Financial competence, risk presentation and

retirement portfolio preferences.

Hazel Batemany, Christine Eckertz, John Gewekex,

Jordan Louviere{, Stephen Satchellkand Susan Thorp

5 March, 2011

Keywords: discrete choice, risk preference, numeracy skills, …nancial

literacy
JEL Classi…cation: G23; G28; D14

We thank Olivia Mitchell and Jack Gray for helpful comments. The authors acknowledge …nan-
cial support under ARC DP1093842, generous assistance with the development and implementation
of the internet survey from PurePro…le and the sta¤ of the Centre for the Study of Choice, Univer-
sity of Technology Sydney; and excellent research assistance from Frances Terlich and Edward Wei.
Part of this work was completed while Bateman visited the School of Finance and Economics at
the University of Technology Sydney. Thorp acknowledges the support of the the Sydney Financial
Forum, the NSW Premier’s O¢ ce, the Association of Superannuation Funds of Australia (ASFA),
the Industry Superannuation Network (ISN), and the Paul Woolley Centre for the Study of Capital
Markets, UTS.
y
Centre for Pensions and Superannuation, University of New South Wales,
[email protected]
z
Centre for the Study of Choice, University of Technology Sydney, [email protected]
x
Centre for the Study of Choice, University of Technology Sydney, [email protected].
{
Centre for the Study of Choice, University of Technology Sydney, [email protected]
k
Trinity College, University of Cambridge, University of Sydney, [email protected]
Corresponding author: Centre for the Study of Choice, University of Technology Sydney,
[email protected], +612 95147784.

1
 Abstract

We investigate risk presentations in retirement savings decisions using a discrete

choice experiment where subjects choose between a bank account, a growth account
and a 50:50 account. Using nine standard formats for investment risk, we analyze
responses to risk per se and to format changes. Switching between formats changes
individuals’investment decisions, given constant risk and return. Choices made un-
der graphical and textual presentations contrast markedly, as do choices based on
formats that emphasize event frequencies rather than return ranges or values at risk.
Less numerate individuals tend to be more susceptible to format changes, but higher-
than-average personal …nancial competence does not eliminate presentation e¤ects.
Respondents with weak basic …nancial literacy are less sensitive to changes in under-
lying risk, regardless of presentation format. Variations in demographics and expec-
tations can impact account choice probabilities at least as much as variation in the
risk/return trade-o¤. Policy-makers and industry need to coordinate over …nancial
product disclosures and initiatives for improving …nancial competence.

2
1 Introduction

Retirement incomes policy is moving away from public pension provision towards
enabling individuals to save through privately organized de…ned contribution schemes.
As a result, many ordinary retirement fund members must choose a portfolio for
their retirement savings, a choice where an appreciation of investment risk is crucial.
Portfolio choices re‡ect the interaction between an individual’s …nancial competence
and the menu of investment information o¤ered by a plan provider, adviser or …nancial
institution. These choices have important and far reaching implications for the long
term adequacy and security of retirement incomes.
Personal …nancial skills and successful retirement planning are strongly comple-
mentary (van Rooij et al., 2007; Lusardi and Mitchell, 2007; van Rooij et al., 2009;
Yoong, 2010) and governments around the world are attempting to improve over-
all levels of …nancial competence.1 In 2010, the G20 Summit in Seoul endorsed nine
‘Principles of Innovative Financial Inclusion’, which, require, inter alia, member coun-
tries to ‘develop …nancial literacy and …nancial capability’(G20 Financial Inclusion
Experts Group, 2010). In addition, the format of information o¤ered by the …nan-
cial services industry matters. Investment menu design (Benartzi and Thaler, 2001;
Madrian and Shea, 2001; Huberman and Jiang, 2006; Brown et al., 2007; Beshears
et al., 2009; Agnew and Szykman, 2005), speci…c presentations of return and risk
(Benartzi and Thaler, 1999; Rubaltelli et al., 2005; Anagol and Gamble, 2008), and
framing (Brown et al., 2008; Agnew et al., 2008) all in‡uence investment decisions
independently of underlying portfolio risk and return. Further, less …nancially liter-
ate individuals are more susceptible to presentation e¤ects in some …nancial decisions
(Agnew et al., 2008; Hastings et al., 2010).
1
A collective term for numeracy skills and …nancial literacy.

3
 Retirement plan providers communicate risk to members in a range of standard
formats. Here we show the extent of variation in retirement savings investment choices
caused by changing from one standard format of risk presentation to another. We
test nine of the most common mass-market risk presentations in a discrete choice
experiment (DCE) on more than 1200 retirement savers. Conditioning on stockmar-
ket expectations and demographics, we assess the importance of risk perception and
presentation sensitivity as they interact with …nancial competence. We identify risk
presentations that tilt choices towards, or away from, high return/high risk accounts,
and pro…le the groups of people most likely to choose safe or risky allocations.
Switching from one standard format to another does change individuals’invest-
ment decisions, even when underlying risk and return are held constant. We …nd a
marked contrast between graphical and textual presentations of investment risk and
between formats that emphasize event frequencies rather than return ranges or values
at risk. Respondents with poor numeracy tend to be more susceptible to presentation
changes whereas respondents with weak basic …nancial literacy are less sensitive to
changes in the underlying risk/return pro…le of investments. Further, age, size of re-
tirement accumulation and expectations about post-crisis stock market recovery are
very important predictors of choices. In fact, variations in these pre-existing personal
characteristics and opinions play at least as large a role in predicting allocations as
the information that theory treats as the key - that is, variation in the risk/return
trade-o¤.
Existing studies of investment risk framing analyze dynamic and static illustra-
tions of risk, including the combination of text and graphs (Vlaev et al., 2009),
portfolio return distribution builders (Goldstein et al., 2008) and ‘experience sam-
pling’, where decision makers receive feedback about outcomes (Haisley et al., 2010).
While distribution builders and ‘experience sampling’are important additions to the

4
…nancial advice tool kit, their role in retirement planning is usually restricted to
professional advisers or proprietary interfaces, out of reach of most retirement fund
members. By contrast, policymakers are concerned with e¤ective communication to
the population at large through intelligible, standardized formats that allow com-
parisons across products. Australian regulators, for example, have recently made
mandatory a short-form …nancial product disclosure document for pension fund and
asset allocation decisions (Minister for Financial Services, 2010), the UK Financial
Services Authority is working to make insurance and pension disclosures easier to
understand (Andrews, 2009), and the US Department of Labor is proposing a stan-
dard simpli…ed disclosure format (Hung et al., 2010). Cross-country coordination of
simpli…ed disclosures is being initiated by multilateral organizations such as the Eu-
ropean Commission and the OECD (European Commission, 2009a,b; Organisation
for Economic Cooperation and Development, 2008). Our results contribute to this
policy debate.
The next section explains the context and design of the discrete choice experiment,
including the survey structure, alternative risk presentations and sample character-
istics. Section 3 describes the underlying model for investment choice. Results are
presented and discussed in Section 4, while Section 5 concludes with a summary of
our …ndings and implications for policy development.

2 Setting and experimental design

We draw an experimental sample from Australian retirement savers who partici-

pate in the publicly mandated but privately managed retirement incomes system,
the Superannuation Guarantee. Most Australian workers hold accounts in privately-
managed, de…ned contribution (DC) retirement savings (‘superannuation’) funds and

5
are required to choose an investment portfolio for their (mandatory) contributions.
Superannuation fund members’investment decisions are very important, since Aus-
tralia does not have a social security system based on individual earnings history
and these DC accounts hold the bulk of mandatory retirement accumulations. Most
funds o¤er expansive investment menus: a typical not-for-pro…t fund o¤ers 10-12 in-
vestment choices, rising to an average of over 200 in the for-pro…t sector (Australian
Prudential Regulation Authority, 2011). The related …nancial product disclosure
statements usually extend to several hundred pages, creating an arduous investment
decision process. This burden is increasingly common around the world, as masses
of workers …nd that individual DC accounts are replacing corporate or public de…ned
bene…t (DB) pension schemes.

2.1 Survey structure and sample.

For the discrete choice experiment, we recruited a sample of 1220 individuals over
the age of 18 from the PurePro…le online web panel of over 600,000 Australians.
PurePro…le …ltered the sample to ensure that all respondents currently held retire-
ment savings (superannuation) accounts, that genders were equally represented and
that the age distribution did not deviate far from population proportions. Of this
sample, 1199 fully completed the survey over the internet, and were paid a ‡at rate
of $3AUD ($3USD) by PurePro…le. Table 1 compares the survey sample with the
general population, demonstrating a reasonable match on most categories, with some
over-sampling of the employed and more educated, a consequence of selecting super-
annuation account holders.
Respondents completed a four-part questionnaire comprised of:
Introductory questions about subjects’ retirement savings, including the

6
name of their superannuation fund and the aggregate amount in their accounts (See
Appendix A);
14 questions to measure …nancial competence as well as questions to elicit
self-assessed knowledge of …nance, access to …nancial education, use of …nancial advice
and con…dence in stock market recovery;
A hypothetical asset allocation task for retirement savings; and
Demographic questions relating to marital status, work status, occupation,
industry/business, education, income, assets, household make-up and number in
household.2
Numeracy questions (Appendix A, Q1 to Q5) test concepts such as fractions, per-
centages, division, multiplication and simple probability (Gerardi et al., 2010). For
measures of …nancial literacy, we follow the American Life Panel (ALP), the Health
and Retirement Survey (HRS) and the Dutch Household Survey (DHS), dividing
questions into ‘basic …nancial literacy’ covering compound interest, in‡ation, time
value of money and money illusion (Appendix A, Q6 to Q10). and ‘sophisticated
…nancial literacy’ relating to the asset allocation decisions frequently required for
retirement saving (Appendix A, Q11 to Q14). These test knowledge of the di¤er-
ences between bonds and stocks and the impact of risk diversi…cation (Lusardi and
Mitchell, 2009; van Rooij et al., 2009). For some questions we adapted the wording
to Australian terminology and practices. Respondents also gave a self-assessment of
their understanding of …nance on a scale of 1 (very low) to 7 (very high) (Appendix
A, P3); reported current (and prior) access to …nancial education at school (P4) and
in the workplace (P5), and indicated they had recently paid for …nancial advice (P6).
Further questions assessed respondents’expectations of the prospect of another stock
market crash in the near and distant future, and prospects of recovery (Appendix A,
2
We provide access to the entire survey at http://survey.con…rmit.com/wix/p1250911674.aspx

7
S1 to S3).
Table 2 summarizes responses to the introductory survey questions. The distribu-
tion of self-reported retirement savings account balances closely matched Australian
Bureau of Statistics survey data (ABS, 2008), showing fairly modest accumulations,
with 70% at $80,000 or less. Self-assessed …nancial competence was high, (partic-
ularly for males) with more than 80% of respondents reporting at least an average
understanding of …nance. Views on the likelihood of another equity market crash in
the near future and speed of recovery from the Global Financial Crisis were di¤use,
and around 20% of respondents could not assign a probability to the prospects.
Most respondents answered numeracy questions correctly, although 17% scored
at least one incorrect response (Table 3). Results for the …nancial literacy questions
were more variable, with only 36.5% of respondents correctly answering all basic
…nancial literacy questions. Similarly, subjects had trouble with the sophisticated
…nancial literacy questions, although disaggregated data not reported here showed
higher scores for older respondents with large superannuation accounts. Only 35.5%
of respondents correctly answered all sophisticated …nancial literacy questions.
We used the responses to the …nancial competence questions to construct nu-
meracy and …nancial literacy indices (Lusardi and Mitchell, 2009) and used these in
econometric modelling of investment choices. We …rst ran a factor analysis on all
14 questions, with the answers recoded as correct and incorrect. This indicated that
there were three factors which were broadly related to the three groups of questions –
numeracy, basic and sophisticated …nancial literacy. (Q9 was grouped with numeracy
questions by the factor analysis although initially listed with basic literacy.) For each
factor we then used the questions which had the highest loading for this factor and
ran three separate factor analyses on these subsets of questions. The result was three
indices which could be used to describe the numeracy and …nancial literacy of each

8
respondent (Table 4).

2.2 Asset allocation task

We designed the discrete choice experiment recognizing that retirement portfolio de-
cisions involve a substantial fraction of lifetime wealth, where risk preferences shape
choices. We asked subjects to choose an investment option for their entire current
superannuation account and future contributions under a (hypothetical) simpli…ca-
tion of retirement savings arrangements by the government. Such regulatory changes
have been enacted in the recent past and made a plausible context for the survey
(Super System Review, 2010).
The instructions for the experiment, reproduced in Appendix B, present a sim-
pli…ed retirement savings (superannuation) scheme in which the only options for the
savings portfolio are a 100% bank account (‘bank’) with a guaranteed real rate of
return, 50% bank account and 50% growth assets (‘50:50’) and 100% growth assets
(‘growth’) comprised of risky …nancial assets such as equities and property. We asked
respondents to select which option he or she would be most likely to choose, and
which option he or she would be least likely to choose, in a series of settings in which
average real returns remain the same but levels of risk vary. There are 36 settings in
total in the experiment, of which each respondent sees 12. Each setting presents the
annual returns net of in‡ation (2% for the bank account, 3.25% for the 50:50 account,
and 4.5% for the growth account) together with a presentation of risk for the 50:50
and growth accounts (Figure 1). The 12 settings are the product of four risk levels
(common to all respondents) and three of the nine risk presentations, which we now
describe.

9
2.2.1 Risk presentation

Presentations: Many presentations of investment risk to retirement savers are possi-

ble. We utilized nine standard alternatives, drawn from prospectuses of the …nancial
services industries in Australia, Europe and the United States, as well as from related
studies (Vlaev et al. 2009). Table 5 sets out the nine presentations. Presentation
9 gives the same information as presentations 1 through 4 but in graphical form,
together with the sure return on the bank account and Figure 1 illustrates presen-
tation 9 using the highest risk level. Presentations 1 through 4 and presentation 9
convey risk through the cumulative distribution function of returns each year. Pre-
sentations 5 through 8 convey risk through the frequency of returns above or below
simple reference points or benchmarks.
We arranged the nine investment risk presentations into four groups (denoted
‘presentation groups’), each made up of three risk presentations with the visual range
graph (F9) common to each group. For one group of respondents, A, the presentations
are 1, 2 and 9; for group B, 3, 4 and 9; C, 5, 6 and 9; and for group D, 7, 8 and 9.
Each presentation group included one format that emphasized the risk of losses (2,4,6,
and 8) and we call these ‘loss frames’, and one format that emphasized gains (1,3,5
and 7), ‘gain frames’. We also distinguish ‘graphical’ (F9) from ‘textual’ (F1-F8)
presentations. In total we collected 14388 selections of most-preferred and least-
preferred superannuation account choices.
Investment risk levels: The standard deviation of returns to the growth asset
enters the choice sets at four levels ranging from 12% to 28% p.a. Every subject
made choices across all four risk levels in each of three of the nine presentations.
Appendix C gives details of the underlying densities for each of the three alternative
investment accounts at each level of risk and Table 6 reports the simulated values

10
that entered the risk presentations.3
Using this experimental design provided nearly 14400 rankings of the three ac-
counts: around 3600 per format for F1-F8, and nearly 1200 for F9. In the next
section we report the e¤ects of risk level, presentation, …nancial competence and
demographics on these choices.

3 Results

We begin by discussing the aggregated decisions of subjects, then set out the econo-
metric model we estimated using the ‘most-preferred’ accounts and demographics
data. We present graphs showing the impact of risk presentation changes on the
account choices of subjects with a range of …nancial competence scores.

3.1 Aggregate Responses

Most preferred account choices: The 50:50 account was chosen most often, with
growth second, and bank, third (Table 7). Respondents were sensitive to risk pre-
sentation: the relative popularity of bank and growth accounts varied considerably
across the nine formats, particularly when the presentation format emphasized losses.4
For example, presenting risk as the chance of large negative returns (F4) encouraged
respondents to choose the bank account signi…cantly more often, and the growth ac-
count signi…cantly less. However choices did not always favor the bank account in
all of the loss frames, in fact, the reverse occurred in one case - compare choices for
the presentation, ‘number of years in 20 with returns below zero’(F6) and ‘number
3
By holding the expected return constant and allowing the risk of the 50:50 and growth accounts
to vary, we created mean-preserving spreads of the returns distributions.
4
We report test results with caution since they assume independence across the populations
viewing each presentation, whereas each respondent saw three of nine.

11
of years with returns below the bank account’ (F8). As well as emphasizing losses
or gains, the choice of textual or graphical presentation also mattered: we observe a
signi…cantly lower proportion of choices given to the bank account when respondents
viewed graphs of ranges of returns (F9), even though identical information about risk
and return was conveyed via text in the range presentations, F1 and F2.
Least preferred account choices: Respondents indicated they were least likely to
choose the bank account in all but three presentations (F1,F2,F4). We reject the
null of equal ‘least preferred’ choice percentages across presentations for all three
accounts, but variation was concentrated around presentation F4, which emphasized
the chance of large negative returns, and F5, which stated the expected number of
years out of 20 when the rate of return will exceed zero.
In the econometric analysis below, we concentrate on explaining ‘most preferred’
choices using investment risk, risk presentation, …nancial competence and demograph-
ics. This allows us to isolate the impact of descriptions of investment risk on stated
preferences for retirement investments. In the experiment, respondents ranked the
three retirement accounts, making six possible rankings, however a full multinomial
logit analysis of all six rankings with the complex model outlined in the next sec-
tion is unlikely to result in reliable estimation. Instead we work with one dimension
of choices. For an analysis of the complete preference rankings, see Bateman et al.
(2010).

3.2 Model

We assume individuals maximize a general linear random utility function in choosing

their most preferred retirement savings (superannuation) accounts. The choice model
is a conditional mixed logit, as described in Revelt and Train (1998), where the mix-

12
ing is due to random individual error components (random intercepts) rather than
random coe¢ cients on attributes. In our model, the alternative-speci…c intercept
varies with investment risk presentation, and observed and unobserved respondent
characteristics. Choices also depend on taste for investment risk adjusted by inter-
actions between the variable risk level and each respondent’s …nancial competence
factor scores.
We write the nth individual’s random utility (Unktj ) from investment option (alter-
native) j (j = 1; ::; 3); for risk presentation k (k = 1; :::; 9), at risk level t (t = 1; :::; 4)
as
0 0 0
Unjkt = j zn + jk ln + j xjt + njkt (1)

where j is a (1 13) vector of alternative-speci…c coe¢ cients on a constant and ob-

served respondent characteristics including age and age squared, reported retirement
savings (superannuation) accumulation, six indicator variables for equity market ex-
pectations, numeracy and …nancial literacy factor scores (zn ); jk is a (1 4) vector
of presentation and alternative-speci…c coe¢ cients on a constant and the interactions
between the presentation risk indicator and three numeracy and …nancial literacy
scores of individual n (ln ); j is a (1 4) vector of alternative-speci…c coe¢ cients on
the risk level of investment alternative j and interactions between the risk of alterna-
tive j and three numeracy and …nancial literacy scores for individual n; (xnjt ); and

njkt is a random error comprised of two components such that njkt = nj + "njkt :
The …rst error component is assumed normally distributed and captures unobserved
heterogeneity among individual respondents that may create serial correlation in the
0
errors. These alternative-speci…c random e¤ects n = ( n1 ; n2 ) are multivariate
normally distributed n ~N (0; ), with =diag(! 21 ; ! 22 ). (The error component of
the third alternative is n3 = ( n1 + n2 ).) In other words, the random e¤ects,

13
 nj ; are draws from a scaled standard normal distribution, where the unit standard
deviation is scaled by the alternative-speci…c parameter ! j : The second error com-
2
ponent "njkt is assumed iid extreme value with zero mean and variance =6 and is
independent of nj :

Given the assumption regarding the error term "njkt , the choice probability for
alternative j can be expressed as

Pnjkt ( n) = Pr(Unjkt > Unikt j n; 8j 6= i)

0z + 0 l + 0x +
e j n jk n j jt nj
= P 0z + 0 l + 0x + ;
i e i n ik n i it ni

and the unconditional choice probability is given by the integral over all possible
values of n

Z 0z + 0 l + 0x +
j n jk n
e j jt nj
Pnjkt = P 0z + 0 l + 0x + f ( )d :
i e i n ik n i it ni

We augment standard maximum likelihood methods for the conditional logit com-
ponent with Gauss-Hermite quadrature for the (normal) integral over the random
e¤ects. Estimation of the parameters of this probability is implemented in Latent
GOLD software (Magidson and Vermunt, 2005). We restricted the set of covariates
to those found to be relevant after pre-testing more general models.

3.3 Estimated choice probabilities

Estimated odds ratios from the maximum likelihood estimation of the model in equa-
tion (1) and standard errors are reported in Table 8. The odds ratios show the change
in the odds of choosing the 50:50 or growth account over the bank account, or the
growth account over the 50:50 account, when the relevant covariate increases by one.

14
For example, the odds ratio 1.023 in the second cell of column one indicates that
when age rises by one, the probability of choosing the 50:50 account over the bank
account increases by 2.3%. For risk and interactions between risk and …nancial com-
petence index scores, Table 8 reports the change in odds associated with an 8% p.a.
increase in growth asset volatility, which implies at 4% p.a. increase in 50:50 account
volatility. Because of interaction terms, a precise interpretation of the marginal ef-
fects of covariates depends on the levels of all covariates - we do this for the median
respondent below - but we can observe that increasing age, retirement wealth, and
stock market optimism all have positive and statistically signi…cant e¤ects on the
odds of selecting the 50:50 and growth accounts over the bank account. Optimistic
expectations of the future path of stock prices and improvements in basic and so-
phisticated …nancial literacy increase the odds of choosing the growth account over
the 50:50 account, whereas improvements in numeracy reduce those odds. Rising
growth account volatility reduces the odds of choosing the 50:50 and growth accounts
over the bank account. Risk level, risk presentations and their interactions with …-
nancial competence are also signi…cant and we turn now to consider these e¤ects on
retirement savings portfolio decisions in detail.

3.3.1 Investment risk, risk presentation and …nancial competence

Our benchmark (median) respondent is 35-44 years of age and has a retirement savings
account balance of $20-39.9K, considers the prospect of another large stock market
crash within the next …ve years to be as likely as not (a ‘toss-up’) and has numeracy
and basic and sophisticated …nancial literacy scores above the (zero) mean at 0.466,
0.227 and 0.272, respectively. Along with all other respondents, this individual eval-
uated the graphical risk presentation (F9). If the volatility of the growth account
is set at 20% p.a., a respondent with this median pro…le chooses the bank account

15
with a 1.9% probability, the 50:50 account with a 76.9% probability and the growth
account with a probability of 21.2%. Figures 2, 3 and 4 show an array of graphs
that capture the e¤ects of risk level and risk presentation on choice probabilities at
varying levels of …nancial competence and we turn to those results now. Each graph
shows the probabilities of choosing the bank, 50:50 and growth accounts as either
numeracy (Figure 2), basic (Figure 3) or sophisticated (Figure 4) …nancial literacy
index score moves from the 1st to the 99th percentile. We computed standard errors
for each choice probability using the delta method and these are summarized in the
table notes.
Variations in Numeracy: Each panel in Figure 2 shows the predicted probability
that the median respondent chooses the bank account (black), 50:50 account (pale
gray) and the growth account (dark gray) as the respondent’s numeracy score in-
creases from the lowest to the highest percentile in the sample. Numeracy scores
depend on correct answers to questions testing fractions, percentages, division, and
multiplication which are general skills with no speci…c relationship with retirement
savings decisions as opposed to other …nancial decisions. Each row of Figure 2 corre-
sponds to one of the nine risk presentations described in Table 5 (F1 to F9) and in
each row, each of the four panels corresponds to the risk level for the growth account
indicated near the head of the table (12% to 28%). (Risk for the 50:50 account is
always half of the growth account volatility by construction.)
The median respondent is risk averse and adjusts choices in response to changes
in volatility; keeping numeracy quantile constant and moving along each row, we see
that the probability of choosing the riskier accounts decrease and the probability of
choosing the bank account increases as volatility rises from 12% p.a. to 28% p.a.
However changes in numeracy have some surprising e¤ects. At very low levels of
numeracy, the model predicts a high probability of choosing the growth account that

16
declines in favor of the 50:50 account as numeracy improves. This pattern can be
seen for all risk presentations except F3 and F4, suggesting that very innumerate re-
spondents may tend to choose concentrated portfolios in preference to diversi…cation.
Compared with the changes in choice probabilities resulting from increasing volatil-
ity, the changes due to switching risk presentation are very large. We see much greater
variation moving down the columns than along the rows of the array. Several general
patterns emerge. First, the graphical format (F9) is associated with the highest prob-
ability of 50:50 and growth account allocations and the least probability of choosing
the bank account. Choices are riskier when respondents view the graph than under
any of the textual presentations, including F3, which emphasized upper tail returns
probabilities. Secondly, the presentations emphasizing losses generally induce higher
probabilities of choosing the bank account and less likelihood of choosing the growth
account - this more conservative pattern holds for F2, F4 and F8, but not for F6.
The pair F5 and F6 present risk as the expected frequency (number years out of 20)
when returns will be above (F5) and below (F6) zero. Contrary to other ‘loss’frames,
choices show more risk tolerance under F5 (a ‘gain’frame) than F6 (a ‘loss’frame).
Thirdly, the largest variation in choice probabilities occurs when respondents see the
upper (F3) and then the lower (F4) tail values at risk, which indicates that respon-
dents do not infer that a high probability of losses also implies a high probability of
gains or vice versa. On the other hand, the smallest changes in choice probabilities
occur when respondents are presented with information about both upper and lower
tails (F1 and F2). Fourthly, the changes in choice probabilities due to presentation
changes are larger as numeracy declines.
Variations in Basic Financial Literacy: Basic …nancial literacy scores are higher
for respondents who correctly answer questions on key …nancial concepts such as
compounding, in‡ation, and the time value of money. In Figure 3 we see predicted

17
changes in choice probabilities as basic literacy improves for each risk presentation
and volatility level. Many of the patterns are consistent with Figure 2 but several
important di¤erences emerge.
First, unlike the innumerate, respondents with poor basic …nancial literacy choose
the bank and 50:50 accounts rather than the growth account, and demonstrate an
increasing probability of choosing the growth and 50:50 accounts in preference to
bank as basic literacy improves. With reference to the parameter estimates, the
main source of increasing preference for riskier portfolios as basic …nancial literacy
improves is shifting intercepts in the individual random utility model, that is, through
the parameters j;i which are positive, implying odds ratios greater than one (Table
8). However, improved basic literacy shifts slope coe¢ cients in the opposite direction
via the interactions with risk j;i :which are negative. In other words, respondents
who score well in the key components of basic …nancial literacy (time value of money,
compounding and in‡ation) are more sensitive to changes in risk than respondents
with lower basic …nancial literacy. A close examination along the rows of Figure
3 shows that choice probabilities change more at the right hand side than at the
left hand side of each graph. In fact, respondents with basic literacy scores two
or more standard deviations below the mean become largely insensitive to changing
risk information, while at the same time choosing more conservative accounts than
their more literate peers. Consequently, individuals with poor basic …nancial literacy
are unlikely to switch to higher yielding portfolios even if investment risk declines
signi…cantly.
Variations in Sophisticated …nancial literacy: Figure 4 shows the same analysis
holding numeracy and basic …nancial literacy constant at median scores and allowing
sophisticated …nancial literacy to vary. Improvements in sophisticated …nancial liter-
acy - knowledge of diversi…cation and the risk and return features of …nancial assets

18
- is associated with a greater probability of choosing growth and diversi…ed accounts.
However, unlike the previous two measures of …nancial competence, risk sensitivity
and preference for diversi…cation is relatively constant across di¤erent sophisticated
…nancial literacy levels. We continue to observe large presentation e¤ects consistent
with the patterns in Figures 2 and 3, even at high levels of …nancial knowledge.
To summarize, graphs of returns ranges induce respondents to choose riskier retire-
ment savings accounts compared with textual descriptions of risk, even when exactly
the same underlying returns volatility information is conveyed. Further, emphasizing
one tail of the returns distribution over another, or one side of a benchmark outcome,
induces greater variation in account choices compared with returns range information,
either graphical or textual. Presentation groups that emphasize the frequency with
which returns exceed or fail a benchmark level also show large changes in account
choices as the presentation emphasizes either losses or gains. This is surprising, since
the statement ‘On average, negative returns occur x years in every 20’ necessarily
implies that ‘On average, positive returns occur (20-x) years in every 20’. Even more
puzzling is that presentations using the zero return benchmark, induce more conser-
vative choices when describing gains than losses, whereas the presentations using the
bank account benchmark induce more conservative choices when emphasizing losses.
Finally, poor numeracy is linked with a low preference for the diversi…ed account and
poor basic …nancial literacy indicates unresponsiveness to increasing risk.

3.3.2 Age, retirement accumulation, and stock market expectations

In Figure 5 we compare the in‡uence of other covariates on choice probabilities. In

these charts, we …x the presentation at F9, since all respondents viewed this graphi-
cal risk presentation, and we …x numeracy, basic and sophisticated …nancial literacy
scores at median levels. Each panel indicates the predicted probability that a re-

19
spondent chooses each of the three accounts as a function of age, current retirement
accumulation, and expectations that the stock market will su¤er a severe crash within
the next …ve years, for varying levels of risk. Increasing age leads to a stronger prefer-
ence for 50:50 account at the expense of both bank and growth accounts. Retirement
account balance increases do not a¤ect choice probabilities greatly for the majority
of respondents. It is not until account balances become very high (towards $1 million
or more) that preferences for the growth account increase sharply. A greater amount
of choice variation is due to optimistic or pessimistic views on the stock market. Re-
spondents who rated the probability of another large stock market crash in the next
…ve years as ‘nearly impossible’or ‘very unlikely’have a higher than 50% probability
of selecting the growth account compared with about 25% for those who thought
the chance of another crash was as likely as not (‘a toss-up’). More pessimistic or
uncertain respondents (who chose ‘don’t know’) exhibited much more conservatism.
The changes in choice probabilities due to these covariates can be large compared
with the impact of changes in investment risk itself. For example, moving from
median wealth to the $500K-$1 million category increases the likelihood of choosing
the growth account much more than decreasing growth account risk by 8 percentage
points. Much larger variations in probabilities are also caused by changes in stock
market expectations than by changes to investment risk. The signi…cance of these
pre-existing expectations on investment choice is evidence that people mix the risk
information o¤ered in the survey (and disclosed product prospectuses) with their own
priors and uncertainty.

3.3.3 Pro…les of respondents by investment account choice

Finally, we pro…le the respondent most likely to select the growth or bank accounts.
Setting aside the e¤ects of risk presentation, the probability of choosing the growth

20
account is maximized at 94.4% for a respondent of any age but of maximum wealth,
who considers the prospect of another stock market crash within the next …ve years
as ‘very unlikely’, who has relatively poor numeracy, but high sophisticated and basic
…nancial literacy5 and who is making choices at the lowest volatility level for the risky
asset. This respondent has a 5.5% chance of selecting the 50:50 account and his or
her probability of choosing the bank account approaches zero. The respondent most
likely to choose the bank account is among the 18-24 years age group, the lowest
wealth level, is uncertain about the probability of a future stock market crash, and
has relatively high numeracy skills but low basic and sophisticated …nancial literacy.
For this pro…le, the probability of choosing the bank account is 93.2%, with a 6%
probability of choosing the 50:50 account and a 0.9% probability of choosing the
growth account.

4 Conclusion

Retirement provision systems around the world increasingly require ordinary retire-
ment fund members to evaluate investment risk. However the evaluation process
is complicated by other factors: …rst, a consumer’s …nancial competence a¤ects his
or her ability to make an evaluation of risk; and second, the details of information
presentation (Saez, 2009) and frame (Brown et al., 2008) can in‡uence outcomes in
unanticipated ways.
Here we address several questions about the use of risk information in retirement
savings decisions using a discrete choice experiment where subjects choose between
a bank account, a growth account and a 50:50 account. Experimental subjects are
5
We …nd 87 respondents who score below the median for numeracy but above the median for
other …nancial literacy measures, and 150 respondents who score above median for numeracy and
below median for the other …nancial literacy scores.

21
retirement savers from Australia, who participate in the publicly-mandated, privately-
managed superannuation system. The majority of Australian workers are members of
de…ned contribution funds and are required to make decisions about the investment
portfolio for their contributions. Consequently, communicating information about the
risk of investment returns is crucial to the e¢ ciency of the retirement savings system
and regulators are deliberating the merits of alternative formats. We choose nine
standard formats for risk information (risk presentations), vary the level of underlying
risk, and analyze the responses of retirement savers to both risk per se, and to the
presentation format. Using detailed questions on demographics, existing retirement
accumulations, stock market expectations, numeracy and …nancial literacy, we can
estimate the conditional value of risk information and presentation format in this
simpli…ed investment task.
Consistent with expected utility theory, we …nd that choice probabilities for the
risky accounts are signi…cantly decreased by increasing investment risk; in other
words, most retirement savers do pay attention to risk information and exhibit risk
aversion. However, their responses are a¤ected by …nancial competence: people who
have a good grasp of such …nancial basics as compounding and in‡ation are more risk
sensitive than those whose understanding is weak. In fact, as knowledge in this area
declines, risk sensitivity also decreases and approaches zero for people of very poor
understanding.
Holding underlying risk and return at constant levels, the discrete choice exper-
iment shows that changing from one standard risk presentation to another changes
choice probabilities, and in some cases, markedly. Our aim here was to test risk pre-
sentations designed for mass communication that were common in prospectuses and
disclosure documentation, so we presented risk as returns ranges, both textual and
graphical, probabilities of tail events, and frequencies of exceeding or failing to meet

22
benchmarks. The graphical format is associated with noticeably higher probabili-
ties of choosing 50:50 or growth accounts, compared with textual presentations that
convey identical information. Presentations that emphasized losses resulted in more
conservative account choices, in all but one format. This tendency to switch to safety
was weaker for presentations that described ranges of returns, and dramatically larger
when presentations gave information about tail events. These results indicate that
respondents do not infer that upside risk implies downside risk and vice versa. At
times, respondents were also confused by benchmark measures. Presentation e¤ects
remain strong for all levels of …nancial competence but innumerate respondents were
especially susceptible to presentation.
Other pre-existing characteristics also a¤ect investment choices. We …nd that age,
retirement accumulation and expectations of another large stock market crash in the
next few years also predict investment choice. Older respondents tend to favor a
diversi…ed 50:50 investment. Wealthier respondents move into the riskiest options,
as do people with optimistic expectations of stock prices. Increasing uncertainty (or
ignorance) about the future path of stock prices generates conservative choices. Re-
spondents with better …nancial literacy skills, both basic and sophisticated, choose
riskier investments than people with weaker skills. Numeracy skills present a more
puzzling outcome: innumerate respondents steer away from the 50:50 account, pre-
ferring the bank or growth extremes. Compared with the impact of information on
the riskiness of returns to di¤erent investment options, these demographic features,
expectations and competencies are at least as important, and sometimes, much more
important, for predicting investment choices.
We conclude that retirement fund providers and regulators cannot assume that
their members digest risk information provided in product disclosures independently
of presentation format, background knowledge, expectations and skills. Fund mem-

23
bers read investment information in a personal context, interpret textual and graph-
ical information di¤erently and integrate provider information with their own pre-
existing opinions. For some people, the provided information is di¢ cult to interpret,
incomplete, or relatively unimportant, so they are guided by other factors almost
entirely. For the majority, the risk information is signi…cant and in‡uential, but large
changes in underlying risk do not change choice probabilities as much as changes in
the way that risk is communicated.
Our …ndings indicate that as governments divest themselves of the responsibility
to …nance and pay public pensions, they need to pay attention to …nancial product
disclosure requirements and information provision practices by industry. As well,
initiatives to improve …nancial competence need to provide retirement savers with
the skills to understand the information provided to guide their investment decisions.

24
Appendix A. Numeracy and Financial Literacy Ques-
tions
Numeracy skills
Q1: In a sale, a shop is selling all items at half price. Before the sale, a sofa costs
$300. How much will it cost in the sale? (Answers: $150; $300; $600; Do not
know; Refuse to answer.)

Q2: If the chance of getting a disease is 10 per cent, how many people out of 1,000
would be expected to get the disease? (Answers: 10; 100; 1000; Do not know;
Refuse to answer.)

Q3: A second hand car dealer is selling a car for $6,000. This is two-thirds of what
it cost new. How much did the car cost new? (Answers: $4,000; $6,600; $9,000;
Do not know; Refuse to answer.)

Q4: If 5 independent, unrelated people all have the winning numbers in the lottery
and the prize is $2 million, how much will each of them get? (Answers: $40,000;
$400,000; $500,000; Do not know; Refuse to answer.)

Q5: If there is a 1 in 10 chance of getting a disease, how many people out of 1,000
would be expected to get the disease? (Answers: 10; 100; 1000; Do not know;
Refuse to answer.)

Basic …nancial literacy

Q6: Numeracy. Suppose you had $100 in a savings account and the interest rate
was 2% per year. After 5 years, how much do you think you would have in the
account if you left the money to grow? (Answers: More than $102; Exactly
$102; Less than $102; Do not know; Refuse to answer.)

Q7: In‡ation. Imagine that the interest rate on your savings account was 1% per
year and in‡ation was 2% per year. After 1 year, how much would you be able
to buy with the money in this account? (Answers: More than today; Exactly
the same; Less than today; Do not know; Refuse to answer.)

Q8: Time value of money. Assume a friend inherits $10,000 today and his sibling
inherits $10,000 three years from now. In three years, who is richer because of
the inheritance? (Answers: My friend; His sibling; They are equally rich; Do
not know; Refuse to answer.)

Q9: Money Illusion. Suppose that in the year 2020, your income has doubled and
prices of all goods have doubled too. In 2020, how much will you be able to buy

25
 with your income? (Answers: More than today; Exactly the same; Less than
today; Do not know; Refuse to answer.)
Q10: Compound interest. Suppose you had $100 in a savings account and the inter-
est rate is 20% per year and you never withdraw money or interest payments.
After 5 years, how much would you have on this account in total? (Answers:
More than $200; Exactly $200; Less than $200; Do not know; Refuse to answer.)

Sophisticated …nancial literacy (Understanding bonds and stocks)

Q11: Risky assets. Is the following statement true or false? Shares are normally
riskier than bonds. (Answers: True; False; Do not know; Refuse to answer.)
Q12: Long period returns. Considering a long time period (for example 10 or
20 years), which asset normally gives the highest return? (Answers: Bonds;
Savings accounts; Shares; Do not know; Refuse to answer.
Q13: Volatility. Normally, which asset displays the highest ‡uctuations over time?
(Answers: Bonds; Savings accounts; Shares; Do not know; Refuse to answer.)
Q14: Risk diversi…cation. When an investor spreads his money among di¤erent
assets, does the risk of losing money? (Answers: Increase; Decrease; Stay the
same; Do not know; Refuse to answer.)

Preliminary Superannuation questions

P1: Which fund manages your main superannuation account in Australia? (Re-
sponses: Please specify name of fund; Don’t know.)
P2: Which of the following ranges best describes the total amount you currently
have in all your superannuation accounts in Australia? (Responses: 13 ranges
from ‘Under $10,000 to $5,000,000 or over’.)
P3: On a scale of 1 to 7, where 1 means very low and 7 means very high, how would
you assess your understanding of …nance?
P4: How much of your …nancial education was devoted to …nancial education, such
as commerce, business studies, …nance or economics? (Responses: A lot; Some;
A little; Hardly at all.)
P5: Did any of the …rms you have worked for (including your current employer) o¤er
…nancial education programs, for example retirement seminars? (Responses:
Yes; No; Not applicable.)
P6: Have you paid for professional …nancial advice about your superannuation over
the past twelve months? (Responses: Yes; No.)

26
Stock market recovery questions
In the global …nancial crisis that began in late 2007 Australian shares lost about half
their value before they began to recover. Since then, they have recovered about half
the value they lost and are worth about 75% of what they were at the market’s high
in September 2007.

S1: How likely is it that Australian share prices will su¤er another similar sized loss
in the next 5 years? (Answers: Nearly impossible (Chance of this happening is
1 in 100 or less); Very unlikely (Chance of this happening is higher than 1 in
100 but less than 1 in 10); Unlikely (Chances of this happening are between 1
in 10 and 1 in 2); Toss-up (Chance is about 1 in 2); Likely (Chance is greater
than 1 in 2); Don’t know; Refuse to answer.)

S2: How likely is it that Australian share prices will su¤er another similar sized loss
in the next 25 years? (Answers: Nearly impossible (Chance of this happening
is 1 in 100 or less); Very unlikely (Chance of this happening is higher than 1 in
100 but less than 1 in 10); Unlikely (Chances of this happening are between 1
in 10 and 1 in 2); Toss-up (Chance is about 1 in 2); Likely (Chance is greater
than 1 in 2); Don’t know; Refuse to answer.)

S3: Since the crisis Australian share prices have recovered about half the value they
lost. How long do you think it will take for them to fully recover? (Answers:
Within 12 months; Within 2 years; Within 5 years; Within 10 years; Don’t
know; Refuse to answer.)

27
Appendix B. Instructions to survey subjects
The Australian Government is concerned about the complexity of superannuation
arrangements and is looking for ways of simplifying superannuation investment choices.
One possibility is to o¤er only three investment options for all superannuation ac-
counts. Each investment option has a di¤erent expected rate of return (the average
rate at which your investment will grow each year), and a di¤erent amount of invest-
ment risk (year to year UPSIDE and DOWNSIDE variation in the return to your
investment).
The options are
Option A: All (100%) of your superannuation account is invested in a guaranteed
bank deposit with a …xed rate of interest paid each year.
Option B: Your superannuation account will be divided half and half (50%/50%)
between the bank account and growth assets. You can anticipate that savings in this
option will grow faster than the bank deposit (Option A), but will grow more slowly
and be less risky than only choosing growth assets (Option C).
Option C: All (100%) of your superannuation account is invested in assets like
shares and property. On average, you can anticipate that savings in this option will
grow at a faster rate than in the bank deposit (Option A) but without a guarantee.
There is some risk that your account will grow faster or slower than average if you
choose this option.
We are going to show you 12 sets of these options for investing your superannu-
ation. Each set includes 3 investment options like the ones described above. Each
investment option has a average rate of return and investment risk. The average
rates of return stay the same in each of the twelve sets; only the risk will change.
Remember that more risk of high returns also means more risk of low returns.
What we want you to do is simple. There are two questions to ask about each set
of options::
1. If these superannuation options were available for you to invest your money
today, which one of the three would you be most likely to choose?
2. If these superannuation options were available for you to invest your money
today, which of the three would you be least likely to choose?
Your choices will inform government and industry about better ways to simplify
superannuation arrangements.

28
Appendix C. Risk and return to investment ac-
counts
In our underlying experimental set up, risky asset returns, r~; are identically and
independently lognormally distributed
2
log (~
r) s N r~; r~ : (2)

The investor can allocate current and future retirement savings wealth to one of
three accounts (funds):
1: A bank deposit, with risk free return rf :
2: A growth account, with return r~ , E(y) = exp( r~+ 12 2r~); var(~
r) = exp (2 r~ + 2r~) [exp( 2r~) 1] :
3: A 50:50 account, made up of equal proportions of risk-free and growth assets
with return r = (rf + r~)=2; E(r) = 0:5 [rf + E(~ r)] ; var(r) = 0:25var(~
r):
The experiment uses four increasingly risky con…gurations for the lognormal risky
asset, denoted r~t , (t = 1; 2; 3; 4), and let rt = (rf + r~t ) =2. We derived numerical
values (quantiles or frequencies) for risk presentations from simulated cumulative
densities of the corresponding lognormal random variable r~t . The number of random
draws was D = 100; 000 and the simulated probability of a return less than a …xed
value r was Pr(~ rt r)

1 X
D
Pr(~
rt;d r) = 1(~
rt;d r) (3)
D d=1

where 1( ) is the indicator function, returning the value 1 when r~t;d r and zero
otherwise. The draws were generated as r~t;d = exp ( r~t + r~t ed ) where ed is a random
draw from a standard normal density. We chose r~ and 2r~ to match the return and
risk parameters in Table 6.

29
Appendix D. Variable de…nitions for mixed logit
choice model
Age – mean centred mid points of response ranges 18-24 years; 25-34 years; 35-44
years; 45-54 years; 55-64 years; 65-74 years; and 75 years and over. Mid point
for respondents age 75 and over was set at 79.5.

Retirement savings accumulation –mean centred and divided by 1000 mid points of
response ranges to the question ‘Which of the following ranges best describes
the total amount you currently have in all your superannuation accounts in Aus-
tralia?’; Ranges where: Under $10,000; $10,000 - $19,999; $20,000 - $39,999;
$40,000 - $59,999; $60,000 - $79,999; $80,000 - $99,999; $100,000 - $149,999;
$150,000 - $199,999; $200,000 - $299,999; $300,000 - $499,999; $500,000 -
$999,999; $1,000,000 - $4,999,999; and $5,000,000 or over. No respondents
selected the highest category.

Equity crash ‘Nearly impossible’–indicator variable for share market recovery ques-
tion S1 taking value 1 if response was ‘Nearly impossible’, 0 otherwise

Equity crash ‘Very unlikely’–indicator variable for share market recovery question
S1 taking value 1 if response was ‘Very unlikely’, 0 otherwise

Equity crash ‘Unlikely’ – indicator variable for share market recovery question S1
taking value 1 if response was ‘Unlikely’, 0 otherwise

Equity crash ‘Toss up’ – indicator variable for share market recovery question S1
taking value 1 if response was ‘Toss up’, 0 otherwise

Equity crash ‘Likely’ – indicator variable for share market recovery question S1
taking value 1 if response was ‘Likely’, 0 otherwise

Equity crash ‘Don’t know/refuse to answer’ – indicator variable for share market
recovery question S1 taking value 1 if response was ‘Don’t know’or ‘Refuse to
answer’, 0 otherwise; this category served as base category in the estimation
with e¤ects-coding of all equity crash indicator variables.

Numeracy index score –mean centred factor weighted responses to answers on nu-
meracy skills questions; coded 1 for correct and 0 for incorrect/‘don’t know’/‘refuse
to answer’.

Basic …nancial literacy index score – mean centred factor weighted responses to
answers on basic …nancial literacy skills questions; coded 1 for correct and 0 for
incorrect/‘don’t know’/‘refuse to answer’.

30
Sophisticated …nancial literacy skills index score – mean centred factor weighted
responses to answers on sophisticated …nancial literacy skills questions; coded
1 for correct and 0 for incorrect/‘don’t know’/‘refuse to answer’.

F1 – indicator variable taking value 1 if risk presentation format showed returns

between 5 and 95 percentiles.

F2 – indicator variable taking value 1 if risk presentation format showed returns

outside 5 and 95 percentiles.

F3 – indicator variable taking value 1 if risk presentation format showed returns

above 95 percentile.

F4 – indicator variable taking value 1 if risk presentation format showed returns

below 5 percentile.

F5 – indicator variable taking value 1 if risk presentation format showed expected

years in 20 with returns above 0.

F6 – indicator variable taking value 1 if risk presentation format showed expected

years in 20 with returns below 0.

F7 – indicator variable taking value 1 if risk presentation format showed expected

years in 20 with returns above bank.

F8 – indicator variable taking value 1 if risk presentation format showed expected

years in 20 with returns below bank.

F9 –indicator variable taking value 1 if risk presentation format showed range graph
of 5 and 95 percentiles; this category served as base category in the estimation
with e¤ects-coding of all frame indicator variables.

Investment risk –mean-centered risk levels of alternatives.

31
References
Agnew, J., Szykman, L., 2005. Asset allocation and information overload: the in‡u-
ence of information display, asset choice and investor experience. Journal of Behav-
ioral Finance 6, 57-90.

Agnew, J., Anderson, L., Gerlach, J., Szykman, L., 2008. Who chooses annuities?
An experimental investigation of the role of gender, framing and defaults. American
Economic Review 98(2), 418-22.

Anagol, S., Gamble, K.J., 2008. Presenting investment results asset by as-
set lowers risk taking. Yale university Working Paper. Available at SSRN:
http://ssrn.com/abstract=1640791.

Andrews, P., 2009. Did life and pensions ‘disclosure’work as expected? Occasional
Paper Series No. 32, UK Financial Services Authority, London, UK.

Australian Bureau of Statistics (ABS) 2008. Employment arrangements, retirement

and superannuation, Australia, ABS 6361.0.

Australian Prudential Regulation Authority (APRA), 2011. Annual Superannuation

Bulletin. June 2010 (released 19 January 2011).

Bateman, H., Ebling, C., Geweke, J., Louviere, J., Satchell, S., Thorp, S., 2010. Eco-
nomic rationality, risk presentation and retirement portfolio choice. CenSoC Working
Paper No. 10-004, University of Technology Sydney.

Benartzi, S., Thaler, R.H., 1999. Risk aversion or myopia? Choices in repeated
gambles and retirement investments. Management Science 45(3), 364-81.

Benartzi, S., Thaler, R.H., 2001. Naïve diversi…cation strategies in de…ned contribu-
tion saving plans. American Economic Review 91(1), 79-97.

Brown, J.R., Liang, N., Weisbenner, S.J., 2007. Individual account investment op-
tions and portfolio choice: Behavioral lessons from 410(k) plans. Journal of Public
Economics 91, 1992-2013.

Brown, J.R., Kling, J.R., Mullainathan, S., Wrobel, M.V., 2008. Why don’t peo-
ple insure late life consumption? A framing explanation of the under-annuitization
puzzle. American Economic Review 98(2, 304-309.

European Commission, 2009a. UCITS Disclosure testing research report, prepared

for the European Commission by IFF Research and YouGov, June 2009.

European Commission, 2009b. Update on Commission work on Packaged Retail In-

vestment Products, December 2009.

32
G20 Financial Inclusion Experts Group, 2010. Principles and report on Innovative
Financial Inclusion from the Access Through Innovation Sub-Group of the G20 Fi-
nancial Inclusion Experts Group, 25 May 2010.

Gerardi, K., Goette, L., Meier, S., 2010. Financial literacy and subprime mortgage
delinquency: evidence from a survey matched to administrative data. Federal Reserve
Bank of Atlanta Working Paper Series No 2010-10, Atlanta, GA.

Goldstein, D.G., Johnson, E.J., Sharpe, W.F., 2008. Choosing outcomes versus choos-
ing products: consumer-focused retirement investment advice. Journal of Consumer
Research 35, 440-456.

Haisley, E., Kaufmann, C., Weber, M., 2010. How much risk can I handle? The role
of experience sampling and graphical displays on one’s risk appetite. Paper presented
at the 1st Boulder Conference on Consumer Financial Decision Making, University
of Colorado, Boulder, June 2010.

Hastings, J., Mitchell, O.S., Chyn, E., 2010. Fees, framing, and …nancial literacy
in the choice of pension manager. Pension Research Council Working Paper PRC
WP2010-09.

Huberman, G., Jiang, W., 2006. O¤ering versus choice in 401(k) plans: equity expo-
sure and number of funds. Journal of Finance 56(2), 763-801.

Hung, A., Heinberg, A., Yoong, J., 2010. Do risk disclosures a¤ect investment choice?
RAND Working Paper Series WR- 788.

Lusardi, A., Mitchell, O.S., 2007. Baby Boomer Retirement Security: The Roles of
Planning, Financial Literacy, and Housing Wealth. Journal of Monetary Economics
54, 205-44.

Lusardi, A., Mitchell, O.S., 2009. How ordinary consumers make complex economic
decisions: …nancial literacy and retirement readiness. NBER Working Paper w15350,
Cambridge MA.

Madrian, B.C. Shea, D.F., 2001. The power of suggestion: inertia in 401(k) partici-
pation and saving behavior. Quarterly Journal of Economics 16(4), 1149-87.

Magidson, J., Vermunt, J.K., 2005. Technical Guide to Latent Gold Software 4.5.
Statistical Innovations.

Minister for Financial Services, 2010. Explanatory statement, select legislative instru-
ment 2010 (No.135), Corporations Amendment Regulations 2010 (No.5), Canberra.

Organization for Economic Cooperation and Development, 2008. Improving …nancial

education and awareness on insurance and private pensions, OECD, Paris.

33
Rabin, M., 1998. Psychology and Economics. Journal of Economic Literature 36,
11-46.

Revelt, D., Train, K., 1998. Mixed logit with repeated choices: Households’choices
of appliance e¢ ciency level, Review of Economics and Statistics, 80:4, 647-657.

Rubaltelli, E., Rubichi, S., Savadori, L., Tadeschi, M., Ferretti, R., 2005. Numerical
information format and investment decisions: implications for the disposition e¤ect
and status quo bias. Journal of Behavioral Finance 6(1), 19-26.

Saez, E., 2009. Details matter: the impact of presentation and information on the
take-up of …nancial incentives for retirement saving, American Economic Journal:
Economic Policy 1:1, 204–228.

Super System Review, 2010. Super System Review Final Report, Part 1, Overview
and Recommendations, Commonwealth of Australia.

van Rooij, M., Kool, C., Prast, H., 2007. Risk-return preferences in the pension
domain: Are people able to choose? Journal of Public Economics 91, 701-722.

van Rooij, M., Lusardi, A., Alessie, R., 2009. Financial literacy and retirement plan-
ning in the Netherlands. DNB Working Paper, November, De Nederlandsche Bank.

Vlaev, I., Chater, N., Stewart, N., 2009. Dimensionality of risk perception: factors af-
fecting consumer understanding and evaluation of …nancial risk. Journal of Behavioral
Finance 10(3), 158-181.

Yoong, J., 2010. Financial illiteracy and stock market participation: evidence from
the RAND American Life Panel. Pension Research Council Working Paper, PRC
WP2010-29.

34
 TABLE 1: SURVEY SAMPLE DEMOGRAPHIC CHARACTERISTICS a

Survey General Survey General

respondent Australian respondent Australian
population (%) population (%) population (%) population (%)
Gender Industry
Male 49.9 50.1 Agriculture, forestry & fishing 0.98 3.17
Female 50.1 49.9 Mining 2.07 1.21
Age (as % of 18‐65 years pop’n) Manufacturing 4.79 10.74
18‐34 years 35.8 37.4 Electricity, gas, water & waste services 1.20 1.01
35‐54 years 43.2 43.6 Construction 5.01 8.00
55‐65 years 21.1 18.9 Wholesale trade 2.18 4.47
Marital status Retail trade 9.80 11.65
Not living with long term partner 42.94 46.72 Accommodation & food services 3.70 6.49
Married or living with long term partner 57.06 53.28 Transport, postal & warehousing 4.79 4.82
Work status Information media & telecommunications 5.45 1.99
Employed full‐time 51.72 40.79 Financial & insurance services 7.19 3.93
Employed part‐time 23.52 18.79 Rental, hiring & real estate services 1.20 1.74
Unemployed 3.44 3.53 Professional, scientific & technical services 6.86 6.79
Not in the labour force 21.31 36.89 Administrative & support services 5.56 3.23
Occupation Public administration & safety 3.70 6.86
Clerical & administrative worker 20.81 15.00 Education & training 11.22 7.87
Community & personal service worker 3.59 8.81 Health care & social assistance 11.55 10.78
Labourer 5.66 10.46 Arts & recreation services 0.98 1.44
Machinery operators & drivers 3.49 6.64 Other services 11.76 3.81
Manager 10.89 13.21
Professional 31.15 19.84
Sales worker 7.63 9.84
Technicians & trades worker 7.41 14.38
Other 9.37 1.82
High School completion Household composition
Year 12 or equivalent 70.49 46.87 Couple family with no children 25.49 25.67
Year 11 or equivalent 9.10 11.08 Couple family with children 37.46 31.20
Year 10 or equivalent 17.13 25.36 One parent family 6.48 10.87
Year 9 or equivalent 2.13 7.74 Other family household 3.44 1.18
Year 8 or below 1.07 7.98 Single person household 13.77 23.38
Did not go to school 0.08 0.96 Group household (i.e. shared) 13.36 7.68
Highest non‐school qualification
Number of people living in household
Postgraduate or equivalent 13.59 6.58 1 10.98 24.36
Graduate Diploma and Graduate Certificate from
8.43 3.64 2 34.67 34.10
University or equivalent
Bachelor Degree or equivalent 30.77 29.33 3 22.95 15.79
Advanced Diploma and Diploma from
20.65 18.01 4 19.92 15.73
University/Vocational College equivalent
Certificate or equivalent 26.55 42.43 5 7.62 6.88
Annual total household gross income (before tax) 6 or more 3.85 3.13
Number of people in family fully/partially
Less than $18,200 pa (i.e. $350 a week) 3.28 4.72
financially supported b
$18,200‐$72,799 pa (i.e.$499‐1,399 a week) 34.33 39.49 None 45.66 50.18
$72,800‐$129,999 pa (i.e. $1,400‐$2,499 a week) 31.64 28.44 1 23.28 17.24
$130,000 pa (i.e. $2,500 a week) or more 16.88 14.93 2 or more 31.06 32.58
Prefer not to answer 13.87 12.42a Net wealth (individual)
Under $10,000 13.93 ‐
$10,000 ‐ $99,999 27.54 18.21
$100,000 ‐ $999,999 35.00 62.44
$1,000,000 or over 6.80 19.35
Prefer not to answer 16.72 ‐
Notes: a Source for population statistics: Australian Bureau of Statistics Census of Population and Housing & Household Wealth and Wealth Distribution, Australia, 2005‐2006. The survey
sample of 1220 is taken from the PureProfile web panel of 600,000 Australians and filtered to ensure that respondents were over 18 and current holders of a superannuation account.
Sampling ensured that genders and age proportions were reasonably close to the Australian population.
 TABLE 2: RETIREMENT SAVING CHARACTERISTICS

Survey Survey
respondent respondent
population (%) population (%)
Knows superannuation fund Workplace financial education programs offered
Yes 80.2 Yes 24.3
No 19.8 No 69.7
Amount in superannuation account(s) NA 6.0
Under $20,000 35.9 Paid for financial advice about superannuation over past 12 months
$20,000 ‐ $79,999 35.0 Yes 11.8
$80,000 ‐ $499,999 26.6 No 88.2
Assessment that the market will suffer the same or an even
$500,000 or over 2.5
greater loss in value sometime in the next 5 years?
Understanding of finance? Nearly impossible (Chance of this happening is 1 in 100 or less) 2.3
Very unlikely (Chance of this happening is higher than 1 in 100 but
1 (very low) 3.7 13.0
less than 1 in 10)
2 5.9 Unlikely (Chances of this happening are between 1 in 10 and 1 in 2) 24.1
3 10.1 Toss‐up (Chance is about 1 in 2) 24.2
4 (about average) 39.1 Likely (Chance is greater than 1 in 2) 16.3
5 23.2 Don't know/ Refuse to answer 19.2
6 13.7 Time for Australian share prices to fully recover
7 (very high) 4.3 Within 12 months 3.8
Education devoted to financial education (e.g ., commerce, Within 2 years 30.2
business studies, finance, economics)?
A lot 9.1 Within 5 years 36.9
Some 26.6 Within 10 years 9.9
A little 25.5 Don’t know/ Refuse to answer 19.2
Hardly at all 38.8
Notes: See notes to Table 1 on survey sample. Table shows responses to survey questions on retirement savings and share market expectations (reproduced in Appendix
A). The entire survey is available at http://survey.confirmit.com/wix/p1250911674.aspx
 TABLE 3: NUMERACY AND FINANCIAL LITERACY RESPONSES

Numeracy questions, per cent correct/incorrect

Q1 Q2 Q3 Q4 Q5
Correct 96.1 95.8 91.6 92.9 91.1
Incorrect a 3.9 4.2 8.4 7.1 8.9
Basic financial literacy, per cent correct/incorrect
Q6 Q7 Q8 Q9 Q10
Numeracy Inflation Time value of Money illusion Compound
money interest
Correct 88.4 78.4 55.2 86.8 72.0
a
Incorrect 11.6 21.6 44.8 13.2 28.0
Sophisticated financial literacy, per cent correct/incorrect
Q11 Q12 Q13 Q14
Risky assets Long period Volatility Risk
returns diversification
Correct 64.3 55.2 76.9 73.4
Incorrect a 35.7 44.8 23.1 26.6
Notes: Table shows proportion of correct responses in aggregate to numeracy and basic and sophisticated
financial literacy questions (Appendix A) a. ‘Incorrect’ includes ‘don’t know’ and ‘refuse to answer’. See notes
to Table 2 for survey sample characteristics.
 TABLE 4: FACTOR LOADINGS ON NUMERACY AND FINANCIAL LITERACY QUESTIONS
Factor 1 Factor 2 Factor 3
Q1: In a sale, a shop is selling all items at half price. Before the sale, 0.648 0.206 ‐0.043
a sofa costs $300. How much will it cost in the sale?
Q2: If the chance of getting a disease is 10 per cent, how many 0.668 ‐0.001 0.073
people out of 1,000 would be expected to get the disease?
Q3: A second hand car dealer is selling a car for $6,000. This is two‐ 0.414 0.372 0.004
thirds of what it cost new. How much did the car cost new?
Q4: If 5 independent, unrelated people all have the winning 0.465 0.384 ‐0.033
numbers in the lottery and the prize is $2 million, how much will
each of them get?
Q5: If there is a 1 in 10 chance of getting a disease, how many 0.497 0.152 0.191
people out of 1,000 would be expected to get the disease?
Q9: Money Illusion. Suppose that in the year 2020, your income has 0.619 ‐0.064 0.118
doubled and prices of all goods have doubled too. In 2020, how
much will you be able to buy with your income?
Q6: Numeracy. Suppose you had $100 in a savings account and the 0.375 0.538 0.082
interest rate was 2% per year. After 5 years, how much do you think
you would have in the account if you left the money to grow?
Q7: Inflation. Imagine that the interest rate on your savings account 0.177 0.542 0.297
was 1% per year and inflation was 2% per year. After 1 year, how
much would you be able to buy with the money in this account?
Q8: Time value of money. Assume a friend inherits $10,000 today ‐0.125 0.671 0.166
and his sibling inherits $10,000 three years from now. In three
years, who is richer because of the inheritance?
Q10: Compound interest. Suppose you had $100 in a savings 0.142 0.648 0.180
account and the interest rate is 20% per year and you never
withdraw money or interest payments. After 5 years, how much
would you have on this account in total?
Q11: Risky assets. Is the following statement true or false? Shares 0.008 ‐0.003 0.739
are normally riskier than bonds.
Q12: Long period returns. Considering a long time period (for ‐0.033 0.242 0.591
example 10 or 20 years), which asset normally gives the highest
return?
Q13: Volatility. Normally, which asset displays the highest 0.192 0.108 0.769
fluctuations over time?
Q14: Risk diversification. When an investor spreads his money 0.155 0.301 0.554
among different assets, does the risk of losing money?
Note: Table shows factor analysis of responses to the 14 numeracy and financial literacy questions where
answers were recoded as correct and incorrect (which included ‘incorrect’, ‘do not know’ and ‘refuse to
answer’). Factor 1 broadly covers the numeracy questions. Factor 2 broadly covers the basic financial literacy
questions while Factor 3 matches the sophisticated financial literacy questions. We then used the questions
which had the highest loading for each factor and ran three separate factor analyses on these subsets of
questions to make three indices. Note that the factor analysis reassigns Q9 on money illusion to numeracy
rather than basic financial literacy skills.
 TABLE 5: ALTERNATIVE PRESENTATIONS FOR INVESTMENT RISK

F1 : There is a 9 in 10 chance of a return between x% and y%.

F2 : There is a 1 in 10 chance of a return outside x% and y%.

F3 : There is a 1 in 20 chance of a return above y%.

F4 : There is a 1 in 20 chance of a return below x%.

F5 : On average, positive returns occur (20‐x) years in every 20.

F6 : On average, negative returns occur x years in every 20.

F7 : On average, returns above the bank account occur (20‐x) years in every 20.

F8 : On average, returns below the bank account occur x years in every 20.

F9 : Three options are shown in the chart below.

 Option A: 100% bank account, the rate of return is always exactly x% (black dot)
 Option B: 50% bank account & 50% growth asset, there is a 9 in 10 chance of a rate
of return within the light blue box
 Option C: 100% growth asset, there is a 9 in 10 chance of a rate of return within the
dark blue box

Notes: Table shows nine alternative formats for investment risk information as inserted into retirement
savings choice sets (see Figure 1). Each respondent answered 12 choice sets (3 out of the 9 presentations x 4
risk levels). The nine investment risk presentations were arranged into four presentation groups then
randomly assigned to one quarter of the sample of respondents: group A, 1, 2 and 9; group B, 3, 4 and 9; group
C, 5, 6 and 9; and group D, 7, 8 and 9. Values of x and y depend on quantities derived from four log normal
growth account returns densities with mean equal to 1.045 and standard deviation taking four values from the
0.12 to 0.28. Table 6 reports all x and y values that entered the risk presentations.
 TABLE 6: INVESTMENT RISK LEVELS AND PRESENTATION PARAMETERS.

Net rates of return Volatility Log‐normal

%, p.a. %, p.a. parameters
Level Bank 50:50 Growth 50:50 Growth μ σ
1 2.0 3.25 4.5 6 12 0.038 0.11
2 2.0 3.25 4.5 8 16 0.032 0.15
3 2.0 3.25 4.5 10 20 0.026 0.19
4 2.0 3.25 4.5 14 28 0.009 0.26
Presentation 1‐4: (x,y) 5‐6: x 7‐8: x
Portfolio 50:50 growth 50:50 growth 50:50 &
growth
Risk level 1 (‐6,14) (‐14,25.5) 6 7 9
Risk level 2 (‐9,17.5) (‐19.5,32.5) 7 8 9.5
Risk level 3 (‐11.5,21) (‐25,40) 8 9 10
Risk level 4 (‐16.5,29) (‐34.5,55.5) 9 10 11.5

Notes: Top panel shows net rates of return entered into retirement account choice sets (see Figure 1), and
four volatility levels of the distributions of growth account returns. Last two columns show log‐normal
parameters use to simulate corresponding growth account returns distributions at each volatility level. Lower
panel shows numerical values entered into choice sets for presentations 1‐8 at four risk levels (see Table 5).
 TABLE 7: AGGREGATE CHOICES BY RISK PRESENTATION

All F1 F2 F3 F4 F5 F6 F7 F8 F9 χ²(8)
Most preferred % % % % % % % % % %
Bank 24 27 31 16 50* 28 22 21 33* 16* 40.45*
50:50 46 51 48 54 35 34* 44 52 41 47 9.19
Growth 30 22 21 30 15* 38 33 27 26 37 17.30*
Least preferred
Bank 49 43 39 57 28* 52 50 49 44 56 14.80*
50:50 9 6 8 8 7 20* 13 10 13 7 19.22*
Growth 42 51 53 36 65* 29* 37 41 42 37 23.50*

Notes: Table reports percentage of best and worst choices allocated to the bank, 50:50 and growth accounts over the entire survey (‘all’) and by 9 risk presentation format
as described in Table 2. The final column reports the test statistic for the chi‐square test of the joint equality of percentages across frames for each row, where the
reference level is the aggregate percentage (‘all’). Individual percentages marked with an asterisk are significantly different (at the 10% level or less) from the aggregate
percentages according to a chi‐square test with 1 d.f.
 TABLE 8: ODDS RATIOS FOR MIXED LOGIT CHOICE MODEL

1199 respondents; 14388 account choices Retirement savings account Retirement savings account
Covariate 50:50/bank Grth/bank Grth/50:50 Covariate 50:50/bank Grth/bank Grth/50:50
constant αj,1 5.599 1.748 0.312 F1 * basic γj1,3 1.067 0.849 0.795
(0.157) (0.087) (0.058) (0.102) (0.068) (0.132)
Age αj,2 1.023 1.017 0.994 F2 * basic γj2,3 1.046 1.007 0.962
(0.003) (0.004) (0.005) (0.108) (0.077) (0.156)
Age^2 αj,3 1.000 1.000 1.000 F3 * basic γj3,3 1.183 1.021 0.863
(0.000) (0.000) (0.000) (0.090) (0.076) (0.117)
Retirement savings accumulation αj,4 1.004 1.004 1.000 F4 * basic γj4,3 0.720 0.382 0.531
(0.000) (0.000) (0.000) (0.086) (0.054) (0.117)
Equity crash ‘Nearly impossible’ αj,5 0.689 1.813 2.631 F5 * basic γj5,3 0.649 0.752 1.159
(0.136) (0.168) (0.414) (0.068) (0.068) (0.140)
Equity crash ‘Very unlikely’ αj,6 2.713 3.969 1.463 F6 * basic γj6,3 0.933 0.990 1.061
(0.187) (0.214) (0.192) (0.084) (0.076) (0.134)
Equity crash ‘Unlikely’ αj,7 1.523 1.307 0.858 F7 * basic γj7,3 1.184 2.719 2.297
(0.107) (0.105) (0.138) (0.111) (0.132) (0.230)
Equity crash ‘Toss up’ αj,8 1.861 0.905 0.486 F8 * basic γj8,3 1.231 1.535 1.247
(0.142) (0.085) (0.097) (0.111) (0.101) (0.173)
Equity crash ‘Likely’ αj,9 0.580 0.427 0.737 F9 * basic γj9,3 1.191 0.965 0.810
(0.066) (0.058) (0.126) (0.049) (0.038) (0.060)
Equity crash ‘Don’t know/refuse to answer’ αj,10 0.325 0.275 0.845
(0.049) (0.049) (0.125)
Numeracy index score αj,11 1.073 0.785 0.731 F1 * sophisticated γj1,4 1.340 1.373 1.024
(0.036) (0.030) (0.049) (0.114) (0.088) (0.150)
Basic financial literacy index score αj,12 2.242 2.390 1.066 F2 * sophisticated γj2,4 1.177 1.617 1.374
(0.072) (0.070) (0.083) (0.113) (0.097) (0.184)
Sophisticated financial literacy index score αj,13 1.451 2.037 1.404 F3 * sophisticated γj3,4 1.486 1.474 0.992
(0.062) (0.072) (0.099) (0.101) (0.098) (0.124)
F4 * sophisticated γj4,4 0.835 0.527 0.631
(0.091) (0.063) (0.127)
F1: returns between 5‐95 percentile γj1,1 1.752 1.924 1.098 F5 * sophisticated γj5,4 0.639 0.840 1.314
(0.116) (0.095) (0.133) (0.066) (0.066) (0.145)
F2: returns outside 5‐95 percentile γj2,1 1.077 1.104 1.025 F6 * sophisticated γj6,4 0.880 1.212 1.376
(0.094) (0.073) (0.134) (0.082) (0.079) (0.152)
F3: returns above 95 percentile γj3,1 6.801 9.464 1.392 F7 * sophisticated γj7,4 1.160 0.929 0.801
(0.226) (0.235) (0.132) (0.108) (0.080) (0.131)
F4: returns below 5 percentile γj4,1 0.085 0.048 0.561 F8 * sophisticated γj8,4 0.771 0.687 0.890
(0.029) (0.017) (0.106) (0.087) (0.068) (0.142)
F5: years in 20 above zero γj5,1 0.281 0.410 1.458 F9 * sophisticated γj9,4 1.015 0.893 0.880
(0.039) (0.047) (0.132) (0.039) (0.040) (0.059)
F6: years in 20 below zero γj6,1 0.810 0.592 0.730
(0.068) (0.056) (0.094)
F7: years in 20 above bank γj7,1 2.604 1.609 0.618 Riska βj,1 0.753 0.965 0.795
(0.140) (0.096) (0.098) (0.040) (0.038) (0.035)
F8: years in 20 below bank γj8,1 0.392 0.349 0.890 Riska * numeracy βj,2 0.980 0.599 0.983
(0.054) (0.044) (0.120) (0.039) (0.041) (0.042)
F9: range graph 5‐95 percentile γj9,1 3.937 7.645 1.942 Riska * basic βj,3 0.853 0.982 0.898
(0.080) (0.095) (0.077) (0.046) (0.055) (0.044)
Riska * sophisticated βj,4 0.991 0.963 0.991
(0.048) (0.045) (0.045)
F1 * numeracy γj1,2 1.113 1.047 0.940
(0.080) (0.068) (0.120)
F2 * numeracy γj2,2 1.071 0.781 0.729
(0.081) (0.060) (0.109)
F3 * numeracy γj3,2 1.385 2.158 1.558
(0.088) (0.088) (0.149)
F4 * numeracy γj4,2 0.779 1.190 1.527
(0.068) (0.071) (0.157)
F5 * numeracy γj5,2 0.779 0.842 1.081
(0.080) (0.076) (0.151)
F6 * numeracy γj6,2 0.713 0.698 0.980
(0.079) (0.069) (0.143)
F7 * numeracy γj7,2 1.296 0.821 0.634 Random effect ωj 4.695 8.009
(0.083) (0.055) (0.093) (0.049) (0.058)
F8 * numeracy γj8,2 0.915 0.722 0.788
(0.067) (0.052) (0.100) Log likelihood (model) ‐9166.6
F9 * numeracy γj9,2 1.181 1.368 1.158
(0.041) (0.039) (0.065)
Notes: Table shows estimated odds ratios for model of retirement savings account choices. Odds ratios show the change in the odds of selecting the 50:50 or growth
accounts over the bank account, or the growth account over the 50:50 account, when the relevant covariate rises by one. For example, a ratio of 1.5 indicates a 50%
increase in the probability of choosing the numerator account over the denominator account. aFor risk and risk interactions, we compute the change in odds associated
with an 8% p.a. increase in growth asset volatility (a 4% increase in 50:50 volatility). Approximate standard errors, computed using the delta method, are in brackets.
 FIGURE 1: ILLUSTRATIVE INVESTMENT CHOICE TASK

Features of Options Option A Option B Option C

50% Bank account &
Option type 100% Bank account 100% Growth assets
50% Growth assets
Average annual rate of return
2% 3.25% 4.5%
(above inflation)
There is a 1 in 20 chance of a There is a 1 in 20 chance of a
Level of investment risk No risk rate of return above rate of return above
14% 25.5%

If these superannuation options above were available for you to invest your money today

1. Which one of the three would you be most likely to choose?

Option A
Option B
Option C
2. Which one of the three would you be least likely to choose?
Option A
Option B
Option C

Note: Figure shows example investment choice task. There are 36 such settings in total, of which each of 1199 respondents sees 12. Each of 1199 survey respondents
answered 12 such tasks where rates of returns stayed constant and risk level (x4) and risk presentation (x3) varied.
 FIGURE 2: Probabilities of account choice as a function of numeracy

Risk 12% Risk 16% Risk 20% Risk 28%

1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F1 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F2 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F3 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F5 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F6 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F7 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F8 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F9 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Notes: Each panel indicates the predicted probability that a respondent chooses the bank account (black), 50:50 account (pale grey)
and the growth account (dark grey) as a function of the quantile of the respondent’s numeracy score (horizontal axis). The panels
in each row correspond to a given risk presentation, indicated at the left end of the row and defined in Table 5. In each row, each
of the four panels corresponds to the risk level indicated near the head of the table. All other respondent characteristics are set to
median values. 90% of the standard errors fell between .0032 and .0502.
 FIGURE 3: Probabilities of account choice as a function of basic financial literary

Risk 12% Risk 16% Risk 20% Risk 28%

1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F1 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0
1.0 1.0 1.0
0.8
0.8 0.8 0.8
0.6 0.6 0.6 0.6
F2 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F3 0.4 0.4 0.4 0.4
0.2 0.2 0.2
0.2
0.0 0.0 0.0
0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8
1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F4 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F5 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0
0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4
F6 0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F7 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F8 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0

0.8 1.0
0.8 0.8 0.8
0.6 0.6 0.6 0.6
F9 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Notes: Each panel indicates the predicted probability that a respondent chooses the bank account (black), 50:50 account (pale grey)
and the growth account (dark grey) as a function of the quantile of the respondent’s basic financial literacy score (horizontal axis).
The panels in each row correspond to a given risk presentation, indicated at the left end of the row and defined in Table 5. In each
row, each of the four panels corresponds to the risk level indicated near the head of the table. All other respondent characteristics
are set to median values. 90% of the standard errors fell between .0028 and .0570.
 FIGURE 4: Probabilities of account choice as a function of sophisticated financial literacy

Risk 12% Risk 16% Risk 20% Risk 28%

1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F1 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F2 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F3 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F4 0.4
0.4 0.4 0.4
0.2
0.0 0.2 0.2 0.2
0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F5 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F6 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F7 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
1.0 1.0 1.0 1.0
0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F8 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8
0.6 0.6 0.6 0.6
F9 0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
0.0 0.0 0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Notes: Each panel indicates the predicted probability that a respondent chooses the bank account (black), 50:50 account (pale grey)
and the growth account (dark grey) as a function of the quantile of the respondent’s sophisticated financial literacy score
(horizontal axis). The panels in each row correspond to a given risk presentation, indicated at the left end of the row and defined in
Table 5. In each row, each of the four panels corresponds to the risk level indicated near the head of the table. All other
respondent characteristics are set to median values. 90% of the standard errors fell between .0034 and .0509.
FIGURE 5: Probabilities of account choice as a function of age, retirement accumulation and stock market expectations

Risk 12% Risk 16% Risk 20% Risk 28%

100% 100% 100% 100%

80% 80% 80% 80%

60% 60% 60% 60%

Agea

40% 40% 40% 40%

20% 20% 20% 20%

0% 0% 0% 0%
18‐24 35‐44 55‐64 >75 18‐24 35‐44 55‐64 >75 18‐24 35‐44 55‐64 >75 18‐24 35‐44 55‐64 >75

100% 100% 100% 100%

80% 80% 80% 80%
Accumulationb

60% 60% 60% 60%

Retirement

40% 40% 40% 40%

20% 20% 20% 20%
0% 0% 0% 0%

100% 100% 100% 100%

80% 80% 80% 80%

60% 60% 60% 60%

Expectationsc
Stock market

40% 40% 40% 40%

20% 20% 20% 20%

0% 0% 0% 0%

Notes: Each panel indicates the predicted probability that a respondent chooses the bank account (black), 50:50 account (pale grey) and the growth account (dark grey) as a function of age, retirement
accumulation and stock market expectations (horizontal axis). The panels in each row correspond to a given risk presentation, indicated at the left end of the row and defined in Table 5). In each row, each of the
four panels corresponds to the risk level indicated near the head of the table. All other respondent characteristics are set to median values.
a
90% of the standard errors fell between 0.0015 and 0.0307.
b
90% of the standard errors fell between 0.0018 and 0.0313.
c
90% of the standard errors fell between 0.0016 and 0.0483.