Reward from public office and selection of politicians by parties
27
REWARD FROM PUBLIC OFFICE AND SELECTION OF POLITICIANS BY PARTIES Fabio Cerina Luca G. Deidda WORKING PAPERS 2014/14 CONTRIBUTI DI RICERCA CRENOS
Transcript of Reward from public office and selection of politicians by parties
REWARD FROM PUBLIC OFFICE AND SELECTION OF POLITICIANS BY PARTIES
Fabio Cerina Luca G. Deidda
WORKING PAPERS
2 0 1 4 / 1 4
C O N T R I B U T I D I R I C E R C A C R E N O S
C E N T R O R I C E R C H E E C O N O M I C H E N O R D S U D ( C R E N O S )
U N I V E R S I T À D I C A G L I A R I U N I V E R S I T À D I S A S S A R I
C R E N O S w a s s e t u p i n 1 9 9 3 w i t h t h e p u r p o s e o f o r g a n i s i n g t h e j o i n t r e s e a r c h e f f o r t o f e c o n o m i s t s f r o m t h e t w o S a r d i n i a n u n i v e r s i t i e s ( C a g l i a r i a n d S a s s a r i ) i n v e s t i g a t i n g d u a l i s m a t t h e i n t e r n a t i o n a l a n d r e g i o n a l l e v e l . C R E N o S ’ p r i m a r y a i m i s t o i m p r o v e k n o w l e d g e o n t h e e c o n o m i c g a p b e t w e e n a r e a s a n d t o p r o v i d e u s e f u l i n f o r m a t i o n f o r p o l i c y i n t e r v e n t i o n . P a r t i c u l a r a t t e n t i o n i s p a i d t o t h e r o l e o f i n s t i t u t i o n s , t e c h n o l o g i c a l p r o g r e s s a n d d i f f u s i o n o f i n n o v a t i o n i n t h e p r o c e s s o f c o n v e r g e n c e o r d i v e r g e n c e b e t w e e n e c o n o m i c a r e a s . T o c a r r y o u t i t s r e s e a r c h , C R E N o S c o l l a b o r a t e s w i t h r e s e a r c h c e n t r e s a n d u n i v e r s i t i e s a t b o t h n a t i o n a l a n d i n t e r n a t i o n a l l e v e l . T h e c e n t r e i s a l s o a c t i v e i n t h e f i e l d o f s c i e n t i f i c d i s s e m i n a t i o n , o r g a n i z i n g c o n f e r e n c e s a n d w o r k s h o p s a l o n g w i t h o t h e r a c t i v i t i e s s u c h a s s e m i n a r s a n d s u m m e r s c h o o l s . C R E N o S c r e a t e s a n d m a n a g e s s e v e r a l d a t a b a s e s o f v a r i o u s s o c i o - e c o n o m i c v a r i a b l e s o n I t a l y a n d S a r d i n i a . A t t h e l o c a l l e v e l , C R E N o S p r o m o t e s a n d p a r t i c i p a t e s t o p r o j e c t s i m p a c t i n g o n t h e m o s t r e l e v a n t i s s u e s i n t h e S a r d i n i a n e c o n o m y , s u c h a s t o u r i s m , e n v i r o n m e n t , t r a n s p o r t s a n d m a c r o e c o n o m i c f o r e c a s t s . w w w . c r e n o s . i t i n f o @ c r e n o s . i t
C R E N O S – C A G L I A R I V I A S A N G I O R G I O 1 2 , I - 0 9 1 0 0 C A G L I A R I , I T A L I A
T E L . + 3 9 - 0 7 0 - 6 7 5 6 4 0 6 ; F A X + 3 9 - 0 7 0 - 6 7 5 6 4 0 2
C R E N O S - S A S S A R I V I A T O R R E T O N D A 3 4 , I - 0 7 1 0 0 S A S S A R I , I T A L I A T E L . + 3 9 - 0 7 9 - 2 1 3 5 3 6 ; F A X + 3 9 - 0 7 9 - 2 1 3 0 0 2
T i t l e : REWARD FROM PUBL IC OFF ICE AND SELECT ION OF POL IT IC IANS BY PART IES F i r s t Ed i t i on : December 2014
Reward from public office and selection of politicians by
parties∗
Fabio Cerina
Universita di Cagliari and CRENoS
Luca G. Deidda
Universita di Sassari, CRENoS, and SOAS (Un. London)
December 2014
Abstract
We investigate the relationship between quality of politicians and reward from public office
in a game between parties and citizens, in which parties play a crucial role in the selection of
candidates. Citizens who wish to become politicians have to become party activists first. Parties
produce information about the quality of potential candidates. An increase in the reward from
public offices leads to two opposing effects on the average quality of politicians: 1) A selection
effect, whereby more skilled citizens enter politics, leading to an increase in average quality; 2)
An information manipulation effect such that unskilled activists will have a higher probability to
be elected since parties find convenient to increase the degree of costly information manipulation
leading to a decrease in average quality. We find that, for a plausible range of parameters
values, the information manipulation effect dominates the selection effect when: 1) The cost of
manipulating information is low enough; 2) The amount of information that parties can provide
at no cost is low enough; 3) The net gains from becoming a politicians for unskilled citizens
is sufficiently larger than that for skilled citizens. These findings provide a rational for the
ambiguous sign of the empirical correlation (relationship) between quality and pay of politicians.
JEL Classification: J33, D72, H70.
Key words: Selection, Public office rewards, Political parties, Quality of politicians, Information
manipulation.
∗ We thank Giovanni Sulis for insightful conversations and suggestions, Michele Argiolas for excellent research
assistance, and seminar participants at Universita di Sassari.
1
1 Introduction
The tradition view in political economy has been for a long time that good politics and good policy are
the result of well-designed institutional incentives (see Besley, 2005 for a general discussion). More
recently, however, in the literature there has been a growing attention to the notion that – aside
from incentives – intrinsic quality of politicians could well matter for quality policies and quality
government, which could have a significant impact on socio-economic outcomes1.
Notwithstanding the potential importance of politicians’ quality, as many theoretical and empirical
contributions show, good quality politicians is often not the equilibrium outcome we observe. In that
respect, our focus is on the relationship between quality of politicians and pay of politicians. That
is, we ask if – and if so – under which circumstances, paying politicians well leads to better quality
politicians. Casual observation suggests that increasing politicians’ pay is not necessarily associated
with higher quality. For instance, figure and 1 and 2 show the trends in MPs pay and education – a
widely used measure of quality in the literature over time – for Italy and United States.2
.7.7
5.8
.85
.9Ita
lian
MPs
with
und
ergr
adua
te d
egre
e or
mor
e
050
000
1000
0015
0000
2000
00Ita
lian
MPs
Pay
1940 1960 1980 2000 2020year
Italian MPs Pay Italian MPs with undergraduate degree or more
Figure 1: MPs reward in real terms (2005 dollars) and percentage of MPs with undergraduate degreein Italy 1948-2007. Source: Fondazione De Benedetti (FRDB), “Italian Members of the Parliament”dataset. Data originally collected by Merlo et al (2008) with support of FRDB.
While in the US the percentage of MPs with undergraduate education or more increases with
pay, this is not the case in Italy, where the two phenomena are strongly negatively correlated.3 A
recent strand of empirical literature addresses the challenging task of identifying the causal effect
1See Besley, 2005, Besley, Montalvo and Ryenal-Querol, 2013, and Besley, Pande, Tao, 2005.2Empirical proxies of competence used in the literature include education attainment (see Besley, 2004, Ferraz and
Finan, 2009, Fisman et al., 2015, Gagliarducci and Nannicini, 2013).3Correlation between MPs pay and education is around 31% in the US and −91% in Italy.
2
.7.7
5.8
.85
.9.9
5%
US
MPs
with
und
ergr
adua
te d
egre
e or
mor
e
5000
010
0000
1500
0020
0000
US
MPs
Pay
1940 1960 1980 2000 2020year
US MPs Pay % US MPs with undergraduate degree or more
Figure 2: MPs reward in real terms (2005 dollars) and percentage of MPs with undergraduate degreein US 1948-2007. Source: Online bibliographical directory of the United States Congress.
that politicians’ pay could have on quality of politicians. Using data from local Italian municipalities,
Gagliarducci and Nannicini, 2013, show that a higher wage attracts more-educated candidates, and
that better-paid politicians size down the government machinery by improving efficiency.4 Similarly,
using municipalities data from Mexico, Dal Bo, Finan and Rossi, 2013, find that announcing a higher
salary results in applicants who are on average smarter, with better personality traits, higher earnings
and a better occupational profile. Similarly, Ferraz and Finan, 2009, find that in the case of brazilian
municipalities, higher wages result in more educated candidates.
These findings support the case for positive causal relationship between pay of politicians and
quality of politicians, at local level. However, things change substantially, when national politics is
considered. Using data on Members the European Parliament (MEPs), Fisman et al., 2015, find that
”[..] high salaries reduce the quality of elected MEPs (as proxied by the quality of the colleges they
attended) [..]” (page 3). Relatedly, Braendle, 2015, finds no significant impact of MEPs’ pay on MEPs’
quality as proxied by formal education, occupational background and political experience. Kotakorpi
and Poutvaara, 2011, study the the relationship between politicians’ wage and quality in Finnish
National Parliament. They report that the wage increase led to more educated female candidates
but had no effect on the composition of male candidates. Finally, Hoffman and Lyons (2013) find
almost no correlation between salary and politician performance or quality of US governors and state
legislators.
Our model provides a possible rational as to why the sign of the relationship between pay and
quality of politicians might turn negative as well as on why it is more likely to be positive at local
4According to their findings, most of this effect can be attributed to the selection of competent politicians.
3
level than at national level.
We model parties’ role in determining quality of politicians by informing citizens about the quality
of prospective candidates in a setup very much related to the theory of quality of elected officials put
forward by Caselli and Morelli, 2004. In the model, citizens who wish to be elected politicians have
to become party activists, first. If elected, as politicians they contribute service duties to their party.5
Citizens are of two qualities, skilled and unskilled. Unskilled politicians lead to a more costly scheme
for the provision of the essential public good compared to skilled ones. Moreover, skilled individuals
earn a higher salary than unskilled ones if working in the private sector. Individual quality is private
information. However, parties observe the quality of each of their activists at no cost, and can produce
an informative public signal about prospective candidates’ quality . More precisely, parties can either
provide perfect information about candidates’ skills at no cost, or manipulate information, which is
costly and results in an imperfect (still informative) signal of candidates’ type.
Our main result is that – under parties’ free entry – whether increasing the pay of politicians will
help improving the quality of politicians depends on how it affects parties incentives to manipulate
information about candidates in order to favor the elections of unskilled (low quality) politicians. If
information about candidates is opaque, due either to low awareness of citizens or scarce availability of
public information, and parties’s cost of manipulating information is low, then increasing politicians’
pay can result in lower quality of politicians.
The intuition is as follows. Increasing politicians’ pay might favor the entry of skilled citizens
in politics, to the extent that they have a higher opportunity cost to enter politics compared to
unskilled ones. Other things equal, this positive selection effect would increase quality of elected
politicians. However, entry of skilled citizens into politics worsens election prospects for unskilled
activists, who would be therefore less willing to contribute service duties to the party if elected.
To counteract that effect, parties manipulate information more (manipulation effect), in order to
increase unskilled activists’ chances to be elected and the service duties that they can extract from
unskilled politicians. Such increase in information manipulation might undo the positive selection
effect, thereby resulting in a negative relationship between pay of politicians and quality. More so if
manipulating information is inexpensive because parties’ manipulation technology is powerful and/or
the informational disadvantage of citizens is significant (unaware society).
According to our model, if quality of politicians and politicians’ pay are negatively correlated in
the case of Italy, and positively correlated in the case of the US – other things equal – we should
expect the cost of manipulating information and/or the level of citizen awareness to be lower in Italy
than in the US. Finding proper proxies for these quantities is not an easy task. Having said that, if
one accepts the idea that parties’ manipulation technology is more effective, and citizens’ awareness
is lower in countries where (i) The press is not entirely free and independent, and (ii) The citizens
are ignorant, respectively, then we could take the Freedom of the Press index, elaborated by Freedom
5These duties can be interpreted both as monetary transfers as well as transfers in kinds.
4
House 6, and the so-called Index of Ignorance elaborated by IPSOS-MORI7 and see how the two
compare in the two countries.
Indeed, according to the Freedom of the Press index , Italy is in 70th position in 2012, 69th in
2013 and 64th in 2014 and its press is defined as ”partly free”. On the other hand, US is 22th in
2012, 23th in 2013, 30th in 2014 and its press is defined ”free”.8 In the same year, US press is ahead
of Italy in terms of freedom also accordingly to the index elaborated by Reporters without borders
association.
As for the Ignorance index, Italy is the more ignorant country among the 14 taken into consid-
eration.9 The behavior of the rough proxies affecting the cost of information manipulation across
countries in in line with the model.
Finally, according to our model, the fact that the relationship between quality and pay of politicians
is generally positive only at local municipality level, might be due to the fact that informational
asymmetries, and therefore citizens’ awareness, are more of an issue at national level, where it is less
likely that citizens have a direct knowledge of the candidates.
Our paper is directly related to the literature on how citizens self-select, or are selected by parties,
into politics. The theory put forward by Caselli and Morelli, 2004, predicts that increasing politicians’
pay always increases politicians’ quality due to a selection effect. We significantly depart from their
setup by introducing parties and endogenizing the information that citizens have about candidates.
Our results show that the selection effect dominates – so that a positive relationship between quality
and pay of politicians emerges – if and only if the society is sufficiently aware and/or parties’ ability
to manipulate information is low enough so that the manipulation effect is sufficiently low.
Using a different setup, Mattozzi and Merlo, 2008, show that increasing the pay for public service
makes politics more attractive at all skill levels, which reduces quality, but also makes more attractive
for skilled politicians to stay in politics longer. As a result the impact of pay for public office on
quality is ambiguous. In our setup, a similar intuition applies that a higher pay makes politics a more
attractive option for both unskilled and skilled agents. Yet, in our case, the ambiguous relationship
between pay and quality of politicians is due to parties’ incentives to manipulate information about
candidates, rather than to opportunity cost kind of considerations by skilled politicians on whether
to stay in politics or leave.
Messner and Polborn, 2004, also show that quality of politicians might go down following an
6Freedom House (freedomhouse.org) describe itself as ”an independent watchdog organization dedicated to theexpansion of freedom around the world.”. Its main supporters and financiers are: Canadian International Develop-ment Agency, Dutch Ministry of Foreign Affairs, European Instrument for Democracy and Human Rights, John D.& Catherine T. MacArthur Foundation, Smith Richardson Foundation, U.S. Agency for International Development,U.S. Department of State but also Ford Foundation, Google, Leon Levy Foundation, The Lilly Endowment, NorwegianMinistry of Foreign Affairs, The Schloss Family Foundation (Edwin Schloss).
7Ipsos-Mori (www.ipsos-mori.com) is a leading UK research company with global reach. According to their descrip-tion, the Ignorance Index [..] “highlights how wrong the public across 14 countries are about the basic make-up of theirpopulations and the scale of key social issues ”[..] (from IPSOS-MORI website.
8There are three categories: ”Free”, ”Partly Free” or ”Not Free”.9The 14 countries are Australia, Belgium, Canada, France, Germany, Hungary, Italy, Japan, Poland, South Korea,
Spain, Sweden, Great Britain and the United States of America. By the way, taking into consideration only these 14countries.
5
increase in politicians’ wage. Their result, though, stems from an intuition very different from ours.
In their setup, citizen care about quality of politicians, and, crucially, each citizen know he can affect
the quality of politicians if entering politics. Hence, while other things equal an increase in salary
makes it politics more attractive for a skilled citizen (who would choose to run if no one else is running)
it also increases the incentive of each skilled citizen to free ride on other skilled citizens, that is not
to run counting on the fact that others will run.
While all the above mentioned contributions abstract from parties, Mattozzi and Merlo, 2011,
present a model in which two parties select politicians trading off costs and benefits of selecting good
vs bad ones. They show that a party may deliberately choose to recruit only mediocre politicians, in
spite of the fact that skilled politicians might enhance party’s electoral prospects. Differently from
what happens in our model, parties might prefer mediocre politicians not because they can extract
more rents form them, but rather because the presence of skilled politicians might discourage other
party members from putting effort. Moreover, that model does not deal with the relationship between
pay and quality of politicians.
The rest of the paper is as follows. Section 2 presents the model. Section 3 describes payoffs and
actions. Section 4 contains the equilibrium analysis. Section 5 presents the main result about quality
and pay of politicians. Section 6 concludes the paper.
2 The model
In this section we describe the environment. We clarify upfront that our modeling approach is to
augment the setup put forward Caselli and Morelli (2004) by introducing parties, whose main role is
to provide public information about candidates as fully explained later on in the discussion.
2.1 Citizens, Candidates, Activists, Politicians and Parties
The economy is populated by a continuum of size 1 + p of risk-neutral citizens and an endogenous
number N of political parties. Citizens are of two types: a fraction s is skilled (s) and a fraction 1− sis unskilled (u). A citizen of type i = s, u, earns wi in the private sector, with ws > wu and, without
loss of generality, wu = 1.
Citizens consume an indispensable public good, whereby citizen’s utility would equal zero, if the
good were not provided. The provision of the public good requires a mass p of citizens being elected
to public office, where we call politicians the elected public officials. Politics is run by parties. A
citizen can become politician only enrolling as activist in a party, first. Parties hire activists and
select citizens who run for election among their activist, where we call candidate an activist who is
running for office. We assume that activists face an infinite cost of switching party, 10 so that parties
operate as local monopolists.
Becoming an activist entails a participation cost φ per unit of gross income. We interpret such
cost as to potentially include not just a possible membership fee but also the services that an activist
10To be precise, in order for our results to go through we do not need the cost to be infinite, but just sufficiently highthat switching party is never an option for activists.
6
contributes to the party. Accordingly, φ could be also taken as a measure of citizens’ willingness to
engage in politics.
Following Caselli and Morelli (2004), in order to eliminate a trivial equilibrium were the entire
population runs for office, we assume that there is a measure v ∈ [p, 1] of citizens who have an infinite
(subjective) cost of becoming activist11, so that, µ ≡ 1 + p − v is the mass of citizens who can
potentially become politicians.
Politicians are elected among candidates by citizens who vote individually. Each politician is
assigned to a public office and receives a monetary reward, π > ws. Crucially, skilled politicians
are assumed to be more productive: the per-capita tax, t(q), necessary to finance the provision of
the public good, is decreasing in the quality of politicians, q, where q is defined as the fraction of
politicians of type s.
2.2 Information and the role of Parties
Parties have a role in politics, since they can produce information about the quality of their activists
– or, perspective candidates. Parties operate thanks to the contributions of their politicians, whereby
a politician of type i pays service duties ei to the party she belongs to.
The information structure is the following. Citizen’s type i is private information. However, parties
learn the true type of their activists, at no cost. Moreover, each party produces (according to the
timing of the game which will be defined below) a public signal j = s, u about the type of its activists
(potential candidates). Let f(i|j) the probability that for a party activist of true type j, the public
signal generated by her party is i. The structure of the public signal generated by parties is as follows,
Definition 1 (Signal’s structure). For a party activist of true type i the public signal generatedby her party is correct (i = j) with probability f(i|i) ≡ σ and incorrect (i 6= j) with probabilityf(i|j 6= i) ≡ 1− σ.
Clearly, the signal is informative if and only if σ ≥ 0.5. Each party k ∈ N controls the informa-
tiveness of her signal by choosing, σk, for all its activists at a fixed cost c(σk). More specifically, a
party can produce a signal i that is correct with probability σk = σ > 0.5 for each of its activists
at no cost, or manipulate information thereby producing a signal i that is correct with probability
σk < σ for each of its activist, at an overall cost, c(σ) > 0. We assume c(σk) to be decreasing in σ, i.e.
c′(σk) < 0. We further assume that manipulating information is increasingly costly, i.e. c′′(σk) > 0.
All parties face the same cost c(σk) of manipulating information. More specifically, for party k, the
cost of manipulating information is as follows
c =
{c(σk) if σk < σ
0 if σk = σ(1)
where σ, which we assume to greater than 0.5, measures the quality of the signal that parties can
produce at no cost. Accordingly, σ can be interpreted as a measure of the quality of information in
11Similarly, to eliminate equilibria in which some offices go unfilled, we assume that φ is paid by a candidate only ifthe measure of candidates is greater than p (otherwise there is no competition, and hence no point in campaigning).
7
the society which – other things equal – can be associated with the quality of the media, for instance,
or the degree of awareness of citizens. An aware society is characterized by a relatively high level of
σ, while an unaware one is characterized by a relatively low level of σ.
In order to avoid trivial results, following Caselli and Morelli (2004) we assume that for either type
and for any feasible set of σk, with k ∈ N , the measure of potential candidates with correct signal is
greater than the measure of public offices to be filled.12 From now on we refer to σk as the amount of
information produced by a party.
2.3 Voting Rules
Voting rules are set in a standard way as follows. Citizen can vote at most one candidate, and votes
to non-candidates are void. For a given mass of public offices x to be filled, the mass x of candidates
receiving the most votes are elected to office. When necessary, ties are broken with a random draw.
2.4 Citizen-Parties game.
Citizens and parties play the following game.
1. Nature decides individual types;
2. Parties enter;
3. Citizens decide whether to become party activists or not;
4. Each party recruits activists, decides service duties to be contributed to its politicians, and gener-
ates information about its activists who are all potential candidates;
5. Candidates are decided, and citizens vote;
6. The outcome of the game is realized and payoffs are assigned.
Note that stages 3 and 4 happen simultaneously.
3 Payoffs and actions
In this section, we analyze parties and citizens’ payoffs and behavior.
3.1 Citizens
Let αj and di be the probability of a candidate to be elected – conditional on a signal j = s, u, and
the decision to become a party activist by a citizen of type i, respectively, where di = 1 in case a
citizen decides to start a political career by becoming a party activist, and di = 0 otherwise. For a
12Restricting attention to σ ≥ 0.5, the relevant condition is µs > 2p and µ(1− s) > 2p.
8
citizen of type i, the expected payoff from choosing a political career in a party k ∈ N that charges
service duties es,k to unskilled politicians and eu,k to unskilled ones, is the following
vi|di=1 ≡ [αsf(s|i) + αu(1− f(s|i))] (π − ei,k) +
[(1− αs)f(s|i) + (1− αu)(1− f(s|i))]wi − φ− t (q)
while the expected payoff if choosing not to engage in politics would be
vi|di=0 ≡ wi − t(q) (2)
Given the mass of C ∈ R of candidates, each citizen of type i expresses a vote bi ∈ R+, where the
vote is void if bi /∈ C. An action for citizen of type i is a pair (di, bi).
3.2 Parties
We define Ak,s, and Ak,u, with Ak,s, Ak,u ∈ R the masses of activists of types s and u, respectively,
recruited by party, k ∈ N . Accordingly, As ≡∫kAk,s, and Au ≡
∫kAk,u, are the aggregate masses
of activists of type s and u, respectively, while A ≡ As + Au is the aggregate mass of activists,
independent of type. Note that the following inequalities must hold
As ≤ µs (3)
Au ≤ µ(1− s) (4)
as the mass of activists cannot be – conditional on type – larger then the mass of the citizens who
can engage in politics.
Given the above, the expected payoff of party k can be written as
vk = es,k(αsσk + αu (1− σk))Ak,s + eu,k(α
s (1− σk) + αuσk)Ak,u − c (σk) (5)
Each party k chooses σk ∈ [0, σ] and eu,k, es,k ∈ R+ in order to maximize the above, taking into
account the optimal decisions of citizens. An action for the party k is a triple (σk, es,k, eu,k).
4 Equilibrium analysis
A strategy for a citizen of type i is a probability function ai that maps actions (di, bi) on [0, 1]. A
strategy for a party k is a probability function ak that maps actions, (σk, eu,k, es,k), to [0, 1]. Given
these definitions,
Definition 2. An equilibrium is a strategy profile for the citizens and the parties such that each agent’sstrategy is best reply at any given stage of the game.
Let Cjk,i be the mass of candidates of type i and signal j belonging to party k, so that, (i) Cj
i ≡∫kCjk,i is the aggregate mass of candidates of type i and signal j across parties; (ii) and Cj ≡ Cj
s +Cju
is the aggregate mass of candidates of signal j across candidate’s types; (iii) C ≡ Cs +Cu = Cu +Cs
is the total mass of candidates. Note that, Ck,i ≤ Ak,i holds, for any i = s, u, and k ∈ N .
Rather than characterizing all possible equilibria, we restrict our attention to a subset of the
possible equilibria that satisfy satisfy the following conditions (properties).
9
1. Symmetry. Activists, candidates and politicians, are equally divided among parties. That is,
for each k ∈ N ,
Ak,i =AiN
(6)
Cjk,i =
Cji
N(7)
with i, j = s, u;
2. Candidates are perceived to be skilled. For all candidates, j = s holds. Since voters prefer to
vote high-signal candidates than low-signal ones, this condition is equivalent to imposing that
there are enough candidates of high signal to cover all seats, i.e.,
Cs > p (8)
3. Reluctancy of skilled citizens to enter politics. All the unskilled citizens become activists, and
only a fraction of skilled ones do so. That is13
(Au = µ(1− s)) ∧ (As < µs) (12)
Our focus on equilibria in which while all unskilled citizens become activists, only some of the
skilled ones do so, is motivated by the fact that such composition of activists delivers the same situation
as the one analyzed by Caselli and Morelli (2004), whereby an increase in the pay of politicians can
only induce a positive selection effect by encouraging more skilled citizens to enter policies. This is
important to us, as it allows a more direct comparison between our results and those by Caselli and
Morelli (2004).
In the following discussion, when we refer to the equilibrium, we always mean an equilibrium that
satisfies properties 1-3 listed above. Since we are assuming symmetry, we drop the k when referring
to parties’ actions.
4.1 Citizens’s behavior and equilibrium values of the probability to beelected
We analyze first citizens’ voting behavior, and then their decision to become activists.
13Concerning the skill-composition of activists (point 3), in principle, possible equilibrium situations alternative tothe one we focus on, would be
i. Skilled citizens all become activists, while some of the unskilled ones do not
(Au < µ(1− s)) ∧ (As = µs) (9)
ii. All citizens, skilled and unskilled, become activists
(Au = µ(1− s)) ∧ (As = µs) (10)
iii. Some skilled and unskilled citizens become activists and some (both skilled and unskilled) do not
(Au < µ(1− s)) ∧ (As < µs) (11)
10
Citizens vote based upon the set C of candidates, whom they observe. Given C, the optimal
strategy of a citizen is as follows. Each candidate votes for himself.14 Furthermore, since citizens
prefer skilled politicians and, for each candidate, they observe the signal j, whenever such signal is
informative, i.e. σ > 0.5, and the mass of candidates with signal j = s, Cs, is positive, each non-
candidate citizen votes for a uniformly drawn element of Cs. So long as the signal is informative,
only if Cs is empty, would non-candidate citizen vote for an uniformly drawn element of the set of
candidates with signal u, Cu.
Citizens’ optimal behavior as voters allows us to characterize the equilibrium probabilities to be
elected for candidates with signals u and s, respectively. In particular, in any equilibrium in which
condition (8) holds (property 2), so that there are enough candidates with a signal s to cover all public
office seats, a candidate with a signal u will never be voted for public office, since citizens prefer skilled
politicians. Then, provided that such condition holds:
αs =p
Cs(13)
αu = 0 (14)
are the equilibrium probabilities for a candidate to be elected conditional on her signal.
Substituting in for the values of αs and αu in the citizen’s payoff expressions (2) and (2), the
equilibrium expected gains from becoming an activist, for citizens of type s and u are
vs|ds=1 − vs|ds=0 ≡p
Csσ(π − ws)︸ ︷︷ ︸
Expected benefit
−
p
Csσes + φ︸ ︷︷ ︸
Expected cost
(15)
and,
vu|du=1 − vu|du=0 ≡p
Cs(1− σ)(π − 1)︸ ︷︷ ︸
Expected benefit
−
p
Cs(1− σ)eu + φ︸ ︷︷ ︸Expected cost
(16)
respectively.
Given the above values for the net gains to become activists , the optimal decision about whether
to become an activist (di = 1) or not (di = 0), for a citizen of type i is as follows,
di ≡
1 if vi|di=1 − vi|di=0 > 0
0, 1 if vi|di=1 − vi|di=0 = 0
0 if vi|di=1 − vi|di=0 < 0
(17)
Note that in a candidate equilibrium in which the net expected gain from becoming a politician
is zero, a citizen will be indifferent about whether to become a political activist or not.
Citizens net expected gain from becoming an activist is given by the difference between an expected
benefit and an expected cost. The expected gain is, for both skilled and unskilled citizens, a positive
function of the probability to be elected and of the difference between the reward as public officer and
14Note that candidates are selected by political parties among their activists.
11
the salary in the private sector. The expected cost, for both skilled and unskilled citizens, is given by
the sum of the expected costs of activism and service duties. Since in the equilibria we look at, only
activists with signal s have a positive probability to be elected, the net expected gains from entering
politics are increasing in the informativeness, σ, of the signal for skilled citizens and decreasing for
unskilled ones.
We note that, in the equilibrium we are interested in, the net expected gain from entering politics
for skilled and unskilled citizen respectively should satisfy the following conditions. First, vu|du=1 −vu|du=0 ≥ 0 must hold, because if all unskilled decide to become activists, this should be the preferred
choice for this type of citizens. Second, vs|ds=1 − vs|ds=0 = 0 holds, because skilled citizens must be
indifferent between becoming activists or not, in order for only a fraction of skilled individuals to
choose to become activists.15
Crucially, since only activists with signal s, have a positive probability of being elected, the mass
of candidates will be the sum of skilled activists with correct signal and of unskilled activists with
incorrect signal. More precisely
Css = σAs (18)
Csu = (1− σ)Au (19)
In the class of equilibria we restrict attention to, where unskilled citizens all become activists, we also
have
Csu = µ(1− s)(1− σ) (20)
And since Cs = Css + Cs
u, in our equilibrium the probability of being elected reduces to
αs =p
Css + µ(1− s)(1− σ)
(21)
where the value of Css will be pinned down endogenously from parties optimal behavior in the next
subsection.
4.2 Parties’ behavior
In this subsection we analyze parties’ choice of the optimal service duties, ei to be extracted from
skilled and unskilled politicians, i = s, u, and of the optimal degree of information manipulation, σ,
for a given number of parties N . As we shall see, the optimal value of ei together with the conditions
relative to citizens’ optimal behavior, pins down the probabilities of being elected and the mass of
candidates as well as its skill composition.
4.2.1 Equilibrium values of the service duties, es and eu, and of the mass of candidates,Cs
The following result holds,
15Since all skilled citizens share the same payoff function, if vs|ds=1 − vs|ds=0 > 0, then all skilled citizens would bewilling to run while if vs|ds=1 − vs|ds=0 < 0 none of them will be willing to become activist.
12
Lemma 1. Consider an equilibrium where unskilled citizens all become activists, while some skilledcitizens do not (Au = µ(1−s) and As < µs). Then, the equilibrium values of eu and es are as follows,
es = 0 (22)
Cs =pσ (π − ws)
φ; (23)
eu = (π − 1)− σ (π − ws)(1− σ)
≥ 0 (24)
Proof. First, we note that, in any equilibrium, parties extract all surplus that citizens could havefrom being activists. This follows from the fact that (i) Parties decide party duties after citizenshave become activists and therefore have chosen their party; (ii) Switching party has an infinite cost;which imply that parties are local monopolists. In other words, ei : vi|di=1 − vi|di=0 = 0. Second, itis immediate to verify that in any equilibrium, es = 0 must hold. By contradiction, let us considera candidate equilibrium in which es > 0 and not all citizens of type s are activists. Then there aretype s citizens who are willing to become activist if paying less than es. Then, there would be roomfor a new party to be formed, with service duties es − ε, where ε → 0+. Note that such party wouldsurely make strictly positive profits by setting σ = σ, which costs zero. Hence, in any equilibriumwhere not all citizens of type s are activists, es = 0 must hold. Using es = 0 in (15) and setting
vs|ds=1 − vs|ds=0 = 0, we find Cs = pσ(π−ws)φ
. Substituting for this value for Cs into (16) and setting
vu|du=1 − vu|du=0 = 0 we finally obtain eu = (π − 1)− σ(π−ws)(1−σ)
.
The above result tells us that in an equilibrium where unskilled citizens all become activists, while
some skilled citizens do not, parties are able to extract (strictly) positive service duties from unskilled
activists only. As already anticipated, this is similar to what happens in Caselli and Morelli (2004),
where bad politicians are more willing to get into politics than good ones.
4.2.2 Optimal information manipulation
First of all, since parties cannot make negative profits, in any equilibrium in which parties are ex-
tracting service duties only from unskilled politicians, the following result holds.
Lemma 2. Consider an equilibrium where there are enough high-signal candidates to cover all seats(Cs > p) and where unskilled citizens all become activists Au = µ(1 − s) while some skilled citizensdo not (As < µs) and let σ∗ the associated equilibrium value of information manipulation by parties.Then,
σ∗ < σ ≡ π − 1
π − ws + π − 1∈(
1
2, 1
)(25)
Proof. Parties cannot make negative profits. Hence, with es = 0, they hire activists if and only ifthey can extract rents from them. Hence, eu ≥ 0 must hold. The result then, comes directly from thefact that eu = (π − 1)− σ(π−ws)
(1−σ)is strictly positive if and only if σ < σ and equals zero if and only if
σ = σ.
σ is the level of informativeness of the signal produced by parties above which skilled citizens are
more willing to become activists than unskilled one. Clearly, an equilibrium where only some skilled
13
citizens enter politics while all unskilled ones become activists cannot exist if the associated level of
σ chosen by parties exceeds σ. It is interesting to note that σ is always decreasing in π since
∂σ
∂π/σ =
1− ws(π − 1 + π − ws)(π − 1)
< 0 (26)
The intuition for this outcome is straightforward. When the level of informativeness of the signal
equals σ skilled and unskilled citizens are equally willing to enter politics. Whenever the signal is
informative an increase in pay of politicians favors more skilled activists than unskilled ones, since
the former have a higher chance to be elected than the latter. To compensate that effect, as the pay
of politicians increases, the value of σ should go down, thereby increasing the chance of unskilled
activists to be elected.
Lemma 2 implies that we need to assume σ ≥ σ > σ∗, in order for the type of equilibria we are
interested in to exist. Finally, combining (23) and (13) we obtain the probability of being elected as
a function of only one endogenous variable16 σ:
αs =φ
σ(π − ws)(27)
Substituting in in (5) for the equilibrium expressions of es, eu, Cs and αs as derived above, and
considering that in a symmetric equilibrium each party is identical, the representative party solves
maxσ
φµ (1− s)N
(π − 1
π − ws1− σσ− 1
)− c (σ) (28)
In the case of an interior solution, the optimal level of σ satisfies the following first-order condition
−φµ (1− s)N
π − 1
π − ws− σ2c′ (σ) = 0 (29)
A rise in σ has then two distinct effects on the party’s objective function: one positive and one
negative. The positive effect stems from a lower cost of information manipulation (remember c′(σ) < 0
so that −σ2c′(σ) > 0). This effect is associated with the second term on the LHS of equation (29).
The negative effect is given by the first term on the LHS of (29). A lower degree of information
manipulation reduces the amount of service duties that a party can extract from politicians. That
happens because of two main effects: 1) The equilibrium mass of unskilled candidates are reduced
(both because more skilled citizens are running for office and because the probability of being elected
for a high-type/low-signal activist is reduced); 2) The maximum level of service duty each unskilled
activist is willing to provide is lower because the expected net gains from politics are reduced. The
optimal level of σ is the one such that these two forces are counterbalanced. Before introducing an
16The fact that αs cannot be larger than unity imposes another restriction on σ which cannot be too small (σ > φπ−ws
)otherwise there would not be enough skilled citizens willing to enter politics and then cover all seats. However, sincewe know that σ can’t be smaller than 1/2 this assumption might be either neutralized by a restriction such thatφ < 2(π−ws) so that the previous assumption is always true; or it can be made unnecessary by assuming that unskilledcitizens with high signal are enough to cover all seats so that no other skilled citizen is needed: µ(1 − s)(1 − σ) ≥ p.However in this case we would need another restriction on the equilibrium level of information which cannot be toolarge otherwise some unskilled citizens will not run for elections: σ ≤ 1− p
µ(1−s) .
14
explicit cost function in order to fully characterize the optimal value of σ∗, we we first analyze how is
the equilibrium number of parties N∗ being determined.
Finally, it is important to emphasize that in order for the second-order condition to be satisfied
for σ∗ to be a maximum, we require the objective function should be globally strictly convex in
σ ∈ (1/2, σ) which in turn requires
∂2vk∂σ2
=2
σ3
AuNφπ − 1
π − ws− c′′ (σ) < 0, ∀σ ∈
(1
2, σ
)Substituting in using (29) the above condition reduces to
−σc′′ (σ)
c′ (σ)= γ(σ) > 2
which tells that the elasticity of the marginal cost of manipulating information should be large enough.
4.3 Number of parties and information manipulation
We assume that entry is free for parties (stage 2 of the game). Accordingly, the following result holds
Lemma 3. Given free entry, the equilibrium number of parties N∗ is such that parties make zeroexpected profits. Accordingly, in an equilibrium where unskilled citizens all become activists, whilesome skilled citizens do not (Au = µ(1− s) and As < µs),
N∗ =1
c (σ)
(φµ (1− s)
(π − 1
π − ws1− σσ− 1
)). (30)
Proof. For given σ, N∗, is found by: (i) substituting in equation (5) for the equilibrium values of es,eu, αs, αu, Ak,u, and Ak,s, then; (ii) imposing vk = 0, and; (iii) solving for N .
It is immediate to verify that the equilibrium number of parties, N∗, and the equilibrium value of
information manipulation are found solving the simultaneous system of equations (29) and (30). By
substituting for N∗ in (29) using (30), we get
σ
σ − σ= −σc
′ (σ)
c (σ)(31)
which defines implicitly the equilibrium value for σ = σ∗ as a function of σ and therefore of π and
ws only.
4.4 Full equilibrium characterization with an explicit cost function
We now fully characterize the equilibrium by assuming the following explicit cost function
c (σ) =
(σ − σσ
)θ(32)
with θ, σ > 0, where we recall that σ, which we assume to be not only greater than 0.5, but also
greater than σ, measures the quality of the signal that parties can produce at no cost.
15
Given (32), condition (31) that implicitly defines σ∗ in the case of an interior solution, becomes,
σ
σ − σ= θ
σ − σσ
(33)
It is important to note that, in order for an interior solution to exist, the following condition needs to
be satisfied,
γ (σ) =σ (θ + 1)− 2σ
σ − σ> 2⇔ θ > 1 (34)
which ensures that the second order condition associated with an interior maximum at σ = σ∗ is
satisfied.
Solving (33) for σ∗ yields,
σ∗ =σ
θσ − σσ(θ − 1) (35)
Note that θσ > σ > σ by assumption, which automatically implies σ∗ < σ, which ensures that σ∗ is
consistent with the existence of the type of equilibrium we analyzing. Moreover, for σ∗ > 0.5 to hold,
we need to impose the additional following restriction on θ
θ >σ(2σ − 1)
σ(2σ − 1)> 1 (36)
If this were not the case, then manipulating information would be so cheap that parties would always
choose the minimum level of information manipulation of σ∗.17 The associated value of N∗ is found
substituting for the value of σ in equation (30) using the above explicit solution for σ∗.
Having fully characterized the equilibrium, we can now turn to the analysis of how changes in the
pay of politicians affect the quality of elected politicians.
5 Quality of politicians and politicians’ pay
Given the equilibrium value of the mass of candidates, Cs, we can express the average quality of
politicians as a function of the equilibrium value of information manipulation, σ∗. As already dis-
cussed, the average quality of politicians q is defined by the fraction of skilled politicians which in a
symmetric equilibrium is – by the law of large numbers – equivalent to the fraction of skilled can-
didates with a signal s, that is, we average quality as q ≡ Css
Cs . By using (23), and recalling that
Css = Cs − µ(1− s)(1− σ) we finally obtain
q = 1− µ (1− s) (1− σ∗)Cs
= 1− φµ (1− s) (1− σ∗)pσ∗ (π − ws)
(37)
where σ∗ is the equilibrium level of σ. In order to see how politicians’ pay affects quality of politicians
we differentiate with respect to π, yielding
∂q
∂π=
φµ (1− s)pσ∗ (π − ws)
1− σ∗
π − ws︸ ︷︷ ︸Selection effect
+∂σ∗
∂π
1
σ∗︸ ︷︷ ︸Information manipulation effect
. (38)
17It is easy to note that σ∗ is, quite intuitively, decreasing in θ.
16
Accordingly, the effect of pay on quality can be viewed as the sum of two different effects: (i)
A Selection effect and; (ii) An information manipulation effect. The first effect – which is always
positive – is the one put forward by CM04. An increase in π attracts skilled citizens into politics,
which other things equal, improves the quality of activists, candidates, and elected politicians. The
second effect is novel and embodies the main contribution of the present paper. It says that increasing
the pay of politicians affects quality of politicians also because of the effect that such increase might
have on the extent to which parties will try to manipulate information in order to increase the chance
of election for their unskilled potential candidates, from whom they might be able to extract higher
service duties compared to those they can extract from skilled ones. Crucially, the following result
holds,
Lemma 4. Consider an equilibrium where unskilled citizens all become activists, while some skilledcitizens do not (Au = µ(1 − s) and As < µs). Then, the information manipulation effect induced byan increase in the pay of politicians, π, is always negative. That is,
∂σ∗
∂π
1
σ∗< 0 (39)
Proof. We evaluate the first-order condition (29) in σ = σ∗ and we write it as follows
c (σ∗) σ ≡ − (σ − σ∗)σ∗c′ (σ∗)
Now apply the implicit function theorem by differentiating both sides of the latter expression withrespect to σ and π and solving for ∂σ∗
∂πto obtain
∂σ∗
∂π=
c (σ∗)
(σ − σ∗) (γ (σ∗)− 2) c′ (σ∗)
∂σ
∂π
(σ∗c′ (σ∗)
c (σ∗)+ 1
)Where γ(σ∗) = −σ∗c′′(σ∗)
c′(σ∗)is the elasticity of the marginal cost of manipulating information evaluated
at the equilibrium value of σ∗.Now we can use again the first order condition (29) so that we can finally write
∂σ∗
∂π/σ∗ =
σ∗
σ − σ∗1
γ (σ∗)− 2
∂σ
∂π/σ
which is clearly negative since: 1) by (26), ∂σ∂π/σ is negative; 2) by the second-order condition,
γ(σ∗) > 2; 3) in the interior solution σ > σ∗.
Why does an increase in politicians’ reward always induce parties to manipulate information
more? The intuition can be easily gained from figure 3 that shows the equilibrium expected benefits
B(σ; π), blue and green lines, and costs c(σ), red line, of the representative party as a function of
σ for σ ∈ (0.5, σ) for two different values of politicians’ pay equal to π0 and π1 respectively, with
π0 < π1 . Notice that while c(σ) does not depend on politicians’ pay π, B(σ; π) does depend on π
because parties’ benefits are affected by politicians’ pay (and market wage ws as well) both through
the service duties that can be charged from each (unskilled) politician, eu, as well as and from the
mass of (unskilled) politicians belonging to the representative party.
17
Figure 3: The information manipulation effect: change in the optimal solution of σ following anincrease in π.
At the initial value of π = π0, the optimal choice σ∗(π0) corresponds to a tangency point of the
benefit curve B(σ; π0) and the cost curve c(σ). Notice that at σ∗(π0) total costs equal total benefits,
i.e. B(σ∗(π0); π0) = c(σ∗(π0)), and marginal costs equal marginal benefits, i.e. B′(σ∗(π0); π0) =
c′(σ∗(π0)).18 It is worth to emphasize the interpretation of the horizontal intercepts for the two curves.
The horizontal intercept for the costs curve is σ which is in fact the amount of information that the
representative party can produce at no cost (c(σ) = 0). The horizontal intercept for the benefits curve
is by contrast σ(π0) = π0−1π0−1+π0−ws
which as we know represents the amount of information that, for
a given π = π0, equalizes the maximum service duties that can be charged from unskilled politicians
to those that can be charged to skilled politicians. Since in this type of equilibrium es = 0, then
eu(σ(π0)) = 0 and therefore B(σ; π0) = eu(σ(π0))αs(σ(π0))(1− σ(π0)) = 0.
Now assume an exogenous increase in politicians’ reward from π0 to π1 > π0. The costs curve
remains unaffected while the benefits curve becomes steeper and the horizontal intercept shifts to
the left. The movement of the benefits curve can be explained as follows. First, party’s marginal
benefits of manipulating information are increased at any level of σ and, second, unskilled politicians
are less willing to run for public office so that the threshold σ falls. As a result, the ”old” equilibrium
value σ∗(π0) is not optimal anymore, because, according to the new benefits curve B(σ; π1), at σ∗(π0),
marginal benefits are larger than the marginal costs, i.e. −B′(σ∗(π0); π1) > −c′(σ∗(π0)), and total
benefits are smaller than total costs (B(σ∗(π0);π1) < c(σ∗(π0))). The equality between marginal and
absolute benefits and costs is restored by reducing the value of σ to σ∗(π1) which is associated to
higher marginal and absolute costs (since c′′(σ) > 0 and c′(σ) < 0) but even higher marginal and
absolute benefits.
18Also notice that, due to free entry, the representative party cannot make profit in equilibrium and this is why thecosts curve always lies above the benefits curve except in correspondence with the optimal value.
18
One may ask how to reconcile an increase in absolute and marginal benefits of manipulating
information following an increase in politicians’ reward, with a smaller propensity of unskilled citizens
to become activists, reflected in smaller service duties that can be charged to them. Here the role of
free-entry in the parties market is crucial. An increase in politicians’ pay, π, has two opposing effects.
First, it reduces the service duties that can be extracted from an unskilled activist , and the probability
that the latter can be elected, reducing both total and marginal benefits of manipulating information,
for a given N∗. Second, by reducing absolute benefits while leaving absolute costs unchanged (for a
given σ), it also reduces the equilibrium number of parties (N∗(σ∗(π1)) < N∗(σ∗(π0))), which in turns
increases total and marginal benefits of manipulating information for incumbent parties. It turns
out that the second effect always dominates so that an increase in π is good news for the incumbent
representative party.
To sum-up, the information manipulation effect can be explained as follows. An increase in
politicians’ pay, π, increases the marginal benefits of manipulating information for the incumbent
parties that are then more willing to bear the cost of manipulating information even further. The
overall effect of increasing the pay of politicians on quality of politicians is the result of the tradeoff
between the selection effect and the information effect. Increasing politicians’ pay always lead to more
information manipulation by parties, according to lemma 4, which –other things equal– would increase
the chance of election for unskilled activists, thereby reducing politicians’ quality. Therefore, for an
increase in politicians’ pay to improve the quality of elected politicians, the selection effect should be
strong enough, as we explore in the next subsection by resorting to the full characterization of the
equilibrium under the cost function (32).
5.1 Pay and quality of politicians with a specific cost function
Given the cost function, (32), the expression for the information manipulation effect can be obtained
by differentiating (35) with respect to π, which yields
∂σ∗
∂π
1
σ∗=
θσ
θσ − σ∂σ
∂π
1
σ(40)
where ∂σ∂π
1σ
is defined by (26). The following results holds
Proposition 1 (Pay and quality of politicians). Consider an equilibrium where unskilled citizens allbecome activists, while some skilled citizens do not (Au = µ(1− s) and As < µs). Assume the cost ofmanipulating information is given by equation ( 32). Then, increasing the pay of politicians reducesthe quality of politicians whenever, (
π − wsπ − 1
)2
<1− σθσ
(41)
and improves it otherwise.
Proof. From (38) we know that quality of politicians, q, is decreasing in π if the information manip-ulation effect is larger, in absolute value, then the information manipulation effect, that is
∂q
∂π< 0⇔ −∂σ
∗
∂π
1
σ∗>
1− σ∗
π − ws(42)
19
Using (40) and (35) we can substitute for the explicit value of ∂σ∗
∂π1σ∗
and σ∗ to obtain
∂q
∂π< 0⇔ − σθ
θσ − σ∂σ
∂π
1
σ>θσ − σ (1 + σ (θ − 1))
θσ − σ1
π − ws(43)
Now simplify and use (26) and (25) to substitute for ∂σ∂π
1σ
and σ as a function of π and ws to obtain
∂q
∂π< 0⇔ σθ
(ws − 1
π − 1
)> θσ − (π − 1)
(π − ws)(1− σ) (44)
Collecting, σθ, multiplying both sides by π − 1 and π − ws, and dividing both sides by θσ we obtain
∂q
∂π< 0⇔
(π − wsπ − 1
)2
<1− σσθ
(45)
Quite intuitively, according to Proposition 1, with the specific cost function (32), an increase in
the pay of politicians’ reduces the quality of politicians whenever, other things equal,
i. θ is relatively small. When parties’ technology for information manipulation is efficient and
manipulation is cost-effective, parties have a role to play in manipulating information, i.e. the
manipulation effect is comparatively stronger;
ii. σ is relatively small. When the society is less aware, then the information manipulation effect
is large and the selection effect is comparatively small.
iii. π−ws
π−1is relatively small. When politicians reward is small compared to the market wage of skilled
citizens, the latter have less incentive to run for politics and the selection effect is comparatively
smaller.
Notice that cases (i) and (ii) might well be self-reinforcing. In other words, θ and σ, might be positively
correlated. On the one hand, it looks more likely that manipulating information is cheaper in less
aware societies, where the quality of media and the average level of citizens’ awareness and ability to
process information correctly is low. On the other hand, when manipulating information is cheap we
expect that parties take advantage of this by choosing a higher degree of information manipulation
(σ∗ is low when θ is low, see equation (35)), and one may expect that exposing citizens to an high
degree of information manipulation for a significant amount of time can affect their ”exogenous” level
of awareness and their ability to process informations correctly. In order to study these intriguing
interactions in detail, a dynamic framework is needed. We leave this analysis for future research
and we simply observe that the alleged positive correlation between σ and θ would strengthen our
argument.
What are the empirical implications of our model? To what extent our results provide a rational
for the relationship between reward for public office and quality of politicians in Italy and in USA
described in the introduction? Assuming that the left-hand side of condition (41) is similar in US and
20
Italy19, our model predicts that if quality of politicians and politicians’ pay are negatively correlated
in the case of Italy, and positive correlated in the case of the US, then we should expect the cost
of manipulating information θ and/or the level of citizen awareness σ to be lower in Italy than in
the US. As already discussed in the introduction, despite finding proper proxies for these quantities
is not an easy task, we propose the Freedom of the Press index as a proxy for θ and the so-called
Index of Ignorance elaborated by IPSOS-MORI as a proxy for citizens’ awareness σ. According to
the Freedom of the Press index, Italy is in 70th position in 2012, 69th in 2013 and 64th in 2014 and
its press is defined as ”partly free”. On the other hand, US is 22th in 2012, 23th in 2013, 30th in 2014
and its press is defined ”free”. Indeed, these observations suggests that manipulating information
is more difficult in US than in Italy which, in our model, translates in a lower value of θ in Italy
than in US. As for the Ignorance index, Italy is the more ignorant country among the 14 taken into
consideration, suggesting that citizens’ awareness (and then the level of σ) is lower in Italy than in
US. These empirical findings seem to support the model’s predictions and suggest that the mechanism
we propose might have a role in explaining the observed data.
Finally, note that taking into consideration only these 14 countries, we have computed the corre-
lation between the Index of Ignorance and the Freedom of the Press index in 2014. Such correlation
is equal to −.49, which confirms our priors on the interaction between these two variables.
6 Conclusions
The literature on how remuneration to politicians affects their quality is inconclusive, both at the
empirical and theoretical level. We analyze how such relationship could be affected when party play
a role in the selection of candidates, in a setup related to Caselli and Morelli, 2004, and show that the
relationship between quality and pay of politicians is ambiguous. In particular, an equilibrium where
increasing political pay reduces the average quality of politicians exists, when the costs of manipulating
information are not too large, and/or the society is less aware, i.e. the quality of information available
about perspective politicians is relatively low, due for instance to the bad quality of the media or the
low level of awareness of citizens. In such contexts, political parties may prefer to bear the additional
cost of distorting information through aggressive political campaign in order to increase the probability
of elections for unskilled candidates which, in case of election, would be willing to provide higher party
duties relative to skilled politicians.
19Using OECD data (OECD, 2011) for earnings premiums from tertiary education, Barro-Lee dataset (Barro andLee, 2010) for the percentage of population with a tertiary degree, Piketty (2014) estimates on labour share and public
informations for politicians pay in US and Italy, we have calibrated the LHS of (41),(π−ws
π−1
)2, for the US and Italy
from 1999 to 2009. The estimates for the two countries turn out to be very close. In US the estimated value of theLHS of condition (41) ranges from 0.66 to 0.72 while in Italy it ranges from 0.79 to 0.89.
21
References
[1] Barro, R. and J. Lee, April, 2010. ”A New Data Set of Educational Attainment in the World,
1950-2010.” Journal of Development Economics, 104: 184-198.
[2] Besley T., 2004. ”Paying Politicians: Theory and Evidence”. Journal of the European Economic
Association, 2: 193-215
[3] Besley T., 2005. ”Political Selection”. Journal of Economic Perspectives, 19: 43-60
[4] Besley, T, Montalvo J. G., and M. Reynal-Querol, 2011. ”Educated leaders matter”. The Economic
Journal, 554: 205-227.
[5] Besley, T., R. Pande, and V. Rao. 2005. ”Political selection and the quality of government: Evi-
dence from South India”. Economic Growth Center, Yale Un. DP N. 921
[6] Braendle, T. 2015. ”Does remuneration affect the discipline and the selection of politicians? Evi-
dence from pay harmonization in the European Parliament”. Public Choice,162: 1-24
[7] Caselli F. and Morelli M. (2004). ”Bad Politicians”. Journal of Public Economics, 88: 759-782.
[8] Dal bo E., Finan F. and Rossi M, 2013. ”Strengthening State Capabilities: The Role of Financial
Incentives in the Call to Public Service”, Quarterly Journal of Economics, 128: 1169-1218.
[9] Ferraz C. and Finan F., 2009, ”Motivating Politicians”, NBER Working Papers 14906
[10] Fisman R., Harmon N., Kamenica E., Munk I., 2015. ”Labor Supply of Politicians”. Journal of
European Economic Association, Forthcoming.
[11] Gagliarducci S. and Nannicini T., 2013, ”Do better paid politicians perform better? Disentangling
Incentives from Selection”. Journal of the European Economic Association, 11: 369-398.
[12] Hoffman M. and Lyons E., Do Higher Salaries Lead to Higher Performance? Evidence from State
Politicians, Working Paper, 2013.
[13] Keane, M. P., A. Merlo, 2010. ”Money, Political Ambition, and the Career Decisions of Politi-
cians”. American Economic Journal: Microeconomics, 2, 186-215.
[14] Kotakorpi, K., and P. Poutvaara, 2011. ”Pay for Politicians and Candidate Selection: An Em-
pirical Analysis”. Journal of Public Economics, 95: 877-885.
[15] Mattozzi A. and Merlo A., 2008, ”Political careers or career politicians?”, Journal of Public
Economics, 92: 597-608.
[16] Mattozzi A. and Merlo A., 2011, ”Meritocracy, Fourth version”, PIER Working Paper Archive
13-010.
22
[17] Merlo A., Galasso V., Landi M. and Mattozzi A., 2008, ”The Labor Market of Italian Politicians,”
Carlo Alberto Notebooks 89, Collegio Carlo Alberto.
[18] Messner M. and Polborn M., 2004, ”Paying Politicians”. Journal of Public Economics, 88: 2423-
2445.
[19] Piketty T. and Zucman G., 2014, ”Capital is back: Wealth-Income Rations in Rich Countries
1700-2010”, Quarterly Journal of Economics, 124, 1255-1310.
23
Ultimi Contributi di Ricerca CRENoS I Paper sono disponibili in: Uhttp://www.crenos.itU
14/13 Robe r t a Me l i s , A l e s sand r o T rudda , “Mixed pens ion sys tems sus ta inab i l i t y”
14/12 Gera rdo Mar l e t t o , “Soc io- techn ica l dynamics and po l i t i ca l ins t i tu t ions : A mul t i l eve l Darwin ian f ramework of sus ta inab i l i t y t rans i t ions”
14/11 Andr ea P inna , Sha l l We Keep Ear ly Diers A l ive?” 14/10 Gianp i e r o Me l on i , D im in t r i Pao l i n i , Juan d e D io s T ena ,
“Amer ican Beauty : an ana lys i s o f U .S . mov ies revenues in the g loba l marke t”
14/09 S i l v i a Ba l i a , R ina ld o B rau , Emanue l a Mar r o cu , “Free pa t ien t mobi l i t y i s not a f ree lunch . Lessons f rom a decentra l i sed NHS”
14/08 Gera rdo Mar l e t t o , “A de l ibera t ive -par t i c ipa t ive procedure for sus ta inab le urban mobi l i t y – F ind ings f rom a tes t in Bar i ( I ta ly )”
14/07 Manue la De idda , “Insu la r i ty and economic deve lopment : a survey”
14/06 Edoardo O t ran t o , Mas s imo Muc c i a rd i , P i e t r o B e r tu c c e l l i , “Spat ia l Effec ts in Dynamic Cond i t iona l Corre la t ions”
14/05 Fran c e s c o Qua t ra r o , S t e f ano Usa i , “Are knowledge f lows a l l a l ike? Ev idence f rom European reg ions”
14/04 Ange l o An to c i , Fab i o Saba t in i , Mauro Sod in i , “Onl ine and of f l ine soc ia l par t i c ipa t ion and soc ia l pover ty t raps . Can soc ia l ne tworks save human re la t ions?”
14/03 Anna Bus su , C laud i o De t o t t o , “The b i -d i rec t iona l re l a t ionsh ip be tween gambl ing and add ic t ive subs tances”
14/02 Ale s sand r o F i o r i , Tada s Guda i t i s , “Opt ima l Ind iv idua l Cho ice of Contr ibut ion to Second P i l l a r Pens ion Sys tem in L i thuan ia”
14/01 Ol i v i e r o A . Carbon i , Pao l o Rus su , Measur ing Env i ronmenta l and Economic Eff ic iency in I ta ly : an Appl ica t ion of the Ma lmquis t -DEA and Grey Forecas t ing Mode l
13/24 Luca De idda , J o s é J . Cao -A lv i r a , “F inanc ia l l ibera l i za t ion and the deve lopment of microcred i t”
13/23 Manue la De idda , “Economic hardsh ip , hous ing cos t burden and tenure s ta tus : ev idence f rom EU-SILC”
13/22 Vania Manue l a L i c i o , Anna Mar ia P inna , “Measur ing insu la r i t y a s a s ta te o f na ture . I s there a case of bad geography?”
13/21 Vania Manue l a L i c i o , Anna Mar ia P inna , “The European f i rms ' expor t ac t iv i ty to the ne ighbour ing countr ies”
13/20 Ka l l i o r a s D im i t r i s , Anna Mar ia P inna , “Trade ac t iv i ty be tween the EU and i t s ne ighbor ing countr ie s : Trends and potent ia l”
13/19 Claud ia C i ga gna , G io vann i Su l i s , “On the potent ia l in te rac t ion be tween l abour marke t ins t i tu t ions and immigra t ion po l ic ie s”
13/18 Romana Gar gano , Edoa rdo O t ran t o , “F inanc ia l C lus te r ing in Presence of Dominant Marke ts”
13/17 Etto r e Pan e t t i , “Financ ia l L ibera l i za t ion w i th Hidden Trades”
13/16 Adr iana Di L ib e r t o , “Length of s tay in the hos t country and educa t iona l ach ievement of immigrant s tudents : the I ta l i an case”
www.crenos.it
