Inflation Expectations and Behavior:

Do Survey Respondents Act on their Beliefs?

Olivier Armantier 1

Wndi Bruine de Bruin 2

Wilbert van der Klaauw 1

Giorgio Topa 1

Basit Zafar 1

January 2013

ABSTRACT
We compare the inflation expectations reported by consumers in a survey with their
behavior in a financially incentivized investment experiment designed such that future
inflation affects payoffs. The inflation expectations survey is found to be informative in
the sense that the beliefs reported by the respondents are correlated with their choices in
the experiment. More importantly, we find evidence that most respondents act on their
inflation expectations showing patterns consistent (both in direction and magnitude) with
economic theory. Respondents whose behavior cannot be rationalized tend to be less
educated and to score lower on a numeracy and financial literacy scale. These findings
help confirm the empirical relevance of inflation expectations surveys, and provide
support to the micro-foundations of modern macroeconomic models.

1
2

Federal Reserve Bank of New York

Carnegie Mellon University, Department of Social and Decision Sciences

Corresponding Author: Olivier Armantier. [email protected]. The views expressed are those of the authors
and do not necessarily reflect those of the Federal Reserve Bank of New York or the Federal Reserve System. We
thank Charles Bellemare, Richard Blundell, Michael Bryan, Jeff Dominitz, Julie Downs, Baruch Fischhoff, Jordi Gali,
Charlie Holt, Eric Johnson, Arthur Kennickell, Chuck Manski, Rosemarie Nagel, Athanasios Orphanides, Onur Ozgur,
Simon Potter, Robert Rich, Justin Wolfers and Ken Wolpin for their advice on this project, as well as Sandy Chien,
Tim Colvin, Daniel Forman, Peter Fielding, Daniel Greenwald, Tanya Gutsche, Mandy Holbrook, Scott Nelson and
Bas Weerman for their help with conducting the research. Thanks to Nicolas Treich for his help on the theory. We also
would like to thank seminar participants at the Federal Reserve Bank of New York, the Federal Reserve Bank of
Boston, the University of Salento, the University of Virginia, the University of Maastricht, Science Po. Paris, the
Banque de France, the Bank of England, the University College London, the 2010 ESA conference, the 2011 IMEBE
conference, the 2011 Florence Workshop on Behavioral and Experimental Economics, the 2011 LeeX International
Conference on Theoretical and Experimental Macroeconomics, the Third French Econometric Conference, the 2011
workshop on Recent Advances on the Role of Beliefs on Decision Theory, the 2012 FUR XV International Conference,
and the 2012 SED conference.

1. Introduction
Expectations, and inflation expectations in particular, are at the center of monetary policy
and much of modern macroeconomic theory (Woodford 2005, Gali 2008, Sims 2009).
Although the academic debate about expectations formation is still open, 1
macroeconomic models are generally built on the assumption that agents maximize
expected utility under a well defined distribution representing their beliefs. In a nominal
world (i.e. with prices denominated in money), any intertemporal decision requires
inflation expectations to convert nominal into real returns. As a result, theory predicts
that inflation expectations are a key determinant of behavior. In particular, households
should take expected inflation into consideration when deciding, for instance, on large
durable purchases, saving, managing debt, mortgage (re)financing, or wage negotiations.
In general equilibrium, these intertemporal decisions in turn affect real economic activity
and therefore realized inflation. Because households in aggregate are an important driver
of economic activity, 2 they play a major role in this transmission mechanism. The
importance of this mechanism and the role played by households are now well
recognized both in academic and in central banking circles (Bernanke 2004, 2007). It is
therefore generally agreed that one of the first steps to controlling inflation consists in
actively managing the publics beliefs about future inflation (Woodford 2004, 2005). 3
Because of the role played by inflation expectations, accurate measurements of the
publics beliefs are important to scholars and policy makers. In particular,
macroeconomists increasingly use outside estimates of inflation expectations as an input
to their models.4 Central banks also need accurate measures of inflation expectations to
calibrate monetary policy. In addition, to monitor the effectiveness of its communication,
a central bank needs to regularly assess the consistency of the publics beliefs with policy
objectives.5
Existing measures of inflation expectations may be partitioned into two broad categories
depending on whether they are direct or indirect. Indirect measures are inferred from
either financial instruments (such as TIPS, the Treasury Inflation-Protected Security), the
term structure of interest rates, or past realizations of inflations rates. Direct measures are
1

Since Muth (1961) and Lucas (1972) inflation expectations have been mostly assumed to be formed
rationally. Over the past 20 years, with mounting empirical evidence rejecting the rational expectations
hypothesis, several alternatives have been introduced including adaptive learning (Sargent 1999, Evans and
Honkapohja 2001), sticky information (Mankiw and Reis 2002), rational inattention (Sims 2006,
Mackowiak and Wiederholt 2009), or asymmetric information (Capistran and Timmermann 2009).
Regardless of how expectations formation is formalized, these models all assume that agents act on their
inflation beliefs.
2
For instance, consumer expenditure represented 71% of U.S. GDP in 2011.
3
In particular, Bernanke (2004) argued that an essential prerequisite to controlling inflation is controlling
inflation expectations.
4
Examples include Roberts (1995), Erceg and Levin (2003), Carroll (2003), Mankiw, Reis and Wolfers
(2003), Nunes (2010), Adam and Padula (2011).
5
Observe that what academic economists and policy makers need are not necessarily accurate predictors of
future inflation, but accurate measures of the publics true beliefs. Of course, the publics true beliefs may
be unbiased predictors. However, unbiasedness is not a requirement for inflation expectations to be
informative about the economic decisions the public makes.

obtained from surveys in which consumers, businesses or professional forecasters are

asked to self-report their subjective beliefs about future inflation. In the U.S., such
surveys include the monthly Reuters/University of Michigan Survey of Consumers, the
Livingston Survey, the Conference Boards Consumer Confidence Survey and the Survey
of Professional Forecasters. In addition, because of the role played by households in
determining aggregate demand, several central banks around the world are now
conducting inflation expectations surveys of consumers.6
Each of these direct and indirect measures has potential weaknesses. While market based
estimates rely on strong modeling assumptions (e.g. about risk and liquidity premia),
surveys of professionals are often based on small samples and may be biased by strategic
misreporting.7 Because of the absence of direct financial incentives, inflation surveys of
households may suffer from a cheap talk problem leading to noisy and possibly
uninformative responses.8 Moreover, although inflation is consistently ranked among the
top economic concerns in public-opinion polls (Shiller 1997), the impact of inflation
expectations on behavior is still not well-understood. In particular, because households
may face more significant risks (related to e.g. employment or health), the extent to
which they account for future inflation prospects when making decisions is unclear.
In this paper, we examine whether consumers act on the inflation expectations they report
in a survey. Because of possible confounds, it is difficult to address this issue with field
(i.e. non-experimental) data. For instance, the timing of a durable good purchase could be
influenced by the respondents inflation expectations, but also by time discounting or
liquidity constraints (which may be difficult to measure). Instead, we propose to use
controlled experimental methods to isolate the impact of inflation expectations. More
specifically, we compare the behavior of consumers in a financially incentivized
investment experiment with the beliefs they self report in an inflation expectations survey.
As further explained below, the survey and experiment were fielded twice with the same
respondents at a five-month interval. Our analysis may be decomposed in two parts.
First, we evaluate the extent to which inflation expectations surveys of consumers are
informative by looking at the correlation between reported expectations and experimental
choices. In essence, we are conducting what the survey literature refers to as a construct
validity exercise, which is a key requirement to validate a survey question (Carmines
and Zeller 1991).9 More generally, our validation exercise is relevant to the literature on
6

Central banks that survey consumers about their inflation expectations include the Bank of England, the
European Central Bank, the Bank of Australia, the Bank of Japan, the Reserve Bank of India, and the
Sveriges Riksbank.
7
For instance, the Survey of Professional Forecasters conducted by the Federal Reserve Bank of
Philadelphia currently consists of 45 respondents on average. Moreover, it has been argued that, because of
strategic and reputational considerations, professional forecasters may have incentives to misreport their
beliefs (Ehrbeck and Waldmann 1996, Laster, Bennett and Geoum 1999, Ottaviani and Srensen 2006).
8
For a discussion of this issue see e.g. Keane and Runkle (1990), Manski (2004, 2006), Pesaran and Weale
(2006), or Inoue, Kilian and Kiraz (2009).
9
For instance, a survey question aimed at eliciting understanding of HIV risks is validated by examining
how responses are correlated with risky sexual activities (Bruine de Bruin et al. 2007). An alternative
validation approach would be to compare survey responses with inflation expectations elicited with a
financially incentivized mechanism such as a proper scoring rule (see e.g. Savage 1971). There is no
guarantee, however, that such an approach would be valid. Indeed, the respondents wealth is likely to

subjective expectations in which expectations surveys are often used to complement

choice data in order to better explain education, labor, retirement or health decisions (see
e.g. Wolpin and Gonul 1985, van der Klaauw and Wolpin 2008, de Paula, Shapira and
Todd 2011, as well as Manski 2004, Pesaran and Weale 2006 and van der Klaauw 2012
for reviews). For this approach to be effective, however, a key requirement is that the
survey responses truly capture the agents unobserved beliefs. We propose to test this
hypothesis and evaluate the information content of an expectations survey by using
controlled experimental methods.
Second, we subject the data to a stricter analysis by examining the extent to which
inflation expectations and experimental decisions comply with economic theory. More
precisely, we exploit the panel structure of the data to investigate whether or not the
survey respondents who report different inflation beliefs in each survey also modify their
experimental choices in a way that can be rationalized. In other words, we conduct a
simple yet formal test of one of the basic assumptions underlying most modern
macroeconomic models. Note that we are not trying to assess how consumers generally
take into account expected inflation when making decisions in their everyday life. Instead,
our objective is to evaluate whether, when given an explicit opportunity to do so,
consumers act on their inflation expectations. Therefore, this study should only be
considered a first step in establishing empirically the role played by inflation expectations
in shaping economic behavior.
The results indicate that the inflation expectations survey is informative. Indeed, stated
beliefs and experimental decisions are found to be highly correlated and consistent, on
average, with payoff maximization. Furthermore, we find that respondents who change
their inflation expectations from one survey to the next, also tend to adjust their decisions
in the experiment in a way consistent (both in direction and magnitude) with economic
theory. Finally, our results suggest that respondents whose behavior is difficult to
rationalize are less educated and score lower on a numeracy and financial literacy scale.
We consider these findings to lend support to the use of survey measures of inflation
expectations for monetary policy, and to provide direct empirical evidence, at the micro
level, on one of the basic assumptions underlying most modern macroeconomic models.
The remainder of the paper is organized as follows. The survey and the respondents are
described in Section 2. The design of the experiment is presented in Section 3. The
responses to the inflation expectations questions and the choices made in the experiment
are analyzed separately in Section 4. In Section 5, we test whether stated beliefs about
future inflation are informative about experimental choices. In Section 6, we exploit the
panel structure of the data to study how respondents who change their predictions from
one survey to the next adjust their experimental choices. Our final comments are
provided in Section 7.

depend on future inflation, which creates a stake in the event predicted. As shown by (e.g.) Karni and Safra
(1995), incentivized beliefs elicitation techniques are only incentive compatible when the respondent has
no stake in the event predicted (the so called no stake condition).

2. The Survey and the Respondents

The survey is part of an ongoing effort by the Federal Reserve Bank of New York, with
support from academic economists and psychologists at Carnegie Mellon University. The
general goal of this initiative is to better understand how the public forms and updates
beliefs about future inflation, and to develop better tools to measure consumers inflation
expectations (Bruine de Bruin 2010a). The survey consists of two sets of questions. The
first set, which is analyzed in this paper, examines the link between self-reported beliefs
and economic behavior. The second set, which is analyzed separately in Armantier et al.
(2012), investigates how individual consumers revise their inflation expectations after
being exposed to new information.

2.1 The Respondents

The survey, which includes the experiment, was conducted over the internet with
RANDs American Life Panel (ALP). Our target population consists of individuals 18 or
older who participated in the Michigan Survey between November 2006 and July 2010
and subsequently agreed to participate in the ALP. 10 Out of a total sample of 972
individuals invited to participate in the survey, 771 did so, for a response rate of 79.3%.
Those who completed the first survey were invited to participate in the second survey, of
which 734 did so, implying a response rate of 95.2%. The first survey was fielded
between July 20, 2010 and August 17, 2010. The second survey was fielded roughly five
months later, between January 3rd, 2011 and February 9, 2011. Respondents received
$20 for each completed survey. As explained in the next section, respondents were also
eligible to earn extra money if they completed the experiment. Although respondents
were allowed to skip questions, those who tried to do so received a prompt encouraging
them to provide an answer.
As indicated in Table 1 (column All Data), respondents reported an average age of
52.1, with a median of 53. In total, 57% of the respondents were female, 15% had no
more than a high school diploma, while 21% possessed a post graduate degree (i.e.
beyond a Bachelor degree). The median reported income range was $60-$75k, with 42%
of the respondents reporting incomes over $75k. The average and median time taken to
complete the survey and experiment were respectively 42 and 26 minutes with no notable
differences between survey 1 and 2. There is, however, substantial heterogeneity across
respondents. While fewer than 1% of the respondents completed the survey and
experiment in less than 9 minutes, other respondents took a considerable amount of time
between the moments they opened and finished the survey and experiment (more than a
week for 5% of the respondents). As we shall see next, it is important to control for these
differences when studying the link between reported beliefs and experimental choices.

10

The Michigan survey is a monthly telephone survey with 500 respondents, consisting of a representative
list assisted random-digit-dial sample of 300, and 200 respondents who were re-interviewed from the
random-digit-dial sample surveyed six months earlier. Our target population is further restricted to active
ALP members, defined as those who either participated in at least one ALP survey within the preceding
year, or were recruited into the ALP within the past year.

2.2 Procedure
Both surveys had a similar structure.11 As explained in more detail below, respondents
first reported their expectations for future inflation. Then, they were asked to explain
what information they used to form their reported inflation expectations (not analyzed
here). The experiment was presented next. After answering questions about how they
update their beliefs about future inflation (not analyzed here), respondents completed
measures of numeracy, financial literacy, and willingness to take risk.
Three features of the design are worth noting. First, up to the experiment, the respondents
were asked the same questions in the same order in both surveys. Second, respondents
did not know about the incentivized experiment before they reach it. As a result,
respondents reported their inflation beliefs without knowing about the subsequent
experiment in which payments depend on future inflation. Third, several questions
separated the experiment from the inflation expectations questions. More precisely, after
a respondent reported his expectations he had to provide more than 30 responses before
reaching the experiment. This delay, along with the financial incentives involved in the
experiment, reduces the possibility for ex-post rationalization, whereby a respondent
makes choices in the experiment not as a reflection of his preferences, but simply to be
consistent with the predictions he stated previously. As we shall see, we find evidence
against the hypothesis that respondents were driven by ex-post rationalization.

2.3 Reported Beliefs

Each respondent was randomly assigned to one of two expectation treatments. In the
Inflation treatment, respondents were asked directly about their expectations for the
rate of inflation. In the Price treatment, respondents were asked about their
expectations for the prices of the things I usually spend money on. In both treatments
and both surveys point predictions were elicited for the same time horizon: between now
and 12 months from now. In addition to point estimates, the respondents in both
expectation treatments were asked to report probabilistic beliefs for a range of inflation
outcomes. More specifically, respondents were asked to state the percent chance that,
over the next 12 months, the rate of inflation or changes in prices would be within
the following intervals: ]-12% or less, [-12%,-8%], [-8%,-4%],[-4%,-2%], [-2%,0%],
[0%,2%], [2%,4%], [4%,8%], [8%,12%], [12% or more. Respondents could press a
button to see the sum of the probabilities entered so far in order to verify that their
answers added to 100%. If it was not the case, they were prompted to go back and make
the appropriate changes.
Following the approach developed by Engelberg, Manski and Williams (2009), a
generalized beta distribution is fitted to each respondents stated probabilistic beliefs. We
then generate two variables that will be used in the econometric analysis. The first, the
Estimated Expected Prediction, is the mean of the respondents beta distribution. The

11

The complete list of questions asked in the first survey may be found in the supplemental material
available online at https://sites.google.com/site/olivierarmantier/.

second, equal to the variance of the respondents fitted distribution, is assumed to capture
the respondents inflation uncertainty.

2.4 Numeracy, Financial Literacy and Self-Reported Risk Tolerance

Six questions were asked in the first survey to measure the respondents numeracy and
financial literacy. The numeracy questions were drawn from Lipkus, Samsa, and Rimer
(2001), while the questions about financial literacy were slightly adapted from Lusardi
(2007).12 We created a variable taking integer values between zero and six depending on
the number of correct answers the respondent gave to these questions. As indicated in
Table 1 (column All Data) respondents answered 4.5 questions correctly on average
with a median of 5. There is, however, some heterogeneity across respondents: While
29.8% answered every question correctly, 11.0% got less than half of the answers right.
In each survey, respondents were also asked to assess their willingness to take risk
regarding financial matters using a qualitative scale ranging from 1 (Not willing at all) to
7 (very willing). This instrument has been shown to produce meaningful measures of risk
preferences. In particular, Dohmen et al. (2011) find that the risk tolerance self-reported
on this qualitative scale is consistent with the risk preference elicited with a financially
incentivized lottery-type experiment developed by Holt and Laury (2002). Other studies
using this measure of risk attitude include Bonin et al. (2007), Dohmen, Khamis and
Lehman (2010), or Caliendo, Fossen and Kritikos (2010). As indicated in Table 1
(column All Data) the average reported risk tolerance across the two surveys is 3.3
with a median of 3. One third of the respondents selected a rating of 1 or 2, thereby
reflecting substantial aversion to risk. In contrast, one respondent out of four indicated a
high tolerance toward risk by selecting a rating of 5, 6 or 7. This distribution is generally
consistent with those obtained in previous work using the same measure. Furthermore,
our risk attitude measure appears to be generally stable over time. Indeed, we find a
correlation of 0.822 between the risk tolerances reported by the same respondents across
the two surveys.13

3. The Financially Incentivized Experiment

3.1 Experimental Design
As shown in Appendix A where the experimental instructions are reported, the
experiment consists of 10 questions with real monetary consequences. For each question
the respondent is asked to choose between two investments. Each investment produces a

12

Here is an illustration of the type of questions we asked: If you have $100 in a savings account, the
interest rate is 10% per year and you never withdraw money or interest payments, how much will you have
in the account after: one year? two years?. The other numeracy and financial literacy questions may be
found in the supplemental material.
13
Reporting different risk tolerance in each survey is not necessarily a violation of standard economic
theory. Indeed, the qualitative measure reflects both the respondents risk preference and the nature of the
risks they face. While standard theory assumes that the former is stable over the time, the latter may have
evolved in the five months separating the two surveys.

specific revenue payable 12 months later.14 Investment B produces a fixed dollar amount
while investment A is indexed on future inflation. More specifically, the respondents
earnings under investment A depend on what realized inflation will be over the next 12
months. The possible earnings under investment A as a function of realized inflation were
presented to the respondents as in Table 2, where the rate of inflation was explicitly
defined as the official annual U.S. CPI rounded to the nearest percentage point.15
As indicated in Appendix A, investment A remains the same in each of the 10 questions.
In contrast, the revenue produced by investment B varies across questions. We conducted
two treatments by changing the order in which investment B was presented to
respondents. In the Ascending scale treatment, the earnings of investment B increase in
increments of $50 from $100 in question 1 to $550 in question 10. In the Descending
scale treatment the earnings of investment B decrease in increments of $50 from $550 in
question 1 to $100 in question 10. To simplify the description of the experiment, we only
refer to the Ascending treatment for the remainder of Section 3.
With respect to predicted behavior, observe in Appendix A that an expected payoff
maximizer with an inflation expectation within [0%,9%], say 5%, should first select
investment A for the first 4 or 5 questions (the respondent is indifferent between the two
investments in question 5 as they both produce $300 in expectation), and then switch to
investment B for the remaining 5 questions. Likewise, it is easy to see that, regardless of
risk attitude, an expected utility maximizer should switch investments at most once and in
a specific direction (i.e. from investment A to B). The analysis conducted in the next
section therefore focuses on a respondents switching point. We only define this
switching point for respondents whose behavior may be rationalized, that is for
respondents who switch at most once from investment A to investment B. For these
respondents, the switching point is set equal to the number of questions for which the
respondent selected investment A. So, for both experimental treatments, the switching
point can take integer values between 0 (the respondent always selects investment B) and
10 (the respondent always selects investment A).16
The participants were informed that two respondents would be paid in each of the two
surveys according to their choices in the experiment. Once a survey was completed, we
randomly picked one of the ten questions, and two survey participants who completed the
experiment. Twelve months later, these two participants were paid according to the
investment choice they made for the selected question. Although the amounts a
respondent could earn were substantial compared to traditional lab experiments (i.e. up to
$600), the respondents were not able to calculate exactly their odds of being selected for
payment since the exact number of participants was unknown at the time the experiment
was conducted. Note also that motivating subjects by paying a few of them large amounts
14

Observe that, regardless of the investment selected, a respondent can only receive money 12 months after
the experiment. As a result, time preference cannot influence the choice between the two investments.
15
Although related to studies on portfolio choices (e.g. Dominitz and Manski 2007), our experiment has a
distinctive feature as it focuses on uncertain returns linked to future inflation prospects.
16
Observe that, although the question addressed is different, the structure of our experiment is akin to the
experiment of Holt and Laury (2002) in which risk attitude is measured by the number of questions after
which a respondent switches from a safer to a riskier lottery. Unlike Holt and Laury (2002), however, the
switching point does not have a direct interpretation here.

with small probabilities is a method used in several lab and field experiments (Camerer
and Ho 1994, Harrison, Lau and Williams 2002, Kobberling, Schwieren and Wakker
2007, Bettinger and Slonim 2007, Dohmen et al. 2011).
Finally, each respondent was randomly assigned to one of the four possible treatment
combinations (i.e. either the Price or the Inflation treatment, and either the
Ascending or the Descending treatment). Once assigned to a treatment, a respondent
remains in the same treatment in the two surveys.

3.2 Economic Considerations

Although presented in terms of terminal payoffs to facilitate the respondents
comprehension, investments A and B both have an economic interpretation. Investment
A corresponds to the following scenario: an agent borrows $5,000 for 12 months at a
rate equal to the inflation rate, and invests the $5,000 for 12 months in an account that
earns a fixed annual rate of 11%. Investment B corresponds to the following scenario:
an agent borrows $5,000 for 12 months at a rate equal to the inflation rate, and invests
the $5,000 for 12 months in an inflation protected account that earns an annual rate equal
to the inflation rate plus k %, where k varies in increments of 1% from 2% in question 1
to 11% in question 10.
In nominal terms, investment B earns $5,000
while investment A earns $5,000
0.11
, where i denotes the inflation rate over the next 12 months. If expressed in real
terms, investment B earns
$5,000 / 1
, while investment A earns $5,000
0.11
/ 1
, where
1.11/
1 and
$5,000. It is then easy to
see that the variance of the earnings with respect to inflation is always lower with
investment B whether one expresses earnings in nominal or in real terms. We can then
derive three propositions that will help us assess whether the behavior observed in the
experiment is consistent with standard economic theory. To do so we consider an
expected utility framework and we assume that the agents utility function over income,
denoted . , is thrice differentiable, strictly increasing, and satisfies the von Neumann
Morgenstern axioms.
Proposition 1: If investment A and investment B have the same expected return
then a risk-averse agent prefers investment B to investment A.17
The proposition therefore shows that, all else equal, and in particular for the same set of
beliefs, a risk-averse (respectively risk-loving) agent has a lower (respectively, higher)
switching point than a risk neutral agent. For instance, consider a respondent who expects
the inflation rate to be 4% over the next 12 months. In question 6 the two investments
produce the same expected return of $350. If this respondent is risk averse (respectively
risk loving) then he should select the safer (respectively riskier) option in question 6, that
is, he should select investment B (respectively investment A).
We now generalize Proposition 1 by showing that, all else equal, a more risk-averse agent
has a lower switching point.

17

The proofs of all the propositions are provided in Appendix B.

Proposition 2: If a risk averse agent is indifferent between investment A and

investment B then, all else equal, a more risk averse agent (in the classical sense
of Pratt 1964) prefers investment B to investment A.
Proposition 2 therefore allows us to rationalize differences in behavior observed in the
experiment. Consider for instance two agents who share the same beliefs about future
inflation and report an expectation of 4%. Furthermore, assume the first agent selects a
switching point of 5 while the second agent selects a switching point of 3. Under
expected utility, we can rationalize this difference in behavior by a difference in risk
aversion, whereby the second agent is more risk averse than the first.
For Proposition 3, we restrict our attention to HARA (hyperbolic absolute risk aversion)
utility functions. This class of utility functions is quite general as it encompasses CRRA
(constant relative risk aversion) and CARA (constant absolute risk aversion) utility
functions. In addition, virtually all the utility functions used in practice (e.g. exponential,
logarithmic, power) belong to the HARA family.
Proposition 3: If a risk-averse agent with a HARA utility function is indifferent
between investment A and investment B, then the agent prefers investment B to
investment A for any increase in risk (in the classical sense of Rothschild and
Stiglitz 1970).
Proposition 3 therefore shows that if a risk-averse agent is indifferent between the two
investments for a given belief distribution, then the agent should strictly prefer the safer
investment (i.e. investment B) for any mean preserving changes of his belief distribution.
In other words, all else equal, a risk averse agent should switch from investment A to
investment B earlier when the risk associated with investment A increases.

4. Responses to the Survey

Out of the 771 respondents who answered at least one of the two surveys, a total of 81
respondents (23 in survey 1 only, 24 in survey 2 only, and 34 in both surveys) failed to
report a point prediction and/or to provide an answer for the 10 questions in the
experiment. Out of the 688 (respectively 676) remaining respondents in survey 1
(respectively survey 2), 598 or 86.9% (respectively 615 or 91.0%) had rationalizable
answers as they switched at most once from investment A to investment B during the
course of the 10 questions. Furthermore, 82.8% (1,004 out of 1,213) of the total number
of rationalizable answers are due to the same 502 repeat respondents who provided
rationalizable answers to both surveys. Note that the ratio of non-rationalizable responses
(11.1% across the two surveys) is lower than those typically obtained in the literature on
measuring risk attitude using the Holt and Laurys instrument. In particular, Holt and
Laury (2002) and Eckel and Wilson (2004) report that respectively 25% and 15% of their
respondents made non-rational choices.
Perhaps not surprisingly, the respondents with missing data and multiple switching points
have specific characteristics. Indeed, a comparison of the columns Group 1 and Group
2 with the columns Group 3 and Group 4 in Table 1 indicates that the first two
groups score significantly lower on our scale of numeracy and financial literacy, are more
likely to be a female, they have lower income and lower education, and they provide
higher and more volatile point predictions. As indicated in the last column of Table 1,
9

however, a probit regression in which the dependent variable is equal to 1 when a

respondent provides rationalizable answers in both surveys reveals that only the measure
of numeracy and financial literacy remains significant after controlling for other
respondents characteristics. More specifically, we find that a respondent is 6% more
likely to make rationalizable choices in both surveys for each additional numeracy and
financial literacy question he answers correctly. As we shall see later on, low numeracy
and low financial literacy are characteristics shared not only by respondents with nonrationalizable choices, but also by respondents whose choices, although rationalizable (in
the sense that they switch at most once from investment A to B), are not consistent with
expected utility theory.
Before examining the possible link between the respondents inflation expectations and
their behavior in the experiment, we look separately at the responses to the inflation
expectations questions and the choices made in the experiment.

4.1 Responses to the Point Predictions Questions

In Figure 1, we plot for each survey the distribution of the inflation point predictions over
the next 12 months combined across the Inflation and the Price treatments. As we
can see, both distributions have similar shapes with the same mode (2% to 4%), the same
median (3%) and the same interquartile range (3%). Observe, however, that the
distribution of point predictions for both treatments shifts to the right in survey 2, thereby
indicating an increase in inflation expectations between the five months that separate the
two surveys. A Mann-Whitney test confirms that the average point prediction in survey 1
(4.1%) is significantly lower (P-value=0.03) than the average point prediction in survey 2
(4.8%). Finally, note that the distributions of point predictions are similar to those
obtained in other inflation expectation surveys we conducted during the same period of
time with different but comparable respondents (Bruine de Bruin et al. 2010b). In other
words, we find no evidence that the inflation expectations reported in each survey were
affected by the presence of the subsequent experiment. This result is not surprising since
respondents only learned of the experiment after they reported their beliefs.
We plot in Figure 2 the distribution of the individual differences in point predictions
across the two surveys. To do so we calculate for each of the 502 repeat respondents with
rationalizable answers in both surveys the difference between the point prediction she/he
made in survey 2 and the point prediction she/he made in survey 1. Although the mode of
the distribution is centered on zero, the majority of respondents reported different
predictions in the two surveys. More precisely, between the five months that separate the
two surveys, 74% of the respondents revised their point prediction by at least 0.5%, and
27% by more than 4.0%.18 Finally, note that, consistent with the increase in average
point prediction observed in Figure 1, the distribution reported in Figure 2 has more
weight in the positive domain. In fact, only 28% of the respondents revised their point
prediction downward in survey 2.

18

For an analysis of the way respondents update their beliefs and whether their updating process may be
rationalized, we refer the reader to Armantier et al. (2012).

10

A year after we conducted each survey, i.e. in August 2011 and February 2012, the
Bureau of Labor Statistics reported that the CPI over the past 12 months was respectively
3.8% and 2.9%. Our respondents therefore made relatively accurate forecasts in the first
survey as their average point prediction (4.1%) is not significantly different from realized
inflation. In contrast, with an average point prediction of 4.8%, respondents significantly
overestimated inflation in the second survey and incorrectly predicted an increase in
inflation. The two surveys therefore provide mixed evidence about the ability of
consumers to forecast accurately future inflation. Without a longer panel, however, our
results should be considered anecdotal evidence. Furthermore, recall that the focus of this
paper is on the link between reported beliefs and economic behavior, not on the
forecasting ability of consumers.
Finally, we explore whether the responses to the point prediction question are affected by
the price versus inflation treatment to which a respondent is assigned. Recall that
roughly half of the respondents are asked about their expectations for the prices of
things I usually spend money on, while the other half is asked about the rate of
inflation. We plot in Figure 3 the distribution of responses for the two expectation
treatments. In both surveys, the different distributions exhibit a similar pattern. Note,
however, that consistent with previous studies we conducted (e.g. Bruine de Bruin et al.
2010b), the question about the prices of things I usually spend money on yields higher
average predictions than the question about the rate of inflation question. More
specifically, the average point prediction for the prices of things I usually spend money
on is 4.29% in survey 1 and 4.91% in survey 2, while the average point prediction for
the rate of inflation question is 3.80% in survey 1 and 4.58% in survey 2. These
differences, however, are well within one standard deviation (between 5% and 6% across
expectation treatments) and a Mann-Whitney test fails to identify a significant difference
between expectation treatments (P-value equals 0.18 in survey 1 and 0.25 in survey 2).

4.2 Choices Made in the Experiment

The distribution of switching points in each survey is plotted in Figure 4. We can see that
although respondents make use of all possible switching points, most choices (61% in
survey 1 and 55% in survey 2) are concentrated between 4 and 7. Note also that the
distribution of switching points shifts to the left in survey 2. A Mann-Whitney test
confirms that this difference across surveys is in fact highly significant (P-value = 3.4E4). As we shall see next, this shift toward lower switching points is consistent with the
fact that respondents reported higher point predictions in the second survey.
We plot in Figure 5 the distribution of individual differences in switching points for the
502 respondents who made rationalizable experimental choices in both surveys. As with
the point predictions, we can see that 78% of the respondents chose a different switching
point in each survey. Furthermore, note that 48% of the 502 respondents who made
rationalizable choices in both surveys selected a strictly lower switching point in survey 2.
In the next section, we will therefore be able to exploit the fact that most respondents
change their predictions and switching points across the two surveys to test whether the
direction and the magnitude of those changes are consistent with expected utility theory.

11

Finally, we explore whether the choice of switching point is influenced by the treatment
combination to which a respondent is assigned. Recall that our sample is segmented in
four groups depending on which expectation treatment (i.e. Price or Inflation) and
which experimental treatment (i.e. Ascending or Descending) a respondent is
assigned to. We plot in Figure 6 the distribution of switching points for each treatment
combination. None of these distributions seems to exhibit a distinguishable pattern. This
absence of treatment effect is confirmed by a series of Mann-Whitney tests (the P-values
range from 0.21 to 0.77).

5. The Link between Beliefs and Behavior

5.1 Are Point Predictions Informative about Experimental Choices?
We now turn our attention to the correlation between the respondents point predictions
and their switching points. In Figure 7, we plot for each switching point between 0 and 10
the average point prediction across the respondents who selected that switching point. For
instance, we can see that the respondents who always selected investment B, and who
therefore have a switching point equal to 0, reported an average point prediction of 9.3%
in survey 1 and 10.2% in survey 2. Observe first in Figure 7 that there is a generally
monotonic decreasing relationship between the reported beliefs and the switching points.
Furthermore, note that this relationship is very similar in both treatments. This result
therefore supports the hypothesis that our inflation expectations survey is informative, in
the sense that the beliefs the respondents reported correlate well, on average, with their
choices in the financially incentivized experiment.
We also plot in Figure 7 a risk-neutral band indicating the range of beliefs that would
rationalize each switching point under risk neutrality. For instance, if a risk-neutral agent
selects a switching point equal to 5, then his point prediction should belong to the interval
[3.5%, 5.5%]. As shown in Proposition 1, switching points below (respectively, above)
the risk-neutral band may be rationalized under risk aversion (respectively, risk loving).
We can see in Figure 7 that, on average, respondents exhibited behavior consistent with
risk neutrality, although a number of switching point and average prediction
combinations are close to the risk averse frontier.
This does not imply, however, that respondents systematically behaved as if risk-neutral.
In fact, the box plot in Figure 8 reveals that most respondents are outside the risk-neutral
band. More precisely, we find that in survey 1 (respectively in survey 2) 41%, 32% and
27% (respectively 37%, 41% and 22%) of the respondents behaved as if risk averse, risk
neutral, and risk loving. This finding therefore provides evidence against the hypothesis
that respondents, driven by a desire for internal consistency, simply chose a switching
point within the risk-neutral band as a way to rationalize ex-post the inflation prediction
they previously reported. Furthermore, we will see in section 5.2 that the departures from
risk neutrality are explained in large part by the respondents self reported risk tolerance.
The general trends observed in Figure 7 seem to be robust. In particular, instead of the
average point prediction, we plot in Figures C1 and C2 (reported in Appendix C) the
median point prediction in one case, and the estimated expected prediction (calculated
with the respondents reported probabilistic beliefs) in the other case. Although slightly
12

flatter, these two additional figures display a similar relationship between point
predictions and switching points. In Figures C3 to C6 (reported in Appendix C) we
reproduce Figure 7 with the data collected in each of the four treatment combinations.
Although the average point predictions are somewhat more volatile across switching
points than in Figure 7 (as can be expected given the reduction in sample sizes) the
general trend does not vary substantially across treatment combinations.
To confirm these observations statistically, we estimate a series of ordered probit models
in which the dependent variable is the respondents switching points. Table 3 shows the
results of these estimations for each survey. In Model 1, the parameter associated with the
variable Point prediction is highly significant and negative. 19 This result therefore
confirms that the respondents reported beliefs are informative about their decisions in
the incentivized experiment. We also find that the parameter associated with the selfreported measure of risk attitude is positive and significant. In other words, consistent
with Proposition 2, respondents who report being more risk-averse tend to select lower
switching points, while respondents who report being more risk-loving have higher
switching points. Furthermore, the parameter associated with inflation uncertainty is
significant and negative. Respondents with more diffuse beliefs, therefore, tend to switch
investment earlier. According to Proposition 3, this result may be rationalized under
expected utility if respondents exhibit risk aversion (which is the case for many
respondents). Finally, note that none of the treatment dummies is significantly different
from zero in Model 1, thereby supporting the absence of treatment effects.
To confirm the robustness of the results, we estimated several additional specifications.
In Model 2 of Table 3, we augment the specification by including demographic variables.
Observe that the parameters estimated in Model 1 remain essentially unchanged. Once
we control for the variables in Model 1, we find that none of the demographic variables
plays a role in explaining when a respondent switches from investment A to investment B.
In Model 3 of Table 3, we replace the point prediction by the estimated expected
prediction (i.e. the mean of the Beta distribution fitted to the respondent reported
probabilistic beliefs). Once again the parameters previously estimated remain essentially
unchanged. The parameter associated with the estimated expected prediction in Model
3 is similar, both in sign and magnitude, to the parameter associated with the point
prediction variable in Model 1. In fact, a log-likelihood ratio test reveals that the two
parameters are statistically indistinguishable at the usual significance levels (the P-value
is 0.182 in survey 1 and 0.354 in survey 2).
Finally, three of the four treatment combination dummies were interacted with the point
prediction in Model 4. As indicated in Table 3, none of the corresponding parameters is
found to be significantly different from zero at the usual significance levels. In other
words, we do not find statistical evidence that the slope of the relationship between the
point predictions and the switching points varies significantly across treatments. This
absence of treatment effect may seem somewhat surprising. Indeed, the payments in the
19

We do not report the marginal effects from the ordered probit regressions because, unlike binary probit
models, they are not directly interpretable. Instead, we report in Appendix D the outcome of simple linear
regressions. These additional regressions not only confirm the robustness of the results presented in this
section, but they also provide a sense of the relative effect of each explanatory variable.

13

experiment depend on a measure of inflation (the CPI), while respondents in the price
treatment are asked to make a prediction about the prices of things I usually spend
money on. As a result, one could have expected a weaker relationship between point
predictions and experimental choices in the price treatment than in the inflation
treatment (where respondents are asked to state their beliefs about inflation). Note,
however, that this lack of treatment effect is consistent with the fact that we failed to
identify significant differences between the point predictions (respectively the switching
points) reported in the inflation and prices treatments.20

5.2 Deviations from Risk Neutrality and Reported Risk Tolerance

Although the estimates in Table 3 confirm that reported beliefs and experimental
decisions are correlated, they fail to assess the extent to which choices in each experiment
comply with standard theory. Under expected utility, the switching point of a respondent
can deviate from risk neutrality only because of his risk attitude and subjective beliefs
about the investments riskiness, as determined by the shape of his inflation distribution.
To test this hypothesis, we calculate each respondents pair of risk neutral switching
points, that is, the pair of switching points the respondent should have selected if risk
neutral given his stated point prediction. For instance, consider a respondent in the
ascending scale treatment who made a point prediction of 5%. If risk neutral, this
respondent should switch to investment B after selecting investment A in the first 5 or 6
questions (the respondent is indifferent between the two investments in question 5 as they
both earn $400). In that case, the respondents pair of risk neutral switching points is
{5,6}. We then define the respondents deviation from risk neutrality as the difference
between the respondents actual switching point and his pair of risk neutral switching
points. For instance, if in the example above the respondent selected a switching point of
4 (respectively 8), then her deviation from risk neutrality is -1=Max{4-5,4-6}
(respectively 2=Min{8-5,8-6}). The deviation from risk neutrality is therefore an integer
between -10 and 10.
Using Proposition 2, we can then classify respondents depending on their revealed risk
attitude. In what follows, we will refer to a respondent as exhibiting risk aversion, risk
loving, or risk neutrality when his deviation from risk neutrality is respectively strictly
negative, strictly positive or equal to zero. Said differently, a respondent exhibits risk
aversion, risk loving, or risk neutrality when his pair of point prediction and experimental
choice is located below, above or within the risk neutral band in Figure 7. Note that a
respondents revealed risk attitude should not be confused with his reported risk tolerance.
The former is fully characterized by the respondents point prediction and experimental
20

We also estimated several alternative models not reported here. For instance, we estimated a model in
which i) a price and ascending dummy variable were introduced instead of interactions between the
expectation and experimental treatment variables, ii) the reported risk aversion is defined as a set of
seven dummy variables, iii) respondents with extreme switching points (i.e. 0 or 10) are excluded, iv) the
reported risk attitude was interacted with other explanatory variable (e.g. the estimated variance), v) the
time taken to complete the survey enters non-linearly in order to test the hypothesis that the respondents
with the shortest completion time did not answer the survey with as much attention as other respondents,
vi) the sample is limited to the 502 respondents that made rationalizable choices in both surveys. These
models produced similar conclusions to the ones we present in this section.

14

choice, while the latter is the respondents self assessment in the survey about his general
willingness to take risk when it comes to financial matters.
To get a complete picture of the factors that influence the deviations from the risk neutral
choices, we report in Table 4 the estimates produced by four ordered probit models. The
dependent variable is a respondents deviation from risk neutrality in Model 1, and the
absolute value of this deviation in Model 2. In the last two models, the dependent
variable is a respondents deviation from risk neutrality, but we restrict the sample to
respondents with revealed risk aversion in Model 3, and to respondents with revealed risk
loving in Model 4. The last two models will help us gauge whether the same factors
influence respondents who exhibit risk aversion and risk loving. Based on the theory, the
hypothesis to be tested with these models is that the reported risk tolerance and the
inflation uncertainty are the only two explanatory variables with statistical power to
explain the deviation from risk neutrality.
The results reported in Table 4 (Model 1) reveal a strong positive relationship between a
respondents reported risk tolerance and his/her deviation from risk neutrality. As
indicated in Model 3 and 4, this effect applies both to respondents that exhibit risk
aversion and risk loving. In other words, we find that, all else equal, respondents who
report being more risk averse (respectively, risk loving) select lower (respectively,
higher) switching points in the experiment thereby revealing higher levels of risk
aversion (respectively, risk loving). This result is confirmed in Model 2, as the effect of
reported risk tolerance on the absolute deviation from risk neutrality is now identified as
being U shaped. Consistent with our hypothesis, we therefore find that reported risk
tolerance plays a role in explaining why a respondent did not behave as if risk neutral in
the experiment.
Inflation uncertainty does not appear to affect choices in Model 1. This result, however,
is misleading. Indeed, observe in Model 2 that inflation uncertainty has a significant
positive impact on the absolute deviation from risk neutrality. The difference between the
results obtained in Model 1 and Model 2 may be explained by the fact that inflation
uncertainty has an opposite effect on respondents depending on whether their
experimental choices reveal risk aversion or risk loving (see Model 3 and 4). Indeed, in
line with Proposition 3, we find that, all else equal, respondents who exhibit risk aversion
(respectively, risk loving) have a lower (respectively, higher) switching point when their
inflation uncertainty is higher. Consistent with our hypothesis, we therefore find that
deviations from risk neutrality may be explained by the respondents reported risk
tolerance and their subjective uncertainty about future inflation.
In contrast with our hypothesis, however, other variables also have explanatory power. In
particular, observe in the last three models in Table 4 that the parameters associated with
the respondents education level, as well as numeracy and financial literacy are highly
significant. Furthermore, the results from Model 3 and 4 indicate that these variables
have an opposite effect on respondents who behave as if risk averse and those who
behave as if risk loving. In other words, it seems that, all else equal, the experimental
choices of respondents with lower numeracy, financial literacy, and education level tend
to deviate further away from those of a payoffs maximizer. This finding has several
possible interpretations. Respondents with lower numeracy, financial literacy or
education i) may be more likely to make optimization errors when they try to select an
15

investment consistent with their beliefs, ii) they may not fully understand the experiment
(although their experimental choices are rationalizable), iii) they may report point
predictions (respectively risk tolerance) in the survey that are not a true reflection of their
inflations beliefs (respectively risk attitude), or iv) they are not expected utility
maximizers.
Finally, we find that all else equal, and in particular after controlling for reported risk
attitude and inflation uncertainty, respondents who took longer to complete the survey
have a behavior in the experiment that is more likely to deviate from that of a risk neutral
optimizer. This result also has several possible interpretations. It could reflect the fact
that, when these respondents finally reach the experiment, the predictions they made at
the beginning of the survey do not accurately represent their beliefs anymore (some
respondents took more than 10 days between the time they started and completed the
survey and experiment). Alternatively, it is possible that respondents who took a longer
to complete the survey did not make predictions and choices in the experiment with the
same attention as the other respondents. Finally, it could reflect a lack of comprehension,
whereby respondents who took longer to complete the survey may have needed the extra
time because it was more difficult for them to understand and answer the questions.

6. Changes in Predictions and Changes in Experimental

Choices across Surveys
We now exploit the panel structure of the data to look at the predictions and experimental
choices of a given respondent across the two surveys. To do so, we concentrate on the
502 repeat respondents who made rationalizable choices in both surveys. Our general
objective in this section is to address the following question: When a respondent reports
different inflation expectations in survey 2 than in survey 1, does this respondent also
adjust his experimental choices in a way that is consistent with standard theory? We
decompose the analysis in two parts: First we look at whether the direction of the
adjustments across surveys may be rationalized; then, we study whether the magnitude of
the adjustments is consistent with a simple expected utility model.

6.1 Is the Direction of Adjustments Consistent with Theory?

6.1.1 Changes in Revealed Risk Attitudes
The first test we conduct is based on the standard assumption that an agents preferences
are invariant over time. As a result, a respondent should not act as if risk averse in one
survey and as if risk loving in the other survey. If a respondent exhibits such behavior,
then we say that the respondents revealed risk attitudes are inconsistent across the two
surveys. In contrast, switches from revealed risk neutrality to revealed risk aversion or
risk loving will not be considered inconsistent with theory. Indeed, given the fact that
experimental choices are discrete, we cannot exclude the possibility that respondents who
make decisions consistent with risk neutrality are in fact slightly risk averse or risk loving.
We report in Figure 9 the distributions of revealed risk attitudes in both surveys. In more
than 50% of the cases a respondent with a specific revealed risk attitude in survey 1 (i.e.
16

risk averse, risk loving or risk neutral) exhibits the same revealed risk attitude in survey 2.
Furthermore, note that only 6% of the repeat respondents have inconsistent revealed risk
attitudes. More precisely, 16 out of the 214 (respectively 14 out of the 123) respondents
who made experimental choices consistent with risk aversion (respectively risk loving) in
survey 1, behaved as if risk loving (respectively risk averse) in survey 2.
To identify whether these respondents have specific characteristics, we estimate a probit
model in which the dependent variable is equal to 1 when a respondents revealed risk
attitudes are inconsistent across the two surveys. In addition to demographic variables we
control for the absolute difference between the risk tolerance reported in surveys 1 and 2.
Thus, we can test whether inconsistent revealed risk attitudes are accompanied by large
changes in reported risk tolerance. Similarly, we control for the difference in inflation
uncertainty across surveys and the total time taken to complete both surveys.
The results reported in Table 5 (Model 1) indicate that inconsistent revealed risk attitudes
are driven by three factors: the respondents i) financial literacy and numeracy, ii)
education level and iii) total completion time. In particular, we find that respondents with
at most a high school diploma are 15.3% more likely to exhibit inconsistent revealed risk
attitudes than respondents with some college but no more than a Bachelor degree.
Likewise, respondents who answer correctly half of the numeracy and financial literacy
questions are 14.4% more likely to exhibit inconsistent revealed risk attitudes than
respondents who answer all six questions correctly. Finally, observe that changes in
reported risk tolerance or in inflation uncertainty across surveys do not explain
inconsistent switches in revealed risk attitudes.

6.1.2 Correlation between Changes in Point Predictions and Changes in

Switching Points
As illustrated in Figure 7, we should expect a negative correlation between changes in
inflation expectations and changes in switching points: All else equal, a respondent
whose inflation expectations increase (decrease) from one survey to the next should
select a lower (higher) switching point in survey 2 than in survey 1. We call this first
order effect, the expectation effect. There is, however, a possible second order effect, a
risk effect, for respondents who are not risk neutral. Indeed, all else equal, and in
particular if expected inflation remains the same, an agent may be led to select a different
switching point if the shape of his beliefs distribution changes from one survey to the
next. For instance, if the inflation uncertainty of a risk-averse agent increases (decreases),
then, as shown in Proposition 3, the agent will be led to select a smaller (higher)
switching point. The first order expectation effect and the second order risk effect
can go in the same or in opposite directions. In the latter case, the exact correlation
between changes in predictions and changes in switching points depends on the relative
strength of the two effects. Simulations, however, indicate that the first order
expectation effect almost always dominates.21
21

The simulations are based on the expected utility model presented in Section 6.2. For almost all levels of
risk aversion, these simulations suggest that the risk effect only dominates when expected inflation
remains virtually identical across the two surveys, while the distribution of beliefs change from extremely
concentrated in one survey to nearly uniform in the other survey.

17

Taking these two effects into consideration, we now define a new variable to study how
changes in switching points across surveys respond to changes in point predictions. Let
{S1;S2} and {P1;P2} denote a respondents pairs of switching points and point predictions
in surveys 1 and 2. We say that this respondent has an inconsistent correlation when the
following three conditions are all satisfied:
1)
(S2-S1) * (P2-P1) 0
2)
{S1;S2} {1;1} and {S1;S2} {11;11}
3)
|S2-S1| > 1 or |P2-P1| > 1%
The first condition reflects a violation of the first order expectation effect just described
(e.g. an increase in point prediction should result in a decrease in switching points).
Condition 2 excludes respondents who pick the same switching point either equal to 1 or
equal to 11 in both surveys. Indeed, regardless of the point predictions made in survey 1
and 2, it is always possible to find a specific risk attitude to rationalize the behavior of
respondents who consistently pick the same extreme choice in both surveys. Note,
however, that only 3.2% of the 502 repeat respondents exhibit such behavior.
The last condition characterizes an indifference zone in which changes in point
predictions and changes in switching points do not exceed respectively 1% and 1.
Respondents with inconsistent correlations should fall outside this indifference zone. In
contrast, three types of respondents fall within the indifference zone. First, the 8.4% (42
out of 502) respondents who provide exactly the same prediction and the same switching
point in both surveys. Second, the respondents with about the same prediction in both
surveys who make slight changes in switching points because of the second order risk
effect. Finally, the indifference zone reflects the discreteness of the experiment. Indeed,
recall that the choice of a switching point may be rationalized by a range of predictions.
For instance, a risk neutral agent could have the same inflation expectation of 4% in both
surveys, but selects a switching point of 6 in one survey and a switching point of 7 in the
other survey as he is indifferent between the two investments in question 6. Likewise, a
risk neutral agent could select the same switching point of 6 in both surveys, and change
his prediction from 4% to 5%. Note, however, that the way in which the indifference
zone is defined may be considered somewhat conservative. Indeed, the behavior of a
respondent with S2=S1 and |P2-P1|=2%, or |S2-S1|=2 and P2=P1 is rational under risk
neutrality. However, such a respondent would fall outside our indifference zone.
To illustrate the adjustments across surveys, we plot in Figure 10 the difference between
a respondents point predictions across the two surveys (X-axis) and the difference
between the respondents switching points across surveys (Y-axis). The area of each
bubble in Figure 10 reflects the number of respondents at that point. In addition, the
indifference zone defined in the previous paragraph is indicated in Figure 10.
As mentioned earlier, we generally expect a negative correlation between the adjustment
in predictions and the adjustment in switching points. If this is the case, then respondents
have consistent correlations and they should be located either in the upper left or bottom
right quadrants of Figure 10. As we can see, this is overwhelmingly the case. More
precisely, a total of 102 out of 502 repeat respondents (20.3%) exhibit inconsistent

18

correlations. Furthermore, we find that the correlation coefficient between the adjustment
in predictions and the adjustment in switching points is -0.49.22
To identify whether respondents with inconsistent switching points have specific
characteristics, we again estimate a probit model in which the dependent variable is equal
to 1 when the respondent exhibits an inconsistent correlation. The results reported in
Table 5 (Model 2) reveal that respondents with lower education, respondents who scored
low on our numeracy and financial literacy scale, and respondents who took more time to
complete the survey are more likely to exhibit inconsistent correlations. More specifically,
we find that, compared to a respondent with the highest possible numeracy and financial
literacy score, a respondent who answered half the questions correctly is 16% more likely
to exhibit an inconsistent correlation. Furthermore, compared to a respondent who went
to college up to a Bachelor degree, a respondent without a college education (respectively
with a post graduate education) is 15% more likely (respectively 10% less likely) to
exhibit an inconsistent correlation.

6.2 Is the Magnitude of Adjustments Consistent with Theory?

We have just established that when respondents change their predictions from one survey
to the next, most of them also adjust their experimental choices in a direction consistent
with the theory. In this section, we explore whether the magnitude of the adjustments in
behavior may be rationalized under a simple expected utility model.

6.2.1 Methodology
Consider respondent i who selects a switching point , in survey 1 and , in survey 2.
We assume that this respondent possesses a specific power utility function over income
of the form
,
, with
0. The parameter , which is assumed to be time
1 the agent is
invariant, therefore characterizes the agents risk attitude. When 0
risk averse with a constant relative risk aversion parameter equal to 1
. The agent is
risk neutral when
1 and risk loving when
1 . Finally, we assume that the
generalized beta distribution fitted to the respondents reported probabilistic beliefs in a
given survey accurately represents the respondents true inflation beliefs at the time. The
expectation operator with respect to the distribution elicited in survey 1 (respectively
survey 2) is then denoted , . (respectively
, . ). Our approach to evaluate the
magnitude of respondent is behavior adjustments proceeds in three steps.
Step 1: The first step consists of inferring a range of risk attitude parameters ,
that
,
the
rationalizes the switching point , . More specifically, denote ,
respondents expected utility under investment A. Recall that the payments under
investment B increase in increment of $50 from $100 in question 1 to $550 in question 10.
We therefore denote
$50 1
the revenue (in nominal terms) generated by

22

A perfectly linear correlation should not be expected because the difference in switching points across
surveys is bounded between -10 and 10.

19

investment B in question 1, ,10 oftheexperiment. Furthermore, we assume that

0
$50 and 11
$600.23
In this framework, the switching point
,

actually selected by the agent should satisfy:

,

1 ,

In other words, given his probabilistic beliefs and his risk attitude parameter, the
respondents should have no strict incentive to switch earlier or later than he actually did
in the experiment.24
We can then define the pair
,
that satisfies:
,

and

1 ,

It must then be the case that, given the probabilistic beliefs expressed by the respondent,
can rationalize the choice of switching point , .
any risk attitude parameter ,
Finally, we denote

/2 the midpoint of the risk attitude parameters.

Step 2: In step 2, we use the pair

,
inferred in step 1 and the probabilistic beliefs
expressed by the respondent in survey 2 to predict a range of possible switching points
, , , . These predicted switching points are defined by the pair of inequalities:
,

As we shall see,
the pair

,
,
,

,
,

,
,

1 ,
1 ,

may be equal to 0, 1 or 2 depending on the values taken by

and by the respondents reported belief distribution in survey 2.

Step 3: The last step consists of comparing the predicted range of switching points
to , , the respondents actual choice of switching point in survey 2. More
, , ,
precisely, we calculate a new variable, the adjustment precision, as the minimum
distance between , and , , , . The adjustment precision is therefore an integer
between -10 and 10. Observe that an adjustment precision equal to 0 does not simply
imply that the respondent choice of switching point in survey 2 is consistent with our
model. Instead, it implies that the magnitudes of all adjustments the respondent made are
consistent with our model. In other words, an adjustment precision equal to 0 requires the
distribution of beliefs expressed in survey 1 and 2, as well the experimental choices made
in survey 1 and 2, to all be consistent with our simple expected utility model.

23

Assuming that $0
0
$50 and 11
$600 does not affect the results meaningfully.
To illustrate, consider a respondent who opts for investment A in the first question and then switches to
investment B for the remaining nine questions. In that case, we have ,
1, and this decision is rational
100,
,
150,
.
in our framework if is such that ,
,
,
24

20

6.2.2

Results

Step 1 of our methodology produces a specific pair of risk attitude parameters

for

each of the 502 repeat respondents. We find that the distribution of the mid-points, , of
these risk attitude parameters has a mean of 0.975 and a standard deviation of 0.342. This
distribution is consistent with the results presented in Section 5. In particular, the average
is close to 1, thereby reflecting the fact that, on average, respondents in survey 1 were
found to behave as if risk neutral (see Figure 7). Furthermore, the relatively large
standard deviation of is consistent with the substantial heterogeneity in revealed risk
attitudes illustrated in Figure 8.
Our methodology also produces a range of predicted switching points , , , for each
respondent. The width of this range is equal to either 0, 1 or 2 for respectively 30%, 64%
and 6% of the respondents. Now imagine the respondents in survey 2 made their choices
in the experiment randomly by assigning the same probability to each possible switching
point. In that case, we would find that on average 16% of the switching points chosen
randomly in step 2 fall within the range of predicted switching points , , , obtained
in step 1. This statistic will therefore serve as a benchmark to evaluate how well our
simple model explains the respondents experimental choices in survey 2.
We plot in Figure 11 the distribution of adjustment precisions in survey 2. This
distribution is highly concentrated around 0. In fact, 41.2% of our respondents have an
adjustment precision of 0, therefore clearly exceeding the 16% benchmark. Accounting
for possible noise (e.g. in reported inflation expectations or in making experimental
choices) we find that 63.7% (respectively 78.8%) of the respondents have an adjustment
precision in the interval [-1,1] (respectively [-2,2]). In other words, for most of the
respondents, the magnitude of behavior adjustment across surveys is consistent, or nearly
consistent, with our simple expected utility model. Several factors may explain why the
remaining respondents have large differences between predicted and actual switching
points: i) their behavior in the experiment is inconsistent with expected utility, ii) they are
expected utility maximizers, but their utility function is different than the one assumed in
our simple model, iii) they incorrectly report their inflation expectations, or iv) their
attitude toward risk changed between surveys.
To identify the characteristics of respondents whose prediction and switching point
adjustments cannot be explained by our model, we estimate an ordered probit model in
which the dependent variable is the absolute value of the adjustment precision. As we can
see in Table 5 (Model 3), the usual three variables seem to have statistical power in
explaining deviations from the pair of predicted switching points: low numeracy and
financial literacy, lower education, and high completion time. We also find that
respondents with an income greater than $75k are better able to adjust their behavior
across surveys. This result may be explained by the fact that these respondents are also
more likely to report the same prediction and the same switching point across surveys.

21

7. Discussion
In this paper, we compare the inflation expectations individual consumers reported in a
survey with their choices in a financially incentivized investment experiment. Our results
show that, on average, there is a tight correspondence between stated beliefs and behavior
in the experiment. Across respondents, we find a substantial amount of heterogeneity in
behavior that can be explained to a large extent by the respondents self reported risk
tolerances. Furthermore, when considering changes in beliefs for the same individual
over time, we find the adjustments in experimental behavior to be mostly consistent (both
in direction and magnitude) with standard economic theory. Finally, the respondents
whose behavior is difficult to rationalize (either because they switch multiple times in the
experiment, or because their behavior is inconsistent with theory) tend to exhibit specific
characteristics: they score lower on a numeracy and financial literacy scale, they are less
educated and they take longer to complete the survey. We now conclude with a brief
discussion of the implications of our results.
Historically, economists have been skeptical about expectations surveys (see Manski
2004 for a discussion). One of the main concerns is that, unlike choices, subjective
beliefs are unobservable. As a result, nothing guarantees that surveys can capture the true
beliefs of each respondent. By showing that the beliefs reported in an inflation survey are
informative, in the sense that they correlated with a decision with financial consequences,
our analysis provides evidence to support the empirical validity of inflation expectations
surveys. More generally, our study illustrates how controlled experimental methods can
be used to establish the information content of an expectations survey.
More importantly, by showing that respondents generally act on their stated inflation
beliefs in a way consistent with standard economic theory, we provide some evidence to
support one of the key assumptions underlying macroeconomic models whereby forward
looking agents make economic decisions that are influenced by their beliefs about future
inflation. The generality of this finding, however, should not be overstated for at least
two reasons. First, our surveys concentrate exclusively on households and therefore
ignore other important economic actors (e.g. firms, professional investors). Nevertheless,
we note that, on aggregate, households are a key component of economic activity with
considerable influence on aggregate demand and therefore equilibrium prices. Second,
our experimental results do not necessarily imply that the decisions made by consumers
in their daily life (e.g., about investments, savings, wages, or large durable purchases) are
systematically influenced by the beliefs they hold about future inflation. As argued in the
introduction, it is very difficult to test this hypothesis directly with real life observations
without other confounding factors (e.g. time discounting, liquidity constraints) playing a
substantial role. What we have shown is that, when presented with an opportunity to act
on their inflation beliefs, most consumers responding to a survey make decisions
consistent with the expectations they report. Our results are therefore a first step in
understanding how inflation prospects affect actual economic behavior.
More generally, this paper fits in a broader research agenda aimed at collecting new
survey data to better measure and understand how agents form and update their
subjective expectations about macroeconomic variables of interest (see, for instance,
Eusepi and Preston 2011, Branch 2004, Carvalho and Nechio 2012, Burke and Manz
22

2011). Agent heterogeneity also plays an increasingly large role in lifecycle models of
consumption and saving (e.g., Mankiw 2000, Kaplan and Violante 2012), in models of
labor, housing and credit market dynamics (Karahan 2011), and in models that study the
links between financial markets and real economic activity (Gertler and Kiyotaki 2010).
Furthermore, although economists implicitly assume that expectations influence behavior,
it is important to confirm empirically the extent to which this is the case in a variety of
contexts. For example, policy makers need to understand how the decision of households
to default on their home mortgages may depend on their expectations of future house
price appreciation or declines. Likewise, it is important to know the extent to which job
search efforts in the labor market are affected by workers expectations about their ability
to find jobs in various occupations or locations, as well as their expectations for wage
growth on a current job. As discussed earlier, this paper provides direct evidence at the
individual level that inflation expectations can affect economic decisions, which is part of
the micro-foundations of modern macroeconomic models.
Beyond testing the assumption that agents act on their beliefs, it is also important to
identify the characteristics of those who fail to do so. Indeed, this may be the only way
for policy makers to address the possible negative externalities generated by agents who
are ill-equipped to make sound economic choices. In that respect, our finding that
consumers with low numeracy and financial literacy are less likely to act on their
reported beliefs in accordance with expected utility theory is particularly relevant in the
context of the rapidly growing literature on the role of numeracy and financial literacy in
economic decisions. For instance, financial illiteracy has now clearly been linked with
lower wealth accumulation (Behrman et al. 2012), insufficient retirement planning
(Lusardi and Mitchell 2007), and inadequate portfolio diversification (Goetzmann and
Kumar 2008). In its 2009 report, the Presidents Advisory Council on Financial
Literacy identified numeracy and financial literacy as prime determinants of the 2008
subprime mortgage crisis. These views are supported by the results of Gerardi, Goette
and Meier (2011) who find a large negative correlation between numeracy and mortgage
delinquency in the U.S. Our study underlies the importance of financial education by
providing further evidence about the role played by financial literacy on sound economic
choices. Furthermore, by identifying a breakdown of the connection between beliefs and
actions, our results suggest a specific channel through which financial literacy affects
economic behavior.

23

REFERENCES
Adam K. and M. Padula (2011): Inflation Dynamics and Subjective Expectations in the
United States, Economic Inquiry, Vol. 49, No. 1, January, 1325.
Armantier O., Bruine de Bruin W., van der Klaauw W., Nelson S., Topa G. and B. Zafar
(2012): The Price is Right: Belief Updating in a Price Information Experiment,
mimeo, Federal Reserve Bank of New York.
Behrman J., Mitchell O., Soo C. and D. Bravo (2012): How Financial Literacy Affects
Household Wealth Accumulation, AER Papers and Proceedings, forthcoming.
Bernanke B. (2004): The Economic Outlook and Monetary Policy speech at the Bond
Market Association Annual Meeting, New York, New York.
Bernanke B. (2007): Inflation Expectations and Inflation Forecasting, speech at the
Monetary Economics Workshop of the National Bureau of Economic Research
Summer Institute, Cambridge, Massachusetts.
Bettinger E. and R. Slonim (2007): Patience Among Children, Journal of Public
Economics, vol.91:1-2, pp. 343-363.
Bonin H., Dohmen T., Falk A., Huffman D. and U. Sunde (2007): Cross-Sectional
Earnings Risk and Occupational Sorting: The Role of Risk Attitudes, Labour
Economics, 14, 926937.
Branch W. (2004): The Theory of Rationally Heterogeneous Expectations: Evidence
from Survey Data on Inflation Expectations, Economic Journal, Vol. 114, 592-621.
Bruine de Bruin W., Downs J.S., Fischhoff B. and C. Palmgren (2007): Development
and Evaluation of an HIV/AIDS Knowledge Measure for Adolescents Focusing on
Misunderstood Concepts, HIV/AIDS Prevention in Children and Youth, 8, 35-57.
Bruine de Bruin W., Potter S., Rich R., Topa G. and W. van der Klaauw (2010a):
Improving Survey Measures of Household Inflation Expectations, Current Issues in
Economics and Finance, Vol. 16(7), Aug/Sept.
Bruine de Bruin W., van der Klaauw W., Downs J.S., Fischhoff B., Topa G. and O.
Armantier (2010b): Expectations of Inflation: The Role of Financial Literacy and
Demographic Variables, Journal of Consumer Affairs, 44, 381-402.
Burke M, and M. Manz (2011): Economic Literacy and Inflation Expectations:
Evidence from a Laboratory Experiment, Public Policy Discussion Paper No. 11-8,
Federal Reserve Bank of Boston.
Caliendo M., Fossen M. and A. Kritikos (2010): The Impact of Risk Attitudes on
Entrepreneurial Survival, Journal of Economic Behavior & Organization, Volume
76, Issue 1, Pages 45-63.
Camerer C. and T-H. Ho (1994): Violations of the Betweenness Axiom and
Nonlinearity in Probability, Journal of Risk and Uncertainty, 8, 167196.
Capistran C. and A. Timmermann (2009): Disagreement and Biases in Inflation
Expectations, Journal of Money, Credit and Banking, 41, 365396.
Carmines E. and R. Zeller (1991): Reliability and Validity Assessment, Thousand Oaks,
CA: Sage Publications.
Carroll C. (2003): Macroeconomic Expectations of Households and Professional
Forecasters, Quarterly Journal of Economics, CXVIII, 269298.
Carvalho C. and F. Nechio (2012): Do People Understand Monetary Policy? mimeo.

24

Dohmen T., Falk A., Huffman D., Sunde U., Schupp J. and G. Wagner (2011):
Individual Risk Attitudes: Measurement, Determinants, and Behavioral
Consequences, Journal of the European Economic Association, Vol. 9, 3, 522550.
Dohmen T., Khamis M. and H. Lehmann (2010): Risk Attitudes and the Incidence of
Informality among Workers: Evidence from a Transition Country, ESCIRRU
Working Paper No. 22, Berlin.
Dominitz J. and C. Manski (2007): Expected Equity Returns and Portfolio Choice:
Evidence from the Health and Retirement Study, Journal of the European Economic
Association, Vol. 5 no2-3, 369379.
Eckel E. and R. Wilson (2004): Is Trust a Risky Decision? Journal of Economic
Behavior & Organization, 55, pp 447-465.
Ehrbeck T. and R. Waldmann (1996): Why are Professional Forecasters Biased?
Agency versus Behavioral Explanations, Quarterly Journal of Economics, 111.
Engelberg J., Manski C. and J. Williams (2009): Comparing the Point Predictions and
Subjective Probability Distributions of Professional Forecasters, Journal of Business
and Economic Statistics, 27, pp. 30-41.
Erceg C. and A. Levin (2003): Imperfect Credibility and Inflation Persistence, Journal
of Monetary Economics, 50, 915-944.
Eusepi S. and B. Preston (2011): Expectations, Learning and Business Cycle
Fluctuations, American Economic Review, Vol. 101, No. 6.
Evans G. and S. Honkapohja (2001): Learning and Expectations in Macroeconomics,
Princeton University Press, Princeton, NJ.
Gali, J. (2008): Monetary Policy, Inflation and the Business Cycle: An Introduction to the
New Keynesian Framework, Princeton University Press.
Gerardi K., Goette, L. and S. Meier (2011): Financial Literacy and Subprime Mortgage
Delinquency: Evidence from a Survey Matched to Administrative Data, Federal
Reserve Bank of Atlanta, working paper.
Gertler M. and N. Kiyotaki (2010): Financial Intermediation and Credit Policy in
Business Cycle Analysis, Handbook of Monetary Economics, Vol. 3, P 547599.
Goetzmann W. and A. Kumar (2008): Equity Portfolio Diversification, Review of
Finance, Vol. 12, 433-463.
Gollier C. and J. Pratt (1996): Weak Proper Risk Aversion and the Tempering Effect of
Background Risk, Econometrica, 64, 1109-1123.
Harrison G., Lau M. and M. Williams (2002): Estimating Individual Discount Rates in
Denmark: A Field Experiment, American Economic Review 92, 16061617.
Holt C. and S. Laury (2002): Risk Aversion and Incentive Effects, American Economic
Review, 2002, 92(5), pp. 164455.
Inoue A., Kilian L. and F-C. Kiraz (2009): Do Actions Speak Louder Than Words?
Household Expectations of Inflation Based on Micro Consumption Data, Journal of
Money, Credit and Banking, 41, 13311363.
Jewitt I. (1989): Choosing Between Risky Prospects: The Characterization of
Comparative Statics Results, and Location Independent Risk, Management Science,
35: 60-70.
Karahan F. (2011): Geographical Reallocation and Unemployment during the Great
Recession: The Role of the Housing Bust, mimeo.

25

Karni E. and Z. Safra, (1995): The Impossibility of experimental Elicitation of

Subjective Probabilities, Theory and Decision, 38, 313-320.
Kaplan G. and G. Violante (2012): A Model of the Consumption Response to Fiscal
Stimulus Payments, NYU Working Paper.
Keane M. and D. Runkle (1990): Testing the Rationality of Price Forecasts: New
Evidence from Panel Data, American Economic Review, Vol. 80, No. 4, pp. 714-735.
Kbberling V., Schwieren C. and P. Wakker (2007): Prospect-Theorys Diminishing
Sensitivity versus Economics Intrinsic Utility of Money: How the Introduction of the
Euro Can Be Used to Disentangle the Two Empirically, Theory and Decision, 63.
Laster D., Bennett P. and I. Geoum (1999): Rational Bias in Macroeconomic Forecasts,
The Quarterly Journal of Economics, 114:293-318.
Lipkus I., Samsa G. and B. Rimer (2001): General Performance on a Numeracy Scale
Among Highly Educated Samples, Medical Decision Making, 21, 374.
Lucas R. (1972): Expectations and the Neutrality of Money, Journal of Economic
Theory, 4, 103-124.
Lusardi A. (2007): Financial Literacy: An Essential Tool for Informed Consumer
Choice? presented at the conference Understanding Consumer Credit: A National
Symposium on Expanding Access, Informing Choices, and Protecting Consumers,
organized by the Harvard Joint Center for Housing Studies, Harvard University.
Lusardi A. and O. Mitchell (2007): Financial Literacy and Retirement Preparedness.
Evidence and Implications for Financial Education, Business Economics, Jan, 35-44.
Mackowiak B. and M. Wiederholt (2009): Optimal Sticky Prices under Rational
Inattention, American Economic Review, 99(3), 769-80.
Mankiw G. (2000): The Savers-Spenders Theory of Fiscal Policy, American Economic
Review, Vol. 90, No. 2, 120-125.
Mankiw G. and R. Reis (2002): Sticky Information versus Sticky Prices: a Proposal to
Replace the New Keynesian Phillips Curve, Quarterly Journal of Economics, 117,
1295-1328.
Mankiw G., Reis R. and J. Wolfers (2003): Disagreement about Inflation Expectations,
NBER Macroeconomics Annual.
Manski C. (2004): Measuring Expectations, Econometrica, Vol. 72, No. 5, 13291376.
Manski C. (2006): Interpreting the Predictions of Prediction Markets, Economics
Letters, 91, 3, pp 425-29.
Muth J. (1961): Rational Expectations and the Theory of Price Movements,
Econometrica, 29, 315-335.
Nunes R. (2010): Inflation Dynamics: The Role of Expectations, Journal of Money,
Credit and Banking, Volume 42, Issue 6, pages 11611172.
Ottaviani M. and P. Srensen (2006): The Strategy of Professional Forecasting,
Journal of Financial Economics, 81, 441-466.
de Paula A., Shapira G. and P. Todd (2010): How Beliefs about HIV Status Affect
Risky Behaviors: Evidence from Malawi, PIER Working Paper 10-023, University
of Pennsylvania.
Pesaran M.H. and M. Weale (2006): Survey Expectations, in Handbook of Economic
Forecasting. Graham Elliott, Clive W.J. Granger, Allan Timmermann.
Pratt J. (1964): Risk Aversion in the Small and in the Large, Econometrica, Vol. 32,
No. 1/2, pp. 122-136.
26

Roberts J. (1995): New Keynesian Economics and the Phillips Curve, Journal of
Money, Credit and Banking, 27, 975984.
Rothschild M. and J. Stiglitz (1970): Increasing Risk: I. A Definition, Journal of
Economic Theory, 2, 225-243.
Savage L.J. (1971): Elicitation of Personal Probabilities and Expectation, Journal of
the American Statistical Association, 66, 783-801.
Sargent T. (1999): The Conquest of American Inflation. Princeton University Press.
Shiller R. (1997): Why Do People Dislike Inflation? in Reducing Inflation: Motivation
and Strategy, Chicago: University of Chicago Press, pp. 13-65.
Sims C. (2006): Rational Inattention: Beyond the Linear-Quadratic Case, American
Economic Review, 96 (2), 158163.
Sims C. (2009): Inflation Expectations, Uncertainty and Monetary Policy, BIS
Working Paper No. 275.
van der Klaauw W. (2012): On the Use of Expectations Data in Estimating Structural
Dynamic Choice Models, Journal of Labor Economics, Vol. 30(3), 521-554.
Wolpin K. and F. Gonul (1985): On the Use of Expectations Data in Micro Surveys:
The case of Retirement. Report on the Employment and Training Administration,
U.S. Department of Labor, Washington, DC, March.
Wolpin K. and W. van der Klaauw (2008): Social Security and the Savings and
Retirement Behavior of Low Income Households, Journal of Econometrics, Vol.
145(1-2), 21-42
Woodford M. (2004): Inflation Targeting and Optimal Monetary Policy, Federal
Reserve Bank of St. Louis Economic Review, 86(4), 15-41.
Woodford M. (2005): Central Bank Communication and Policy Effectiveness,
Proceedings, Federal Reserve Bank of Kansas City, August issue, 399-474.

27

Table1: Descriptive Statistics

All
Data

Group 1

Group 2

Group 3

Group 4

Probit a

Missing Data for Point

Prediction or Experiment

Non-Rationalizable b
Experimental Choices

Rationalizable
Experimental Choices

Rationalizable Choices
in Both Surveys

Probability to
be in Group 4

-0.001
(0.001)
-0.043
(0.032)
0.051
(0.048)
0.017
(0.038)
0.032
(0.031)
0.058***
(0.010)
0.011
(0.010)

Age

52.126
(14.049)

51.765
(16.794)

51.291
(13.657)

52.519
(13.736)

52.592
(13.422)

Gender (female)

57.3%

73.0%

68.2%

54.4%

52.6%

15.1%

17.4%

18.6%

14.0%

12.8%

20.9%

23.5%

13.25%

21.6%

22.5%

41.8%

23.5%

29.1%

44.9%

47.4%

4.502
(1.481)
3.335
(1.586)
5.382
(6.903)
1,479
(100%)

3.3487
(1.692)
3.601
(1.823)
9.394 e
(12.842)
115
(7.8%)

3.629
(1.553)
3.308
(1.745)
8.705
(6.387)
151
(10.2%)

4.746
(1.335)
3.368
(1.567)
4.777
(3.190)
1,213
(82.0%)

4.986
(1.234)
3.403
(1.548)
4.563
(3.167)
1,004
(67.9%)

Education: No more
than High School
Education:
More than Bachelor
Income
greater than $75k
Numeracy and
Financial Literacy c
Reported Risk
Tolerance d
Point prediction
Number of
Observations
a

__
1,479
2

Marginal effects are reported. The endogenous variable equals one when a respondent is in group 4. The log Likelihood is -269.79. The pseudo R is 0.080.
A respondent is said to make non rationalizable choices when he switches more than once from investment A to investment B in the ascending scale treatment.
c
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.
d
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
e
This statistic is based on 85 observations.
b

Table 2: Earnings under investment A

Rate of
inflation

-1%
or less
(deflation)

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%
or
more

Earnings

$600

$550

$500

$450

$400

$350

$300

$250

$200

$150

$100

$50

Table 3: Factors Influencing the Choice of the Switching Point

Ordered probit estimations where the dependent variable is the switching point
(an integer between 0 and 10) for each respondent

Point prediction
Estimated Expected
Prediction d

Model 1
Survey 1 Survey 2
-0.108*** -0.104***
(0.009)
(0.008)

-0.096***
(0.010)
0.165***
(0.028)
0.011
(0.032)
-0.007***
(0.001)
-0.014
(0.019)
0.001
(0.003)
-0.126
(0.088)
-0.059
(0.087)
0.036
(0.128)
0.081
(0.105)
-0.069
(0.121)
0.034
(0.118)
0.096
(0.119)

-0.108
(0.120)
0.026
(0.118)
0.104
(0.116)

0.184
(0.029)
0.001
(0.032)
-0.004***
(0.001)
-0.008
(0.019)
0.001
(0.003)
-0.094
(0.088)
-0.095
(0.087)
0.052
(0.127)
0.147
(0.105)
-0.108
(0.120)
0.033
(0.118)
0.091
(0.119)

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

-0.018
(0.027)

-0.026
(0.023)

589
-1,227.0

607
-1,266.9

589
-1,225.5

607
-1,265.3

589
-1,232.9

607
-1,265.2

589
-1,224.4

607
-1,277.7

***

__
***

Age

__

__

Gender (female)

__

__

Income greater than $75k

__

__

__

__

-0.130
(0.119)
0.050
(0.119)
-0.007
(0.118)

Education: No more
than High School
Education:
More than Bachelor
Prices *
Ascending
Prices *
Descending
Inflation *
Descending
Point prediction * Prices
* Ascending
Point prediction * Prices
* Descending
Point prediction *
Inflation *
Descending
N
Log Likelihood

__

0.191
(0.030)
-0.002
(0.035)
-0.007***
(0.002)
0.008
(0.024)
0.003
(0.003)
-0.060
(0.089)
-0.027
(0.106)
0.124
(0.125)
0.078
(0.106)
-0.129
(0.119)
0.053
(0.116)
-0.011
(0.118)

__

Log of Time Taken to

Complete the Survey

Inflation Uncertainty c

__

Model 4
Survey 1 Survey 2
-0.065*** -0.091***
(0.017)
(0.016)

-0.090
(0.012)
0.163***
(0.030)
-0.001
(0.035)
-0.005**
(0.002)
0.009
(0.024)
0.004
(0.003)
-0.101
(0.088)
0.046
(0.089)
-0.001
(0.124)
0.043
(0.106)
-0.146
(0.121)
-0.013
(0.116)
0.039
(0.118)

__

0.183
(0.028)
0.005
(0.032)
-0.004***
(0.001)
-0.012
(0.019)

Numeracy and Financial

Literacy Score b

Model 3
Survey 1 Survey 2
***

0.188
(0.029)
0.003
(0.035)
-0.007***
(0.002)
0.007
(0.024)

Reported Risk Tolerance a

Model 2
Survey 1 Survey 2
-0.108*** -0.104***
(0.009)
(0.009)
__
***

***

__

__
***

0.193
(0.030)
-0.002
(0.035)
-0.007***
(0.002)
0.008
(0.024)
0.003
(0.003)
-0.063
(0.089)
-0.035
(0.089)
0.128
(0.125)
0.077
(0.107)
-0.022
(0.152)
0.150
(0.157)
0.056
(0.148)
-0.028
(0.025)
0.025
(0.024)

0.183***
(0.029)
-0.001
(0.031)
-0.004**
(0.001)
-0.007
(0.019)
0.001
(0.003)
-0.094
(0.088)
-0.099
(0.088)
0.070
(0.127)
0.177
(0.105)
0.045
(0.170)
0.031
(0.152)
0.199
(0.165)
-0.033
(0.026)
-0.002
(0.022)

The standard deviations are robust and clustered at the treatment combination level. Significance: * = 10%, ** =5%, *** = 1%.
a
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
b
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.
c
As explained in Section 2.3, this variable is equal to the standard deviation of the Beta distribution fitted to the respondents reported probabilistic beliefs.
d
As explained in Section 2.3, this variable is equal to the mean of the Beta distribution fitted to the respondents reported probabilistic beliefs.

Table 4: Explaining the Deviations from Risk Neutrality

Ordered Probit estimations based on the difference between a respondent
switching point and her/his pair of risk neutral switching points.

Reported Risk Tolerance

Square of Reported
Risk Tolerance
Inflation Uncertainty g
Numeracy and Financial
Literacy Score f
Log of Time Taken to
Complete the Survey
Gender (female)
Age
Income greater than $75k
Education: No more
than High School
Education:
More than Bachelor
Prices *
Ascending
Prices *
Descending
Inflation *
Descending
N
Log Likelihood

Model 1 a

Model 2 b

Model 3 c

Model 4 d

Dependent = Deviation

Dependent = Absolute
Deviation

Dependent = Deviation
when Deviation < 0

Dependent = Deviation
when Deviation > 0

Survey 1
0.247***
(0.030)

Survey 2
0.196***
(0.028)

Survey 1
0.258***
(0.056)

Survey 2
0.220***
(0.053)

Survey 1
0.356***
(0.059)

Survey 2
0.217***
(0.059)

__

__

Survey 1
-0.550***
(0.128)
0.077***
(0.017)
0.004**
(0.002)
-0.146***
(0.036)
0.075***
(0.024)
0.043
(0.091)
-0.001
(0.003)
-0.125
(0.092)
0.535***
(0.128)
-0.273**
(0.112)
0.106
(0.127)
0.022
(0.127)
0.002
(0.123)
589
-1,030.1

__

__

__

__

0.002
(0.002)
-0.034
(0.035)
-0.036
(0.025)
0.101
(0.087)
0.004
(0.003)
-0.092
(0.088)
0.298**
(0.125)
0.019
(0.105)
0.003
(0.121)
0.180
(0.120)
-0.138
(0.118)
589
-1,295.6

0.001
(0.011)
-0.057*
(0.031)
-0.029
(0.019)
0.011
(0.088)
0.001
(0.003)
-0.143
(0.087)
0.085
(0.126)
0.102
(0.105)
-0.039
(0.122)
0.164
(0.119)
-0.173
(0.118)
607
-1,259.7

Survey 2
-0.543***
(0.124)
0.071***
(0.017)
0.004***
(0.001)
-0.118***
(0.033)
0.070***
(0.020)
0.007
(0.093)
0.004
(0.003)
-0.048
(0.092)
0.591***
(0.129)
-0.350***
(0.116)
0.086
(0.128)
-0.001
(0.126)
-0.084
(0.125)
607
-1,008.2

***

-0.011
(0.004)
0.234***
(0.059)
-0.097**
(0.036)
0.206
(0.148)
0.001
(0.006)
0.185
(0.149)
-0.467**
(0.218)
0.463**
(0.187)
-0.110
(0.206)
-0.086
(0.216)
-0.141
(0.194)
242
-369.8

**

-0.003
(0.002)
0.191***
(0.054)
-0.064**
(0.031)
-0.071
(0.152)
0.006
(0.006)
0.306**
(0.152)
-0.666***
(0.203)
0.487**
(0.198)
-0.115
(0.211)
-0.066
(0.214)
-0.078
(0.197)
134
-204.1

***

0.010
(0.003)
-0.175***
(0.068)
0.087*
(0.048)
-0.014
(0.184)
0.005
(0.007)
-0.131
(0.184)
0.820***
(0.217)
-0.575**
(0.257)
0.088
(0.255)
0.021
(0.251)
-0.230
(0.268)
161
-255.4

0.008***
(0.003)
-0.239***
(0.067)
0.091**
(0.041)
0.236
(0.206)
0.011
(0.007)
-0.106
(0.205)
0.950***
(0.244)
-0.831***
(0.287)
0.330
(0.284)
-0.286
(0.270)
-0.422
(0.310)
134
-213.9

The standard deviations are robust and clustered at the treatment combination level. Significance: * = 10%, ** =5%, *** = 1%.
a
In Model 1 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points.
b
In Model 2 the dependent variable is the absolute value of the difference between a respondent actual switching point and his/her pair of risk neutral switching
points.
c
In Model 3 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points, but the
sample is restricted to respondents who behaved as if risk averse (i.e. the deviations from risk neutrality is strictly negative).
d
In Model 4 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points, but the
sample is restricted to respondents who behaved as if risk loving (i.e. the deviations from risk neutrality is strictly positive).
e
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
f
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.
g
As explained in Section 2.3, this variable is equal to the standard deviation of the Beta distribution fitted to the respondents reported probabilistic beliefs.

Table 5: Are Changes in Predictions and Changes in

Switching Points across Surveys Consistent with Theory?

Absolute Difference in Reported Risk

Tolerance across Surveys d
Absolute Difference in Inflation
Uncertainty across Surveys e
Numeracy and Financial Literacy Score f
Log of Total Time Taken to
Complete the Two Surveys
Gender (female)
Age
Income greater than $75k
Education: No more
than High School
Education:
More than Bachelor
Prices *
Ascending
Prices *
Descending
Inflation *
Descending
N
Log Likelihood

Model 1 a

Model 2 b

Model 3 c

Probit (Marginal Effects)

Dependent = Inconsistent
Revealed Risk Attitudes

Probit (Marginal Effects)

Dependent =
Inconsistent Correlation

Ordered Probit
Dependent = Absolute value of
Adjustment Precision

-0.009
(0.014)
0.001
(0.001)
-0.048***
(0.014)
0.013**
(0.005)
-0.026
(0.024)
-0.001
(0.001)
-0.009
(0.025)
0.153***
(0.060)
-0.052
(0.027)
-0.026
(0.021)
0.025
(0.032)
-0.028
(0.035)
495
-88.5

0.032
(0.022)
0.001
(0.001)
-0.054***
(0.014)
0.027***
(0.007)
-0.062
(0.073)
-0.001
(0.001)
-0.054
(0.036)
0.150***
(0.064)
-0.096**
(0.038)
-0.054
(0.047)
-0.035
(0.049)
-0.020
(0.048)
495
-222.1

-0.027
(0.060)
0.001
(0.001)
-0.153***
(0.042)
0.054**
(0.021)
0.089
(0.100)
0.001
(0.004)
-0.203**
(0.101)
0.722***
(0.148)
-0.353***
(0.125)
-0.146
(0.144)
0.098
(0.141)
-0.105
(0.137)
495
-791.5

The standard deviations are robust and clustered at the treatment combination level. Significance: * = 10%, ** =5%, *** = 1%.
a
A respondent is said to have inconsistent revealed risk attitudes when he behaves as if risk averse in one survey and as if risk loving in the other survey.
b
A respondent is said to have an inconsistent correlation when his pair of switching points and his pair of point predictions satisfy all the conditions described in
Section 6.1.2.
c
The adjustment precision is the difference between a respondents switching point in survey 2 and his pair of predicted switching points (see Section 6.2).
d
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
e
As explained in Section 2.3, this variable is equal to the standard deviation of the Beta distribution fitted to the respondents reported probabilistic beliefs.
f
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.

Figure1:DistributionofPointPredictions
40%

30%

20%

10%

0%
<0%

[0%,2"[

[2%,4%[

[4%,6%[

[6%,8%[

[8%,12%[

>12%

PointPrediction
Survey1(N=598)

Survey2(N=615)

Figure2:DistributionofIndividualDifferences
inPointPredictions(N=502)
30%
25%
20%
15%
10%
5%
0%

Figure3:HistogramofPointPredictions
foreachSurveyTreatment
40%
35%
30%
25%
20%
15%
10%
5%
0%
<0%

[0%,2"[

[2%,4%[

[4%,6%[

[6%,8%[

[8%,12%[

>12%

PointPrediction
"Prices"(Survey1)

"Inflation"(Survey1)

"Prices"(Survey2)

"Inflation"(Survey2)

Figure4:DistributionofSwitchingPoints
25%
20%
15%
10%
5%
0%
0

10

SwitchingPoint
Survey1(N=598)

Survey2(N=615)

Figure5:DistributionofIndividualDifferences
inSwitchingPoints(N=502)
25%

20%

15%

10%

5%

0%
<5

>5

Figure6:HistogramofSwitchingPointsforeachTreatmentCombination
25%

20%

15%

10%

5%

0%
0

10

SwitchingPoint
"Prices"*"Ascending"(Survey1)

"Prices"*"Descending"(Survey1)

"Inflation"*"Ascending(Survey1)

"Inflation"*"Descending"(Survey1)

"Prices"*"Ascending"(Survey2)

"Prices"*"Descending"(Survey2)

"Inflation"*"Ascending"(Survey2)

"Inflation"*"Descending"(Survey2)

Figure7:ChoicesandPredictions
11%
10%
9%
8%

PointPrediction

7%
6%
5%
4%
3%
2%
1%
0%
0

10

1%

SwitchingPoint
AveragePredictionSurvey1(N=598)
LinearTrendLineSurvey1

AveragePredictionSurvey2(N=615)
LinearTrendlineSurvey2

RiskNeutralBand

Figure8:ChoicesandDispertionofPredictions(N=502)
25

20

Point Prediction

15

10

10

SwitchingPoint

10

15

AveragePredictionSurvey1(N=598)

AveragePredictionSurvey2(N=615)

RiskNeutralBand

Figure9:DistributionofRevealed
RiskAttitudesineachSurvey(N=502)
70%
60%

50%
40%

30%
20%

10%
0%

AsifRAinSurvey2

AsifRNinSurvey2

AsifRAinSurvey1

AsifRNinSurvey1

AsifRLinSurvey2
AsifRLinSurvey1

Figure10:PredictionandBehaviorAdjustment(N=502)
10

54"inconsistentcorrelations"
are inthisquadrant
5

PredictionDifference
0
30

20

10

"Indifference"Zone
5

48"inconsitentcorrelations"
are inthisquadrant
10

SwitchingPoint
Difference

10

20

30

Figure11:AdjustmentPrecisioninSurvey2
50%

40%

30%

20%

10%

0%
10

DifferencebetweenActualandClosestPredictedSwitchingPointinSurvey2

10

Appendix A: Experimental Instruction (Ascending Scale Treatment)

You can earn extra money by answering the following 10 questions. In each
question, you are asked to choose between 2 investments, investment A and
investment B.
If you choose investment A, then how much you earn depends on what the rate
of inflation will be over the next 12 months. Your earnings under
investment A depending on the rate of inflation are summarized in the table
below:
Earnings under investment A
Rate of
inflation

-1%
or less
(deflation)

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%
or more

Earnings

$600

$550

$500

$450

$400

$350

$300

$250

$200

$150

$100

$50

For example, we can see in the table that your earnings under investment A
will be $50 if the rate of inflation over the next 12-months is 10% or more.
Alternatively, your earnings under investment A will be $600 if the rate of
inflation over the next 12-months is -1% or less (deflation).
If you choose investment B, then how much you earn will not depend on the
rate of inflation. Exactly how much you earn under investment B will be
specified in each of the 10 questions below.

Once the survey is completed, we will randomly pick 1 of the 10

questions, and 2 survey participants. Twelve months from now,
these 2 participants will be paid extra money according to the
investment choice they made for the selected question. So answer
every question carefully, as you may earn up to several hundred
dollars. For investment A, the inflation rate over the next 12
months will be based on the official U.S. CPI index (Consumer
Price Index) and it will be rounded to the nearest percentage
point.

Earnings under investment A

Rate of
inflation

-1%
or less
(deflation)

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%
or more

Earnings

$600

$550

$500

$450

$400

$350

$300

$250

$200

$150

$100

$50

For every question, please choose between investment A and investment B.

Question 1: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $100?
Question 2: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $150?
Question 3: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $200?
Question 4: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $250?
Question 5: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $300?
Question 6: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $350?
Question 7: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $400?
Question 8: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $450?
Question 9: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $500?
Question 10: Which one of these two investments do you choose?
( ) Investment A: your earnings are determined by the table above.
( ) Investment B, your earnings are exactly $550?

Appendix B: Demonstrations of the Propositions in Section 2

Proposition 1: If investment A and investment B have the same expected return then a riskaverse agent prefers investment B to investment A.
Proof: We consider a standard expected utility framework. The agents utility function over income,
denoted
. , is thrice differentiable, strictly increasing, and satisfies the von Neumann
Morgenstern axioms. Assume that the agents beliefs are such that the two investments have the
same expected returns, so that
or equivalently
. In that case, we
have
1
and the earnings under investment A are then
1
.
Let

with

1
1 . Note also that

0 . Observe that

1
1

while

Now, let

1
so that

1
. When
. is strictly concave,
0, which implies that
0 for all risk averse agents. Then,
because
1, we have 1
or equivalently
. Conversely,
0
when . is strictly convex, so that
0 for all risk loving agents, in which case 1
.
or equivalently
Proposition 2: If a risk averse agent is indifferent between investment A and investment B,
then, all else equal, a more risk averse agent (in the classical sense of Pratt 1964) prefers
investment B to investment A.
Proof: Consider two risk-averse agents, the first with a utility function . and the second with a
utility function . . The second agent is more risk-averse (in the classical sense of Pratt 1964) if
his utility function verifies .

. with .
0 and .
0. Assume the two agents
share the same beliefs about the distribution of the random variable X defined above. Let us also
. and
. (respectively
. and
. ) the cumulative and probability distribution
denote
functions associated with investment A (respectively investment B). To simplify, we assume that
the support of investments A and B is the real line.
Proposition 2 may then be written

. 2.1

Based on Theorem 1 of Jewitt (1989), the implication in (2.1) is satisfied when

satisfy the single crossing property:

. and

. 2.2

Observe that
.
Thus, when
exists and

0 because

we have
, or equivalently
1 and
0. Now observe that

. Thus

.
. The cumulative distributions
. and
. therefore
Because
1 we have
satisfy the single crossing property in (2.2), which implies that the implication in (2.1) is also
satisfied.
Proposition 3: If a risk-averse agent with a HARA utility is indifferent between investment A
and investment B, then the agent prefers investment B to investment A for any increase in
risk (in the classical sense of Rothschild and Stiglitz 1970).
Proof: Consider a risk-averse agent with a utility function . . The agent initially believes that
investments A and B are characterized by a random variable X. For this distribution of beliefs,
assume that the agent is indifferent between investment A and investment B, so that
or equivalently
. Now, assume that the agent faces an increase
in risk (in the classical sense of Rothschild and Stiglitz 1970), that is, he now believes that
investments A and B are characterized by a random variable
, where is a mean zero
random variable.
Let us denote
be written

and

the investments that earn respectively

and . Proposition 3 may then

. 3.1

Let us also denote the investment that earns

show that

.
Let
Furthermore, note that

Now, let
concave then
1, we have

0. Observe that

with

. To prove Proposition 3, let us first

so that
0, which implies that
or equivalently

and

1 .

. When
. is strictly
0 for all risk averse agents. Then, because

Next, consider the following implication:

. 3.2

Because

, it is sufficient to verify that the implication in (3.2) is satisfied in

order to demonstrate Proposition 3 as stated in (3.1).
Let us denote the indirect utility function
be written

. The implication in (3.2) can then

. 3.3

From Proposition 2 we know that the implication in (3.3) is satisfied when . is more risk averse
than . , that is, .

. with .
0 and .
0. From Gollier and Pratt (1996),
. is more risk averse than . if . satisfies the vulnerability condition. This condition,
which imposes constraints up to the fourth derivative, is not satisfied by every risk averse utility
function. Nevertheless, Gollier and Pratt (1996) show that the vulnerability condition is satisfied by
the family of HARA utility functions.

Appendix C: Additional Figures

FigureC1:ChoicesandMedianPredictions
11%
10%

ReportedInflationExpectation

9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
0

10

1%

SwitchingPoint
MedianPredictionSurvey1(N1=598)

MedianPredictionSurvey2(N2=615)

FigureC2:ChoicesandEstimatedExpectedPredictions
11%
10%

ReportedInflationExpectation

9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
1%

SwitchingPoint
ExpectedPredictionSurvey1(N1=598)

ExpectedPredictionSurvey2(N2=615)

10

FigureC3:ChoicesandPredictions
"Prices"*"Ascending"
12%
11%

ReportedInflationExpectation

10%
9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
1% 0

10

2%
3%
4%

SwitchingPoint
AveragePredictionSurvey1(N1=144)

AveragePredictionSurvey2(N2=144)

FigureC4:ChoicesandPredictions
"Prices"*"Descending"

12%
11%
10%

ReportedInflationExpectation

9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
1% 0

2%
3%
4%

SwitchingPoint
AveragePredictionSurvey1(N1=145)

AveragePredictionSurvey2(N2=154)

10

FigureC5:ChoicesandPredictions
"Inflation"*"Ascending"

12%
11%
10%

ReportedInflationExpectation

9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
1% 0

10

2%
3%
4%

SwitchingPoint
AveragePredictionSurvey1(N1=160)

AveragePredictionSurvey2(N2=165)

FigureC6:ChoicesandPredictions
"Inflation"*"Descending"
12%
11%
10%

ReportedInflation Expectation

9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
1%

2%
3%
4%

Numberof"A"choices
AveragePredictionSurvey1(N1=149)

AveragePredictionSurvey2(N2=152)

10

Appendix D: Linear Regressions

Table D.3: Factors Influencing the Choice of the Switching Point
Linear regressions where the dependent variable is the switching point
(an integer between 0 and 10) for each respondent

Point prediction

Model 1
Survey 1 Survey 2
-0.182*** -0.213***
(0.020)
(0.017)

Model 2
Survey 1 Survey 2
-0.182*** -0.213***
(0.020)
(0.018)

Model 3
Survey 1 Survey 2
__

__
-0.226***
(0.019)
0.376***
(0.064)
0.040
(0.072)
-0.013***
(0.003)
-0.032
(0.044)
-0.001
(0.007)
-0.307
(0.202)
-0.117
(0.200)
0.074
(0.290)
0.248
(0.241)
-0.208
(0.276)
0.032
(0.270)
0.161
(0.272)

__

__

__

__

Log of Time Taken to

Complete the Survey

0.436***
(0.065)
0.012
(0.078)
-0.015***
(0.004)
0.025
(0.054)

0.417***
(0.061)
0.042
(0.080)
-0.009***
(0.003)
-0.025
(0.043)

Age

__

__

Gender (female)

__

__

Income greater than $75k

__

__

__

__

-0.288
(0.269)
0.101
(0.269)
-0.052
(0.268)

-0.294
(0.272)
0.017
(0.268)
0.182
(0.268)

0.435***
(0.067)
0.010
(0.079)
-0.014***
(0.004)
0.029
(0.054)
0.005
(0.007)
-0.172
(0.201)
-0.056
(0.202)
0.245
(0.284)
0.172
(0.241)
-0.288
(0.270)
0.109
(0.273)
-0.063
(0.279)

0.417***
(0.064)
0.029
(0.072)
-0.009***
(0.003)
-0.017
(0.043)
-0.001
(0.007)
-0.180
(0.200)
-0.197
(0.198)
0.065
(0.288)
0.437
(0.239)
-0.292
(0.273)
0.038
(0.269)
0.153
(0.270)

-0.199***
(0.025)
0.379***
(0.068)
0.017
(0.081)
-0.011***
(0.004)
0.031
(0.055)
0.006
(0.007)
-0.284
(0.203)
-0.122
(0.204)
0.017
(0.287)
0.088
(0.246)
-0.354
(0.274)
0.060
(0.274)
-0.036
(0.272)

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

__

5.659***
(0.534)
589
0.221

5.327***
(0.478)
607
0.287

5.450***
(0.719)
589
0.218

5.450***
(0.672)
607
0.287

5.594***
(0.732)
589
0.192

5.714***
(0.680)
607
0.277

Estimated Expected
Prediction d
Reported Risk Tolerance a
Numeracy and Financial
Literacy Score b
Inflation Uncertainty c

Education: No more
than High School
Education:
More than Bachelor
Prices *
Ascending
Prices *
Decreasing
Inflation *
Decreasing
Point prediction * Prices
* Ascending
Point prediction * Prices
* Decreasing
Point prediction *
Inflation * Decreasing
Constant
N
Adjusted R2

Model 4
Survey 1 Survey 2
-0.188*** -0.180***
(0.024)
(0.031)
__

__

0.403***
(0.072)
0.023
(0.085)
-0.019***
(0.004)
-0.006
(0.060)
0.003
(0.008)
-0.325
(0.208)
-0.116
(0.208)
0.160
(0.310)
0.263
(0.246)
0.063
(0.375)
-0.249
(0.329)
-0.073
(0.331)
-0.078
(0.084)
0.038
(0.035)
-0.033
(0.040)
5.934***
(0.778)
589
0.240

0.418***
(0.064)
0.026
(0.071)
-0.010**
(0.003)
-0.014
(0.044)
0.001
(0.007)
-0.179
(0.201)
-0.203
(0.199)
0.092
(0.289)
0.426*
(0.239)
0.021
(0.371)
0.128
(0.341)
0.459
(0.343)
-0.057
(0.050)
-0.022
(0.045)
-0.068
(0.047)
5.295***
(0.686)
607
0.287

The standard deviations are robust and clustered at the treatment combination level. Significance: * = 10%, ** =5%, *** = 1%.
a
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
b
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.
c
As explained in Section 2.3, this variable is equal to the standard deviation of the Beta distribution fitted to the respondents reported probabilistic beliefs.
d
As explained in Section 2.3, this variable is equal to the mean of the Beta distribution fitted to the respondents reported probabilistic beliefs.

Table D.4: Explaining the Deviations from Risk Neutrality

Linear regressions based on the difference between a respondent
switching point and her/his pair of risk neutral switching points.

Reported Risk Tolerance e

Square of Reported
Risk Attitude
Inflation Uncertainty g
Numeracy and Financial
Literacy Score f
Log of Time Taken to
Complete the Survey
Gender (female)
Age
Income greater than $75k
Education: No more
than High School
Education:
More than Bachelor
Prices *
Ascending
Prices *
Decreasing
Inflation *
Decreasing
Constant
N
Adjusted R2

Model 1 a

Model 2 b

Model 3 c

Model 4 d

Dependent = Deviation

Dependent = Absolute
Deviation

Dependent = Deviation
when Deviation < 0

Dependent = Deviation
when Deviation > 0

Survey 1
-1.029***
(0.221)
0.146***
(0.029)
0.008***
(0.003)
-0.262***
(0.062)
0.139***
(0.043)
0.074
(0.156)
-0.001
(0.006)
-0.246
(0.175)
1.120***
(0.222)
-0.477**
(0.189)
0.240
(0.217)
0.086
(0.216)
-0.049
(0.210)
4.012***
(0.654)
589
0.172

Survey 1
0.361***
(0.073)

Survey 2
0.335***
(0.816)

Survey 1
0.495***
(0.080)

Survey 2
0.294***
(0.081)

__

__

__

__

Survey 1
0.651***
(0.074)

Survey 2
0.497***
(0.069)

__

__
***

0.015
(0.004)
-0.085
(0.088)
-0.072
(0.060)
0.271
(0.220)
0.009
(0.008)
-0.201
(0.222)
0.805**
(0.312)
-0.008
(0.266)
0.009
(0.308)
0.425
(0.306)
-0.282
(0.287)
-2.768***
(0.787)
589
0.140

-0.001
(0.003)
-0.115
(0.077)
-0.071*
(0.047)
0.052
(0.217)
-0.001
(0.008)
-0.295
(0.215)
0.250
(0.311)
0.251
(0.258)
-0.140
(0.301)
0.322
(0.295)
-0.427
(0.292)
1.121
(0.718)
607
0.083

Survey 2
-1.086***
(0.213)
0.143***
(0.030)
0.008***
(0.002)
-0.239***
(0.056)
0.130***
(0.034)
0.009
(0.175)
0.004
(0.003)
-0.097
(0.157)
1.217***
(0.226)
-0.588***
(0.189)
0.213
(0.218)
0.059
(0.214)
-0.077
(0.212)
3.391***
(0.595)
607
0.204

***

-0.016
(0.005)
0.307***
(0.080)
-0.119**
(0.049)
0.161
(0.201)
0.001
(0.008)
0.273
(0.205)
-0.705**
(0.310)
0.593**
(0.250)
0.078
(0.282)
0.033
(0.297)
0.292
(0.263)
-5.016***
(0.747)
242
0.252

**

-0.007
(0.003)
0.346***
(0.086)
-0.102**
(0.050)
-0.004
(0.241)
-0.007
(0.009)
0.458**
(0.242)
-1.013***
(0.332)
0.851**
(0.305)
-0.210
(0.332)
-0.102
(0.340)
-0.009
(0.313)
-4.694
(0.836)
134
0.270

***

0.014
(0.005)
-0.212**
(0.099)
0.131**
(0.073)
0.121
(0.268)
0.007
(0.010)
-0.142
(0.268)
1.236***
(0.317)
-0.819**
(0.357)
0.218
(0.372)
0.190
(0.363)
-0.168
(0.386)
0.868
(0.910)
161
0.331

0.011***
(0.004)
-0.322***
(0.092)
0.107*
(0.057)
0.241
(0.286)
0.013
(0.010)
-0.217
(0.285)
1.249***
(0.344)
-0.990***
(0.368)
0.431
(0.396)
-0.344
(0.375)
-0.523
(0.429)
1.632
(0.791)
134
0.309

The standard deviations are robust and clustered at the treatment combination level. Significance: * = 10%, ** =5%, *** = 1%.
a
In Model 1 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points.
b
In Model 2 the dependent variable is the absolute value of the difference between a respondent actual switching point and his/her pair of risk neutral switching
points.
c
In Model 3 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points, but the
sample is restricted to respondents who behaved as if risk averse (i.e. the deviations from risk neutrality is strictly negative).
d
In Model 4 the dependent variable is the difference between a respondent actual switching point and his/her pair of risk neutral switching points, but the
sample is restricted to respondents who behaved as if risk loving (i.e. the deviations from risk neutrality is strictly positive).
e
Self reported willingness to take risk regarding financial matters on a scale from 1 (Not willing at all) to 7 (very willing).
f
The variable takes integer values between 0 and 6 depending on the number of correct answers the respondent gave to the six questions asked to measure
numeracy and financial literacy.
g
As explained in Section 2.3, this variable is equal to the standard deviation of the Beta distribution fitted to the respondents reported probabilistic beliefs.