SSRN Id4230236

Straw donor schemes and corporate political participation
Michele Castiglioni*
September 27, 2022
Abstract
Curtailing corporate participation in electoral campaigns can improve political accountability; as

a result, campaign finance regulation interventions are widespread. However, in settings of high
returns to political connections, corporate actors can undermine their effectiveness by legal and
illegal means. I focus on the latter, addressing the emergence of straw donor schemes – in which
corporate actors circumvent restrictions by employing private donors as intermediaries. Empiri-
cally, I focus on a corporate donation ban in Brazil and study how it affected corporations’ ability
to signal support using private-for-corporate donations coming from their employees. Using a
Difference-in-Differences strategy and leveraging pre-intervention heterogeneity in firms’ dona-
tion patterns, I show that firms with a history of political participation orchestrate a more precise
signal to politicians with respect to comparable firms even after the ban. Overall, I show that
the ban imposed high transparency costs with a minor impact on the ability of firms to maintain
speech in the political arena.
Key words: Money in politics; Campaign Finance Reform; Corruption Displacement
* Michele Castiglioni is a Max Weber Fellow, Department of Political Science, European University Institute, Via della
Badia dei Roccettini 9, 50014 Fiesole, Italy (michele.castiglioni@eui.eu). The author thanks the Harvard-MIT Data Center
(HMDC) for technical support; Scott Desposado, Makoto Fukumoto, Vincenzo Galasso, Natalia Garbiras-Díaz, Tommaso
Nannicini, Massimiliano Onorato, Piero Stanig, Eleanor Woodhouse, and the seminar audiences at UCSD, EUI, as well as
MPSA meetings for helpful suggestions. Any errors that remain are the author’s sole responsibility.
Electronic copy available at: https://ssrn.com/abstract=4230236

1 Introduction
Interest groups’ campaign contributions bring with them, alongside the promise of free expression, the
risk of disproportionate influence of corporate actors in the democratic process. In political contexts
where money’s role is critical, corporate participation in politics has been perceived as a threat to the
integrity of electoral dynamics and governmental processes (Stratmann 2005; Scarrow 2007). Interest
groups making large contributions have been found to gain undue influence on electoral outcomes, on
legislation shaping, and on the regulatory proceedings outcomes (Wright 1990; Romer and Snyder Jr
1994; Ansolabehere, Gerber, and Snyder Jr 2001; Shapiro 2009; Hasen 2016). Regulation of corpo-
rate participation in election is, as a result, unsurprisingly regulated worldwide (IDEA 2018). Despite
practitioners’ rush to implement contribution caps to limit these quid pro quo exchanges , the academic
literature is divided on the extent to which these reforms are effective with respect to their stated scope
(Levitt 1995; Che and Gale 1998; Smith 2009). Presently, I study the focus on corruption displacement
dynamics, specifically focusing on strategic responses corporate actors give to regulation attempts.
I try to overcome the normative attitude underpinning financial regulation reform, towards a more
comprehensive analyses, relevant for political scientists and practitioners alike. If – in an attempt to
preserve the status quo– actors strategically respond to regulation, this should be key in corruption
curbing policy evaluations, as well as in jurisprudence (Hasen 1998; Butrymowicz 2009). Presently,
I explore to what extent an all-encompassing corporate campaign ban, in a context of high-return to
contributions, have stimulated straw donor schemes. This novel unexpected outcomes of reforms lays
the ground to discuss to what extent these strategies negatively effect transparency electoral account-
ability.
Studying responses to policy regulations that formally comply, while de facto bypassing them, appears
beyond the scope of campaign contribution regulation assessment and modeling. A rigorously dis-
cussion of legal loopholes exploitation is crucial to be able to provide the meaningful overlook of a
phenomenon that is both diffuse and worrying. This piece presents a novel approach to the study of
straw-man schemes, that rests on data-mining of campaign finance disclosure reports. The method is

novel to the discipline and to public discourse alike. In a context with high-returns to political con-
nections (Boas, Hidalgo, and Richardson 2014), straw-man donations substitute for special interest
donations when these are banned. I discuss the elusive practice, its diffusion, and its relevance in pre-
serving the rent-seeking ability of interest groups. The analysis rests on the development of different
minimal and scalable measures that are usable in multiple contexts. Brazil is an exceptionally fit con-
text in which to study straw donor schemes. Despite having a sufficiently comprehensive jurisprudence
(Samuels 2001a, 2001b), the country displays overt signs of a dysfunctional campaign finance system
- for corruption, (Samuels 2001c; Fleischer 2002), costs (Aguiar 1994; Zovatto 2003) and returns on
investment (Boas, Hidalgo, and Richardson 2014). On top of this, in the wake a major political scan-
dal, the Country formally ruled out corporate donations, creating the perfect setting in which to study
whether coordinated private signals did substitute for corporate ones, and to what extent.
On top of this, a major political scandal in the Country led to a drastic, all-encompassing reform that
banned all corporate donations that just before fueled the whole system. Pundits hold that the change
pushed corporations to develop ‘creative’ strategies to maintain the status quo, among which the use
of indirect donations was prominent (Douglas 2015; Benjamin and Caruso 2016; Borges and Marcela
2016). To study the universal ban treatment, I explore firms pre-intervention donation patterns match-
ing private and corporate donations within a firm the determine pseudo-control groups. Employing
a Difference-in-Differences strategy, I gauge the impact of the intervention observing the changes in
within- and across-candidates concentration contribution indexes, taken as a proxy for donations or-
chestration. I create both indexes modeling on the Herfindhal-Hirschman Index (Dharmapala and
Palda 2002). Tracing back individual donations to specific corporations is made possible thanks to
of unique identifiers’ present in a dataset spanning the universe of formal employment contracts of
Brazil.
Restricting corporate campaign contributions in the manner described pushes firms to restructure
their donation portfolios with the goal to maintain their political and speech ability, pivotal to extract
rents from politicians (Koger and Victor 2009; McMenamin 2013; Bertrand, Bombardini, and Trebbi

2014). Firms with a history of campaign contributions – those actually affected by the treatment –
display significantly higher concentration indexes than the rest of the firms – both within- and across-
candidates. As the measures are taken as a proxy of forced donations orchestration, whether or not in
overt violation of the law, the piece concludes that treated firms were better able to navigate the post-
scandal political landscape, maintaining their level of signaling. The use of private donations, could
have been key to maintain speech, but also unlawful favors exchange relation down the line. In a sea of
private donations, concentrated donations can be recognized and decoded by a politician as a unitary
signal of support and, thus, can substitute for a corporate contribution in its scope.
To advance an ownership claim over its employees’ donations, at a minimum, a firm needs to make
sure these individual donations display a peculiar pattern. Producing concentrated donations is a suf-
ficiently easy and effective way to produce such pattern. A firm can secure it by means of simply
orchestrating its employees donations or proactively investing funds to achieve such a goal. The pro-
duction of the pattern precedes and is irrespective of a politician’s ability to decode the signal unaided.
My original measures of the phenomenon provide a rigorous, quantitative, and externally valid solu-
tion to detect such efforts. They provide a good indication of the effectiveness of a ban, going beyond
the mechanical elimination of a source of donations.
Blindly advocating for campaign finance regulations often fails to take into consideration several un-
intended consequences of the policy - even in an ideal implementation scenario. The present paper
discusses the effects of such an intervention in a relevant scenario and delves into how legislation’s
shortfalls are taken advantage of by corporate actors. The paper presents a more realistic picture of
the issues and challenges open to any political scientist modeling campaign finance welfare effects,
and any legislator planning to curb donors’ influence. Overlooking strategic reactions to donation
bans comes at the cost for voters of reduced information, transparency and electoral accountability.
This is unacceptable when designing effective legislation.

2 Literature Review
The present paper contributes to a recent body of research on campaign finance regulations, the role of
money in politics, and the ability of actors to displace corruption practices. The study aim at expand-
ing our current understanding of how campaign funding and corruption interlink, especially in Latin
American settings. The specificity of the Latin American case has been explored in a general com-
parative perspective (Austin and Tjernstrom 2003) and in isolation (Castillo Vera and Zovatto 1998;
Carrillo, Lujambio, and Navarro 2003). In both a clear connection emerges between the way in which
political parties fund their election campaigns and the degrees of corruption they experience; with
transparency and reporting being positively correlated to corruption, and restriction to access of dif-
ferent actors negatively correlated to it. These studies mainly present campaign funding systems in
terms of the limitations they impose on private contributions - at the center of corruption dynamics,
as in the present research. Brazil emerges as a remarkably interesting case, with both limits on sources
of private contribution, and contribution ceilings - making it one of the most restrictive systems in
the region - though still reporting high level of corruption. Zovatto (2003) presents some of the ma-
jor problems regarding enforcing anti-corruption restrictions, and explains some of the limitations of
the ‘textbook solutions’ to corruption. In particular, it states that in the context of Brazil “imposing
restrictions generally encourages parties to resort to ‘creative’ accounting and practices designed to
stretch these limits”. This dynamic, in a form that has previously been overlooked, will be at the center
of the present study.
Considering how corporations revert to straw-man donations in response to a all-encompassing pol-

icy intervention, the paper presents a further example of how reforms might produce unexpected out-
comes. Here, no new loci of corruption are generated –as in Mishra (2002), or Yang (2008)–, instead, the
same kind of activity is carried out through a new channel. The corporate ban, in this sense, becomes
one of those topic solutions unable to affect informal institutions, nor the incentives that political and
non-political actors face Kunicová 2006; as such, it is revealed to be ineffective. Despite focusing on
a specific form of political participation of corporations, the paper contributes to a broader literature

discussing the rationale and dynamics behind more classical forms of political participation of corpo-
rations, both in general (Ansolabehere, De Figueiredo, and Snyder Jr 2003; McMenamin 2013), and
with specific reference to Brazil (Samuels 2001a; Arvate, Barbosa, and Fuzitani 2013; Bourdoukan
2010; Boas, Hidalgo, and Richardson 2014).1 The paper, moreover, contributes to the exiguous liter-
ature examining the effects of regulating campaign finances from an empirical perspective, including
Stratmann, Francisco, et al. 2006 that discusses how campaign limits lead to closer elections for fu-
ture incumbents other than the promoters, and to an increase in candidates in the US state elections;
Milligan and Rekkas (2008) that finds that large limits on spending leads to less competitive elections,
fewer candidates, and lower voter turnout in Canadian federal elections; and Avis et al. (2017) discusses
the increase in political competition and incumbency advantage reduction brought about by spending
limits, in the same context as the present one.
In addition, the paper contributes to a vast scholarship in jurisprudence that discusses the potential
undemocratic consequences of campaign finance reforms. The US-centered contributions focus sub-
stantially on the Federal Election Campaign Act of 1971 and the subsequent Supreme Court per curiam
opinion (Buckley v. Valeo 1976) that struck down limits on independent expenditures.2 At the core of
this literature lies the discussion of the role that political contributions play in expressing citizens’ opin-
ions in US politics (Smith 1995, 2009). This speaks directly to the “donations-as-speech” perspective
in political science (McMenamin 2013). Besides this, the scholarship addresses the regularity accord-
ing to which excessive regulation incentivizes the creation of new loci of corruption – aptly labeled the
‘First Law of Political Thermodynamics’: “[...] every reform effort to constrain political actors produces
a corresponding series of reactions by those with power to hold onto it” (Issacharoff and Karlan 1999).
The literature is generally skeptical regarding the possibility to destroy the desire for political power,
raising the issue of monitoring with respect to practices that circumvent fundraising restrictions. The
issue of entrenchment is also explored to a certain, though informal degree (Klarman 1996). Several
contributions in this strand of literature suggest to cope with the phenomenon modest legal reforms
1. For a comprehensive review of more classical literature, see: Stratmann (2005).
2. For an overview of more recent reforms in U.S. campaign-finance policy, as well as an overview of the role political
scientists played in the process, please refer to Mann (2003).

– such as the introduction of public financing and disclosure measures – as a panacea. The present
work, however, calls into question the effectiveness of such measures, at least for some contexts.
To conclude, given that straw-man donations rely on the mobilization of employees, the paper also
speaks to a strand literature in the field of management on the role played by corporations in employ-
ees’ activism, as well as in fostering ideological alignment among them (Beyer 1981; Rosenstone and
Hansen 1993; King 2008; Gupta, Briscoe, and Hambrick 2017). However, the sudden change in pol-
icy and the substantial impact it had on corporations’ signaling ability, suggests that the coordination
effects I find (as revealed by the increase in donations’ concentration) should be attributed to the will
of corporations to tinker with their employees’ donations rather than to an ideological realignment
produced through culture change or nudging.
3 Institutional Context
3.1 Brazilian municipalities
Brazil has 5,570 municipalities3 , which are constitutionally part of the Brazilian Federation and divided
across 26 states. Brazilian municipalities are governed by a mayor (prefeita/prefeito, and a municipal
chamber, made up council members (vereadores whose number varies according to population. The
mayor and the chamber are elected at the same time and simultaneously across the Country. Mayors
are elected by plurality rule and face a limit of two consecutive terms. Their mandate lasts 4 years. The
mayor performs both political and administrative duties, with no other appointed public manager in
charge (Avellaneda and Gomes 2017). The mayor performs not only political but also administrative
functions and is considered the most prominent and identifiable political actor among elected offi-
cials: the mayor is the key decision maker in Brazilian municipalities (Avellaneda 2008; Avellaneda
and Gomes 2017; Rocha, Orellano, and Bugarin 2018). In addition to this, Brazilian municipalities
display a high degree autonomy – among the highest in South America (Nickson 1995; Samuels 2004;
3. Municipalities vary substantially in population, development, and other political and socio-economic indicators (Ge-
ografia e Estatística 2010). These factors are kept into consideration in the analyses that follow.

Titiunik 2009) – and local governments are directly responsible for the provision of key goods and ser-
vices - including management of health services, primary education, transportation, and sanitation.4 It
is not surprising that the practice of corporate actors to cozy up to mayors in search for a an economic
return in the form of contracting is widespread.5
3.2 Campaign finances
Formally, Brazil strictly regulates campaign finance: bans have been historically in place on dona-
tions coming from government contractors, trade unions, foreign entities, and anonymous sources
(Falguera, Jones, and Ohman 2014). Donations limits are in place and historically were set at 2% of
a one’s gross annual revenues for private donations, and at 10% of previous-year gross revenues for
corporate donations (Zovatto 2003). Regulations apply to all races across levels of government, and
candidates are requested o publicly report both campaign donations and spending. Legal source of
finances include corporations, private individuals and parties – self-financing is allowed. However,
given a number of flaws in the implementation of political finance controls, incentives for reckless
fundraising practices has emerged in the Country making these commonplace across the country
(Casas-Zamora 2016). In particular, Brazil’s electoral regulations have been deemed permissive to-
ward excessive corporate donations. As a result, campaign financing has long emerged as a market-
place fo firms to buy political influence with high returns (Boas, Hidalgo, and Richardson 2014, p.
417. Campaign finance regulation changed abruptly in 2015 with the new Electoral Reform (Law No.
13.165 / 2015, or ‘minirreforma’, Eleitoral (2016) . In March of the previous year the largest-ever cor-
ruption scandal in the Country was exposed: the Operation Car wash - Operação Lava Jato uncovered
a complex collusion and corruption network linking Brazilian business and political elites, which led
to an extensive and successful police investigation of even some of the most prominent figures in the
country (Kamm 2015; Sotero 2018). The scandal opened a window of opportunity for the Supreme
4. Brazilian Constitution of 1988, Article 30, Chapter IV establishes municipal responsibilities. See: Avellaneda and
Gomes (2017) for an in-depth discussion.
5. For anecdotal evidence access: https://www.bbc.co.uk/news/world-latin-america-34104124,https:
//g1.globo.com/rio-de-janeiro/noticia/obras-da-odebrecht-foram-superfaturadas-em-mais-de-
r-3-bilhoes-no-rj.ghtml,https://www1.folha.uol.com.br/poder/2015/08/1668585-pf-deflagra-op
eracao-de-combate-ao-superfaturamento-em-estadio-da-copa.shtml.

Court (Supremo Tribunal Federal to pronounce unconstitutional the involvement of corporate actors in
campaign financing (‘Ação Direta de Inconstitucionalidade’ 4650 Federal (2015). Corporate donations, it
was claimed, distort electoral fairness and translate in undue influence over elected officials. On this
basis, the subsequent electoral reform banned corporate donations at all levels of government for both
candidates and parties. The ban was effective beginning from the 2016 municipal election.6
3.3 Straw donor schemes
The corporate donation reform was met with skepticism from the onset.7 In a country with the ex-
tremely expensive elections (Zovatto 2003; Casas-Zamora 2016) and where a system of economic do
ut des practices underpinned the electoral machine, the fear of corrupt practices displacement was very
high.8 Both the proliferation of off-the-books slush funds (‘caixa dois’), and the use of straw man donors
(‘laranjas’) – two sides of the same coin – were discussed both in the local (Borges and Marcela 2016;
Ramalho 2016; Av.Vv. 2016) and international press (Douglas 2015; Benjamin and Caruso 2016). In-
vestigative pieces on the phenomenon only appeared in the subsequent federal election, in 2018, when
it became clear that some businessmen, as private individuals, were participating in the campaign do-
nations in open substitution of their firm (Breno and André 2018; Josette 2018). Straw man donations
are the focus of the present analysis. Despite its enormous reach, the ban leaves open more than one
channel through which parties can still funnel money to candidates. In the presence of alternative
donation channels a legal loophole opens up for corporations that are (still) willing to influence politi-
cians. The extent to which said loophole has been taken advantage of by corporations is precisely what
the present piece aims to assess. Displacement of this kind raises a number of key issuer with respect to
the transparency and accountability trade-offs that come with all-encompassing regulation.9 As these
6. the pronouncement was briefly upheld by the Congress, and then passed into low thanks to the veto by President
Rousseff (Ramalho 2015; Passarinho and Calgaro 2015; Matoso 2015).
7. Questioned on the efficacy of the policy, the President of the Chamber of Deputies Eduardo Cunha, replied: ‘People
will go out ‘hiring’ individuals to donate. Will the campaign cost less? No. You will have to pay the hiring’ (Borges and
Marcela 2016). Whereas, Justice José Celso de Mello declared: “If some donation is made from now on all that we can be
sure of is that we won’t know about it, but we can never be sure that it does not exist." (Douglas 2015)
8. In popular culture, the displacement was framed as an example of ‘Jeitinho Brasileiro’, or ’the Brazilian way’, referring
to way to accomplish something by circumventing or bending the rules or social conventions (Wikipedia 2019).
9. Similar considerations were raised with respect to the ban on public contractors’ donations: Falguera, Jones, and
Ohman (2014) underlines how the challenge of regulating ‘undesired donors’ is to determine whether a ban would merely

kind of regulation appear to be the go to option for legislators with small window of opportunities
in otherwise gridlocked corrupt scenarios (Casas-Zamora 2016), it is important to thorough discuss
their advantages, by design and test measure to ascertain the phenomenon. In other contexts, data
mining campaign finance disclosure reports is started being listed among the options available to an-
alysts to pick up irregularities in the donation process (Kelner 2018). Presently we try to use a novel
methodology to approach the detection of straw donor schemes.
4 Data
4.1 Electoral data
Electoral data for all political races in Brazil are available in the Repositório de dados eleitorais, a reposi-
tory provided by the Superior Electoral Court (Tribunal Superior Eleitoral, TSE). Presently, we will focus
exclusively on the local races for the years 2004-2016, because of data availability and quality concerns
further explained below. The 16,058 candidates running for one of the 5568 mayoral positions open
are identified by an 11-digits social security number (Cadastro de Pessoas Físicas) number. The data
report a number of socio-economic characteristics, including: gender, age, education, marital status,
employment (profession and status). Among the race-specific information, the data report party to
which the candidate is directly affiliated, the coalition of parties that supports her, and the result of
the election. Races contestability changes substantially averaging out at 2.91, and reaching a peak of
12 candidate for São Paolo. In addition to these, the IBGE (Instituto Brasileiro de Geografia e Estatística)
and IPEA (Instituto de Economia Aplicada) provide socio-demographic data on Brazilian municipalities.
Data are public and openly available.10 These are integrate in the analyses to control for some aspects
drive an donor to operate further into the shadows.
10. Data can be accessed at the institutions’ respective websites: www.ibge.gov.br and www.ipeadata.gov.br
10

4.2 Campaign finance data
Contribution reports for any political race in Brazil are publicly available through the Electoral Data
Repository (Repositório de dados eleitorais),that is maintained by the Superior Electoral Court (Tribunal
Superior Eleitoral, or TSE). Since law No. 8713/1993, it is a requirement for candidates to submit to
the Court a detailed overview of all the campaign contributions received in a given electoral cam-
paign. Presently, only electoral reports from years 2004-2016 will be employed: data quality for ear-
lier reports is poor, and using them would introduce too much noise in the analysis.11 . With respect to
donors, the reports include two identifiers: name and tax identification number – for both individuals
(natural person, or ’pessoa física’) and corporations (legal person, or ’pessoa jurídica’). Tax identification
numbers are unique and distinct across the two legal entities: for individuals is the already mentioned
CPF number, for firms is the 14-digits CNPJ (Cadastro Nacional da Pessoa Jurídica) number – each will
be used as personal identifiers for the respective category. Additional information is reported with
respect with the kind of donations received. For in-kind donations a brief (non-standardized) descrip-
tion of the service provided is listed. For private donations, no information is reported with respect to
the origin of the donation or labor market information of the donor, as it is customary in other settings
(i.e. the US).
Candidates in mayoral races have historically received support from both corporations and private in-
dividuals. Regarding private donations (self-donations excluded), they accounted for 20.8% of the total
value of donations, amounting to BRL 474m (USD 231m). This is particularly impressive as only 0.28%
of Brazilians (562,619 people) appeared as individual donors in last election prior to the reform (2012),
used for these calculations. The reform shrunk total donation by 34.6.After the electoral reform the
volume of donations contracted, at around -34.6% for mayors, despite the number of individual donors
surged by 17% – accounting for population change. Banned corporate donations, individual ones ac-
counted for 41.9% of donations. Regarding corporate donations, they accounted for 28.9% of the total
11. Contribution data for 2004 partially suffer form the same issues: for election year 2004 reports data from around
20% of municipalities are not reported. Missing municipalities will be dropped in a robustness check aiming at producing
a more balanced panel data-set.
11

value of donations, amounting to BRL 658m (USD 202m) in the last election prior to the reform. In
the campaign 47,169 (47,670) unique corporations (corporation branches) participated by means of
119,848 distinct contributions. The remaining donations in municipal races came from party transfer
and self-donations. Candidate-level statistical research on campaign finance in Brazil are sufficiently
thorough and, in the years considered, easily available. The ever-growing comparative interest on the
setting, at least since Samuels (2001a, 2001b) contribution, comes at no surprise.
4.3 Labor Market Data
The major source of employment data for Brazil is the administrative database produced by the Brazil-
ian Ministry of Labor and Employment (Ministério do Trabalho e Previdência Social, MTE/MTPS): the
RAIS (Relação Anual de Informações Sociais) which covers the universe12 of individual formal market
contracts for the whole country, including both the private and the public sector. It is considered a
high quality census of the Brazilian labour market (Dix-Carneiro and Kovak 2017; Brollo, Forquesato,
Gozzi, et al. 2017). The data-set made available by the MTE to research institutions upon request. The
RAIS provides detailed demographic and labor market information about each person who holds a
formal labor market contract in a given year. This information includes demographic characteristics,
employment status, and educational achievement. What is essential for the present analysis is that the
data-set provides information on each worker’s individual tax identification number (CPF), as well
as that of the corporation she was employed by (CNPJ). As mentioned in Section 4.2, these numbers
are unique and distinct across individuals and corporations. In the electoral years considered (2004
- 2016), the RAIS includes around 67m contracts in the Brazilian formal market that can be traced
back to about 54m uniquely identifiable employees. On average, employees sign 1.2 formal contracts
per head, with only 18% of the employees’ population having more than one contract registered in
the data-set each year. The RAIS dataset will be used to match individual and corporate donations,
and it is crucial for the resent analysis. The RAIS data-set is limited in one major respect: in the data-
12. The Brazilian institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística, or IBGE) maintains
that RAIS covers about 97% of the Brazilian formal market. http://ces.ibge.gov.br/base-de-dados/metadado
s/mte/relacao-anual-de-informacoes-sociais-rais.html
12

set only formal employment contracts are recorded. Business owners, consultants and self-employed
workers are not present.13 Given that it is of fundamental importance to my research to be able to
track donations back to each firm using the database, their absence leads to a substantial amount of
unmatched donations, which are dropped from the analyses (See: Table 3). Besides this, it is impor-
tant to note that in some firms the higher up a worker moves in the firm’s hierarchy, the more likely
it is that she moves from a dependent worker’s contract to an independent one. This means that the
people likely to have more power and access in a given firm are also those missing from the data-set,
which prevents me from tracing back to a firm specifically their (potentially high) donations. If the
straw-man donation dynamics involves this category of workers – as anecdotal evidence shows –, it
is important to realize that my ability to fully capture the dynamic is limited. The present needs to be
taken as a lower-bound estimate, though still very much informative regarding the straw-man dona-
tion dynamic. Anything captured by my estimations is likely to be representative of a larger and more
significant phenomenon.14
4.4 Primary effects
In assessing the effect of banning corporate actors from directly participating in electoral campaigns,
two levels of effectiveness should be assessed: on one hand, one needs to provide evidence that the ban
hasprima facie effectiveness, meaning corporate donations do disappear from the data-set and no clear
substitution effect occurs across funding channels. On the other hand, one needs to move beyond this
superficial level and try to assess whether this level of effectiveness is stable upon deeper scrutiny –
this will be done in the next Session, discussing straw man donations.
As discussed in Section 4.2, municipal election resources come from five main sources: personal do-
13. Self-employed professionals were not affected by the court ruling banning corporate participation, so that one might
hypothesize that the reform did not affected their signaling ability – as discussed in the next section –, nor their behavior.
14. People involved in straw-man donations are – according at least to one story – might be workers with a history
within a firm. It appears unlikely that non-vetted newcomers, though in a position of power in a firm, would be involved
in illegal arrangements of the kind described. Reputation and transaction costs loom very large in this respect. Because of
this, it might be interesting to look into long term or fast-rising employees that disappear from a firm organization without
appearing in another one and assume they stayed on in a role that is not considered by the data-set. However, the risk of
false positive attribution is extremely high and does not appear to be worth pursuing at this time.
13

nations15 , party donations, donations from other candidates, private donations, and corporate dona-
tions.16 Figure 1 presents the trends for the total amount of donations across all municipal races for the
years 2004-2016. From the graphical evidence, the ban appears to be effective in eliminating corpo-
rate donations, with no corporate actor remaining in the data also under other denomination.17 This
evidence also proves well-founded of some of the fears among pundits (Douglas 2015): the ban left the
system lacking resources, with parties and prominent candidates incapable of redistributing resources
– also indirectly coming from corporations. Overall, the policy intervention managed to substantially
shrink the contribution pie. As reported in the Appendix, the number of candidates who received any
form of funding diminished as a result of the reform (See: Figure A.8). This is clearly the case for en-
trenched parties, and particularly so for the Partido dos Trabalhadores (PT) – the party at the center of
the scandal, that more than any other suffered from the negative shock, and for whom the number of
candidates taking part in races plummeted (Figure A.7)).18
From a global perspective, the ban does not appear to induce any substitution across supporting chan-
nels: lost corporate donations do not appear to be funneled to candidates through other (legal) channels
at this level of analysis. This also seems to be true for both of the offices. Subdividing the sample into
population ventiles, it is possible to appreciate how heterogeneous the picture is different for munic-
ipalities in the right tail of the population distribution. Figure 2 reveals that corporate donations play
a disproportionate role only in large and very large municipalities. Moreover, the same Figure reveals
an up-tick in private donations when all the other channels are shrinking or are unaffected. This sug-
gests the presence of marginal substitution within large municipalities. These figures demonstrate the
15. These are formally registered as such donations only from the 2008 campaign onward. These donations can be
detected – with a certain degree of error – also for previous years by constructing a vocabulary of donors’ names.
16. In Brazil campaign PACs or similar organizations do not exist, and the law explicitly prohibits contributions from
peak associations (Samuels 2001a), as well as from any other body receiving public funding. Through-party donations were
also a feature of the Brazilian campaign context, these donations were produced by individuals to parties and were legally
anonymized. The piece of legislation presently considered as intervention, also eliminated such aspect of the ban
17. The law is mechanically effective otherwise, because after the elections firm cannot donate and are formally bound
not to appear in such graph.
18. This response can be explained in two ways: either PT candidates, at the time of the election, were under investiga-
tion/arrest and the party was not in the position to have them running nor had the time to substitute them; or previously
PT-affiliated candidates fled the party to avoid a personal negative shock. This last hypothesis is in line with the known
fluidity of party affiliation in Brazil, though it is not presently discussed.
14

Image
Table 1: Descriptive statistics - Campaign contributions, Mayor
necessity to distinguish dynamics specific to large municipalities, those attracting more candidates,
parties, and resources.19 .
19. Figure A.1 and A.2 also suggest that the results, and especially those in the last decile of the population might be
driven by outliers such as the municipality of São Paulo, and other large cities present in the South-East region of Brazil,
which is hardly representative of the rest of the country in terms of population, wealth, and political relevance (in the top
ventile, municipalities from the South-East amounts to 49.8% of entries, despite only accounting for 29.9% of the total)
Robustness checks and subgroup analyses will be devised to account for this.
15

Image
Figure 1: Total campaign donations, by source and office
16

Image
Table 2: Mayor - Total campaign donations, by source (population ven-
tile)
It is important to notice that whatever up-tick is registered at the margin, is very limited in the great
scheme of things. As mentioned in the Section, the ban imposed a contraction of resources available
to candidates running in municipal elections.20 At this level of analysis only visual evidence can be
provided to describe the ban’s effectiveness in eliminating corporate donations, as well as in prevent-
ing substantial substitution effects across channels. It is however not possible to estimate a credible
treatment effect: any of the non-corporate channels is equally subjected to the policy treatment and
one should refrain from carrying out the analysis at this level. No control group really exists for pri-
vate donations among the ones already presented – not even self-donations that might appear to be
generally unaffected by the policy. A more refined solution to address the re-channeling of donations
is then required; to move beyond a face-value assessment of policy effectiveness.
20. The resource shrinking should not be underestimated. It might have substantially changed the set of incentives dif-
ferent donors were facing in entering the political game, as well as the modalities according to which they eventually par-
ticipated. This will be discussed at length later, especially with reference to the change in the amount of resources needed
to produce through donations, a detectable support signal to an individual politician.
17

5 Measurement
5.1 Operationalization
In this section, develop a novel approach to proxy straw-man donation and devise a methodology to
identify their emergence as a locus of corruption displacement. In discussing the effect of the 2015
campaign finance reform, the evidence provided in Section 4.4 only considers corporations as able to
participate directly and openly in the political game. This overlooks corporations’ ability to take part
also indirectly in the game, as mobilizing agents: creating the conditions for third parties to partici-
pate, or altering the ways in which they participate, corporations can achieve their original campaign
finance goals having other to take action on their behalf. Presently, I want to explore specific subset
of this indirect form of participation and for which the mobilizing dynamic happens within the firm
itself and by means of donations orchestration. In the US context, such dynamics has been best prac-
tically discussed with respect to the Thornton Law scandal uncovered by the Boston Globe’s Spotlight
Team and Center for responsive politics (Andrea Estes and Novak 2017). In 2017, a Massachusetts law
firm was discovered having its employees’ private donations to prominent politicians timely matched
to the dollar21 by the firm itself in bonuses. The multi-million dollar over-billing scheme remains the
largest straw donor scandal in the country to date. The scheme at its minimum, involves a firm passing
funds down to its employees with the sole scope of using them as figurehead private donors. A firm
might want to explore this solution if, though interested in participating in the political arena, finds
itself legally unable to do so, or anticipates excessive costs to directly participate – for example, during
a political scandal, when displaying support for a politicians might negative affect the business’ im-
age. Brazil presents a clear set of incentives for firms to explore straw-man donation strategies of this
kind, especially when the ban on corporate donations – described in Section 3.2 – was imposed. The
concerns raised among pundits further justify this exploration: to uncover a straw donor scheme, the
legal literature has recently hinted at the possible use of automated analysis of campaign finance dis-
21. This was true for a substantial part of the donations, not for the universe (Adrea Estes and Novak 2017).
18

closure reports (Kelner 2018), on top of routine review by law enforcement agents. Data mining offers
significant advantages, especially when coupled with other red-flagging systems, in terms of speed of
detection and scalability of efforts. In its basic form it rests on then identification of an ‘initiator’ and
explores to what extent in her influence network a number of individuals – or ‘conduits’ – are display-
ing unusual contribution patterns: donating disproportionate amounts of money, given their contri-
bution history or their job position. In extremely naive cases, conduits can be identified as by checking
whether they are maxing out their contribution capacity or even hyper-coordinate donations in space
and time22 In the context presented, it appears reasonable to focus on a firm’s body of employees as its
influence network, and study alterations in their donation pattern of donations produced within each
firm. To do so, one needs to trace each individual donations to the corporate environment in which
it was ‘produced’; namely each donor needs to be matched with the firm she was working for at the
time of her participation in the campaign. To do so, I match the TSE private donations data-set and the
RAIS data-set using employees’ individual tax identification numbers. Given the coverage of the RAIS,
to minimize false positives, I consider only contracts active between the beginning of September and
the end of October - the window in which citizens can contribute to a local level electoral campaign.
In case of multiple contracts, moreover, i keep the contract reporting the highest salary. Robustness
checks are performed by selecting the contract reporting the longest hours, or produced by picking it
at random23 . On average, I am able to trace donations back to a specific firm 32.8% of the time across
years – with the share of matching improving with the quality of the labor market dataset. As men-
tioned before, individuals unmatched are likely owners, self-employed or people working outside of
the formal market – specific consideration will be given to these later in the study.
Once private donations are linked to their corporate setting, it is possible for each firm to be labeled
as having a ‘corporate’, ‘private’, or ‘mixed’ profile - according to whether historically the firm partici-
pated in the political contribution marked only directly, only indirectly, or using both channels. In my
analyses, corporations with a ‘corporate’ profile are those firms that throughout the available election
22. The investigation of Dinesh D’Souza – another famous example in the US context – appears to have used some of
these flags (FBI 2014).
23. Available upon request
19

2004 2008 2012 2016
No. private donations 68,958 247,251 240,190 310,233
No. private donations matched 20,832 64,072 72,022 139,889
Share private donations matched 30.21% 25.91% 29.98% 45.09%
No. distinct corporation 15,516 50,910 67,139 93,568
Table 3: Mayor - Donations matches (cleaned)
years never had employees who donated; on the other hand corporations with a ‘private’ profile are
those that never donated as a corporation. ‘Mixed’ profiles are those that fall in between these two cat-
egories24 . For the years considered, table 4 illustrates in detail the different donation profiles for firms
in the Brazilian electoral contribution market. Across the election years, on average, slightly more than
a half of the companies (ranging from 40.81% to 57.56%) have ’private’ profile, slightly less than half
has a ‘corporate’ one (ranging from 39.46% to 57.10%) , and the remaining few (ranging from 2.09% to
3.26%) a ‘mixed’ profile. The results are even more consistent if one considers only the 2008 and 2012
elections to attribute a firm to either profile25
The matching of individuals and corporate donations creates room for a heterogeneity in pre-
treatment donation patterns to emerge. In the Section 6, this pre-treatment heterogeneity will be
used to define a treatment and a pseudo-control group to overcome the identification challenge posed
by the universal reach of the reform. The idea is that companies with a ‘private’ profile – so, playing
a strategy that did not involve ‘direct’ corporate donations – should not have been affected by the
corporate donation ban - the only strategy sanctioned by the policy.
24. Refer to Table A.1 in the Appendix for an overlook of different donation profiles – were different options are explored
in terms of definition (consider less years, to be ascribed to a profile, be less categorical with the definitions, etc.).
25. As it has been mentioned elsewhere, the quality of data prior to 2008 is significantly lower, and all the analyses will
be conducted both including and excluding said data as a robustness check.
20

Table 4: Donation profiles II
2012 2008 2004
N 112,330 98,797 39,818

Private 64,660 56,163 16,249
Corporate 44,328 39,409 22,738
Mixed 3,342 3,225 831
Private (%) 57.56 56.85 40.81

Corporate(%) 39.46 39.89 57.10
Mixed (%) 2.98 3.26 2.09
The construction of these profiles helps one understand how donors react to the ban, going beyond
superficial evaluation. Figures 2 and 3 present contributions from employees whose firm has been
attributed to one of the three profiles26 . As expected, donations coming from ‘private’ firms are not
affected by the introduction of the ban. On the other hand, ‘mixed’ and ‘corporate’ profiles display -
respectively - a negative and a positive deviation from their trajectory.
Before focusing on the ‘private’ and ‘mixed’ profiles only, it is important to underline how the ban
clearly induced a contribution strategy change in some companies: private donations from firms with
a history of only ‘corporate’ contributions should have remained at 0, whatever change in the regula-
tion. Instead, the graph shows an up-tick of circa 9000 private donations – amounting to more than
BRL 33m – for the corporate profile. This involvement remains a change at the margin, covering only
1.66% of the count of donations. Nevertheless, it remains among the tangible unforeseen effects of the
ban. In all the analyses that follow, donations coming from ‘corporate’ profile will not be considered.
To study straw donor schemes, one needs to move beyond generic consideration based on volume of
26. In the Appendix, Table A.10 reproduce the analyses though excluding from the analysis the private donations coming
from the public administration. Donations from within the public sector can be dropped from the analyses as they are not
openly responding to the investment dynamics that guides the rest of the firms in the market, even if this remains open to
discussion.
21

Image
Figure 2: Total donations by donation profile
donations: simply looking at donation amounts one cannot disentangle patterns of genuine employ-
ment engagement from an orchestrated one. In this, the overall pattern of donation is crucial, and
should be central to the analysis of the phenomenon. To pinpoint the use of straw man donation, one
must look into the characteristics of a firm’s donations as a whole. In particular, one should try to get
a sense of the extent to which donations are orchestrated in a firm, towards the goal of producing a
support signal for the recipient politician. Not all signals are equal: coordinating a costly signal is the
key – one can argue – to keep intact the spoils system connecting politicians and firm. The following
Section sheds light on how to tackle these issues and effectively measure possible secondary effects of
the reform.
5.2 Measurement
As discussed in the previous Sections, devising a credible methodology that enables a researcher to
distinguish straw-man donations from simple, private ones is crucial to provide evidence of campaign
22

Image
Figure 3: Total donations by donation profile (by population ventile)
regulation fraud and, thus, to evaluate whether campaign finance reforms might have generated cor-
ruption displacement dynamics of this kind. If a firm wants to enter into a profitable do ut des rela-
tion with a politician, it needs to produce a sufficiently strong and clear signal of its political support
through its donations - that is to say, it must make money ‘talk’ (McMenamin 2013). In general, the
size of a firm’s contributions achieve the signaling job just fine (Boas, Hidalgo, and Richardson 2014).
However, changes in the campaign finance regulation may force firms to adopt different strategies to
achieve the same goal under a new set of constraints. Universal ban on corporate donations is at the
extreme of the possible constraints. A straw donor scheme has the potential to achieve exactly that,
under certain conditions. Simply dividing up a large corporate donation into smaller ones to be paid
by employees – granted their willingness to do so – solves the issue of conforming to the regulation,
but not per se the signaling one. To de facto conserve the status-quo, a firm needs to produce a suffi-
23

ciently strong signal, so that whomever receives it is capable of decode its message27 , to re-aggregate
a bunch of sounds into speech. Such signal, coming form a careful orchestration of donations, and
lost in the extreme noise of small donations, can be detected approaching donation data systemat-
ically. At a minimum, donations coordinated to produce a political signal by firm (noise included)
should result more ‘concentrated’ than random donations, across and within candidates. Firms are
likely allocate resources on one candidate among the contenders – the one they would have otherwise
directly financially supported. This allocation would produce more concentrated donations for that
candidate, and across candidates, than random allocation would have.Grasping this pattern is key to
figure out which companies might have to proactively produce a signal, and it is the basis for straw-
man donations here developed. To do so, I adapt the Campaign Contribution Concentration Index
(CCCI) described in Dharmapala and Palda 2002, modeled on the classic Herfindhal-Hirschman Index
(Hirschman 1964). I develop two different indexes to try to gauge the two concentrations dynamics just
explained: a Within Candidate Coordination Index (WCC), and an Across Candidates Coordination
Index (ACC). The measure clearly does not allow one to qualify individual donations as the product of
illegal exchange – as only direct prosecution could –, however it allows one to uncover broad donation
patterns that reveal heterogeneous behaviour across groups affected by the policy, and grasps with no
fail the emergence of orchestration practices in the within-firm marketplace. Though the measure re-
mains an imperfect proxy for actual straw-man donations, it is nonetheless a solid way to start the use
of such an elusive practice. A description of the indexes follows.
Within candidate contributions concentration index (W CCcf ).
n
X Cicf
W CCcf = ( Pn )2 (1)
i=1 j=1 Cjcf
Where, Cicf is a private (physical person’s) contribution made to candidate c by an individual working
for a firm f ; n is the number of donations a candidate c receives from firm f . W CCcf ∈ [0, 1].
27. This is particularly true in contexts where small donations are a staple of electoral campaign, as in the setting (Zovatto
2003).
24

Across candidate contributions concentration index (ACCf ).
m Pn
C
Pn icf
X
ACCf = ( Pm i=1 )2 (2)
c=1 d=1 C
j=1 jdf
Where Cicf is a private (physical person’s) contribution made to a candidate c by an individual work-
ing for firm f ; n is the number of donations candidate c receives from firm f ; m is the number of
candidates to whom employees of the firm f give a contribution. ACCf ∈ (0, 1] In isolation, the
indexes help characterize different results of the financing activities a firms engages into through its
employees, intentionally or not. Despite the attractive straightforwardness of the indexes, some warn-
ings are necessary: (i) index values carry little meaning in absolute terms: high scores can be reached
by simply having few employee-donors, up to degenerate cases of only one donation for whom the
index always takes the value of 128 ; (ii) donations should be sector- and region-specific, as the required
strength of the signal produced via a donation depends on the strength of the signal of all the compet-
ing firms (as signal-producers), and is to be controlled for accordingly; and (iii) it might be useful to
consider the indexes described in combination, especially when trying to identify individual indirect
donations, such as straw-man donations – though how to practically do so remains to be studied and
in not presently explored. Discussing how indexes’ levels and the magnitude of donations affect the
strength that a donation produces it is not straightforward. Tables A.5 to A.6 present an overview of in-
dex density according to the donation portfolio of a firm, focusing on the number of donors. Donations
by straw-man are likely those donations that display a high index despite a large number of donations,
or precisely where donations are numerous and the mechanics of the calculation would otherwise pro-
duce a low index. Focusing on general orchestration dynamics, this aspect is left to a specific inquiry
to come. To use campaign finance concentration indexes to gauge a phenomenon of this kind presents
many risks, however, such a methodology remains at the same time extremely promising in terms of
automating the analysis of campaign donation in terms of pattern search. This could very much be the
starting point of any inquiry into straw donor schemes. Highly concentrated donations produce by
28. Robustness checks are produced including and excluding these degenerate cases.
25

large donor sets can be red-flagged, no matter what the amount of money involved is. This instrument
can be pivotal in a next phase of wrongdoing detection.
6 Empirical strategy
6.1 Identification strategy
In the present section, I analyze how the policy intervention affected within- and across-candidates
concentration index, using a standard causal identification strategy whose application is reasonable
based on the discussion on treated and pseudo-control group carried out previously. These results
reveal insightful with respect to the corruption displacement dynamics triggered by the electoral cam-
paign reform. Relying on the discussion in Section 5, the effect of the corporate donation ban is ex-
plored by assessing how firms’ contribution concentration indexes change, comparing firms with and
without a pre-treatment history of corporate donations. Indexes are generated using private donations
only, as they are the sole possible vector of straw-man donations as presently discussed. Changes in
concentration indexes across groups should reveal how the introduction of a corporate ban changed
corporations’ strategies for political participation – formally, if not substantively. To measure the av-
erage treatment effect, I estimate a Difference-in-Differences model, with lags of the treatment and
Fixed Effects.
0
X
Yist = γs + λt + δj (T reatis ∗ P ostit+j ) + ϵist (3)
j=−3
Where i indexes a firm, s its treatment status, and t an election year. Y is either the WCC index or the
ACC one. T reatis is an indicator taking the value of 1 for treated units; P ostit an indicator taking the
value of 1 in the post-reform period; δj is the coefficient of interest and captures the effect of being a
firm with an affected portfolio in the election year t+j relative to the introduction of the reform. Leads
are accounted for to formally support the parallel trend assumption, as a complement to graphical
evidence. γs indicates State Fixed Effects, and λt time Fixed Effects. among the robustness checks a
26

vector of fixed effects are introduced, to control for both firm, race and municipal characteristics. The
introduction of Fixed Effects is justified by the necessity to control for idiosyncratic trends that might
affect contribution practices especially across an extremely varied economy such as the one of Brazil.
In the estimation, each firm enters with a unique, year-specific, value that accounts for a firm’s dona-
tion structure and move away from possible criticisms regarding the undue influence of degenerate
indexes. Regarding the within-candidate concentration index (WCC), the present analysis considers
two weighting scheme for the multiple candidate- and firm- specific WCCs, where the weights are
respectively determined by the number of donations given in a race to a candidate (WCC 1), or the
amount of donations given in the same race to the same candidate (WCC 2).Some firms, through their
employees donations, appear to donate in multiple races. The two index are therefore further weighted
using a similar scheme (number of donations, or value of donations) across races when the case arises29 .
Regarding across-candidate concentration index (ACC), the present analysis considers three specifica-
tion, each firm- and year-specific. The indexes are calculated as weighted means of race-specific ACCs,
where the weights are determined, respectively, by the number of donations (ACC 1), the number of
candidates (ACC 2) and the overall amount of donations given in a race (ACC 3). The same caveat as
before is valid. Results are substantially robust across specifications. Results and discussion are re-
ported in the Appendix. In the analyses, donations under the level of BRL 100 are dropped because
of noise reduction. Results are robust to the inclusion of these donations, at the only cost of reduced
precision in the estimates. In Table 6.1 and 6.1 are reported descriptive statistics that highlight how
the two groups are not substantially different from each other despite the great unbalance in respective
market-share. A more refined version of the paper will call for the adoption of a matching strategy on
pre-treatment characteristics, especially sector and engagement ability in the donation market. For
the time being, given the exploratory nature of the piece, this is not yet done.
29. Between 80-85% of firms support candidates in one race only, so that the across races weighting is of little impact.
Robustness checks would see the definition of a primary race per firm, based on historic data, to eliminate the necessity of
such a scheme.
27

Image
28
Image
6.2 Graphical evidence
Figures 4a and 4b present the Difference-in-Differences parallel trend graph for the WCC and ACC
indexes, as they appear according to the preferred specification – WCC 1 and ACC 1 –, as discussed in
Section ??. The graphs are reported to visually demonstrate pre-treatment parallel trends assumption.
This can be also checked in the Model 1 of Table 5 and 6, where the leads are included with that specific
goal in mind. In the Tables, the lagged interaction despite being reported as statistically significant are
substantial30 irrelevant, amounting, respectively, to 5% and 3% of a SD, when the main effect amounts
to 16% of a SD – for the WCC – , and to 14% and 11% of a SD, when the main effect amounts to 72% of
a SD – for the ACC. Both Figure 4a and 4b shows how companies with a ‘private’ profile, after the re-
form display substantially lower levels of concentration in their donations, an a drop for post-scandal
election. This is justified with reference to the post-scandal political scenario in Brazil – described in
Section 3. The scenarios was one in which many (relatively) prominent politicians disappeared from
30. Please refer to the discussion on how to interpret statistical significance in extremely large dataset in the next para-
graph.
29

the political arena because of their involvement in the corruption scandal. To make things worse, the
parties more involved in the scandal – PT and PTB – went through mayor party organization disrup-
tion that translated into them struggling31 to have mayors and city council candidates running in the
first place (Costa 2016). Before the scandals, these reference politicians were potentially focal points
for donations by employees from ‘private’ firms. Their presence had kept high the concentration index
of a firm despite the lack of any proactive effort to orchestrate donations. The deep change in the po-
litical scenario, with the (self-)removal of these figures left to distribute their donations evenly across
candidates – with the consequent drop in concentration indexes across and within candidates. On the
other hand, companies with a ‘mixed’ profile, on average, managed to maintain a significantly stable
level on both indexes in the face of a political scenario altered in the way described. Firms were able to
maintain unaltered support signaling abilities after the reform. This ability can be attributed to dona-
tions orchestration – as both kind of companies were facing the same abrupt changes in the political
landscape, and private donations produced within the companies with a ‘mixed’ profile should have
otherwise responded similarly to their counterparts . To what extent this ability can be pinpointed
to new coordinating schemes in the form of straw man donation, remains to be proven. However, the
measure devised provides an interesting insight in the coordination ability of firms in terms of political
signaling.
31. PT candidates across the election shrunk by -47,19%. See: Table A.7
30

Image
(a) WCC
Image
(b) ACC
Figure 4: Concentration Indexes, preferred specification
6.3 Estimation
Moving into a more formal analyses, it is worth remarking that given the number of observations is
very large, the standard errors of estimates become mechanically extremely small. On the one hand,
this means that the model used has power to detect even very small changes, which is useful given the
nature of the present dependent variable; on the other, however, t-statistic numerators are so inflated
that even very small changes are deemed statistically significant even when no practical significance is
there to be detected. This presents a trade-off that must be addressed with a pinch of salt32 : the dis-
cussion on the Difference-in-Differences interaction coefficients is carried out looking primarily at
the magnitude of coefficients, and in the present case in terms of standard deviation. The estimation
is consistent with the graphical analysis of Section 5, and substantially robust to the inclusion of State
32. Their statistical significance – partially void of meaning with respect to more classical analyses – is assessed by means
of considering series of ever smaller random re-samples to see up to what point the estimations remain significant. These
are available upon request.
31

and Sector Fixed Effects. Focusing, on the general interaction therm for the WCC index this, regis-
ters – in the most complete specification – a value of 0.0202*** (0.00730), which in substantive terms,
account for a 16% of a standard deviation; whereas the ACC registers at value of 0.0150*** (0.00221)
and amounting to a 72% of a standard deviation. By design index values are hard to make sense by
themselves and disregarding the overall pictures that they give. Nonetheless, the results appear con-
sistent with the data presented and the anecdotal evidence discussed. In the same specification, it is
worth underline the impact of municipality development (HDI) in accounting for the base line value
of the indexes. This is not surprising as mode developed contexts are also those in which competi-
tion is higher and participation broader. The same idea is conveyed looking at the margin of victory
of the village in which firm are located and towards whose municipal race mainly donates. In these
scenarios, highly concentrated indexes are hard to achieve the first place, despite results are consistent
accounting for them. The general results appear worth of further inquiries.
32

Image
Table 5: Difference-in-Difference - WCC, preferred specification
33

Image
Table 6: Difference-in-Difference - ACC, preferred specification
Overall, the estimation reveals an untold ability by firms already active in the donation market (’mixed’
profiles) to produce private donations in a more coordinated fashion that their counterpart (’private’
profiles). Donations that display such features can be pivotal in a busy political speech arena where hav-
ing one’s voice heard can easily translate in economic returns. The effectiveness of concentrated signal
to generate returns is left to future examinations. The results presented here reveal to a point the inef-
fectiveness of the policy intervention aimed a containing the corporate influence in politics. Through
donation orchestration corporate actors appeared to have managed to maintain to an extent the upper
hand within the campaign donation market with respect other firms. In the sea of private donations
the ones produced by coordination stand out and can be used as non-generic prove of commitment
and support of a firm – not only of its employees – towards a politicians. This might have opened a
channel of communication between the two, leading to exchanges corrupt in nature and scope. This is
done whilst firms de jure complied with the new restrictive regulatory environment, and without con-
sidering the wealth of illegal channels otherwise open to both actors to engage in the same practice.
34

The electoral reform and its corporate donation ban could become even less effective and meaningful.
Donation patterns of this kind cannot be openly uncovered by the public who lack easy access to de-
tailed data, as a result, the practical consequence of the ban is a clear reduction in transparency with
no overt gain in accountability.
7 Discussion and Conclusion
The paper studies straw donor schemes, a strategy corporations use to indirectly contribute to politi-
cal campaigns when banned from directly doing so. Empirically, I explore the issue using the context
of Brazilian local government elections, where a relevant reform was implemented. I conceive straw
donors schemes as a form of corruption displacement and I raise the questions on the ultimate effec-
tiveness of all-encompassing reforms in corporate financing. First, I analyze donation portfolios by
firms matching individual donations records and labor market information to link each donor to a
specific firm. Then, I divided companies in treated and pseudo-control groups leveraging their pre-
treatment heterogeneous donation pattern and solely focusing on private donations – key in identify
the effect of a universal corporate donation ban. Second, I constructed a measure of donations orches-
tration within firms, using and adapted version of the the Herfindahl-Hirschman Index. The measure
is minimal, scalable and exportable. Third, using a Difference-in-Differences identification strategy,
I show how firms with a history of direct corporate donations are able to maintain high donation
coordination, thus signaling ability, in the face of a disrupted political scenario. Given information
asymmetries, the ability to produce clear and decodable signals is not irrelevant, especially on a large
scale and when return to investment is expected. Overall, straw-man donations are a phenomenon
difficult to observe, trace, and account for. They are a cunning answers to abrupt changes in campaign
finance law, emerging in contexts already negatively affected by a dysfunctional relation between cor-
poration and politicians. The study of the phenomenon deserve more attention – across settings –
from academics and practitioners alike.
This piece advocates for a closer consideration of the phenomenon and produce usable instruments
35

and strategies to account for it. The use of data mining, coupled with solid identification strategy, ap-
pears an ideal instrument to better design interventions, to contain corruption displacement, and to
finally question the value of all-encompassing interventions usually ultimately detrimental to trans-
parency and electoral accountability.
References
Aguiar, Roberto. 1994. “The cost of election campaigns in Brazil.” Comparative Political Finance among
the Democracies.
Ansolabehere, Stephen, John M De Figueiredo, and James M Snyder Jr. 2003. “Why is there so little
money in US politics?” Journal of Economic perspectives 17 (1): 105–130.
Ansolabehere, Stephen, Alan Gerber, and James M Snyder Jr. 2001. “Corruption and the growth of
campaign spending.” A Users Guide to Campaign Finance Reform, 25–46.
Arvate, Paulo Roberto, Klenio Barbosa, and Eric Fuzitani. 2013. “Campaign donation and government
contracts in Brazilian states.”
Austin, Reginald, and Maja Tjernstrom. 2003. Funding of political parties and election campaigns. IDEA -
International Institute for Democracy / Electoral Assistance.
Av.Vv., Da Redação. 2016. Caixa 2 pode ganhar força nas eleições 2016. Edited by EXAME. Online, 06
Mar 2016. https://exame.abril.com.br/brasil/caixa-2-pode-ganhar-forca-nas-eleicoes-2016/.
Avellaneda, Claudia N. 2008. “Municipal performance: does mayoral quality matter?” Journal of Public
Administration Research and Theory 19 (2): 285–312.
Avellaneda, Claudia N, and Ricardo Corrêa Gomes. 2017. “Mayoral quality and municipal performance
in Brazilian local governments.” Organizações & Sociedade 24 (83): 555–579.
Avis, Eric, Claudio Ferraz, Frederico Finan, and Carlos Varjão. 2017. Money and politics: The effects of
campaign spending limits on political competition and incumbency advantage. Technical report. Na-
tional Bureau of Economic Research.
Benjamin, Teo, and Felipe Caruso. 2016. Here’s what happened when Brazil banned corporate donations in
elections. Edited by World Economic Forum. Online, 09 Dec 2016. https://www.weforum.org/
agenda / 2016 / 12 / here - s - what - happened - when - brazil - banned - corporate - donations - in -
elections/.
Bertrand, Marianne, Matilde Bombardini, and Francesco Trebbi. 2014. “Is it whom you know or what
you know? An empirical assessment of the lobbying process.” American Economic Review 104 (12):
3885–3920.
Beyer, Janice M. 1981. “Ideologies, values, and decision making in organizations.” Handbook of organi-
zational design 2:166–202.
36

Boas, Taylor C., F. Daniel Hidalgo, and Neal P. Richardson. 2014. “The spoils of victory: campaign
donations and government contracts in Brazil.” The Journal of Politics 76 (2): 415–429.
Borges, Laryssa, and Mattos Marcela. 2016. Eleições 2016: novas regras elevam risco de doações ilegais.
Edited by Abril Veja. Online, 10 Jan 2016. https://veja.abril.com.br/brasil/eleicoes-2016-novas-
regras-elevam-risco-de-doacoes-ilegais/.
Bourdoukan, Adla Y. 2010. “Financiamento público para partidos políticos e campanhas eleitorais no
Brasil e seus efeitos sobre o sistema partidário: História e discussão.” In International Congress of
the Latin American Studies Association, Toronto, Oct, 6–9.
Breno, Pires, and Borges André. 2018. Atuação no Parlamento motiva doação de empresários. Edited by
Universo Online S.A. Online, 14 Sep 2018. https : / / noticias . uol . com . br / politica / eleicoes /
2018 / noticias / agencia - estado / 2018 / 09 / 14 / atuacao - no - parlamento - motiva - doacao - de -
empresarios.htm.
Brollo, Fernanda, Pedro Forquesato, Juan Carlos Gozzi, et al. 2017. To the Victor Belongs the Spoils?
Party Membership and Public Sector Employment in Brazil. Technical report. University of Warwick,
Department of Economics.
Buckley v. Valeo. 1976, No. 75-436.
Butrymowicz, Daniel W. 2009. “Loophole. com: How the FEC’s failure to fully regulate the Internet
undermines campaign finance law.” Colum. L. Rev. 109:1708.
Carrillo, Manuel, Alonso Lujambio, and Carlos Navarro. 2003. Dinero y contienda político-electoral: reto
de la democracia.
Casas-Zamora, Kevin. 2016. The state of political finance regulations in Latin America. International IDEA
Discussion Paper 12/2016.
Castillo Vera, Pilar del, and Daniel Zovatto. 1998. La financiación de la política en Iberoamérica. Instituto
Interamericano de Derechos Humanos.
Che, Yeon-Koo, and Ian L Gale. 1998. “Caps on political lobbying.” The american economic review 88 (3):
643–651.
Costa, Fabiano. 2016. Candidatos a prefeito e vereador pelo PT são quase metade de 2012. Edited by G1 -
Globo. Online, 22 Aug 2016. http://g1.globo.com/politica/eleicoes/2016/noticia/2016/08/
candidatos-prefeito-e-vereador-pelo-pt-sao-menos-da-metade-de-2012.html.
Dharmapala, Dhammika, and Filip Palda. 2002. “Are campaign contributions a form of speech? Evi-
dence from recent US House elections.” Public Choice 112 (1-2): 81–114.
Dix-Carneiro, Rafael, and Brian K. Kovak. 2017. “Trade liberalization and regional dynamics.” Ameri-
can Economic Review 107 (10): 2908–46.
Douglas, Broce. 2015. Brazil bans corporations from political donations amid corruption scandal. Edited by
The Guardian. Online, 18 Sep 2015. https : / / www . theguardian . com / world / 2015 / sep / 18 /
brazilian-supreme-court-bans-corporate-donations-political-candidates-parties.
37

Eleitoral, Tribunal Superior. 2016. Doações de pessoas jurídicas estão proibidas nas Eleições 2016. Edited
by Assessoria de Comunicação TSE. Online, 10 Aug 2016. http://www.tse.jus.br/imprensa/
noticias-tse/2016/Agosto/doacoes-de-pessoas-juridicas-estao-proibidas-nas-eleicoes-2016.
Estes, Adrea, and Viveca Novak. 2017. Boston law firm could be charged on donation. Edited by The Boston
Globe. Online, 02 Mar 2017. https://www.bostonglobe.com/metro/2017/03/02/campaign-
regulators - say - thornton - law - firm - violated - state - law - could - face - criminal - prosecution /
fJYLSQEwS0wL7EAJEqIudM/story.html.
Estes, Andrea, and Viveca Novak. 2017. Campaign regulators say Thornton firm violated state law, federal
grand jury continues probe. Edited by Center for Responsive Politics - Opensecrets.org. Online, 02
Mar 2017. https : / / www . opensecrets . org / news / 2017 / 03 / regulators - say - thornton - firm -
violated-state-law/.
Falguera, Elin, Samuel Jones, and Magnus Ohman. 2014. Funding of political parties and election cam-
paigns: A handbook on political finance. Idea.
FBI. 2014. Dinesh D’Souza Sentenced in Manhattan Federal Court to Five Years of Probation for Campaign
Finance Fraud. Edited by Inside Political Law. Online, 23 Sep 2014. https://www.fbi.gov/contact-
us / field - offices / newyork / news / press - releases / dinesh - dsouza - sentenced - in - manhattan -
federal-court-to-five-years-of-probation-for-campaign-finance-fraud.
Federal, Supremo Tribunal. 2015. AçãO Direta de inconstitucionalidade (ADI) 4650. http://portal.stf.jus.
br/processos/detalhe.asp?incidente=4136819.
Fleischer, David V. 2002. Corruption in Brazil: Defining, Measuring, and Reducing: a Report of the CSIS
Americas Program. Center for Strategic & International studies.
Geografia e Estatística, Instituto Brasileiro de. 2010. Pesquisa de informações básicas municipais.
Gupta, Abhinav, Forrest Briscoe, and Donald C Hambrick. 2017. “Red, blue, and purple firms: Orga-
nizational political ideology and corporate social responsibility.” Strategic Management Journal 38
(5): 1018–1040.
Hasen, Richard L. 1998. “Campaign finance laws and the Rupert Murdoch problem.” Tex. L. Rev.
77:1667.
. 2016. Plutocrats united: Campaign money, the Supreme court, and the distortion of American elec-
tions. Yale University Press.
Hirschman, Albert O. 1964. “The paternity of an index.” The American Economic Review 54 (5): 761–762.
IDEA. 2018. Political Finace Database. https : / / www . idea . int / data - tools / data / political - finance -
database.
Issacharoff, Samuel, and Pamela S Karlan. 1999. “The hydraulics of campaign finance reform.” Texas
Law Review 77:1705–1738.
Josette, Goulart. 2018. Grandes empresários evitam financiar presidenciáveis. Edited by Universo Online
S.A. - Folha de S.Paulo. Online, 14 Sep 2018. https://www1.folha.uol.com.br/poder/2018/09/
grandes-empresarios-evitam-financiar-presidenciaveis.shtml.
38

Kamm, Thomas. 2015. Making sense of Brazil’s Lava Jato scandal. Brunswick São Paulo.
Kelner, Robert. 2018. “Straw Donor” Cases Are In The News Again Today. Edited by Inside Political Law.
Online, 31 May 2018. https://www.insidepoliticallaw.com/2018/05/31/straw-donor-cases-
news-today/.
King, Brayden G. 2008. “A political mediation model of corporate response to social movement ac-
tivism.” Administrative Science Quarterly 53 (3): 395–421.
Klarman, Michael J. 1996. “Majoritarian judicial review: the entrenchment problem.” Geo. LJ 85:491.
Koger, Gregory, and Jennifer Nicoll Victor. 2009. “Polarized agents: Campaign contributions by lob-
byists.” PS: Political Science & Politics 42 (3): 485–488.
Kunicová, Jana. 2006. “Democratic institutions and corruption: incentives and constraints in politics.”
International handbook on the economics of corruption 1:140–160.
Levitt, Steven D. 1995. “Policy watch: Congressional campaign finance reform.” Journal of Economic
Perspectives 9 (1): 183–193.
Mann, Thomas E. 2003. “Linking knowledge and action: Political science and campaign finance re-
form.” Perspectives on Politics 1 (1): 69–83.
Matoso, Filipe. 2015. Dilma sanciona reforma política, mas veta doação de empresa a campanha. Edited by
G1 - Globo. Online, 29 Sep 2015. http://g1.globo.com/politica/noticia/2015/09/dilma-veta-
doacao-de-empresas-campanhas-eleitorais.html.
McMenamin, Iain. 2013. If money talks, what does it say?: corruption and business financing of political
parties. OUP Oxford.
Milligan, Kevin, and Marie Rekkas. 2008. “Campaign spending limits, incumbent spending, and elec-
tion outcomes.” Canadian Journal of Economics/Revue canadienne d’économique 41 (4): 1351–1374.
Mishra, Ajit. 2002. “Hierarchies, incentives and collusion in a model of enforcement.” Journal of Eco-
nomic Behavior & Organization 47 (2): 165–178.
Nickson, R Andrew. 1995. Local Government in Latin America. L. Rienner Publishers Boulder.
Passarinho, Nathalia, and Fernanda Calgaro. 2015. Dilma sanciona reforma política, mas veta doação de
empresa a campanha. Edited by G1 - Globo. Online, 9 Sep 2015. http://g1.globo.com/politica/
noticia/2015/09/camara-aprova-texto-que-restabelece-doacao-de-empresas-partidos.html.
Ramalho, Renan. 2015. Supremo proiíbe doação de empresas para campanhas eleitorais. Edited by G1 -
Globo. Online, 17 Sep 2015. http://g1.globo.com/politica/noticia/2015/09/supremo-decide-
proibir-doacoes-de-empresas-para-campanhas-eleitorais.html.
. 2016. Procurador diz que caixa 2 continuará e defende maior punição a partidos. Edited by EXAME.
Online, 06 Aug 2016. https://exame.abril.com.br/brasil/caixa- 2- pode- ganhar- forca- nas-
eleicoes-2016/.
Rocha, Fabiana, Veronica Ines Fernandez Orellano, and Karina Bugarin. 2018. “Elected in a close race:
Mayor’s characteristics and local public finances.” EconomiA 19 (2): 149–163.
39

Romer, Thomas, and James M Snyder Jr. 1994. “An empirical investigation of the dynamics of PAC
contributions.” American journal of political science 38 (3): 745–769.
Rosenstone, Steven J, and John Hansen. 1993. Mobilization, participation, and democracy in America.
Macmillan Publishing Company.
Samuels, David. 2001a. “Does money matter? Credible commitments and campaign finance in new
democracies: theory and evidence from Brazil.” Comparative Politics, 23–42.
. 2001b. “Incumbents and challengers on a level playing field: assessing the impact of campaign
finance in Brazil.” Journal of Politics 63 (2): 569–584.
. 2001c. “Money, elections, and democracy in Brazil.” Latin American Politics and Society 43 (2):
27–48.
. 2004. “The political logic of decentralization in Brazil.” Decentralization and democracy in Latin
America, 67–93.
Scarrow, Susan E. 2007. “Political finance in comparative perspective.” Annu. Rev. Polit. Sci. 10:193–
210.
Shapiro, Ian. 2009. The state of democratic theory. Princeton University Press.
Smith, Bradley A. 1995. “Faulty assumptions and undemocratic consequences of campaign finance
reform.” Yale LJ 105:1049.
. 2009. Unfree speech: The folly of campaign finance reform. Princeton University Press.
Sotero, Paulo. 2018. Petrobras scandal. Edited by Encyclopædia Britannica. Online, 10 Apr 2018. https:
//www.britannica.com/event/Petrobras-scandal.
Stratmann, Thomas. 2005. “Some talk: Money in politics. A (partial) review of the literature.” In Policy
Challenges and Political Responses, 135–156. Springer.
Stratmann, Thomas, J Francisco, et al. 2006. “Competition policy for elections: Do campaign contri-
bution limits matter?” Public Choice 127 (1-2): 177–206.
Titiunik, Rocio. 2009. “Incumbency advantage in Brazil: Evidence from municipal mayor elections.”
University of California-Berkeley, mimeo.
Wikipedia, The Free Encyclopedia. 2019. Jeitinho. Online, 22 July 2018. https://pt.wikipedia.org/wiki/
Jeitinho_brasileiro.
Wright, John R. 1990. “Contributions, lobbying, and committee voting in the US House of Represen-
tatives.” American Political Science Review 84 (2): 417–438.
Yang, Dean. 2008. “Can enforcement backfire? Crime displacement in the context of customs reform
in the Philippines.” The Review of Economics and Statistics 90 (1): 1–14.
Zovatto, Daniel. 2003. “The legal and practical characteristics of the funding of political parties and
election campaigns in Latin America.” In Handbook on Funding of Political Parties and Election Cam-
paigns, edited by Reginald Austin and Maja Tjernstrom, 95–114. Stockholm, Sweden: IDEA - In-
ternational Institute for Democracy / Electoral Assistance.
40

A Online Appendix (Not for print publication)
A.1 Geographical specification
Image
Figure A.1: Contribution by population
41

Image
Figure A.2: Mayor contributions, by municipality
42
A.2 Donation Profiles
Table A.1: Donation profiles
Option I Option II Option III
N 112,330 186,178 211,111

Private 64,660 103,164 109,769
Corporate 44,328 74,819 90,907
Mixed 3,342 8,195 10,435
Private (%) 57.56 55.41 52.00

Corporate (%) 39.46 40.19 43.06
Mixed (%) 2.98 4.40 4.94
Matches 2016 23,766 30,033 31,357

Matches 2016 (%) 20.70 26.15 27.31
These profiles are used in the identification strategy to track how banning one of these profiles actually
affects some measures of coordination. Profiles are assessed before the introduction of the ban. There
are three different options to attribute a firm to specific profile: Option I - using the firm profile in the
election prior to the ban (in this case 2012); Option II - using a constructed profile that considers two
elections prior to the ban (in this case 2012 and 2008); Option III - using all three elections prior to the
ban (in this case going back to election year 2004)33 . Going back to consider more than one election to
assign a firm to a participation profile, the probability that a firm is assigned a ‘mixed’ profile over one
of the other two increases. This follows by construction, as any firm that displays an inconsistent be-
haviour over time is labelled thus. This has the potential to (marginally) muddle the profile assignment,
for example assigning a ‘mixed’ profile to an otherwise ‘corporate’ player because back in time one of
33. ‘Pure’ profiles were also considered. In that case, a firm was attributed to a specific profile only in the case where it
was absolutely consistent throughout the elections in its pattern of donations. The specification was, however, extremely
demanding on the data, driving the number of companies in all profiles but the ‘private’ one down substantially.
43

its employees donated even a little amount of money34 . However, considering more than one year to
define a firm’s profile has the advantage to increase the sample of firms one is able to assign to groups
in the post-treatment election. This gain is potentially irrelevant: a significant number of firms - given
the time-span between elections (let alone more than two elections) - go through a complete life-cycle
and disappear from the economy35 . However, it might also be the case that changes in political land-
scape over a long period of time might discourage some companies to participate in the political game
whilst attracting others to it, in a cyclical fashion. If this is the case, considering a longer period of time
might be more informative. Table A.1 presents synthetically the company profile options alongside
the number of matches they are capable to produce in the 2016 post-treatment election. Presently,
analyses will be conducted with the Option II specification. Using this specification, a firm having a
‘private’ profile has not engaged in donations as a company (‘direct’ donations) both in 2012 and 2008
elections, a firm having a ‘corporate’ profile has not seen its employees engaged in donations(‘indirect’
donations) in the same elections. All other have experienced a mix of the two at least once, or being
inconsistent in its participation behaviour.
34. This will be mitigated by excluding small donations in the robustness checks
35. According to the SDBS Business Demography Indicators, the 3-year survival rate of Brazilian firms is 60.9% (data for
2015). Data are made available by the OECD and can be accessed at https://stats.oecd.org/Index.aspx?data-
setCode=SDBS_BDI_ISIC4
44

Image
Figure A.3: Total donations by donation profile - No Public Administra-
tion
45

Image
Figure A.4: Total donations by donation profile (by population ventiles)
- No Public Administration
46

Image
Figure A.5: ACC - Hexbin, >100 BRL
47

Image
Figure A.6: ACC - Hexbin, >100 BRL
48

A.3 Party dynamics
Image
Figure A.7: Total candidates per party, by office (Top 5 parties)
49

Image
Figure A.8: Total donations per party, by office (Top 5 parties)
50

Image
Figure A.9: Parties representation frequency
51

A.4 Public sector exclusion
Image
Figure A.10: Total donations by profile, ventiles (no public sector)
A.5 Additional Indexes
52

Image Image
(a) WCC (2) (b) ACC (2)
Image
(c) ACC (3)
Figure A.11: Concentration Indexes, alternative specification
Image
53

SSRN Id4230236

Uploaded by

Copyright:

Available Formats

SSRN Id4230236

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SSRN Id4230236

Uploaded by

Copyright:

Available Formats

Straw donor schemes and corporate political participation

September 27, 2022

Curtailing corporate participation in electoral campaigns can improve political accountability; as

Key words: Money in politics; Campaign Finance Reform; Corruption Displacement

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Considering how corporations revert to straw-man donations in response to a all-encompassing pol-

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

3.1 Brazilian municipalities

Electronic copy available at: https://ssrn.com/abstract=4230236

3.2 Campaign finances

Electronic copy available at: https://ssrn.com/abstract=4230236

3.3 Straw donor schemes

Electronic copy available at: https://ssrn.com/abstract=4230236

4.1 Electoral data

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

4.3 Labor Market Data

Electronic copy available at: https://ssrn.com/abstract=4230236

4.4 Primary effects

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Table 3: Mayor - Donations matches (cleaned)

Electronic copy available at: https://ssrn.com/abstract=4230236

2012 2008 2004

N 112,330 98,797 39,818

Private (%) 57.56 56.85 40.81

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Within candidate contributions concentration index (W CCcf ).

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

6.1 Identification strategy

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

6.2 Graphical evidence

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Figure 4: Concentration Indexes, preferred specification

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

7 Discussion and Conclusion

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236

Electronic copy available at: https://ssrn.com/abstract=4230236