Behavioral Political Economy - cesifo-group.de fileFrancesco Passarelli-Guido Tabellini, JPE, 2017...

Behavioral Political Economy

Francesco Passarelli-Guido Tabellini, JPE, 2017

University fo Turin and Bocconi

CES - LMU, July 2018

Francesco Passarelli-Guido Tabellini, JPE, 2017 (Institute)Emotions and Political Unrest CES - LMU, July 2018 1 / 35

Introduction

Many empirical puzzles in politics are hard to reconcile with fullrationality.

Incentives to "invest" in rational decision-making are weak when itcomes to collective decisions.

Cognitive biases should be more relevant in politics than in themarketplace.

As a voter is unlikely to be pivotal, citizens who make poor politicalchoices do not su§er worse consequences than those who make goodpolitical choices.This reduces a citizenís ability to learn of her bias.Behavioral traits that may be detrimental in markets may facilitecollective action.

Behavioral Economics can address some of these puzzles.

Small literature in the Öeld.


Presentation Outiline 3

Emotions and Political Participation

Political entitlements as reference points - Passarelli and Tabellini(JPE, 2017): individuals treated poorly by government get glow fromprotesting.Social preferences and voter turnout - DellaVigna, List, Malmendier,Rao (REStud, 2017): individuals get social image from voting.Voter reciprocity - Finan and Schechter (EMA, 2012): Politicianstarget voter reciprocity.



Information processing

OverconÖdence - Ortoleva and Snowberg (AER, 2015):underestimating signal correlation leads to increased voter turnout, andstronger partisan identiÖcation.Correlation neglect - Levy and Razin (AER, 2015): correlation neglectleads to extreme beliefs, which may induce ideologically biased votersto base their vote on information rather than on political preferences.Rule-of-thumb in performance Öltering - Wolfers (2007): voters(irrationally) reward incumbent state governors for economic outcomesthat are plausibly unrelated to their policies.



Information processing

Limited attention - Eisensee and Stromberg (QJE, 2007): USA reliefsto foreign natural disasters are lower when American public isdistracted by a major news event.Personal appeal - Benjamin and Shapiro (RES, 2009): winnerpredictions based on a 10-second silent video clips of unfamiliargubernatorial debates outperform predictions based on economicvariables.Information availability - Boxell, Gentzkow and Shapiro (NBER WP,2017): growth in polarization in recent years is largest for thedemographic groups least likely to use the internet and social media.



Voter Biases

Present bias - Bisin, Lizzeri and Yariv (AER, 2015): governmentaccumulates debt to respond to individualsí desire to undo theircommitments.Loss aversion - Alesina and Passarelli (NBER WP, 2017): it explains astatus quo bias, lower polarization, intergenerational conáict, and ìlongterm cyclesî in policies with self-supporting movements to the right orthe left.Loss aversion and overconÖdence - Attanasi, Corazzini and Passarelli(JPubE, 2017): the former leads individuals to prefer more protectivevoting rules; the latter leads them to prefer overly low (high) protectionwhen they receive good (bad) news about how others will vote.Focusing - Nunnari and Zapal (mimeo, 2017): focusing more on theattribute in which options di§er more leads to higher polarization in theelectorate.


Passarelli and Tabellini, 2017 - Introduction

Emotions and Political Unrest

Key question: What induces riot participation and how is it policyináuenced?

Starting point: Political unrest is largely motivated by emotions,rather than by instrumental motives.

Individuals participate in costly protests because they are aggrieved andfeel unfairly treated.Other than in this emotional reaction, individuals are assumed to berational.



Who riots?Riots, general strikes and anti-government demonstrations, not entailingany military violence (Banks 2012)OECD vs Non-OECD countries



Who riots?

Demonstrators are more likely to

have extreme political preferences,be attached to speciÖc political parties,know for which party they will vote in the next election (opposite ofthe swing voter in theories of probabilistic voting),have voted in the last election,belong to a minority that feels discriminated,have low income,be dissatisÖed with the government or with speciÖc public policies andto be generally dissatisÖed.

Moreover, demonstrators tend to be educated, males, to be in thelabor force or students, and to be less than 50 years of age.

Survey data from the European Social Survey and the World Value Survey.34(36) countries and 5 waves during the period 2002-2012(2005-2009).



Individuals choose rationally whether to participate in collectiveactions weighing the pros and cons.

Participation provides a psychological reward, which iscommensurate to the feeling of aggrievement.Aggrievement is the di§erence, if positive, between entitlements andoutcomes.BeneÖt of participation depends on how many other individuals alsoparticipate. Hence, a complementarity is at work.Complementarity ampliÖes the mass reaction to controversial policydecisions.



Policy entitlements provide reference points for individualsí feelingsof aggrievement.

They are endogenously determined in equilibrium.

If the government becomes more constrained, individuals scale downtheir entitlements (reference points) accordingly: they become moreresigned.

This resignation e§ect creates an incentive for the policymaker toconstraint today its future choice set.

Individuals become less demanding, and this mitigates future socialconáict.



Insights:1. Even if the government is benevolent and all groups in society areidentical in their propensity to riot, equilibrium policy can be distorted.2. The resignation e§ect leads to an intertemporal distortion: thegovernment Önds it optimal to constrain its future decisions, in order tomitigate future unrest (e.g. issuing debt today).


Passarelli and Tabellini, 2017 - The economy

N groups indexed by i , where li 2 (0, 1) is relative size of group i .V i (q) " Indirect utility of group i , as a function of policy q.Social welfare function:

W (q) = Âi

li V i (q)


Passarelli and Tabellini, 2017 - The economy

Benevolent govtís objective function:

Maxq [W (q)#Âi

Vili pi ]

where:

pi = Pi (q) " riot participation in group i

Vi " social harm caused by riot of group i

Equilibrium policy (govt sets policy, then individuals choose riotparticipation)

Wq(q) = Âi

Vili Piq(q)

Key question:

What induces riot participation, P(q), and how is the policyináuenced?


Passarelli and Tabellini, 2017 - riot participation

Individual j in group i participates in riot if

pili ai # µ# #ij > 0

ai = Ai (q) feeling of "aggrievement" (a beneÖt in riot participation)pi share of participating people in the groupµ Cost of riot participation, known and common to all#ij idiosyncratic component of riot participation, with #ij > 0

Individual j participates if #ij $ pili ai # µ " #i

Equilibrium participation:

pi = Pr(#ij $ #i ) = F i (pili ai # µ) " Gi (pi )

#ij s F i (.), continuous, with density f i (.). Distribution F i (.) iscommon knowledge.



pi = Pr(#ij $ #i ) = F i (pili ai # µ) " Gi (pi )

pi =si # µ

2si # li aiif F i (.) is uniform

ip0

1)( µλ −iiii apF

°45

ip*

Equilibrium participation rate

)( µ−iF

initiators



Comparative statics:

if F i (.) is uniform

pi =si # µ

2si # li ai

larger groups riot more often (large li ); more aggreived groups protestmore (large ai ), more homogeneus groups protest more (low si ).non linear (explosive) reaction of protests to aggrievement.


Passarelli and Tabellini, 2017 - entitlements

Group i aggrieved if actual policy di§ers from "fair policy"Fair policy maximizes a distorted social welfare function:

Wi (q) = Âk pikV k (q)

Self serving bias: group i receives weight pii = li (1+ di ), all othergroups pik = lk (1# di )

qi = Arg max[Wi (q)]

Entitled Utility (Entitlement):

Ri = V i (qi )

Aggrievement:

ai =wi

2Max [0,Ri # V i (q)]2 " Ai (q)

Equilibrium participation rate: p%i = si#µ

2si#li Ai (q)" Pi (q)


Passarelli and Tabellini, 2017 - aggrievements andentitlements

Government trades o§ the direct welfare e§ects of the policies againsttheir impact on political unrest:

W (q)#Âi li Vi Pi (q)

Equilibrium policy (FOC):

Âi

li [1+ ViFi (q%)]V iq(q%) = 0

Fi (q) = li [Pi (q)]2wi!Ri # V i (q)

"/!si # µ

"> 0 if i is aggrieved,

Fi (q) = 0 otherwise


Passarelli and Tabellini, 2017 - aggrievements andentitlements

Favoured groups?

if riots have more disruptive e§ects (Vi )if aggrievement induces wider participation (more homogeneous groupsor larger groups).with stronger sense of entitlements (wi ) or stronger self serving bias(di )

Only aggrieved groups (Fi > 0) receive extra weight and exert somepolicy ináuence.

Groups wanting e¢cient policies (Fi = 0) are under-represented atoptimum

Distorted policies even if all groups can riot (unlike lobbying orprobabilistic voting).


Passarelli and Tabellini, 2017 - redistribution

Two groups, rich and poor, equal size.

Rich: exogenous unit income and pay a tax t.Poor: no income and receive a subsidy s.

Deadweight loss of taxation: t2/2a,

Consumer utility and social welfare:

V r (t) = 1# t , V p(t) = t # t2/2a

W (t) "12V r (t) +

12V p(t) =

12[1# t2/2a]

In the absence of any political constraints, the e¢cient policyminimizes tax distortions and sets t0 = s0 = 0.



Entitled policies

tr = 0 and tp = 2da/(1+ d) > 0

Lemma: Ppt $ 0 $ Prt, with strict inequality if and only if group i isaggrieved.Poor:aggrieved if they do not get the positive subsidy they feel entitled to.Rich: feel aggrieved if taxes are raised above 0.Equilibrium solves:

t = #aV[Ppt (t) + Prt(t)]

Proposition: in equilibrium tp > t% > 0 and the poor protest more thanthe rich: p%p > p%r .Intuitively, mitigating political unrest by the poor is costly in terms of taxdistortions, and so the government stops short of equating marginalaggrievement across the two groups.



Comparative statics: equilibrium taxes and subsidies increase with

the degree of self-serving bias of the poor (dp),the sensitivity of their aggrievement to deprivation (wp),the disruptions caused by their riots (Vp),the homogeneity of their group (sp)

The reverse applies to the rich (with the exception of dr ).


Passarelli and Tabellini, 2017 - dynamics

Two periods, t = 1, 2, indirect utility: Ât Vit (qt , b)

b is a policy variable in t = 1, and an endogenous state variable int = 2.

Government sets policy fq1, b, q2g to maximize:

Ât Wt (qt , b)#Ât Âi li Vi Pit (qt , b)

No discounting.



Sequential rationality in entitlement formation: qit and bi arechosen sequentially over time.

In period 2: qi2 = Qi (b) maximizes:

Wi2(q2, b) =

N

Âk=1

pikV k2 (q2, b)

In period 1: qi1 and bi , maximize:

Wi1(q1, b) +W

i2(Q

i (b), b) =N

Âk=1

pikV k1 (q1, b) +N

Âk=1

pikV k2 [Qi (b), b]



Reference points: Entitled residual lifetime utilities (denoted withRit ) in each period:

Ri1 = Vi1(q

i1, b

i ) + V i2(Qi (bi ), bi )

Ri2(b) = Vi2(Q

i (b), b)

Aggrievements

Ai1(q1, b) =wi

2Max [0, Ri1 # V

i1(q1, b)# V

i2(G (b), b)]

2

Ai2(q2, b) =wi

2Max [0, Ri2(b)# V

i2(q2, b)]

2

ìResignation e§ectî: Ri2b(b) = Vi2qQ

ib +V

i2b < 0. Accumulation of

the state variable b reduces period 2 reference utility, makingindividuals in group i willing to accept a lower level of welfare withoutfeeling aggrieved.



Ai1(q1, b) =wi

2Max [0, Ri1 # V

i1(q1, b)# V

i2(G (b), b)]

2

Ai2(q2, b) =wi

2Max [0, Ri2(b)# V

i2(q2, b)]

2

What is going on?

Possible that in period 2 individuals will not be aggrieved by a loss ofwelfare (lower V i2(q2, b)) due to the state variable b, if this welfare lossis deemed unavoidable and also reduces their reference utility (lowerRi2(b)).


Passarelli and Tabellini, 2017 - redistribution and debt

Consumption and fair utilities:

V rt (tt ) = 1# tt t = 1, 2

V p1 (t1, b) = t1 + b# F (t1), V p2 (t2, b) = t2 # b# F (t2)

Rr2(b) = 1# Tr (b) Rp2 (b) = Max [0, 2da/(1+ d)2 # b]

Period 2: fair policies

Rich: want zero subsidies for the poor sector, s r2 = 0, and a tax ratewhich is just su¢cient to service the debt, tr2 = T

r (b) - where Tr (b)solves the budget constraint, t2 # t22/2a# b = 0.Poor: a) if b is small, want tp2 = 2da/(1+ d) and

sp2 = tp2 #!tp2"2 /2a# b (subsidies are decreasing in b). b) if b is

large, the poor are forced to accept sp2 = 0, and their fair tax ratecoincides with that of the rich.



Period 2Lemma (Resignation e§ect): Pp2b(t2, b) = 0 ( P

r2b(t2, b)

Proposition (Equilibrium): (i) In the second period, the equilibrium taxrate is strictly positive and increasing in b: T (b) > 0 and Tb > 0. (ii)The equilibrium subsidy S(b) is weakly decreasing in b and it is positive orzero, depending on the level of b. (iii) The total equilibrium incidence ofriots in period 2 decreases with b, and it reaches a minimum at b = b.




Period 1: Fair policies and utilities

Rr1 = 1# tr1 + 1# Tr (br ) = 2

Rp1 = tp1 # F (tp1 ) + tp2 # F (t

p2 ) = 4da/(1+ d)2

Aggrievements

Ar1(t1, b) =w

2Max [0, t1 + T (b)]2

Ap1 (t1, b) =w

2Max [0, 4da/(1+ d)2 # t1 + F (t1)# T (b) + F (T (b))]2

Lemma: Pr1t(t1, b) ( 0 ( Pp1t(t1, b) and

Pr1b(t1, b) ( 0 ( Pp1b(t1, b)

Issuing debt aggrieves the rich, since they realize that futureequilibrium taxes will be raised.For a given tax rate t1, the aggrievement of the poor is dampened bymore debt, as they realize that future equilibrium taxes will be raised.The poor are the only beneÖciaries of debt. It will be paid by the richin the future. Equilibrium subsidies go down as b increases, but lessthan one for one.Francesco Passarelli-Guido Tabellini, JPE, 2017 (Institute)Emotions and Political Unrest CES - LMU, July 2018 31 / 35


Period 1Proposition (equilibrium): b% > 0, and 0 < t%1 < t%2.

Issuing government debt has two political beneÖts.

First, it provides an additional instrument to reduce social conáict inperiod 1, besides distorting taxes.Rich more angry in period 1, but the government responds by reducingperiod 1 tax rates.Rich anticipate higher taxes in the future, but, and this is the secondpolitical beneÖt, the resignation e§ect reduces their future riots.The government Önds this policy optimal until tax distortions arehigher in the second period than in the Örst one (i.e. t%1 < t%2):excessive redistribution and intertemporal distortion.


Passarelli and Tabellini, 2017 - evidence

Puzzle: Alesina, Carloni and Lecce (2013) show that governmentsthat achieve large reductions in the budget deÖcit are not punished atthe subsequent elections.

If dependent variable is unrest, the estimated coe¢cient on thechange in primary deÖcit is always statistically signiÖcant and with anegative sign, meaning that a deÖcit reduction increases politicalunrest.

Political unrest is mitigated by a higher initial public debt.


Passarelli and Tabellini, 2017 - conclusion

Threat of social unrest is relevant constraint on public policy

More constrained govt => more resigned citizens => less unrest

This creates incentive for procrastination.


Passarelli and Tabellini, 2017 - Future directions

Rioting for

"Economic rights": collective action ináuences public policy, as groupsseek to defend their ìeconomic rightsî."Political rights": the same ingredients can be adapted to study theendogenous evolution of political institutions, when citizens Öght todefend their ìpolitical rightsî

Anger relevant in other forms of political participation (Protest vote)

Approach useful to study impact of di§erent other emotions (fear,insecurity?) on political participation (Populism)

Manipulating voterís expectations/entitlements: media andpersuasion, not just information provision.


Table 2: Political Unrest and Fiscal Retrenchment

Dependent variable Political unrest

(1) (2) (3) (4) (5) (6) (7) (8)

Δ cyclically adj. primary deficit -0.19*** -0.30*** -0.22*** -0.38*** -0.11 -0.19*** -0.09 -0.36***

(0.042) (0.053) (0.042) (0.062) (0.11) (0.04) (0.09) (0.05)

lagged debt -0.01*** -0.01** (0.005) (0.005) GDP Growth 0.01 -1.44 0.06 -1.79 -0.30*** 0.12* -0.36*** 0.35

(0.074) (1.225) (0.071) (1.158) (0.09) (0.07) (0.12) (0.23)

unemployment growth 0.01 0.00 0.02* 0.02** -0.02** 0.02*** -0.03*** 0.03***

(0.010) (0.010) (0.011) (0.008) (0.01) (0.01) (0.01) (0.01)

Inflation -0.00 -0.03 0.03 0.06 0.04* -0.002 0.01 0.06

(0.040) (0.035) (0.060) (0.042) (0.03) (0.07) (0.04) (0.05)

GDP growth dev. G7 -0.12*** 1.37 -0.09** 1.82 0.02 -0.12** 0.22 -0.32

(0.036) (1.261) (0.042) (1.200) (0.08) (0.06) (0.15) (0.27)

unemployment growth dev. G7 0.07 0.09** 0.12** 0.16** 0.14** 0.14** 0.11 0.17**

(0.055) (0.042) (0.061) (0.065) (0.06) (0.06) (0.09) (0.07)

inflation dev. G7 0.10** 8.95 0.01 12.25 0.14*** 0.06 0.18** -0.10

(0.052) (10.743) (0.081) (10.170) (0.05) (0.08) (0.08) (0.10)

Sample Debt (t-1)

above 90% Debt (t-1)

below 90% Debt (t-1)

above 90% Debt (t-1)

below 90% Year Dummy variables No Yes No Yes No No Yes Yes Observations 599 599 508 508 168 423 168 423 Number of countries 19 19 19 19 10 18 10 18 NOTE. - The dependent variable is defined as the sum of riots, general strikes and anti-government demonstrations. The variables ending with the name dev. G7 are expressed in deviation from the average of the G7 countries. Some year dummy variables are dropped due to collinearity. Estimation: Conditional Poisson Regression; Country FE always included. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

Date post:	26-Oct-2019
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

Behavioral Political Economy - cesifo-group.de fileFrancesco Passarelli-Guido Tabellini, JPE, 2017...

Documents