Explanatory Variables Denition

Sample

Proportion

Change in perceived risk tolerance

Less risk tolerance (more risk

averse)

5 1 if the response to the risk tolerance question in 2009 indi-

cated less willingness to take nancial risk than the response

in 2007; 0 otherwise

0.27

More risk tolerance (less risk

averse)

5 1 if the response to the risk tolerance question in 2009

indicated more willingness to take nancial risk than the

response in 2007; 0 otherwise

0.15

No change in risk tolerance/

aversion

5 0 if the response to the risk tolerance question in 2009

indicated the same willingness as reported in 2007; in the

constant

0.58

UseFinProfAdvisor

5 1 if the household used a nancial professional (lawyer, ac-

countant, banker, broker, nancial planner) in 2007 to make

decisions about savings and investments; 0 otherwise

0.50

Education level Highest level of education earned by household head in 2009

High school

5 1 if high school diploma or less; 0 otherwise

0.32

Some college

5 1 if attended college, but a degree not earned; 0 otherwise

0.16

College degree

5 0 if earned college degree, excludes trade school certicate;

in the constant

0.29

Graduate degree

5 1 if earned a graduate degree; 0 otherwise

0.23

Life cycle Family structure in 2009

Single

5 1 if single, age younger than 55 years, no children;

0 otherwise

0.10

Couple

5 1 if married or with partner, age younger than 55 years,

no children; 0 otherwise

0.08

Full Nest I

5 1 if married or with partner, age younger than 40 years, with

children; 0 otherwise

0.08

Full Nest II

5 0 if married or with partner, age 40 years or older, with

children; in the constant

0.23

Empty nest couple

5 1 if married or with partner, age 55 years or older,

no children; 0 otherwise

0.26

Solitary survivor

5 1 if single, age 55 years or older; 0 otherwise

0.14

Single parent

5 1 if single, any age, with children; 0 otherwise

0.11

Employment status Employment status of household head in 2009

Employed by others

5 0 if work for someone else; in the constant

0.46

Self-employed

5 1 if self-employed or partnership; 0 otherwise

0.26

Not employed

5 1 if unemployed, retired, or otherwise not in labor force;

0 otherwise

0.28

(Continued)

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016 273

TABLE 5. Variable Denitions and Descriptive Statistics

(Continued)

Control Variables Denition Median

Income percentage change

5 percentage change in reported household income between

2007 and 2009; income derived from all sources, including

withdrawals from IRAs and pension accounts

25.2%

Wealth percentage change

5 percentage change in wealth calculated as the difference be-

tween 2007 and 2009 wealth, divided by the absolute value

of 2007 wealth. If a household had zero wealth in 2007, a

value of $1 is in the denominator. Wealth is measured by

household net worth—the total value of all real and nancial

assets owned, including business equity, less the value of all

mortgage and consumer debt outstanding.

219.2%

PortRisk07

5 percentage of nancial assets held in stock in 2007

28.7%

Note. n 5 3,621; excludes 235 households that had no nancial assets in 2007 and/or 2009. IRAs 5 individual retirement

accounts.

Descriptive statistics are unweighted because the weight variable is not used in the regression.

TABLE 6. Regression Analysis of Change in Ratio of Stock to Financial Assets: 2007 Minus 2009

Explanatory variable Estimated Coefcient F Statistic p Value

Less risk tolerant (more risk averse) 0.87 0.75 .385

More risk tolerant (less risk averse)

24.02***

10.82 .001

UseFinProfAdvisor 0.40 0.21 .649

Education level

High school 5.39*** 18.24 .000

Some college 3.57* 5.74 .020

Graduate degree

22.30

2.65 .111

Life cycle

Single 3.89* 4.95 .028

Couple

20.11

0.01 .957

Full Nest I 2.99 2.49 .118

Empty nest couple 0.22 0.03 .860

Solitary survivor 5.46*** 12.33 .000

Single parent 5.17*** 10.53 .001

Employment status

Self-employed 2.12

†

3.44 .065

Not employed 6.38*** 28.33 .000

Income percentage change 0.00 2.60 .107

Wealth percentage change 0.00 0.03 .858

PortRisk07 0.54*** 984.31 .000

Constant

221.23***

159.95 .000

Note. n 5 3,621. Overall F statistic 5 72.9 with p value 5 .000. The adjusted R

ranged from 0.28 to 0.29 for the ve

separate imputation regressions.

*p , .05. ***p , .001.

†

p , .10.

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016274

professional. That is, additional variables were added to

capture whether the household did not use a nancial plan-

ner in 2007, but did in 2009, and vice versa. All variables

had an insignicant impact on the dependent variable.

Further investigation showed that the use of a nancial

professional was highly correlated with the education sta-

tus of the household head; the probability that a nancial

professional was used increased with education level. Thus,

education level may have captured that effect. And, consis-

tent with the results found in other studies (Griesdorn et al.,

2014; Hallahan et al., 2004; Hartog et al., 2002; Yao & Curl,

2011; Yao et al., 2004), education had the expected positive

effect on risk tolerance in a portfolio.

Those households whose head had less education had a

signicantly larger reduction in the percentage of nan-

cial assets held in stock—compared to those with a college

degree. Those with a graduate degree were not signicantly

different.

The stage of life and family structure proved to be some-

what signicant. Although the presence of children or

age of the household head was not signicantly related to

the change in asset allocation, being single was. Singles,

with or without children, exhibited a signicantly larger

reduction in the percentage of nancial assets in stock

than those who were married or had a partner. This result

runs counter to studies that have shown that being married

negatively impacts risk tolerance (Hallahan et al., 2004;

Yao et al., 2004). Perhaps this was because of the fact that

single households do not have the possibility of a second

source of savings and investment and so were more cau-

tious in investment behavior. The regression analysis was

conducted replacing the life cycle variables with variables

measuring age of the household head and the presence of

a partner or spouse. Similar to the life cycle results, age

proved to be insignicant while being single—without a

spouse or partner—had a signicantly positive impact on

the size of the reduction in the percentage of nancial as-

sets held in stock.

Households with self-employed or nonemployed heads in

2009 exhibited a signicantly larger reduction in the per-

centage of nancial assets held as stock—as compared to

those employed by others. This employment status effect

is in addition to the impact of an income differential and so

is likely a psychological phenomenon. Although previous

studies (Colombier et al., 2008; Hartog et al., 2002; Sung &

Hanna, 1996; Yao et al., 2004) found that the self-employed

were more risk tolerant, those in this study became less risk

tolerant after the nancial crisis. The severity of the nan-

cial crisis and its devastating impact on the job market are

the likely causes of this result.

Discussion

The question of how risk tolerance changed because of

the 2008 nancial crisis is not a simple one to answer. The

study nds that the change in risk tolerance is related to

the change in wealth experienced. A decrease in wealth is

associated with a higher probability of reporting less risk

tolerance, whereas an increase in wealth is associated with a

higher probability of reporting more risk tolerance. In their

study of the SCF panel data, Bricker, Bucks, Kennickell,

Mach, and Moore (2012) investigated changes in household

wealth and portfolio composition during the nancial crisis.

They suggested that a large majority of households passively

accepted changes in their portfolio composition that were

because of asset revaluation. However, this study found

evidence that refutes the suggestion that households were

passive. The way in which portfolio composition changed

depended on whether wealth increased or decreased.

As also suggested by Bricker et al. (2012), declining stock

values could explain why the percentage of nancial assets

held in liquid assets increased. However, it is only for those

who reported less risk tolerance that the median percent-

age of risk-free assets increased. For those households

who reported a higher tolerance for risk, the median value

remained relatively unchanged. These results reveal a more

than passive response to the nancial crisis. It would be an

odd coincidence if those who reported more risk tolerance

actively managed their portfolios to increase risk, whereas

those who reported less risk tolerance passively accepted

the decreased risk in their portfolios resulting from changes

in market valuations.

Those who perceive that they became less risk tolerant hold

less risky nancial portfolios than they did before the nan-

cial crisis. The danger in decreasing portfolio risk (holding

less stock as a percentage of nancial assets) after an event

such as the 2008 nancial crisis is that investors end up sell-

ing stock when prices are low. In doing so, they could fall

further behind in achieving their nancial goals.

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016 275

Implications

Single households, those households whose head has not

earned a college degree or is either self-employed or not

employed, may be especially vulnerable to decreasing

their portfolio risk during major negative economic events.

Financial advisors need to revisit with their clients about

their risk tolerance, especially in light of the call for in-

creased diligence because of the recent nancial crisis.

Although changes in demographic circumstances such as

marriage or having children are overt signs that risk toler-

ance may have changed, market conditions can also change

a client’s tolerance for risk.

Commitment strategies can be used to inuence households’

saving and investment behavior. For example, Smith and

Griesdorn (2014) found that the self-employed were more

likely to make retirement contributions when they employed

savings rules as a commitment strategy. Because those with

more self-control have been shown to be more risk tolerant

(Griesdorn et al., 2014), nancial counselors may be able to

help them become more comfortable taking risk by suggest-

ing some commitment strategies. For example, they could

suggest that their clients make a commitment to maintain a

particular level of risk in their portfolios. As markets uctu-

ate, they may be less tempted to adjust their portfolio risk

accordingly if they are committed to holding a portfolio

with a particular amount of risk.

Clients need to be comfortable with the risk level they are

taking, especially during economic downturns. Advisors can

help investors overcome the recency effect so that market

uctuations do not cause them to change their allocations

dramatically. They can assist clients in maintaining a sense

of perspective so that they can overlook the latest news and

keep a long-term focus.

Guillemette and Finke (2014) suggested that nancial planners

can be of great value by assisting their clients in developing a

long-run strategy to deter them from selling low and buying

high because their risk aversion varies in the short run.

Limitations and Future Research

Although the SCF risk aversion measure is widely used

in risk tolerance literature, readers should interpret results

with some caution. Some studies support the SCF risk mea-

sure as an effective proxy for risk tolerance. For example,

Gilliam, Chatterjee, and Grable (2010) found that the SCF

question indicated an individual’s investment risk tolerance

reasonably well. However, it has been suggested that this

single-item measure may not capture risk tolerance as well

as a multiquestion measure (Grable & Lytton, 2001). Also,

individuals may have limited ability to self-assess their risk

tolerance. The SCF question may serve as a better measure

of investment risk tolerance than nancial risk tolerance.

Even so, the measure does provide some information about

risk tolerance. Future research could further study the re-

sponse to the SCF question as well as explore better mea-

sures for assessing risk aversion.

References

Arrow, K. J. (1965). Aspects of the theory of risk-bearing.

Yrjö Jahnsson Lectures. Helsinki, Finland: Yrjö Jahns-

son Foundation.

Bojanic, D. (1992). A look at a modernized family life cycle

and overseas travel. Journal of Travel and Tourism

Marketing, 1(1), 61–78.

Bricker, J., Bucks, B., Kennickell, A., Mach, T., & Moore,

K. (2011). Surveying the aftermath of the storm:

Changes in family nances from 2007 to 2009 (Finance

and Economics Discussion Series 2011-17). Washing-

ton, DC: Board of Governors of the Federal Reserve

System. Retrieved from http://www.federalreserve

.gov/pubs/feds/2011/201117/201117pap.pdf

Bricker, J., Bucks, B., Kennickell, A., Mach, T., & Moore,

K. (2012). The nancial crisis from the family’s per-

spective: Evidence from the 2007-2009 SCF panel. The

Journal of Consumer Affairs, 46(3), 537–555.

Caliendo, M., Fossen, F., & Kritikos, A. S. (2014). Person-

ality characteristics and the decisions to become and

stay self-employed. Small Business Economics, 42(4),

787–814.

Colombier, N., Boèmont, L., Loheac, Y., & Masclet, D.

(2008). Risk aversion: An experiment with self-

employed workers and salaried workers. Applied Eco-

nomics Letters, 15(10), 791–795.

Dagher, V. (2011, June 20). Come on Gen X, take some

chances: Financial advisers worry that their younger

clients are playing it too safe with their money. The

Wall Street Journal, p. R3.

De Bondt, W., & Thaler, R. (1985). Does the stock market

overreact? The Journal of Finance, 40(3), 793–805.

Federal Reserve Board. (2009). Survey of consumer

nances. Retrieved from http://www.federalreserve

.gov/econresdata/scf/scrf_2009psurvey.htm

Journal of Financial Counseling and Planning, Volume 27, Number 2, 2016276

Finke, M., & Huston, S. (2003). The brighter side of nan-

cial risk: Financial risk tolerance and wealth. Journal

of Family and Economic Issues, 24(3), 233–256.

Gibson, R., Michayluk, D., & Van de Venter, G. (2013).

Financial risk tolerance: An analysis of unexplored

factors. Financial Services Review, 22(1), 23–50.

Gilliam, J., Chatterjee, S., & Grable, J. (2010). Measuring

the perception of nancial risk tolerance: A tale of two

measures. Journal of Financial Counseling and Plan-

ning, 21(2), 30–43.

Grable, J. E., & Lytton, R. H. (2001). Assessing the con-

current validity of the SCF risk tolerance question.

Financial Counseling and Planning, 12(2), 43–53.

Griesdorn, T., Lown, J., DeVaney, S., Cho, S., & Evans, D.

(2014). Association between behavioral life-cycle con-

structs and nancial risk tolerance of low-to-moderate-

income households. Journal of Financial Counseling

and Planning, 25(1), 27–40.

Guillemette, M., & Finke, M. (2014). Do large swings in

equity values change risk tolerance? Journal of Finan-

cial Planning, 27(6), 44–50.

Guiso, L., Sapienza, P., & Zingales, L. (2013). Time varying

risk aversion (NBER Working Paper No. 19284). Cam-

bridge, MA: National Bureau of Economic Research.

Halek, M., & Eisenhauer, J. (2001). Demography of risk

aversion. The Journal of Risk and Insurance, 68(1),

1–24.

Hallahan, T. A., Faff, R. W., & McKenzie, M. D. (2004).

An empirical investigation of personal nancial risk

tolerance. Financial Services Review, 13(1), 57–78.

Hartog, J., Ferrer-i-Carbonell, A., & Jonker, N. (2002).

Linking measured risk aversion to individual charac-

teristics. Kyklos, 55(1), 3–26.

Kennickell, A. (2011, August). Look again: Editing and im-

putation of SCF panel data. Paper presented at 2011

Joint Statistical Meeting, Miami, FL. Retrieved from

http://www.federalreserve.gov/econresdata/scf/files/

ASA2011.1.pdf

Kuhnen, C. M., & Knutson, B. (2011). The inuence of

affect on beliefs, preferences, and nancial decisions.

Journal of Financial and Quantitative Analysis, 46(3),

605–626.

Light, J., Pilon, M., & Silver-Greenbert, J. (2011, Novem-

ber 5). The young and the riskless. The Wall Street

Journal, p. B7.

Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch,

N. N. (2001). Risk as feelings. Psychological Bulletin,

127(2), 267–286.

Malmendier, U., & Nagel, S. (2011). Depression babies:

Do macroeconomic experiences affect risk taking? The

Quarterly Journal of Economics, 126(1), 373–416.

McInish, T. (1982). Individual investors and risk-taking.

Journal of Economic Psychology, 2(2) 125–136.

National Bureau of Economic Research. (2010). US busi-

ness cycle expansions and contractions. Retrieved

from http://www.nber.org/cycles.html

Necker, S., & Ziegelmeyer, M. (2014). Household risk tak-

ing after the nancial crisis (MEA Discussion Paper

Series 14279). Munich, Germany: Munich Center for

the Economics of Aging at the Max Planck Institute for

Social Law and Social Policy.

Riley, W., & Chow, K. (1992). Asset allocation and indi-

vidual risk aversion. Financial Analysts Journal, 48(6),

32–37.

Schooley, D., & Worden, D. D. (1996). Risk aversion mea-

sures: Comparing attitudes and asset allocation. Finan-

cial Services Review, 5(2), 87–99.

Smith, H., & Griesdorn, T. (2014). Do saving rules in-

uence self-employed households’ participation in

tax-deferred retirement plans? Family & Consumer

Sciences Research Journal, 43(1), 47–60.

Sung, J., & Hanna, S. (1996). Factors related to risk toler-

ance. Financial Counseling and Planning, 7(1), 11–19.

Tversky, A., & Kahneman, D. (1973). Availability: A heu-

ristic for judging frequency and probability. Cognitive

Psychology, 5(2), 207–232.

Yao, R., & Curl, A. (2011). Do market returns inuence

risk tolerance? Evidence from panel data. Journal of

Family and Economic Issues, 32(3), 532–544.

Yao, R., Hanna, S., & Lindamood, S. (2004). Changes in

nancial risk tolerance, 1983–2001. Financial Services

Review, 13(4), 249–266.

Yousuf, H., Wang, P., & Derousseau, R. (2011). The young

and the riskless. Money, 40(1), 146–151.

Reproduced with permission of the copyright owner. Further reproduction prohibited without

permission.