PreprintPDF Available

Do Expert Experience and Characteristics Affect Inflation Forecasts?

Abstract and Figures

Each person's characteristics may influence that person's behaviors and their outcomes. We build and use a new database to estimate experts' performance and boldness based on their experience and characteristics. We classify experts providing inflation forecasts based on their education, experience, gender, and environment. We provide alternative interpretations of factors affecting experts' inflation forecasting performance, boldness, and pessimism by linking behavioral economics, the economics of education, and forecasting literature. An expert with previous experience at a central bank appears to have a lower propensity for predicting deflation.
Content may be subject to copyright.
Bank of Israel Research Department
Do Expert Experience and Characteristics
Affect Inflation Forecasts?*
c
and Yossi Saadon
b
Shagi-, Makram El
a
Jonathan Benchimol
Discussion Paper 2020.11
October 2020
_______________________
Bank of Israel - http://www.boi.org.il
* This paper does not necessarily reflect the views of the Bank of Israel. The authors thank
Philippe Andrade, Lahcen Bounader, Itamar Caspi, Paul Fontanier, Xavier Gabaix, Paul
Goldsmith-Pinkham, Ori Heffetz, Ben Schreiber, Michel Strawczynski, Matthias Weber,
Osnat Zohar, and participants at the Bank of Israel and Tsinghua University research
seminars for their useful comments.
a Bank of Israel, Jerusalem, Israel.
Email: jonathan.benchimol@boi.org.il
b Center for Financial Development and Stability, Henan University, Kaifeng, China.
Email: makram.el-shagi@cfds.henuecon.education. Corresponding author.
c Bank of Israel, Jerusalem, Israel.
Email: yosis@boi.org.il
Any views expressed in the Discussion Paper Series are those of the authors
and do not necessarily reflect those of the Bank of Israel
ד"ת לארשי קנב ,רקחמה תביטח780 םילשורי91007
Research Department, Bank of Israel. POB 780, 91007 Jerusalem, Israel
Do Expert Experience and Characteristics Affect Inflation Forecasts?
Jonathan Benchimol, Makram El-Shagi and Yossi Saadon
Abstract
Each person's characteristics may influence that person's behaviors and their outcomes. We
build and use a new database to estimate experts' performance and boldness based on their
experience and characteristics. We classify experts providing inflation forecasts based on their
education, experience, gender, and environment. We provide alternative interpretations of
factors affecting experts' inflation forecasting performance, boldness, and pessimism by linking
behavioral economics, the economics of education, and forecasting literature. An expert with
previous experience at a central bank appears to have a lower propensity for predicting deflation.
םאה ןויסינה לש םינייפאמהו םיאזחה םיעיפשמ לע תויזחת היצלפניאה?
לומיש ןב ןתנוי ,םארקמ לא-יגאש וןודעס יסוי
ריצקת
םינייפאמה לש םיאזחה הלכלכה ימוחתב עםילול עיפשהל לע תויזחתה לעו םהלש תובוגתה יאנתל עקרה .
רמאמב-הדמע הז ונא םינוב דסמ םינותנ םישמתשמו וב ידכ ןוחבל תא העפשהה שיש ינייפאמללש ם םיאזחה
הראב"ב
1
לע קויד תדימ תויזחת היצלפניאה ו תדימ לעהזועתה םהלש, תא עובקל םאובב תיזחתה .ןיב
םינייפאמה שםתעפשה הנחבנןתינ תונמל תא םוחתהלכשהה ,ףקיה הלכשהה , השכרנ םהבש תודסומה
הלכשהה , ןויסינהיעוצקמה יללכה ,רדגמה ,הביבסה דועו .ונא םיעיצמ תויונשרפ לקויד תדימ יבג התיזחת
ו תדימהזועתה לש םיאזחה תועצמאב םהינייפאמ, ךות רושיק תורפסל ימוחתמ הלכלכה תיתוגהנתהה ,
תלכלכ הךוניח הויוזיח הילכלכ .ןיב םיאצממה הניחבב ולעש :םידיקפתל םימדוק ולניןויס לשיאזחה שי
העפשה לע קויד תדימ תיזחתה לעו תדימהזועתה ולש ןתמב תיזחת הנוש וזמ לש רתי םיאזחה ;יאזח םע
ויסינן הדובע קנבב יזכרמ הטונ תוחפ תוזחל היצלפד ;םיאזח תויזחתה קויד תדימש ,התוחפ םהלש םיטונ
קיספהל ןתמ תויזחת ךשמהב ;ונא םיהזמ יונישםי בקוידה תודימ לש הזועתהותויזחתה ינפל רבשמה
יסנניפה לש 2008 ולוירחא; ומכ ןכ ,אל ונאצמ רצוי רדגמשלדבה בלש הזועתהו קוידה תודימ תויזחתה.
1
ידכ גישהל תואצות תונימא ונרחב ןוחבל תא תולאש רקחמה לע הרא"ב ,םש שי תואמ םיאזח וחווידש תא םהיתויזחת ךשמב
הברה םינש.
1 Introduction
There has been increasing interest in professional economic forecasts in recent
decades, partly due to the increasing importance of expectations in modern work-
horse macroeconomic models. In a few of the most active …elds, authors have com-
pared private-sector and central bank professional forecasters (Romer and Romer,
2000; El-Shagi et al., 2016), and professional forecasts to market-implied forecasts
(Adeney et al., 2017; Benchimol and El-Shagi, 2020). Others have assessed whether
forecasts (or forecast spreads) incorporate information regarding macroeconomic
uncertainty (Bachmann et al., 2013; Bloom, 2014; Rossi and Sekhposyan, 2015).
While there is ample evidence regarding the factors driving the overall (relative)
performance of professional forecasts (Giacomini and Rossi, 2010; El-Shagi et al.,
2016), there is, to our knowledge, no assessment of the factors driving performance
at the individual level. Instead, studies looking at individual-level forecasts have
so far mostly looked at general time-series properties of individual forecasts and
their interaction with each other (Andrade and Le Bihan, 2013). The main reason
seems to be that there is no information regarding individual forecasters linked to
the available major forecast databases.
This paper overcomes this limitation. We combine a nowcast dataset for US
CPI in‡ation with clearly identi…ed forecasters with two new original databases
with–mostly web-sourced–detailed information about forecasters and institutions.
We contribute to the literature in several ways. First, we show that experts with
central bank experience are less likely to predict de‡ation. These experts are less
pessimistic, but this is mitigated when pessimism turns out to be justi…ed. Sec-
ond, we highlight the implications and nonlinearities of the role of experience and
traits in experts’forecasting performance and boldness. Third, we con…rm that
the in‡uence of experts’traits on forecasting performance and boldness changed
following the Global Financial Crisis (GFC). Fourth, we show that underperform-
ing experts are less likely to survive in our expert database, while boldness does
not signicantly in‡uence this survival rate.
A deeper understanding of professional forecasts is crucial for three reasons.
First, economic agents, including policymakers such as central banks, strongly rely
on forecasts (Piotroski and Roulstone, 2004). Since expert forecasts are generally
better than market-based forecasts (Adeney et al., 2017; Benchimol and El-Shagi,
2020), the economic agents mostly rely on the average of expert forecasts1(Genre
1Although it is impossible to certify if the Fed decides according to expert forecasts, it is
interesting to know how often the Fed often mentions expert forecasts. However, we can reason-
ably assume the Fed actively considers expert forecasts since the GFC, among other indicators,
in their decision and communication processes. The Federal Open Market Committee (FOMC)
meeting minutes detail the record of the committee’s policy-setting meetings and o¤er detailed
insights regarding the FOMC’s stance on monetary policy. They mention the word "forecaster"
2
et al., 2013; Budescu and Chen, 2015). Usually, the forecasts are aggregated in
very simplistic ways, such as the simple average, and policymakers rank them
without considering the experts’characteristics (Alessi et al., 2014; Coibion et al.,
2020). However, as pointed out by Giacomini and Rossi (2010), understanding the
(time-varying) conditional relative forecast performance of several forecasts can
help to select the appropriate one and/or generate better forecast combinations.
Second, understanding professional forecasts helps us to understand the behav-
ioral foundations of expectations. There is a small but growing literature exploring
this …eld. Contrary to our approach, the previous literature has focused on insti-
tutional characteristics–in particular the location and type of institution–due to
data limitations. For example, regarding location, Bae et al. (2008) show that
the earnings forecasts of local …nancial experts are more precise–the local analyst
advantage. Berger et al. (2009) demonstrate that institutions based in Frank-
furt (or with a subsidiary in Frankfurt) are signi…cantly better at predicting the
ECB’s interest rate decisions. With respect to the type of institution, Mitchell and
Pearce (2007) …nd evidence that predictions by some economists covered in the
biannual Wall Street Journal survey are consistently above the survey mean, while
those of others are consistently below, depending on the industry of the econo-
mists’employers. In this vein, economists with a public mission–e.g. academics,
central bank and government employees–demonstrate a tendency towards being
pessimistic, whereas bankers in general are overly optimistic about future stock
market developments (Veress and Kaiser, 2017).
Third, ination expectations from learning-to-forecast experiments2are in line
with experts, households (Michigan), and industry (Livingston) survey forecasts
(Cornand and Hubert, 2020). Understanding the characteristics that in‡uence
the outcomes of expert forecasts should contribute to identifying potential factors
driving the formation of expectations.
We construct two original databases on expert and institutional characteristics.
Our …rst database includes the key individual characteristics of experts, such as
their past job experience (location, type and duration), gender, educational attain-
ment (including the quality of their alma mater), and a¢ liation (type and place).
only six times over the pre-GFC decade (88 meetings) with a more than fourfold increase over
the post-GFC decade (25 times over 82 meetings). The word "survey" does not appear in the
pre-GFC decade interest rate announcements, while it appears 29 times during the post-GFC
decade. The di¤erence in the Chairman’s speeches is even more spectacular than the di¤erence
in interest rate announcements or monetary policy committee minutes. The word "forecaster"
appeared only eight times in the 216 governor speeches during the pre-GFC decade. However,
it appeared 58 times in only 160 governor speeches over the post-GFC decade–almost ten times
more than during the pre-GFC decade.
2Experiments well-incentivized by remuneration. Subjects are asked to submit an in‡ation
forecast and are rewarded solely based on their forecast’s ex-post accuracy (Marimon and Sunder,
1993).
3
Our second database includes institutional characteristics, most importantly, lo-
cation and type, which have been found to matter in the previous literature. We
merge these databases with CPI nowcast data extracted from the Bloomberg sur-
vey of professional forecasters, which include both the name and institution of
each forecaster covered. To our knowledge, this is the …rst paper exploiting such
interconnected data.
Linking this data to the forecast data allows us a unique insight into the be-
havioral aspects of forecasting. We can assess the role of the individual and in-
stitutional characteristics in forecast performance, and boldness at the individual
level. We understand boldness as deviation from the consensus, which can also be
interpreted as overcon…dence (Bordalo et al., 2020). In an extension, we consider
pessimism, which we de…ne as the tendency to predict de‡ation.
In addition to being the …rst to account in such detail for forecaster character-
istics, we deviate from the previous literature by considering a range of nonlineari-
ties and interactions between relevant indicators, and in particular, how forecaster
characteristics interact with growing experience.
Our study contributes to the exciting debate in cognitive sciences and behav-
ioral economics about the role of education, its level, …eld, and quality, in expert
behaviors. It also contributes to the debates in human resources about how expe-
rience within or between institution types matters in terms of forecasting perfor-
mance, or boldness. We …nd several traits that inuence forecasting performance,
herding behaviors (opposite of boldness), and expert survival.
The remainder of the paper is organized as follows. Section 2 outlines literature
relevant to our theoretical background. Section 3 describes a simplistic theoretical
model of expectation formation that demonstrates the importance of accounting
for nonlinarities. Section 4 discusses our data. Section 5 outlines our empiri-
cal methodology and empirical results, including characteristics-based forecasting
ability tests and expert characteristics related to pessimism, with their interpre-
tation. Section 6 presents the policy implications of our results and concluding
remarks.
2 Literature Review and Theoretical Background
With forecasts being both the result of forecasters’labor and, to some degree, a
measure of their expectations, our research question is at the intersection of labor
economics and behavioral economics.
In labor economics and adjacent …elds, in particular economics of education
(Mincer, 1974), there has been extensive discussion regarding what employee traits
improve (or limit) his or her labor productivity. This includes innate traits such
as gender and origin, acquired permanent traits such as the …eld of education and,
4
last but not least, experience. It seems plausible that those factors that drive
labor productivity also drive forecasters’ performance and, thus, the quality of
their forecasts. However, as mentioned above, forecasts are not merely a product
analytically derived by the forecaster using his human capital, but also a re‡ection
of his expectations. As such, they are subject to a plethora of factors that psy-
chologically and rationally a¤ect the forecaster, including attention to economic
variables (Gabaix, 2019, 2020), expert behavioral biases (Thomas, 1999; Davis and
Lleo, 2020), and asymmetric information (Keane and Runkle, 1990), among oth-
ers (Lim, 2001; Coibion and Gorodnichenko, 2015). For example, Malmendier and
Nagel (2011) …nd that people who experienced low stock market returns invest
less in the stock market, indicating how personal experience shapes optimism and
pessimism. Education not only equips a forecaster with the necessary knowledge
to perform his job but also imbues him with a speci…c world view. Speci…c life
experiences are much more (or less) likely depending on someone’s gender, ori-
gin, etc., making the role those factors play in forecasts far less evident than the
traditionally estimated e¤ects on productivity.
What complicates matters is that the reported forecasts are not necessarily the
experts’true expectations. The objective of the forecasters’employer is typically
to obtain the best possible forecasts.3However, due to agency problems, the
forecaster’s objective function does not necessarily mirror the objective function of
the employer.4While there might be intrinsic motivation to provide good forecasts,
the main incentive for the forecaster to provide good forecasts is to maximize
expected lifetime income.5
There has been discussion in the herding literature that failing "alone" is much
worse than failing as part of a group. Indeed, it seems likely that an employer
reads the failure of its experts as the experts’fault when they alone fail, while
he might interpret them as inevitable if everybody fails. Consequently, forecasters
who deviate from the herd (following their own beliefs) have a lot to lose and might
indeed be …red for their supposed incompetence, and little to gain. In this case,
their loss function–based on the underlying objective to maximize–would include
3Their speci…c loss function can di¤er to some degree, depending on their use of the forecast.
For instance, risk-neutral investors might aim to minimize absolute forecast errors rather than
squared forecast errors that are more common in the literature. However, all those loss functions
that aim for unbiased forecasts with minimal errors are highly correlated, making the di¤erences
mostly inconsequential.
4The outcomes achieved by any institution depend on its ability to take action today to
achieve its objectives tomorrow. Institutions use expert forecasts to shape their economic deci-
sions (Bernanke, 2007; Campbell and Sharpe, 2009). Consequently, expert forecasting accuracy
generates opportunity costs (Laster et al., 1999). Ful…lling these objectives depends on the
performance of the forecasts they use, including expert forecasts (See Footnote 1).
5More precisely, the present value of expected lifetime income.
5
both the forecast loss and deviation from the herd of other forecasters.6
This is why we go beyond merely looking at forecast performance in this paper
and consider boldness, which we de…ne as the willingness to deviate from the
herd. Additionally, further highlighting the aforementioned psychological aspects
of forecasts, we consider pessimism.7
We extend the literature considerably by adding a battery of traits that might
ect either of those aspects of forecasts, including educational attainment (Bach-
elor, Master, and PhD), …eld (Economics, Finance, both, or other) and quality
(based on the Academic Ranking of World Universities, also known as Shanghai
Ranking). As the geography (i.e., location) and the institution type of both the
expert and the institution (i.e., its a¢ liation) matter for forecasting (Batchelor,
2007; Hong and Kacperczyk, 2010), we also contribute to the literature by ex-
tending the expert’s experience characteristics with institution type (central bank,
academic, …nancial, or other) and location (of both the expert and the institutional
a¢ liation).
3 Model
In this section, we present a stylized model that motivates the inclusion of nonlin-
earities and interactions. We show that even using a straightforward and standard
form of learning–namely Bayesian learning–experience has nonlinear e¤ects that
depend strongly on initial conditions. Consider a forecaster aiming to forecast
in‡ation who is exposed to both public and private signals. We assume both
signals are drawn from a normal distribution, they are isolated8and mutually
independent, and both are unbiased but noisy. In other words:
su
t N t;  1
u;(1)
and
sr
t N t;  1
r;(2)
where su
tand sr
tare the public and private signals at time t, respectively.
We assume that the precision of the public signal, u, is known and constant.
The precision of the private signal, r, is unknown to the forecaster initially.
Rather, starting from a prior assumption, he learns about the quality (i.e., the
precision) of his private signal over time through Bayesian learning.
6See Section 4.1, Eq. 8 and 9, for more details.
7For details on our de…nition, refer to Sections 3 and 5.4.
8Each expert cannot observe other forecasts when extrapolating this one-agent model to a
multiple-agent model, a simplifying assumption corresponding to the …ndings of Bordalo et al.
(2020).
6
For simplicity, we assume that the prior regarding the precision of the private
signal is Gamma distributed, where the initial prior has a mean of ~0and a variance
of 2
 ;0.9
Since the Gamma distribution is a conjugate prior for the precision of a normal
distribution with a known mean (in this case 0), the Bayesian updating yields a
new Gamma distribution with lower variance and a more accurate estimate of the
true mean every period.
More precisely, the mean of new distribution is given by:
~t=~2
t1=2
 ;0+ 1=2
~t1=2
 ;t1+ 1=2sr
t1t12;(3)
and the variance by
2
t=~2
t1=2
 ;0+ 1=2
~t1=2
 ;t1+ 1=2sr
t1t122:(4)
The Bayesian point estimate for the average forecasteri.e. the representative
forecaster that is repeatedly experiencing errors of a magnitude of exactly one
standard deviationwho starts forecasting in 0and does so every period at time t
is thus given by:
~t=~2
0=2
 ;0+ (t1) =2
~0=2
 ;0+t=2r
:(5)
A forecaster who aims to maximize the expected precision of his forecast, will
then provide a weighted forecast
ft=r
r+ ~t
su
t+~t
r+ ~t
sr
t;(6)
implying a precision of
t= r
r+ ~t2
1
u+~t
r+ ~t2
~1
t!1
:(7)
The dynamics implied by this model are quite straightforward:
1. Unless ~0=u, the forecast performance will improve over time, since any
misconception–whether it is overly optimistic or pessimistic–regarding the
quality of the private signal will lead to a suboptimally weighted forecast.
9In the more common ,parameterization, this corresponds to 0= ~2
0=2
 ;0and 0=
~0=2
 ;0. While this notation makes the updating equations more convoluted, it allows for deriving
the interest variable’s law of motion more straightforwardly.
7
2. Precision converges monotonically to =u+r.
We can now imagine a range of factors that can potentially drive forecast
quality rand the initial optimism (or pessimism) regarding one’s own forecast
quality ~0.
Figure 1 shows some core scenarios–namely a good, a bad, and two average
forecasters, with the latter two di¤ering in the degree of optimism (or pessimism)
regarding the quality of their private signal–linking expected forecast performance
to time, i.e. experience.
Figure 1: Evolution of forecast precision and variance over time
1.0
1.5
2.0
2.5
3.0
12345678910
Precision
Experience
Good Bad Optimistic Pessimistic
0.3
0.4
0.5
0.6
0.7
0.8
12345678910
Variance
Experience (log)
Good Bad Optimistic Pessimistic
Note: The good forecaster has a precision of two with a prior of one. The bad forecaster has a
precision of 0.5 with the same prior. The optimistic forecaster has a precision of one with a prior
of two, and the pessimistic forecaster has a precision of one with a prior of 0.5. All priors have
a variance of zero, and u= 1.
In the left panel, we show the “raw numbers”, i.e., time (experience) tand pre-
cision t, obtained from substituting Eq. 5 into Eq. 7. In the right panel, we show
the data using the transformations that will be applied in the empirical approach,
i.e., the variance (corresponding to the expected values of squared forecast errors)
and the natural logarithm of time (experience). The di¤erent trajectories of perfor-
mance in response to time - i.e., experience - are visible. That is, factors that drive
forecast performance–through both actual quality and suboptimal weighting–also
ect the performance’s response to experience (here shown as time). In other
words, it is necessary to include forecast experience and constant factors driving
forecast performance in a model, but also to consider possible interactions.
The model predicts that both heterogeneity in prior expectations and under-
standing of news across forecasters may drive dispersion in forecasts. Heteroge-
neous interpretation of the incoming information might exacerbate the dispersion
8
of forecasts. We assume this interpretation of incoming information is in‡uenced
by experts’characteristics such as experience and education for example.
To some degree, it is always good to include your own information10 so forecasts
do not converge–worse forecasters stay worse. The two experts who have the
same "true" quality of the private signal–the pessimist and the optimist–eventually
converge as they learn their signal’s quality.
4 Data
In this paper, we combine three unique datasets: individual forecasts for the US
CPI (Section 4.1), web-sourced information and characteristics about the fore-
caster’s CV (Section 4.2), and web-sourced information about institutions where
a forecaster worked (Section 4.3). These datasets cover the period from 1997:Q1
to 2017:Q4.
4.1 Forecaster Behaviors
Most of the data, including individual point forecasts and the name and a¢ lia-
tion of the experts, come from Bloomberg. Each expert can submit and update
US in‡ation forecasts until the …rst day of the corresponding month before the
publication of the e¤ective US CPI in‡ation. Since forgoing the chance to update
despite new information being available is irrational, we assume that the …nal
forecasts are considered the best possible forecasts by the submitting experts.
The expert forecast updates are accessible to other experts at any point of time,
allowing forecasters to react to each other. Therefore, any deviation from the herd
can be considered as deliberate. This allows us to capture both the quality (Fig.
2) and boldness (deviating from the herd) of forecasts.
First, we look at performance, measured through squared forecast errors as the
most commonly used loss function, i.e.,
Li;t =tEi
t[t]2;(8)
where iis a specic expert. We use tto denote a point in time before tbut
clearly after t1, since forecasters can update until the last moment.
Second, we assess boldness, which we de…ne as deviating from the “herd” of
other forecasters. In other words,
Bi;t =Ei
t[t]Mi;t Ei
t[t]2;(9)
where Mi;t is the median operator over all expert forecasts iat time t.
10 Since the signal error is unrelated, even a high variance signal is meaningful.
9
Figure 2: Expert Forecast Performance
-1.20
-0.90
-0.60
-0.30
0.00
0.30
0.60
0.90
%
Note: The blue line is the median of expert forecast errors, and the gray area represents the
di¤erence between the maximum and minimum forecast errors for each period.
Recall that an expert forecast can be updated until the last minute, and is
public after submission. Since forecasts are typically not submitted in the last
second, but we take the last available forecast, this would allow forecasters to
adjust to the perceived consensus (median), causing perfect coincidence of forecasts
by the time the data is published. Therefore, we can assume that any deviation
from this median is an intentional deviation from the group, which corresponds to
our model’s assumption that each expert is isolated from others (and thus cannot
observe other forecasts) and are overcon…dent in their forecasts since they disregard
information from other forecasts.
There is, however, one caveat. Since we use nowcasts, where an excellent infor-
mation set is already available for all forecasters, the median forecast is typically
very close to the actual outcome. The correlation between Band Lis 0.573,
making it problematic to fully attribute variation in Las pure boldness.
These two measures will mostly be used as the dependent variable in our empir-
ical exercise. However, they will also serve as an explanatory variable in a survival
analysis, where we assess the in‡uence of those measures on the probability of
10
submitting another forecast.
4.2 Forecaster Characteristics
Our databases on expert characteristics are based on an automated collection,
which is manually augmented where needed. The primary source is LinkedIn.11
To rule out technical problems, we manually checked LinkedIn itself where the
automated search did not yield results, before looking for alternative sources for
a CV, such as private and institutional websites. Partial database entries from
LinkedIn were completed in the same fashion.
In essence, the data we collected about experts encompasses 151 experts and
164 institutions that published at least one US CPI in‡ation forecast. Since our
panel speci…cation includes forecaster …xed-e¤ects, we drop all forecasters provid-
ing only a single forecast, leaving us with 112 experts.
For each forecaster, we collect data on past job experience (type and dura-
tion12), separated in to (a) academic experience, (b) central bank experience, and
(c) experience in …nancial institutions. We also gather data on crucial expert char-
acteristics such as their location (local or foreign), education (level, type, and
quality), gender (male or female), and a¢ liation (type and location). Their high-
est degree (Bachelor’s, Masters, or PhD) and the corresponding …eld, where we
only distinguish between Economics, Finance, and other …elds, are also identi…ed.
For the highest degree, our database also includes the Shanghai Ranking clas-
si…cation for the respective university following both the economics and …nance
rankings.13
Last but not least, we collect additional information, such as origin and age.
However, since we want to be consistent across all database entries, and the date of
birth is typically not reported in US-style CVs, age is proxied using total experience
plus 18 years plus the typical time necessary to obtain the …nal degree (three years
for a Bachelor’s degree, …ve for a Master’s, and 11 for doctoral degrees).
4.3 Institutional Information
The data about experts’a¢ liations describe the institution type and its primary
location (local or foreign headquarters). We classi…ed all forecast providers into
11 LinkedIn is a US employment-oriented Internet service, founded in 2002. It is mainly used
for professional networking, including employers posting jobs, job seekers posting their CVs, and
people who want to broaden their network.
12 The expert’s experience corresponds to the full period in which the forecaster published
forecasts.
13 We use the Shanghai Ranking’s Global Ranking, Academic Subjects, 2017. We decompose
each ranking (global, economics, and …nance) to four levels: rst tier, second tier, third tier, and
not ranked.
11
several types: retail bank, investment bank, private bank, insurance company, eco-
nomic and …nancial analysis …rm, fund, investment management …rm, brokerage,
credit union, savings and loan …rm, academia, central bank, and others. Although
we built an in-depth database separating several relevant hosting institution types,
we rely only upon the simple (and more relevant) di¤erence between private …nan-
cial, academic, and monetary institutions in our analysis.
5 Results and Interpretation
This section uses the data described in Section 4 to detect the characteristics
leading to forecasting performance and boldness, and experts sentiments. We
apply a proportional hazards model to examine expert’s survival (Section 5.1),
a cross-sectional analysis to identify the role of inherent traits (Section 5.2), a
panel estimation to explore the role of experience (Section 5.3), a probit model to
assess expert’s pessimistic and optimistic behaviors (Section 5.4), and forecasting
ability tests to identify the forecasting performance of characteristics-based groups
of experts.
5.1 Survival Analysis
In our …rst set of analyses, we assess forecasters’probability of providing a further
forecast in the future, i.e., to survive in the market. Our dataset does not allow
us to distinguish the reasons for possible removal from the dataset. Being …red,
or at least removed from this particular responsibility, are possibilities, but merely
retiring or even being promoted, are equally possible. However, the main reason for
us to conduct this analysis is that bad forecast performance may lead to exclusion
from the sample, thereby creating endogeneity issues as outlined above.
The approach we chose is a proportional hazards model pioneered by Cox
(1972). In this model, we estimate:
hit =h0(t)e0f(Yit1)+Xi;(10)
where Yit1is the past history of yit,h0(t)corresponds to the time-speci…c e¤ect,
is a vector of regression coe¢ cients, and 0is a regression coe¢ cient. We use
three di¤erent transformations f(Yit1)describing slightly di¤erent possibilities of
how performance review might be conducted by the hiring institutions.14 First, we
simply take the last forecast’s value, i.e. Yit1=yi;t=1. Second, we look at a mov-
ing average over the past …ve months, indicating a slightly longer horizon rather
14 Raw, MA5, and wMA5. Raw stands for the last period dependent variable, MA5 for …ve-
period moving average (backward-looking) and wMA5 for weighted moving average where the
most recent periods have a higher weight.
12
than penalizing extremely bad forecasts immediately, i.e. Yit1= 1=5P5
s=1 yits.
Finally, we look at a weighted moving average, i.e., Yit1=P5
s=1 wsyits, where
the weights decline exponentially by a power of two.
Since we want to stay true to our general multivariate framework that includes
continuous time-varying variables, we abstain from traditional Kaplan and Meier
(1958) type survival plots, which allow assessing actual expected survival time
for distinct subgroups. Therefore, we cannot interpret our results in terms of
additional months of survival in the job, but merely interpret the sign and relative
magnitude of coe¢ cients.
Before presenting our survival analysis, we present preliminary results in Ta-
ble 1 to capture our research question’s intuition. Tables 1 to 3 present three
transformations for each dependant variable. The Raw column stands for the last
period performance or boldness15 (raw data), MA5 for …ve-period moving average
(backward-looking), and wMA5 for weighted moving average where the most re-
cent periods have a higher weight. In these tables, positive signi…cant coe¢ cients
show that a low forecasting performance or boldness increases the probability of
removal (i.e., not being a forecaster in the next period).
Table 1: Survival Estimates: Intuition.
Performance
Raw MA5 wMA5
Dependent variable 1.801 5.119** 3.59**
Local forecaster (LF) -0.425** -0.417** -0.419**
Education: Economics (EC) -0.586** -0.577** -0.568**
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively. Raw
stands for the last period dependent variable, MA5 for …ve-period moving average (backward-
looking) and wMA5 for weighted moving average where the most recent periods have a higher
weight.
Table 1 presents the survival estimates for US CPI experts, without interac-
tions, related only to the expert’s location and education …eld. It shows that these
variables matter even in this simpli…ed model. Although the expert’s survival de-
pends on his past performance, especially MA5 and wMA5, it also depends on his
education …eld and current location.
A local in‡ation expert may survive longer than a foreign expert, which may
re‡ect employment protection or information advantage e¤ects, in line with the
nancial analysts literature (Malloy, 2005). We also …nd that an expert with
15 See Eq. 8 and Eq. 9, respectively, for the de…nitions of performance and boldness.
13
a degree in economics may survive longer than an expert without an economics
degree, a result interpreted below (Table 2).
Table 2 presents the survival estimates for US CPI experts without interactions
with all the variables at our disposal.16
Table 2: Survival Estimates Without Interactions.
Performance Boldness
Raw MA5 wMA5 Raw MA5 wMA5
Dependent variable 2.657 6.87** 4.655* 4.139 8.577 7.395
Age >50 years old 0.561 0.555 0.554 0.586* 0.573* 0.583*
Local forecaster (LF) -0.929** -0.939** -0.927** -0.935** -0.954** -0.943**
Local institution (LI) -0.063 -0.026 -0.046 -0.048 -0.038 -0.041
Financial institution (FI) -0.5 -0.459 -0.482 -0.465 -0.444 -0.452
Experience: Central bank (CB) -0.318 -0.304 -0.312 -0.298 -0.277 -0.279
Experience: Academia (AC) 0.691* 0.667* 0.692* 0.69* 0.657* 0.687*
Gender (G) 0.546 0.614 0.57 0.521 0.544 0.525
Education: MA (MA) 0.167 0.172 0.181 0.157 0.182 0.171
Education: PhD (PhD) -0.795 -0.828 -0.795 -0.784 -0.766 -0.773
Education: Economics (EC) -1.548*** -1.528*** -1.521*** -1.562*** -1.551*** -1.544***
Education: Finance (EF) -1.288* -1.332* -1.273* -1.358** -1.366** -1.34**
Ranking: Economics (RE) 0.287* 0.288* 0.283* 0.263* 0.261* 0.257
Ranking: Finance (RF) -0.188 -0.183 -0.19 -0.177 -0.175 -0.178
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively. Raw
stands for the last period dependent variable, MA5 for …ve-period moving average (backward-
looking) and wMA5 for weighted moving average where the most recent periods have a higher
weight.
Table 2 shows that a low past forecasting performance (MA5 and wMA5) in-
creases the probability of being removed. However, providing bolder forecasts does
not seem to decrease expert survival signi…cantly.17 According to these results, the
expert’s future seems to be determined partly by current and past performance,
but not necessarily by herding behavior. Herding does not seem to signi…cantly
in‡uence the experts’survival rate.18
Table 2 also shows that having a degree in economics immunizes the expert from
being removed as an economist. Interestingly, a degree in …nance does not o¤er
the same protection as having graduated in economics. Experts with a degree in
16 See Section 4 for more details.
17 The presented results use the exact partial likelihood. Under the Breslow and Chatterjee
(1999) approximation, low past performance in forecasting in‡ation (MA5) increases the proba-
bility of being removed.
18 In general, most experts follow others or are moderately bold. It is evident that, as long
as the consensus provides the best forecast on average, an expert always far from the consensus
(bold) will have a lower survival rate.
14
economics may provide more convincing explanations to back their information
processing and subsequent forecasts of the US CPI, an economic variable par
excellence, than experts with a degree in …nance,19 increasing their survival rate.20
Having graduated from a top-ranked university in economics21 or having experience
in academia seems to deteriorate the expert’s survival rate.
Several theories of reputation and herd behavior indicate that agents’perfor-
mance and boldness may vary with career concerns (Scharfstein and Stein, 1990;
Zwiebel, 1995). Our results suggest that increased reputational capital may in-
crease the labor market attractiveness of top-ranked pro…les, thus leading experts
from top-ranked universities or academia to leave the profession more frequently
than other experts. This might also mean that these experts are …red more often
because, for instance, they are hired based on higher expectations than others,
which they may not meet.
The expert’s age appears to be a crucial factor for survival when dealing with
boldness. Bolder experts will survive less the farther they are above …fty. While
this result is not necessarily related to accumulated experience, it shows that an
old expert forecasting outside the consensus has a higher chance of not being a
forecaster in the next period than one who is in the consensus or younger.
Table 2 also shows that being a foreign expert, i.e., not living in the US, a¤ects
the survival of the expert. This bias in favor of local experts re‡ects the average
inability of experts not living in the US to survive as a forecaster.
Table 3 presents the survival estimates for US CPI experts with interactions.
Table 3 provides additional insights about expert survivals. First, it shows
that having graduated with a Master’s degree from a top …nance institution plays
a positive role in the experts survival. This result sharply contrasts with Master’s
degrees from top institutions in economics. Second, this result is mitigated when it
interacts with a higher education level. PhD graduation from a top institution in
economics increases the survival probability of the expert. However, this increase
in the experts survival rate is less signi…cant for a PhD from a top institution in
economics than for a Master’s degree from a top institution in …nance.
One might think the expert’s institution is more an intermediary than a factor.
An expert with a PhD might increase his chances of working in a better institu-
tion, with better information-gathering function, than an expert without a PhD.
19 Whatever the expert’s outcomes, providing more convincing economic explanations about
their in‡ation forecasts helps.
20 In a previous version of this paper, an application to Fed fund rates shows that experts with
a degree in …nance survive better than those with a degree in economics. We attributed this
result to the fact that the nominal interest rate is both an economic and a …nancial variable, and
is thus better explained by experts with an education in …nance or with an education in both
economics and …nance.
21 See Section 4.2.
15
Table 3: Survival Estimates with Interactions.
Performance Boldness
Raw MA5 wMA5 Raw MA5 wMA5
Dependent variable 2.661 6.569** 4.584 4.083 6.785 6.803
Age >50 years old 0.969*** 0.915*** 0.938*** 0.958*** 0.914*** 0.93***
Local forecaster (LF) -1.48*** -1.481*** -1.472*** -1.501*** -1.502*** -1.499***
Local institution (LI) 0.387 0.413 0.394 0.385 0.379 0.377
Financial institution (FI) -0.63 -0.592 -0.61 -0.716* -0.7* -0.699*
Experience: Central bank (CB) -0.231 -0.183 -0.21 -0.219 -0.176 -0.19
Experience: Academia (AC) 1.228*** 1.183*** 1.217*** 1.15*** 1.105*** 1.134***
Gender (G) 0.345 0.46 0.391 0.335 0.397 0.355
Education: MA (MA) -0.251 -0.225 -0.237 -0.25 -0.223 -0.241
Education: PhD (PhD) -1.574** -1.579** -1.565** -1.512** -1.488** -1.497**
Education: Economics (EC) -1.409*** -1.342** -1.361** -1.424*** -1.41*** -1.393***
Education: Finance (EF) -1.448* -1.491** -1.429* -1.521** -1.548** -1.493**
Ranking: Economics (RE) 0.188 0.172 0.178 0.196 0.19 0.186
Ranking: Finance (RF) -0.043 -0.019 -0.035 -0.043 -0.035 -0.04
Interactions
MA RE 0.384 0.421 0.39 0.339 0.341 0.329
MA RF -0.982*** -0.997*** -0.992*** -0.968*** -0.959*** -0.97***
PhD RE -0.927** -0.941** -0.937** -0.9** -0.92** -0.918**
FI RF 0.383 0.424 0.398 0.378 0.394 0.385
FI LI 0.635 0.41 0.523 0.503 0.382 0.409
GLI 0.25 0.328 0.324 0.219 0.347 0.258
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively. Raw
stands for the last period dependent variable, MA5 for …ve-period moving average (backward-
looking) and wMA5 for weighted moving average where the most recent periods have a higher
weight.
Financial institutions may have a better information-gathering functions, data,
and private information access. Also, experts with a PhD may be attracted by
these institutions, and these institutions may prefer recruiting people with a PhD,
making a case for …rm-e¤ects that are both an intermediary and a factor. Although
we do not have enough companies with more than one forecaster in the sample
(otherwise we would be able to control for …rm …xed-e¤ects), we believe that it is
unlikely that the institution is an intermediary rather than a factor. Indeed, as
long as we use Bloomberg forecasts, meaning all our experts have access to at least
a Bloomberg terminal, we can reasonably assume all the forecasters have access
to almost the same information and necessary equipment to build their forecasts.
Hence, the e¤ect of the PhD is more prevalent than …rm-ects.
Interestingly, Table 3 shows that the experts age signi…cantly a¤ects the ex-
pert’s survival rate for both low performances and bolder forecasts. As in Table
2, it indicates that low average performance (MA5) signi…cantly decreases the
expert’s survival rate.
16
Tables 2 and 3 mitigate existing career-concern-motivated herding theories
(Hong et al., 2000). While our results show that experts are more likely to lose their
jobs after providing inaccurate forecasts, older and probably more experienced ex-
perts are more likely to lose their jobs than younger and perhaps less experienced
experts. Among other things, our results con…rm that underperformers face higher
employment risk than outperformers (Clarke and Subramanian, 2006). The theory
linking analysts’boldness with career concerns and ability (Scharfstein and Stein,
1990; Jacob et al., 1999) is partially veri…ed for CPI in‡ation experts.
5.2 Cross-Section Estimation
Because most of the forecaster traits we collect in our data do not vary over
their forecast history, we opt for a simple cross-sectional approach to assess their
relevance in this second set of analyses. However, we have to account for the fact
that a large part of forecast errors comes from unpredictable shocks to in‡ation,
ecting all forecasts simultaneously. This would not be an issue if we had a
balanced panel (i.e., all forecasters were a¤ected by this equally), but we have
some forecasters only active during the so-called Great Moderation, while others
provided forecasts during the turbulent times of the GFC.
Essentially mirroring the idea of time-…xed-e¤ects in panel analysis, we control
for the average performance of other forecasters during the period in which the
forecaster under consideration was active. For simplicity, we refer to this control
as the active period-speci…c e¤ect22 for the remainder of the paper. This yields the
regression equation:
yi=Xi + 0Pt2TiPj2Ftyj t
Pt2Ti(jFtj  1) +i;(11)
where Xiis the set of forecaster traits, Tiis the set of periods when forecaster i
was active, Ftis the set of forecasters active at time t, and jFtjis the cardinality
of that set. The bar operator indicates the arithmetic mean. 0is a regression
coe¢ cient and is the error term.
By controlling for institutional characteristics, we also guarantee that the e¤ect
of forecaster characteristics is not merely driven by some institution’s easier access
to forecasters with speci…c traits (e.g., forecasters with a PhD or from the best
institutions). We decompose our sample (1997:Q1-2017:Q4) into two subsamples,
the pre-GFC (1997:Q1-2008:Q1) and post-GFC (2008:Q1-2017:Q4) subsamples.
Table 4 presents the cross-section estimations for the post-GFC US CPI in‡ation
expert forecasts, with the correction for time-e¤ects outlined in Eq. 11, for the
variables presented in Section 4. The results presented in this section are based
22 Namely, Dependent variable (mean).
17
on demeaned variables so that the point estimate of the non-interacted variables
corresponds to the marginal e¤ect at the mean.
Table 4: Cross-Section Estimates: Intuition
Performance
Dependent variable (mean) 0.45
Local forecaster (LF) -0.006*
Local institution (LI) 0.003
Financial institution (FI) -0.007*
Experience: Central bank (CB) -0.001
Experience: Academia (AC) 0.001
Gender (G) 0.003
Education: MA (MA) -0.011**
Education: PhD (PhD) -0.015***
Education: Economics (EC) -0.006
Education: Finance (EF) -0.008
Ranking: Economics (RE) -0.002*
Ranking: Finance (RF) 0.001
Observations 58
Adjusted R20.068
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively.
Table 4 shows that education level and …eld, and institution type and location,
could have a role in the expert’s forecasting performance. An expert with a Mas-
ter’s or PhD degree in Economics, local or a¢ liated with a …nancial institution,
may provide better in‡ation forecasts. These results, which provide a clue about
our research question, are developed and discussed below.
Table 5 presents the cross-section estimations for the US CPI in‡ation expert
forecasts with the correction for time-e¤ects.
The cross-section estimates presented in Table 5 show that performance and
boldness are in‡uenced di¤erently.
As far as forecasting performance is concerned, working in a …nancial (FI) or
local institution (LI) improves the expert’s forecasting ability more (and more
signi…cantly) before the GFC than after. However, experts at local …nancial insti-
tutions (LI FI) slightly mitigate this result as this interaction is less signi…cant
than FI or LI e¤ects alone. Interestingly, having graduated from a top university
(…nance ranking) appears to improve expert forecasts before the GFC, while this
is not the case after or over the full sample. This is mitigated by experts working
18
Table 5: Cross-Section Estimates
Performance Boldness
Full sample Pre-GFC Post-GFC Full sample Pre-GFC Post-GFC
Dependent variable (mean) 1.23*** 0.471 1.303*** 0.969*** -1.011 0.978***
Local forecaster (LF) -0.012 0.006 -0.013 -0.002 0.011 -0.001
Local institution (LI) -0.012 -0.105*** -0.005 -0.007 -0.077*** -0.004
Financial institution (FI) -0.015 -0.101** -0.007 -0.003 -0.055* 0
Experience: Central bank (CB) 0 -0.005 0.002 -0.003 0.009 -0.002
Experience: Academia (AC) -0.001 0.003 0 -0.003 -0.001 -0.003
Gender (G) -0.003 -0.015 0.001 -0.002 -0.027 0
Education: MA (MA) 0.007 -0.013 0.008 -0.001 -0.026 -0.002
Education: PhD (PhD) -0.003 -0.018 -0.005 -0.005 -0.012 -0.006
Education: Economics (EC) -0.014 0.004 -0.022 -0.008 0.003 -0.016*
Education: Finance (EF) -0.016 0.025 -0.027 -0.013 0.05* -0.022**
Ranking: Economics (RE) -0.001 -0.006 -0.002 -0.003 -0.022 -0.003
Ranking: Finance (RF) -0.008 -0.089** -0.009 0.001 -0.054* 0.002
Interactions
MA RE -0.002 -0.007 -0.002 0.003 0.031 0.004
MA RF -0.007 0.025 -0.011 -0.006 -0.004 -0.008*
PhD RE -0.002 0.015 0.001 0.001 0.02 0.003
FI RF 0.011 0.081** 0.013 0 0.06*** -0.001
FI LI 0.022 0.101* 0.014 0.01 0.046 0.006
GLI 0.023 0.143** 0.003 0.021** 0.138*** 0.007
Observations 67 30 63 67 30 63
Adjusted R20.35 0.53 0.38 0.11 0.53 0.08
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively.
at a …nancial institution and graduated from a top university (nance ranking),
who achieve lower performance with their forecasts. Before the GFC, male experts
appeared to provide better forecasts than females, but this is not true after the
GFC or over the full sample. The fact that male experts were more numerically
dominant prior to the GFC, while the percentage of female experts has increased
since the GFC, mitigates this result. Our results may also re‡ect that women
bene…t less than men from connections in job performance, herding behavior, and
subjective evaluation by others (Fang and Huang, 2017).
The situation is di¤erent regarding the expert’s boldness (herding behavior).
First of all, working at a local or …nancial institution decreases the expert’s bold-
ness only before the GFC. However, having graduated in economics or …nance
decreases bold expert forecasts after the GFC, while having graduated in …nance
before the GFC increased the expert’s boldness. Interestingly, the boldness of
female experts from local institutions was greater before the GFC, while the ro-
bustness of this result is still questioned due to the paucity of female experts before
the GFC. Over the full sample, female experts from local institutions herd more
19
than male experts. After the GFC, having graduated with a Master’s degree from
a top university in …nance slightly but signi…cantly reduced the experts boldness.
In line with Clarke and Subramanian (2006), performance and boldness results
in Table 5 are often similar in terms of signi…cance and sign,23 con…rming that
signi…cant underperformers are more likely to issue bolder forecasts and vice-versa.
Like …nancial analysts who also tend to exhibit herding behavior, which sometimes
compromises accuracy, our results suggest that social forces (ranking, institution
type, and location), education (type and level), and experience (type and duration)
in‡uence an expert’s rational economic logic and cognitive biases–an interpretation
close to Christo¤ersen and Stæhr (2019) that is presented in our next results
(Section 5.3).
Interestingly, the di¤erence between pre- and post-GFC may re‡ect the change
in the expert’s attention or biases induced by the crisis shock on characteristics’
ects (Andrade and Le Bihan, 2013; Christo¤ersen and Stæhr, 2019). Although
these results consider a time-…xed-e¤ect, results considering individual …xed-ects
like experience are presented in Section 5.3.
5.3 Panel Estimation
In our third set of analyses, we assess the e¤ect of experience on the di¤erent
output measures. Since experience is the only time-varying trait we consider, this
essentially boils down to a simple univariate panel regression with time (t) and
forecaster (i) speci…c e¤ects:
yit =0ln xit +ui+vt+"it;(12)
where yit is one of our two loss functions discussed in Section 4.1 (performance and
boldness) and xit is one of our experience measures discussed in Section 4. 0is
the regression coe¢ cient. ui,vt, and "it represent the forecaster …xed-e¤ect, time
xed-e¤ect, and idiosyncratic component of the error term, respectively.
There are, however, potential endogeneity issues with this speci…cation, since
the ability to gain experience (or in other words “to keep your job”) might depend
on forecast performance. We assess this possibility in detail as outlined in Section
5.1. While the evidence for the existence of such an e¤ect is mixed, and if it exists
it seems to be only moderately sized, we correct for it in a robustness test by
including a dummy for ‡agging the last …ve forecasts submitted by any forecaster.
While much simpler, this follows the spirit of selection estimators24 (Heckman,
23 This is also the case with our next results (Table 7).
24 In a full-‡edged selection model, one would include a transform of the “inclusion”probability
rather than a determinant of the latter. The key problem is that we cannot truly estimate
inclusion probabilities here, but merely relative risks of being removed from the sample.
20
1979). The resulting equation is given by:
yit =0ln xit +11lt2last5(i)+ui+vt+"it ;(13)
where last5(i)is the set of the last …ve periods in which forecaster isubmits a
forecast to our dataset. 0and 1are regression coe¢ cients.
The model controls for individual speci…c and time e¤ects. Table 6 presents
simpli…ed panel estimations for the pre-GFC US CPI in‡ation expert forecasts
with time and experience …xed-ects. The results presented in this section are
also based on demeaned variables to identify the partial e¤ects.
Table 6: Panel Estimates: Intuition
Performance
Experience (E) -0.031***
Age >50 years old 0.17***
Two-way interactions with experience
Local institution (LI) -0.019**
Financial institution (FI) 0.026***
Gender (G) -0.031***
Education: MA (MA) -0.041***
Education: Economics (EC) -0.031*
Ranking: Economics (RE) 0.015***
Ranking: Finance (RF) -0.011***
Observations 339
Adjusted R20.01
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively.
The simpli…ed results presented in Table 6 show that experience generally im-
proves the expert’s forecasting performance. It also provides an intuition about
the results discussed below, namely the positive e¤ects of a top-ranked educational
institution in Finance or the a¢ liated institution location on the expert’s perfor-
mance. These initial results con…rm the relevance of our research question and
results.
Table 7 presents our full results, the panel estimations for the US CPI in‡ation
expert forecasts with time and experience …xed-e¤ects.
The panel estimation presented in Table 7 shows that being a local forecaster,
working in a …nancial institution, or having experience at a central bank or in
21
Table 7: Panel Estimates
Performance Boldness
Full sample Pre-GFC Post-GFC Full sample Pre-GFC Post-GFC
Experience (E) -0.017*** -0.008 0.007 -0.009*** 0.009 -0.001
Age >50 years old 0.004 0.195** 0.008 -0.003 0.119** -0.002
Two-way interactions with experience
Local forecaster (LF) 0.033** 0.014 0.199* 0.002 0 0.091
Local institution (LI) -0.088*** -0.066 -0.292*** -0.036*** -0.042 -0.131***
Financial institution (FI) 0.031*** 0.061 0.077*** 0.016*** 0.015 0.041***
Experience: Central bank (CB) 0.018*** 0.036 0.019*** 0.007*** 0.028 0.006
Experience: Academia (AC) 0.074*** 0.011 0.306* 0.029*** 0.009 0.17**
Gender (G) -0.045*** -0.053* -0.032 -0.029*** -0.035* -0.002
Education: MA (MA) -0.012 -0.005 -0.01 -0.007 -0.021 0.013
Education: PhD (PhD) -0.025*** 0.02 -0.037*** -0.012*** -0.006 -0.015***
Education: Economics (EC) 0.056* 0.039 0.128 0.016 0.034 0.094
Education: Finance (EF) -0.081*** 0.087 -0.288*** -0.036*** 0.142*** -0.124***
Ranking: Economics (RE) 0.027*** -0.007 0.059*** 0.014*** -0.001 0.025***
Ranking: Finance (RF) -0.01*** -0.002 -0.006 -0.004*** -0.01 0.002
Three-way interactions with experience
FI RE 0.005 0.056 -0.018 0.004 0.027 -0.007
MA RE -0.128*** -0.046 -0.163*** -0.07*** -0.048 -0.05***
PhD RE -0.132*** -0.022 -0.19*** -0.073*** -0.027 -0.061***
EC RE -0.074*** -0.043* -0.272*** -0.031*** -0.031** -0.128***
FI RF -0.006 -0.093** 0.023 -0.002 -0.07*** 0.004
GRF -0.065*** 0.001 -0.029 -0.047*** -0.005 0.003
MA RF -0.003 -0.017 -0.007 -0.002 -0.005 -0.007**
EF RF 0.004 -0.027 0.008 0.009*** -0.087*** 0.002
FI LI -0.029** -0.062 -0.064*** -0.014* 0.028 -0.001
AC LI -0.362*** 0.059 -1.434** -0.156*** 0.027 -0.8**
MA LI 0.246*** -0.099* 0.35*** 0.145*** -0.052 0.111***
PhD LI 0.245*** 0 0.28*** 0.165*** 0 0.139***
FI LF -0.106*** 0 -0.147*** -0.057*** 0 -0.057***
GLF -0.085** 0 0.093 -0.084*** 0 0.06
AC LF 0.357*** 0 1.458** 0.155*** 0 0.805**
MA LF 0.152*** 0 0.191*** 0.081*** 0 0.072***
PhD LF 0.167*** 0 0.263*** 0.067*** 0 0.049**
EC LF 0.182*** 0 0.598*** 0.082*** 0 0.273***
GFI -0.349*** 0 -0.413*** -0.178*** 0 -0.144***
AC FI -0.018 0 0.106*** 0.002 0 0.017
EC FI -0.014 0 -0.415*** 0.001 0 -0.22***
GCB 0.137*** 0 0.211*** 0.072*** 0 0.066***
EC AC 0.073*** 0.04 0.295*** 0.034*** 0.016 0.135***
EC MA -0.13** 0 -0.206 -0.023 0 -0.152
Observations 3073 350 2723 3077 350 2727
Adjusted R20.04 0.08 0.07 0.02 0.02 0.03
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively.
22
academia decrease the expert’s forecasting performance after the GFC or over the
full sample. It also shows that experience, working in a local institution, or being
educated in …nance increase the expert’s forecasting performance after the GFC
or over the full sample. Experts may underreact less to prior CPI information as
their experience increases, suggesting one reason why experts, like analysts, be-
come more accurate with experience (Mikhail et al., 2003). Similar to analysts’
rm experience, which is strongly and positively associated with analysts forecast
boldness (Clarke and Subramanian, 2006; Huang et al., 2017), our results comple-
ment these …ndings for CPI experts by di¤erentiating the experience type, size,
and somehow the reputation accumulated through their education.
The pre-GFC period signi…cantly di¤ers from the post-GFC one. Over the
pre-GFC period, only being a local forecaster appeared to explain the expert’s
forecasting performance signi…cantly. Being a female expert over the full or pre-
GFC sample decreases the forecasting performance, but like all these results, they
are mitigated by interactions. However, our results show that the GFC contributed
signi…cantly to change the distribution of the e¤ects of characteristics on forecast-
ing performance and boldness, a loosely documented fact.25
Interactions show that having a Master’s degree or a PhD in economics, ed-
ucation in a top university (economics ranking), working in a …nancial and local
institution, having experience in academia and working in a local institution, or
being a local forecaster working in a local institution, generally improve the ex-
pert’s forecasting performance over the full sample but also during the post-GFC
period. Before the GFC, having a degree in economics from a top university (eco-
nomics ranking), a degree in …nance from a top university (…nance ranking), or
holding a Master’s degree and working in a local institution improved expert’s
forecasting performances.
Putting aside interactions, the results reported in Table 7 are close to the
herding results except for one interesting instance: Having a degree in …nance led
to greater boldness before the GFC but to more herding behaviors after the GFC
and over the full sample. As a result of cognitive biases and an intuitive reaction
to uncertainty and …nancial instability, experts with lower risk tolerance may herd
more (Christo¤ersen and Stæhr, 2019). However, unlike forecasting performance,
being a local forecaster does not seem to in‡uence experts’ herding behaviors,
which slightly mitigates Clarke and Subramanian (2006).
Interactions reveal a more detailed picture. Better education quality in eco-
nomics26 improves experts’forecasting performance among those with a Masters
25 There is a broad literature describing how economic conditions a¤ect (expert) forecasts
(Adeney et al., 2017) or in‡ation perception, but to our knowledge, very few document how eco-
nomic conditions in‡uence the changing e¤ect of experts’characteristics on forecasting accuracy
and boldness.
26 Measured with the ShanghaiRanking’s Global Ranking of Academic Subjects in Economics,
23
or a PhD degree. This was not the case for experts who graduated with a Mas-
ter’s degree from a non-top-ranked university. However, better education quality
in economics and …nance generally accentuates experts’herding behaviors, except
for experts who graduated in economics from a top-ranked university in …nance.
The latter characteristics tend to increase experts’boldness.
Working in a …nancial and local institution improves experts’forecasting per-
formances while it weakly in‡uences herding behaviors. However, working in a
local institution while having experience in academia increases both forecasting
performance and herding behaviors. This is mitigated if local experts working
in a local institution are considered. Working in a local institution and having
a Masters degree or a PhD decreases both forecasting performance and bolder
behaviors. The previous …ndings are con…rmed for local experts having a Masters
degree or a PhD. Local forecasters with a degree in economics tend to provide low-
quality forecasts that are far from the consensus. Interestingly, having experience
in academia and working in a …nancial institution does not signi…cantly in‡uence
experts’herding behavior, but does a¤ect their forecasting performance. Simi-
larly, having a degree in economics and having a Masters degree a¤ect forecasting
performance and herding behaviors derently. The …ndings that mix geography
and education relate to several strands of the literature. Our results demonstrate
that the likelihood of boldness increases with the experts forecasting performance
and experience, and is in‡uenced by institution (Clement and Tse, 2005). Our
results also show that experience (Hong et al., 2000; Mikhail et al., 2003) and
education (De Franco and Zhou, 2009) in‡uence social interactions, cognitive bi-
ases, and intuitive reaction to uncertainty, an interpretation partially shared with
Christo¤ersen and Stæhr (2019).
Female experts having central bank experience outperform men having central
bank experience (E GCB), which may con…rm a labor market entry selection
bias. Female experts in a market segment in which their concentration is lower
(central banking) appear to have better-than-average skills due to self-selection
(Kumar, 2010).
While di¤erences in views may persist through time, di¤erences in information
sets only cannot explain such di¤erences in opinion. Patton and Timmermann
(2010) show they stem from heterogeneity in priors or models and that di¤erences
in opinion move countercyclically. Although this heterogeneity is strongest under
recessions, our results bring another layer to this conclusion. The GFC not only
changed di¤erences in opinion but also modi…ed the inuence of expert’s charac-
teristics on their forecasting performance and boldness.
see Section 4.
24
5.4 Pessimism
In our last analysis, we estimate "pessimism." When looking at the behavioral side
of forecasts, it seems evident that one of the most relevant questions is whether
forecasts–i.e., expectationsare optimistic or pessimistic. While distinguishing op-
timism from pessimism is quite straightforward for business cycle forecasts, it is
less so for in‡ation, where "good" and "bad" are less clearly de…ned. One might
look at the deviation from target in‡ation, but it is hard to argue that 1.9% in‡a-
tion is worse than 2.0%.
However, our sample includes two brief episodes when the US economy was
endangered by–and in some months experiencing–de‡ation. Unlike low in‡ation,
de‡ation is almost universally considered highly problematic in economics, allowing
us to use those periods to assess the question of optimism vs. pessimism. Figure
3 shows how in the early period of the GFC (late 2008 to early 2009), and then
again over most of 2015, forecasters disagreed on whether or not there would be
de‡ation.
Figure 3: Forecast spread and periods of mixed in‡ation/de‡ation expectations.
-3,0
-2,0
-1,0
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
%
Note: The gray area represents the spread of forecasts (disagreement). The blue shaded back-
ground highlights the situations where both de‡ation and in‡ation were considered possible by
forecasters.
For those subsamples we estimate a panel probit model explaining the proba-
blity that the forecast Eit (it)would be below zero, taking the form:
25
p(Eit (it)<0) =  ( 0ln xit +Xi) ;(14)
where is the cumulative distribution function of the standard normal distribu-
tion, and is a vector of regression coe¢ cients.
We then split this sample even further into the subsamples when de‡ation
was observed (i.e., when "pessimism" was justi…ed, or–in other terms–the lack
of de‡ation expectation was overly optimistic) vs. periods where no de‡ation was
observed, i.e., when expecting in‡ation can genuinely be seen as overly pessimistic.
Table 8 presents the probit panel regressions for a general situation (i.e., during
both in‡ationary and de‡ationary periods) and during only deationary periods.
Table 8: Probit Estimates for Pessimism
Performance
All De‡ation
Experience -0.628*** -0.557***
Local forecaster (LF) -0.165 -0.088
Local institution (LI) -0.228 -0.272
Financial institution (FI) -0.172 -0.067
Experience: Central bank (CB) -0.965*** -0.576
Experience: Academia (AC) 0.598 0.616
Gender (G) 0.116 -0.166
Education: MA (MA) 0.398 0.138
Education: PhD (PhD) 0.375 0.294
Education: Economics (EC) 14.538 15.208
Education: Finance (EF) 14.668 15.542
Ranking: Economics (RE) 0.129 0.184
Ranking: Finance (RF) -0.095 -0.197
Observations 336 212
Note: ***, **, and * indicate signi…cance at the 1%, 5%, and 10% levels, respectively. All stands
for both in‡ation and de‡ation periods, and De‡ation stands for only de‡ation periods. Both
samples (All ) are restricted to periods where there was disagreement, i.e., some experts predicted
de‡ation while others did not. The second sample (De‡ation ) is a subsample of All, where the
pessimists turned out to be correct.
Table 8 shows that experts with more experience or with central bank ex-
perience are less likely to predict de‡ation. However, this propensity becomes in-
signi…cant under de‡ation for experts with central bank experience, which strongly
in‡uences private information. These experts are less pessimistic, but this is mit-
igated when pessimism turns out to be justi…ed.
26
In line with our model (Section 3), these results show that the type and dura-
tion of an expert’s experience drive his or her sentiment. Their degree of pessimism
(optimism) may interact with their private signal processing. According to Man-
zan (2011), heterogeneous sentiments may in‡uence experts in deciphering newly
available information, involving a positive relationship between the interpretation
of the mean signal and the prior sentiment (pessimistic or optimistic). Hence, the
ect of prior sentiment on an expert’s information processing, exposed in Section
3, may depend on the expert’s experience, both duration and type (central bank
or not).
We also complement the view stating that disagreement stems from hetero-
geneity in prior sentiments and moves countercyclically, with heterogeneity being
strongest during recessions, when forecasters appear to place greater weight on
their prior beliefs (Patton and Timmermann, 2010). Our results may con…rm that
the weight of experts’prior beliefs, and the beliefs themselves, are in‡uenced by
both experience and prior experience in a central bank, as well as the current state
of in‡ation (or de‡ation).
5.5 Forecasting Ability
The characteristics identi…ed in Section 5 should improve the predictive ability
of experts. We analyze time variation in the out-of-sample relative forecasting
performance to test this hypothesis. More precisely, we test for relative forecast
performance of characteritics-based expert groups in unstable environments, as
proposed by Giacomini and Rossi (2010). The null hypothesis is that the fore-
casts under consideration perform equally well at every point in time. Exceeding
the critical value does not imply that one group of expert forecasts constantly
outperforms the other, but merely that there is a meaningful di¤erence in predic-
tive ability for a subsample. Fig. 4 presents these ‡uctuation tests for forecasts
grouped by characteristics.
The test rejects only for education level characteristics (MA and PhD, see Fig.
4) that is, experts with a PhD di¤er from the others signi…cantly at least once,
and so do forecasters where the top degree is a Masters’s degree. Due to the low
number of experts holding neither an MA nor a PhD, those two tests capture
almost the same information. Thus, it is unsurprising that the rejection is driven
by the same period in the mid-2000s when PhD educated experts underperformed
the competition (mostly MAs), and MAs outperformed their competition (mostly
PhDs).
Interestingly, the test does not reject for the remaining characteristics, con…rm-
ing that experts’characteristics may have a signi…cant role in their out-of-sample
forecasting outcomes even in unstable environments. Our …ndings may have impli-
cations for policymakers. They may decide to select their inputs (multiple expert
27
Figure 4: Forecasting Ability Tests
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-2
0
2Local forecaster
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-2
0
2Local institution
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-2
0
2Financial institution
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-2
0
2Experience: central bank
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-5
0
5Education: MA
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-5
0
5Education: Ph.D.
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
-2
0
2Gender
Note: The red dashed lines represent the critical values of the predictive ability test.
forecasts) according to the expert’s characteristics to maximize their forecasting
precision and optimize their decisions.
Our results also provide a possible basis for an alternative to conventional fore-
cast combination methods in the literature. While the bias-adjusted combination
method is found to work well in practice (Capistrán and Timmermann, 2009b), we
demonstrate that a characteristics-based forecast combination is potentially more
desirable than equal-weighted or bias-adjusted forecast combination methods.
In‡ation forecasts from the Survey of Professional Forecasters are biased, pre-
senting positive serial correlation in forecast errors, cross-sectional dispersion, and
predictability patterns depending on in‡ation variance. As we control for time
xed-e¤ects, we interpret experts’shifts in forecasting performance not explained
28
by asymmetric loss and rational expectations (Capistrán and Timmermann, 2009a)
through their characteristics rather than in‡ation variance.
All in all, combining forecasts with respect to the expert’s characteristics gen-
erally improves out-of-sample forecasting performance.
6 Policy Implications and conclusion
In line with our model, the characteristics we identify in‡uence the expertsper-
formance, boldness, forecasting ability, and sentiment (optimism or pessimism).
Investors and policymakers use forecasts to design or explain their decisions,
and sometimes, the e¢ ciency of these decisions depends on forecasts (European
Central Bank, 2011, 2014; de Vincent-Humphreys et al., 2019). Identifying the
characteristics of the best experts could help …rms and policymakers to achieve
their objectives e¢ ciently. For instance, Carvalho and Nechio (2014) show that
households’macroeconomic forecasts–about interest rates, in‡ation, and unemployment–
are not uniform across income and education levels. Forecasts also constitute an
essential information channel leading investment portfolio decisions. The more
accurate the forecast, the less a surprise could occur, minimizing the required ad-
justment costs of the investment portfolio and the corresponding market volatility
when the data will be made public (Laster et al., 1999).
Policy institutions extensively use expert forecasts for both decision making and
forecasting purposes (Piotroski and Roulstone, 2004; Adeney et al., 2017). While
policymakers generally aggregate these forecasts in simplistic ways and rank them
without considering the expert characteristics (Alessi et al., 2014; Coibion et al.,
2020), the main takeaway from our results is that experts’ characteristics drive
forecasting outcomes, boldness, and sentiment.
Consequently, policymakers may use our results to group forecasters with re-
spect to some of their characteristics (Section 5.5) to increase the reliability of
their in‡ation forecasts compared to simple averaging, thus improving their policy
decision making processes. This should also hamper the spillover e¤ect of pes-
simistic or optimistic behaviors on in‡ation forecasts, which is somehow frequent
during speci…c periods such as de‡ation, if policymakers group expert forecasts
according to experience (Section 5.4).
The current state of in‡ation (or de‡ation) and of the economy (after or be-
fore a crisis like the GFC) in‡uence experts’behaviors and beliefs, and thus the
transmission channels from their characteristics to their forecasting performance
and boldness as demonstrated here.
Underperforming experts are more likely to no longer be part of our expert
database, i.e., they are less likely to be in charge of the in‡ation forecasts con-
tributed to the Bloomberg database, while boldness does not signi…cantly inu-
29
ence the experts’survival rate. Related to career concerns and institutional labor
market expectations, degrees in …nance or economics do not o¤er the same protec-
tion, while having graduated from a top university decreases the expert’s survival
rate.27 Our survival analysis also sheds light on the expert’s age, unrelated to
accumulated experience, which appears to decrease expert survival. The older the
expert (above 50 years old), the less a bold expert will survive.
The comparison of our results from the two subsamples reveals that the GFC
changed both …nancial institutions and the expert’s labor market. After the GFC,
expert’s experience, location, institution type, or education …eld or quality change
their forecasting performance. Our panel estimations also show that before the
GFC, only the expert’s age, institution type, education …eld, and quality matter
for forecasting performance and boldness.
The expert’s location, institution location and type, experience type, and gen-
der a¤ect the expert’s forecasting ability. The expert’s past experience or previous
experience in a central bank signi…cantly in‡uence the experts sentiment.
We interpret our results as evidence of a characteristics e¤ect in in‡ation expert
outcomes. One implication of our analysis is that experts’ characteristics and
experience matter for policymakers as long as expert forecasts are considered in
their decision-making process.
References
Adeney, R., Arsov, I., Evans, R., 2017. In‡ation expectations in advanced
economies. RBA Bulletin March Quarter, Reserve Bank of Australia.
Alessi, L., Ghysels, E., Onorante, L., Peach, R., Potter, S., 2014. Central bank
macroeconomic forecasting during the global …nancial crisis: the European Cen-
tral Bank and Federal Reserve Bank of New York experiences. Journal of Busi-
ness & Economic Statistics 32 (4), 483–500.
Andrade, P., Le Bihan, H., 2013. Inattentive professional forecasters. Journal of
Monetary Economics 60 (8), 967–982.
Bachmann, R., Elstner, S., Sims, E. R., 2013. Uncertainty and economic activity:
evidence from business survey data. American Economic Journal: Macroeco-
nomics 5 (2), 217–249.
Bae, K.-H., Stulz, R. M., Tan, H., 2008. Do local analysts know more? A cross-
country study of the performance of local analysts and foreign analysts. Journal
of Financial Economics 88 (3), 581–606.
27 See MA RF and PhD RE in Table 7.
30
Batchelor, R., 2007. Bias in macroeconomic forecasts. International Journal of
Forecasting 23 (2), 189203.
Benchimol, J., El-Shagi, M., 2020. Forecast performance in times of terrorism.
Economic Modelling 91, 386–402.
Berger, H., Ehrmann, M., Fratzscher, M., 2009. Forecasting ECB monetary policy:
Accuracy is a matter of geography. European Economic Review 53 (8), 1028–
1041.
Bernanke, B. S., 2007. Ination expectations and in‡ation forecasting. Speech 306,
Board of Governors of the Federal Reserve System (U.S.).
Bloom, N., 2014. Fluctuations in uncertainty. Journal of Economic Perspectives
28 (2), 153–176.
Bordalo, P., Gennaioli, N., Ma, Y., Shleifer, A., 2020. Overreaction in macroeco-
nomic expectations. American Economic Review 110 (9), 2748–2782.
Breslow, N., Chatterjee, N., 1999. Design and analysis of two-phase studies with
binary outcome applied to Wilms tumour prognosis. Journal of the Royal Sta-
tistical Society Series C 48 (4), 457–468.
Budescu, D. V., Chen, E., 2015. Identifying expertise to extract the wisdom of
crowds. Management Science 61.
Campbell, S. D., Sharpe, S. A., 2009. Anchoring bias in consensus forecasts and its
ect on market prices. Journal of Financial and Quantitative Analysis 44 (2),
369–390.
Capistrán, C., Timmermann, A., 2009a. Disagreement and biases in in‡ation ex-
pectations. Journal of Money, Credit and Banking 41 (2-3), 365–396.
Capistrán, C., Timmermann, A., 2009b. Forecast combination with entry and exit
of experts. Journal of Business & Economic Statistics 27 (4), 428–440.
Carvalho, C., Nechio, F., 2014. Do people understand monetary policy? Journal
of Monetary Economics 66 (C), 108123.
Christo¤ersen, J., Stæhr, S., 2019. Individual risk tolerance and herding behaviors
in …nancial forecasts. European Financial Management 25 (5), 13481377.
Clarke, J., Subramanian, A., 2006. Dynamic forecasting behavior by analysts:
Theory and evidence. Journal of Financial Economics 80 (1), 81–113.
31
Clement, M. B., Tse, S. Y., 2005. Financial Analyst Characteristics and Herding
Behavior in Forecasting. Journal of Finance 60 (1), 307–341.
Coibion, O., Gorodnichenko, Y., 2015. Information rigidity and the expectations
formation process: a simple framework and new facts. American Economic Re-
view 105 (8), 2644–2678.
Coibion, O., Gorodnichenko, Y., Kumar, S., Pedemonte, M., 2020. In‡ation ex-
pectations as a policy tool? Journal of International Economics 124, 103297.
Cornand, C., Hubert, P., 2020. On the external validity of experimental in‡ation
forecasts: A comparison with …ve categories of …eld expectations. Journal of
Economic Dynamics and Control 110, 103746.
Cox, D. R., 1972. Regression models and life-tables. Journal of the Royal Statistical
Society: Series B 34 (2), 187202.
Davis, M., Lleo, S., 2020. Debiased expert forecasts in continuous-time asset allo-
cation. Journal of Banking & Finance 113 (C), 105759.
De Franco, G., Zhou, Y., 2009. The performance of analysts with a CFA desig-
nation: The role of human-capital and signaling theories. Accounting Review
84 (2), 383–404.
de Vincent-Humphreys, R., Dimitrova, I., Falck, E., Henkel, L., Meyler, A., 2019.
Twenty years of the ECB Survey of Professional Forecasters. Economic Bulletin
Articles 1.
El-Shagi, M., Giesen, S., Jung, A., 2016. Revisiting the relative forecast perfor-
mances of Fed sta¤ and private forecasters: A dynamic approach. International
Journal of Forecasting 32 (2), 313–323.
European Central Bank, 2011. Monthly Bulletin. Executive Board 01/2011, Euro-
pean Central Bank.
European Central Bank, 2014. Monthly Bulletin. Executive Board 01/2014, Euro-
pean Central Bank.
Fang, L., Huang, S., 2017. Gender and connections among Wall Street analysts.
Review of Financial Studies 30 (9), 3305–3335.
Gabaix, X., 2019. Behavioral inattention. In: Bernheim, B. D., DellaVigna, S.,
Laibson, D. (Eds.), Handbook of Behavioral Economics - Foundations and Ap-
plications 2. Vol. 2 of Handbook of Behavioral Economics: Applications and
Foundations 1. Elsevier, pp. 261–343.
32
Gabaix, X., 2020. A behavioral New Keynesian model. American Economic Review
110 (8), 2271–2327.
Genre, V., Kenny, G., Meyler, A., Timmermann, A., 2013. Combining expert
forecasts: Can anything beat the simple average? International Journal of Fore-
casting 29 (1), 108–121.
Giacomini, R., Rossi, B., 2010. Forecast comparisons in unstable environments.
Journal of Applied Econometrics 25 (4), 595620.
Heckman, J. J., 1979. Sample selection bias as a speci…cation error. Econometrica
47 (1), 153–161.
Hong, H., Kacperczyk, M., 2010. Competition and bias. Quarterly Journal of
Economics 125 (4), 1683–1725.
Hong, H., Kubik, J. D., Solomon, A., 2000. Security analysts’career concerns and
herding of earnings forecasts. RAND Journal of Economics 31 (1), 121–144.
Huang, R., Krishnan, M., Shon, J., Zhou, P., 2017. Who herds? Who doesn’t?
Estimates of analysts’herding propensity in forecasting earnings. Contemporary
Accounting Research 34 (1), 374–399.
Jacob, J., Lys, T. Z., Neale, M. A., 1999. Expertise in forecasting performance of
security analysts. Journal of Accounting and Economics 28 (1), 5182.
Kaplan, E. L., Meier, P., 1958. Nonparametric estimation from incomplete obser-
vations. Journal of the American Statistical Association 53 (282), 457–481.
Keane, M. P., Runkle, D. E., 1990. Testing the rationality of price forecasts: new
evidence from panel data. American Economic Review 80 (4), 714–735.
Kumar, A., 2010. Self-selection and the forecasting abilities of female equity ana-
lysts. Journal of Accounting Research 48 (2), 393–435.
Laster, D., Bennett, P., Geoum, I. S., 1999. Rational bias in macroeconomic fore-
casts. Quarterly Journal of Economics 114 (1), 293–318.
Lim, T., 2001. Rationality and analysts’forecast bias. Journal of Finance 56 (1),
369–385.
Malloy, C. J., 2005. The geography of equity analysis. Journal of Finance 60 (2),
719–755.
Malmendier, U., Nagel, S., 2011. Depression babies: do macroeconomic experiences
ect risk taking? Quarterly Journal of Economics 126 (1), 373416.
33
Manzan, S., 2011. Di¤erential interpretation in the survey of professional forecast-
ers. Journal of Money, Credit and Banking 43 (5), 993–1017.
Marimon, R., Sunder, S., 1993. Indeterminacy of equilibria in a hyperin‡ationary
world: experimental evidence. Econometrica 61 (5), 1073–1107.
Mikhail, M. B., Walther, B. R., Willis, R. H., 2003. The ect of experience on
security analyst underreaction. Journal of Accounting and Economics 35 (1),
101–116.
Mincer, J. A., 1974. Schooling, Experience, and Earnings. NBER Books. New
York, NY: National Bureau of Economic Research.
Mitchell, K., Pearce, D. K., 2007. Professional forecasts of interest rates and ex-
change rates: Evidence from the Wall Street Journal’s panel of economists.
Journal of Macroeconomics 29 (4), 840–854.
Patton, A. J., Timmermann, A., 2010. Why do forecasters disagree? Lessons from
the term structure of cross-sectional dispersion. Journal of Monetary Economics
57 (7), 803–820.
Piotroski, J. D., Roulstone, B. T., 2004. The e¤ects of analyst forecast properties
and country-level institutions on the cost of debt. Accounting Review 79 (4),
1119–1151.
Romer, C. D., Romer, D. H., 2000. Federal reserve information and the behavior
of interest rates. American Economic Review 90 (3), 429–457.
Rossi, B., Sekhposyan, T., 2015. Macroeconomic uncertainty indices based on
nowcast and forecast error distributions. American Economic Review 105 (5),
650–655.
Scharfstein, D. S., Stein, J. C., 1990. Herd behavior and investment. American
Economic Review 80 (3), 465–479.
Thomas, L. B., 1999. Survey measures of expected U.S. ination. Journal of Eco-
nomic Perspectives 13 (4), 125–144.
Veress, A., Kaiser, L., 2017. Forecasting quality of professionals: Does a¢ liation
matter? Quarterly Review of Economics and Finance 66 (C), 159–168.
Zwiebel, J., 1995. Block investment and partial bene…ts of corporate control. Re-
view of Economic Studies 62 (2), 161–185.
34
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Governments, central banks, and private companies make extensive use of expert and market-based forecasts in their decision-making processes. These forecasts can be affected by terrorism, a factor that should be considered by decision-makers. We focus on terrorism as a mostly endogenously driven form of political uncertainty and assess the forecasting performance of market-based and professional inflation and exchange rate forecasts in Israel. We show that expert forecasts are better than market-based forecasts, particularly during periods of terrorism. However, the performance of both market-based and expert forecasts is significantly worse during such periods. Thus, policymakers should be particularly attentive to terrorism when considering inflation and exchange rate forecasts.
Article
Full-text available
We assess the prospects for central banks using inflation expectations as a policy tool for stabilization purposes. We review recent work on how expectations of agents are formed and how they affect their economic decisions. Empirical evidence suggests that inflation expectations of households and firms affect their actions but the underlying mechanisms remain unclear, especially for firms. Two additional limitations prevent policy-makers from being able to actively manage inflation expectations. First, available surveys of firms' expectations are systematically deficient, which can only be addressed through the creation of large, nationally representative surveys of firms. Second, neither households' nor firms' expectations respond much to monetary policy announcements in low-inflation environments. We provide suggestions for how monetary policy-makers could pierce this veil of inattention through new communication strategies as well as the potential pitfalls to trying to do so.
Article
We study the rationality of individual and consensus forecasts of macroeconomic and financial variables using the methodology of Coibion and Gorodnichenko (2015), who examine predictability of forecast errors from forecast revisions. We find that individual forecasters typically overreact to news, while consensus forecasts under-react relative to full-information rational expectations. We reconcile these findings within a diagnostic expectations version of a dispersed information learning model. Structural estimation indicates that departures from Bayesian updating in the form of diagnostic overreaction capture important variation in forecast biases across different series, yielding a belief distortion parameter similar to estimates obtained in other settings. (JEL C53, D83, D84, E13, E17, E27, E47)
Article
This paper analyzes how bounded rationality affects monetary and fiscal policy via an empirically relevant enrichment of the New Keynesian model. It models agents’ partial myopia toward distant atypical events using a new microfounded “cognitive discounting” parameter. Compared to the rational model, (i) there is no forward guidance puzzle; (ii) the Taylor principle changes: with passive monetary policy but enough myopia equilibria are determinate and economies stable; (iii) the zero lower bound is much less costly; (iv) price-level targeting is not optimal; (v) fiscal stimulus is effective; (vi) the model is “ neo-Fisherian” in the long run, Keynesian in the short run. (JEL E12, E31, E43, E52, E62, E70)
Article
Expert forecasts are an essential component of asset management and an important research topic. However, the effect of behavioral biases on expert forecasts is generally ignored. This paper examines the effect of biased expert forecasts on asset allocations. We find that biases have a significant impact on portfolios, explaining nearly 70% of excess risk-taking in our implementation. To address the effect of behavioral biases, we propose an integrated behavioral continuous-time portfolio selection model which we solve in closed form. The model applies general principles to identify and reduce the impact of five main behavioral biases. This paper concludes with a new personal fractional Kelly decomposition to account for the effect of opinions on the optimal asset allocation.
Article
Establishing the external validity of experimental inflation forecasts is essential if laboratory experiments are to be used as decision-making tools for monetary policy. Our contribution is to document whether different measures of inflation expectations, based on various categories of agents (participants in experiments, households, industry forecasters, professional forecasters, financial market participants and central bankers), share common patterns. We do so by analyzing the forecasting performance of these different categories of data, their deviations from full information rational expectations, and the variables that enter the determination of these expectations. Overall, the different categories of forecasts exhibit common features: forecast errors are comparably large and autocorrelated, and forecast errors and forecast revisions are predictable from past information, suggesting the presence of some form of bounded rationality or information imperfections. Finally, lagged inflation positively affects the determination of inflation expectations. While experimental forecasts are relatively comparable to survey and financial market data, more heterogeneity is observed compared to central bank forecasts.
Article
In this essay, the author addresses four questions about uncertainty. First, what are some facts and patterns about economic uncertainty? Both macro and micro uncertainty appear to rise sharply in recessions and fall in booms. Uncertainty also varies heavily across countries - developing countries appear to have about one-third more macro uncertainty than developed countries. Second, why does uncertainty vary during business cycles? Third, do fluctuations in uncertainty affect behavior? Fourth, has higher uncertainty worsened the Great Recession and slowed the recovery? Much of this discussion is based on research on uncertainty from the last five years, reflecting the recent growth of the literature.
Article
Economists fulfilling a public mission - namely academics, Fed and government employees - demonstrate a tendency towards being pessimistic, whereas bankers in general are overly optimistic about future stock market developments. We show that these characteristics are of particular relevance and statistically significant during economic recessions and stock market downturns. Whilst investment bankers have always shown a tendency towards being optimistic, other affiliations are increasingly following their footsteps and most dominantly so for short-term forecasts. Especially during the most recent crises, their expectations of fast rebounds remained largely unsatisfied, questioning the applicability of economists' forecasts as we move forward.
Article
We develop parametric estimates of the imitation-driven herding propensity of analysts and their earnings forecasts. By invoking rational expectations, we solve an explicit analyst optimization problem and estimate herding propensity using two measures: First, we estimate analysts' posterior beliefs using actual earnings plus a realization drawn from a mean-zero normal distribution. Second, we estimate herding propensity without seeding a random error, and allow for nonorthogonal information signals. In doing so, we avoid using the analyst's prior forecast as the proxy for his posterior beliefs, which is a traditional criticism in the literature. We find that more than 60 percent of analysts herd toward the prevailing consensus, and herding propensity is associated with various economic factors. We also validate our herding propensity measure by confirming its predictive power in explaining the cross-sectional variation in analysts' out-of-sample herding behavior and forecast accuracy. Finally, we find that forecasts adjusted for analysts' herding propensity are less biased than the raw forecasts. This adjustment formula can help researchers and investors obtain better proxies for analysts' unbiased earnings forecasts.