Cognitive abilities and inflation expectations - KIT

Page created by Enrique Hammond
 
CONTINUE READING
Cognitive abilities and inflation expectations - KIT
Cognitive abilities and
                                   inflation expectations

                                   by Francesco D‘Acunto, Daniel Hoang,
                                   Maritta Paloviita and Michael Weber

                                   No. 126 | JANUARY 2019

                                   WORKING PAPER SERIES IN ECONOMICS

KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft             econpapers.wiwi.kit.edu
Impressum

Karlsruher Institut für Technologie (KIT)
Fakultät für Wirtschaftswissenschaften
Institut für Volkswirtschaftslehre (ECON)

Kaiserstraße 12
76131 Karlsruhe

KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft

Working Paper Series in Economics
No. 126, January 2019

ISSN 2190-9806

econpapers.wiwi.kit.edu
Cognitive Abilities and Inflation Expectations
  By Francesco D’Acunto, Daniel Hoang, Maritta Paloviita, and Michael Weber∗

  Over the last few years, interest has re-                   the variation in their macroeconomic ex-
vived among economists in understanding                       pectations. Individuals with mid-to-low IQ
how households form and update their ex-                      levels have absolute forecast errors for 12-
pectations, as well as the determinants of                    month-ahead inflation rates 2.5 times as
the cross-sectional variation in economic                     large as individuals with high IQ levels, and
expectations across households (for a dis-                    about 4 times as large as the average re-
cussion, see Gennaioli and Shleifer (2018);                   alized inflation rate throughout the sam-
D’Acunto, Prokopczuk and Weber (2019);                        ple period. D’Acunto et al. (2018b) further
D’Acunto et al. (2018c)). Inflation expec-                    document mid-to-low IQ individuals are
tations have been at the center stage of this                 about 4 times more likely to provide round
strand of research, because of their rele-                    numbers when asked about their inflation
vance as a policy tool, especially when in-                   forecasts – which might reflect higher un-
terest rates are low (Coibion et al. (2018)).                 certainty in the forecasts – and are twice as
  Forming and updating expectations                           likely to report implausible values for their
about future macroeconomic variables such                     average forecast.
as inflation requires the use of cognitive                       In a related paper, D’Acunto et al.
resources at several stages, ranging from                     (2018a) further show cognitive abilities are
collecting information about the prevailing                   an important friction to the effectiveness
inflation rates to forecasting future poten-                  of economic policies that aim to affect the
tial states of the world and their likelihood.                overall economy through managing house-
Cognitive abilities thus appear as a natural                  holds’ expectations (D’Acunto, Hoang and
potential determinant of the cross-sectional                  Weber (2016) discuss unconventional fiscal
variation in inflation expectations.                          policy as one example). The facts that a
  Indeed, D’Acunto et al. (2018b) docu-                       substantial part of a representative popu-
ment in a representative population that                      lation of economic agents (i) barely form
the variation in cognitive abilities across in-               plausible inflation expectations and (ii) do
dividuals is an important determinant of                      not react to changes in their inflation expec-
                                                              tations when making plans about consump-
  ∗  Carroll School of Management, Boston College,            tion and saving choices drive this result.
Chestnut Hill, MA, USA. e-Mail: dacuntof@bc.edu.                 A direct consequence of policies that aim
Hoang: Department for Finance and Banking, Karl-
sruhe Institute of Technology, Karlsruhe, B-W, Ger-
                                                              to stimulate consumption expenditure via
many.      e-Mail:    daniel.hoang@kit.edu.     Palovi-       managing inflation expectations is that the
ita: Bank of Finland, Helsinki, Finland.        e-Mail:       policy measure might be less effective than
Maritta.Paloviita@bof.fi. Weber: Booth School of Busi-        a representative-agent model implies. An
ness, University of Chicago, Chicago, IL, USA and
NBER. This research was conducted with restricted ac-
                                                              indirect, unintended consequence of policies
cess to data from the Finnish Armed Forces and Statis-        that aim to affect behavior via managing
tics Finland. The views expressed here are those of the       expectations is, instead, that such policies
authors and do not necessarily reflect the views of the       might imply a redistribution of resources
Bank of Finland, Finnish Armed Forces, or Statistics
Finland. We thank the project coordinator at Statis-
                                                              from the mid-to-low parts of the distribu-
tics Finland, Valtteri Valkonen, for his help with the        tion by cognitive ability to the higher end
data and for very insightful comments. We gratefully          of the distribution.
acknowledge financial support from the Deutsche Bun-
desbank. Weber also gratefully acknowledges financial                  I.   IQ: Subcomponents
support from the University of Chicago Booth School of
Business, the Fama Research Fund at the University of
Chicago Booth School of Business, and the Fama-Miller           The research described so far has treated
Center.                                                       cognitive abilities as a catch-all concept.
                                                          1
2

Cognition though is a complex human char-          fective or required substantial investment
acteristic and includes several facets that        in financial education, cheap advice mech-
are not necessarily highly correlated. For         anisms such as robo-advising tools indi-
instance, individuals who have strong quan-        viduals could easily access through their
titative cognitive skills need not have also       phones might improve the quality of low-
strong verbal or visuospatial skills.              IQ households’ expectations and choice
   At the same time, one might expect that         (D’Acunto, Prabhala and Rossi (forthcom-
all these three facets of cognition are im-        ing) and D’Acunto et al. (2019)).
portant in the determination of expecta-
tions regarding future economic variables.                     II.    Data and Results
Quantitative skills are crucial for individu-
als to map changes in price levels into infla-        In this paper, we propose a first step to
tion rates. Visuospatial skills are important      understand which types of cognitive abili-
because they allow individuals to abstract         ties might matter more or less to explain
from their personal situation and form plau-       the heterogeneity in inflation expectations
sible scenarios about future general infla-        across individuals.
tion and other macroeconomic variables as             We build on the empirical setting of
well as attach plausible probabilities to          D’Acunto et al. (2018b) and D’Acunto et al.
these future states of the world. Ver-             (2018a), who merge administrative data
bal skills might matter because individuals        on cognitive ability test scores from the
need to obtain information about current,          Finnish Armed Forces (FAF) for a repre-
past, and potentially future states of the         sentative sample of men with unique infor-
world through sources such as newspaper            mation on their inflation expectations from
articles, television, or family and friends.       Statistics Finland.1 Grinblatt, Keloharju
   Understanding which types of cognitive          and Linnainmaa (2011) and Grinblatt et al.
abilities might matter to explain the ob-          (2015) use the data on cognitive abilities in
served differences in expectations is still a      finance research and discuss in detail the in-
widely open question. Answering this ques-         centives test takers have to put effect into
tion is important to assess the viability of       answering the questions.
potential interventions that aim to increase          For this paper, we exploit the fact that
the effectiveness of economic policies.            the 120 questions from the FAF test cog-
   Because short-term interventions are un-        nitive abilities across three types, namely,
likely to affect the cognitive abilities of eco-   quantitative, verbal, and visuospatial abili-
nomic agents, policymakers need to under-          ties. FAF aggregate the scores within each
stand the specific skills agents lack, in or-      category and standardizes them within co-
der to design policies that could effectively      horts of test takers so that the IQ rank-
substitute for such skills. For instance,          ings follow a stanine distribution. Stanine
if agents cannot process the information           (STAndard NINE) is a method of scaling
about economic variables from newspaper            test scores on a nine-point standard scale
articles or official central-bank statements,      with a mean of five and a standard devi-
policymakers might invest in communica-            ation of two. The respondents with the
tion strategies that aim to make such con-         lowest 4% of test scores are at least 1.75
cepts easy to grasp for the broader popu-          standard deviations from the mean and are
lation (Coibion, Gorodnichenko and Weber           assigned a standardized IQ score of 1, and
(2019)). Alternatively, information might          the 4% with the highest test scores are as-
be provided to economic agents in a vivid          signed a standardized score of 9. We thus
fashion that does not require high cognition       observe three scores per individual between
or knowledge about basic economic con-             1 and 9 based on the relative performance
cepts but relies on agents’ comparison of          in each part of the test.
their conditions with those of their peers
(D’Acunto, Rossi and Weber (2019)). If,               1 D’Acunto et al. (2018b) and D’Acunto et al. (2018a)

instead, information treatments were inef-         contain detailed descriptions of these data.
3

   The verbal part of the test asks individu-   cast error across individuals in the same bin
als to compare words and pairs of words and     by arithmetic, verbal, and visuospatial cog-
find synonyms and antonyms. In the arith-       nitive abilities.2
metic part, individuals have to solve simple       Figure 1 plots the average absolute fore-
arithmetic operations, solve verbal prob-       cast error by bin of cognitive abilities or-
lems, or compare pairs of numbers. The          dered from the lowest level of cognitive abil-
visuospatial part is similar to the Raven’s     ities (IQ=1) to the highest level of cognitive
progressive matrices test and requires indi-    abilities (IQ=9). Panel A refers to arith-
viduals to identify a missing item that com-    metic abilities, panel B to verbal abilities,
pletes a pattern.                               and panel C to visuospatial abilities.
   One might be concerned that the three           Note we only have one absolute forecast
scores we obtain are highly correlated          error per individual. Hence, to the extent
within individuals and that the different       we observe different patterns in average ab-
portions of the test are ultimately unable      solute forecast errors across different sub-
to capture different sources of variation in    categories of IQ, we conclude that individu-
cognitive abilities across individuals. In      als perform differently in different subparts
fact, we find the point estimates for the       of the IQ test.
pairwise correlations of the three scores in       The three panels of Figure 1 beget a
the individual-level data range from 0.56 to    few comments. First of all, across all
0.66. These statistics suggest that even if,    three types of cognitive abilities, we detect
indeed, a positive correlation exists between   a monotonically decreasing association be-
each pair of scores, we still detect enough     tween IQ levels and average absolute fore-
variation across the scores of each individ-    cast errors. This fact suggests none of the
ual to make our test meaningful.                types of cognitive abilities behaves differ-
   We merge this individual-level cognitive-    ently from the aggregate score for overall
ability information with the micro-data un-     cognitive abilities D’Acunto et al. (2018b)
derlying the Consumer Survey of Statistics      discuss in earlier research.
Finland, in which respondents report their         The second notable feature of Figure 1 is
numerical 12-month-ahead forecast for in-       the three scatter plots have different slopes,
flation on top of several other macro and in-   especially for the levels of cognitive abili-
dividual economic expectations and a large      ties up to the median bins (IQ=5). Specifi-
set of demographic characteristics. Every       cally, the curve is steepest at lower values by
month, the survey asks a representative re-     arithmetic IQ followed by verbal IQ and is
peated cross section of approximately 1,500     flattest at lower values by visuospatial IQ.
Finnish individuals questions about general     One way to interpret this fact is that the
and personal economic conditions, infla-        variation in visuospatial cognitive abilities
tion expectations, and willingness to spend     across individuals is less relevant in explain-
on consumption goods. Statistics Finland        ing the cross section of inflation forecasts
also collects additional information through    relative to the variation in arithmetic and
supplementary questions about households’       verbal IQ.
plans to save and borrow. The sample pe-           A third relevant result – not depicted in
riod is from January 2001 to March 2015.        Figure 1 – relates to statistical inference.
We use the numerical inflation forecasts at     We can reject the null hypothesis that any
the individual level to construct our main      of the forecast errors in any bin are equal
outcome of interest – the absolute forecast     to zero, as well as the null hypothesis that
error for 12-month-ahead inflation. We de-
fine this variable as the absolute value of         2 Because we compute the absolute value, we treat

the difference between the 12-month-ahead       deviations from the realized inflation rate in either direc-
inflation forecast of individual i in a given   tion identically. Our results are qualitatively unchanged
                                                if we repeat the whole analysis computing the average
month and the ex-post realized inflation        forecast error within each bin and hence allowing for
rate observed in Finland 12 months later.       positive and negative forecast errors across individuals
We then compute the average absolute fore-      within the same IQ bin to wash away.
4

                                       Figure 1. Mean Absolute Forecast Error for Inflation by Subcomponenents of IQ
                                            Panel A. IQ Arithmetic                                                                                         Panel B. IQ Verbal

                 5

                                                                                                                              5
    Mean Absolute Forecast Error

                                                                                                                 Mean Absolute Forecast Error
                        4

                                                                                                                                     4
         3

                                                                                                                      3
                 2

                                                                                                                              2
                                   0        2        4               6                          8        10                                     0     2        4             6      8   10
                                                  Normalized IQ Arithmetic                                                                                   Normalized IQ Verbal

                                                                                                Panel C. IQ Visuospatial
                                                                          5
                                                             Mean Absolute Forecast Error
                                                                  3       2      4

                                                                                            0   2        4              6                             8       10
                                                                                                     Normalized IQ Visuospatial

This figure plots the average absolute forecast error for inflation across IQ levels of Finnish men. Forecast error
is the difference between the numerical forecast for 12-month-ahead inflation and ex-post realized inflation. IQ
is the standardized test score from the Finnish Armed Forces that obtains integer values between 1 and 9.
Panel A reports results for arithmetic IQ, panel B reports results for verbal IQ, and panel C reports results for
visuospatial IQ. The sample period is from January 2001 to March 2015.

the averages across any adjacent bins across                                                                                                        III.   Concluding Remarks
each of the three sorting schemes are equal
for most bins. This result suggests that                                                                                Large heterogeneity exists in how indi-
even though the relevance of different types                                                                          viduals form, update, and act upon their
of cognitive abilities in explaining the cross                                                                        expectations, which previous research re-
section of forecast errors for inflation might                                                                        lates to individual-level cognitive abilities.
be higher or lower, higher scores in any                                                                              We show that all three subcomponents of
of the three components of IQ are system-                                                                             cognitive abilities we observe – arithmetic,
atically associated with lower forecast er-                                                                           verbal, and visuospatial – matter for the
rors on average, and hence each component                                                                             forecast accuracy for inflation, with the ef-
might matter.                                                                                                         fect of arithmetic cognitive abilities being
   Based on our data, we conclude that                                                                                strongest.
arithmetic, verbal, and visuospatial cogni-                                                                             The results could suggest policymak-
tive abilities are all likely to be relevant in                                                                       ers should design targeted communication
explaining the cross section of inflation ex-                                                                         strategies for different subpopulations. For
pectations in a representative population of                                                                          instance, short and vivid messages such
men, even though arithmetic cognitive abil-                                                                           tweets might be more successful in reaching
ities seem to be most relevant, especially for                                                                        the lowest part of the population by cog-
those scoring lowest.                                                                                                 nitive abilities, whereas more detailed and
COGNITIVE ABILITIES AND INFLATION EXPECTATIONS              5

technical reports could still be relevant to   D’Acunto,     Francesco,      Marcel
shape the expectations and choices of high-     Prokopczuk, and Michael We-
IQ individuals.                                 ber. 2019. “Historical Antisemitism,
  Moreover, policymakers might propose          Ethnic Specialization, and Financial
simple and salient policy measures to en-       Development.” Review of Economic
sure large parts of the targeted population     Studies.
react to the policy and to alleviate con-
cerns about unintended consequences, such      D’Acunto,    F, Thomas Rauter,
as a redistribution from lower parts of the     Christop Scheuch, and Michael
IQ distribution to higher parts (D’Acunto,      Weber. 2019. “Mobile Overdraft Fa-
Hoang and Weber (2018)).                        cilities and Consumption Behavior:
                                                Evidence from FinTech.” Working
             REFERENCES                         Paper.

Coibion,   Olivier,   Yuriy Gorod-             D’Acunto, F, U Malmendier, J Os-
 nichenko,    Saten Kumar,        and           pina, and M Weber. 2018c. “Salient
 Mathieu Pedemonte. 2018. “Infla-               Price Changes, Inflation Expectations,
 tion Expectations as a Policy Tool?”           and Household Behavior.” Working Pa-
 National Bureau of Economic Research.          per.
Coibion,    Olivier,    Yuryi Gorod-           Gennaioli, N, and A Shleifer. 2018. “A
 nichenko, and Michael Weber. 2019.             Crisis of Beliefs – Investor Psychology
 “Monetary Policy Communication and             and Financial Fragility.” Princeton Uni-
 Inflation Expectations.” Working Paper.        versity Press.
D’Acunto, F, Alberto Rossi, and                Grinblatt, Mark, Matti Keloharju,
 Michael Weber. 2019. “Crowdsourcing            and Juhani Linnainmaa. 2011. “IQ
 Financial Information to Change Spend-         and Stock Market Participation.” The
 ing Behavior.” Working Paper.                  Journal of Finance, 66(6): 2121–2164.
D’Acunto, F, N Prabhala, and                   Grinblatt, Mark, Seppo Ikäheimo,
 Alberto Rossi. forthcoming. “The               Matti Keloharju,    and Samuli
 Promises and Pitfalls of Robo-Advising.”       Knüpfer. 2015. “IQ and Mutual
 Review of Financial Studies.                   Fund Choice.” Management Science,
D’Acunto, Francesco, Daniel Hoang,              62(4): 924–944.
 and Michael Weber. 2016. “The effect
 of unconventional fiscal policy on con-
 sumption expenditure.” National Bureau
 of Economic Research.
D’Acunto, Francesco, Daniel Hoang,
 and Michael Weber. 2018. “Unconven-
 tional Fiscal Policy.” AEA Papers and
 Proceedings, 108: 519–23.
D’Acunto, Francesco, Daniel Hoang,
 Maritta Paloviita, and Michael We-
 ber. 2018a. “Human Frictions to the
 Transmission of Economic Policy.” Work-
 ing Paper.
D’Acunto, Francesco, Daniel Hoang,
 Maritta Paloviita, and Michael We-
 ber. 2018b. “IQ, Expectations, and
 Choice.” Working Paper.
Working Paper Series in Economics
recent issues

  No. 126 Francesco D‘Acunto, Daniel Hoang, Maritta Paloviita and Michael Weber:
                Cognitive abilities and inflation expectations, January 2019

  No. 125 Rebekka Buse, Melanie Schienle and Jörg Urban: Effectiveness of policy
          and regulation in European sovereign credit risk markets - A network
                analysis, January 2019

  No. 124 Chong Liang and Melanie Schienle: Determination of vector error correc-
                tion models in high dimensions, January 2019

  No. 123 Rebekka Buse and Melanie Schienle: Measuring connectedness of euro
                area sovereign risk, January 2019

  No. 122 Carsten Bormann and Melanie Schienle: Detecting structural differences
                in tail dependence of financial time series, January 2019

  No. 121 Christian Conrad and Melanie Schienle: Testing for an omitted multiplica-
                tive long-term component in GARCH models, January 2019

  No. 120 Marta Serra-Garcia and Nora Szech: The (in)elasticity of moral ignorance,
          December 2018

  No. 119 Thomas Mariotti, Nikolaus Schweizer, Nora Szech and Jonas von Wangen-
                heim: Information nudges and self-control, November 2018

  No. 118 Andranik S. Tangian: Methodological notes on composite indicators for
                monitoring working conditions, October 2018

  No. 117 Andranik S. Tangian: Testing the improved third vote during the 2018
                election of the Karlsruhe Institute of Technology student parliament,
                September 2018

  No. 116 Yuri Golubev and Mher Safarian: On robust stopping times for detecting
          changes in distribution, May 2018

  The responsibility for the contents of the working papers rests with the author, not the Institute. Since working papers
  are of a preliminary nature, it may be useful to contact the author of a particular working paper about results or ca-
  veats before referring to, or quoting, a paper. Any comments on working papers should be sent directly to the author.
You can also read