Bologna 2016, The Effect of Informal Employment and Corruption On Income Levels in Brazil

Journal of Comparative Economics 44 (2016) 657–695
Contents lists available at ScienceDirect
Journal of Comparative Economics

journal homepage: www.elsevier.com/locate/jce
The effect of informal employment and corruption on income

levels in Brazil
Jamie Bologna∗
College of Business and Economics, West Virginia University Morgantown, WV 26506-6025, USA
a r t i c l e i n f o a b s t r a c t
Article history: Bologna, Jamie—The effect of informal employment and corruption on income levels in Brazil
Received 21 September 2014
Revised 13 October 2015 This paper exploits a unique dataset on corruption and informal sector employment in 476
Available online 9 December 2015 Brazilian municipalities to estimate whether corruption impacts GDP or income levels once
variation in informal economic activity is taken into account. Overall, I find that higher lev-
JEL Codes: els of corruption and a large informal economy are generally associated with poor economic
D73 outcomes. However, only the size of the informal economy has a statistically significant effect.
O17 This effect is robust to the inclusion of a variety of controls and fixed effects, as well as an
O43 instrumental variable analysis. Further, these effects are large in magnitude. For example, a
Keywords: one standard deviation increase in the share of total employees that are informally employed
Corruption explains a decrease in GDP per-capita of about 18%. Journal of Comparative Economics 44 (3)
Informal economy (2016) 657–695. College of Business and Economics, West Virginia University Morgantown,
Income levels WV 26506-6025, USA.
Growth © 2015 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights
reserved.
1. Introduction
Public corruption, commonly defined as “the misuse of public office for private gain” (Svensson, 2005, p. 20), is considered
to be a major detriment to development, particularly in low-income countries. In a review of the literature, Olken and Pande
(2012) summarize the ways in which corruption has been measured and show that no matter how dramatically the estimates
differ there is a strong negative association between income levels and corruption.1 A major explanation for this finding is that
corruption induces uncertainty, decreasing incentives for investment and thereby reducing growth (Shleifer & Vishny, 1993; Wei,
1997; Campos et al., 1999; Bologna, 2015).
An alternative, less direct, explanation of this finding is that corruption has an impact on the size of the informal sector
(shadow economy) and the size of the informal sector subsequently impacts income levels and growth rates (La Porta and
Shleifer, 2008). Specifically, in lower-income countries, corruption and the informal sector are found to be complements and
therefore a higher level of corruption translates into a larger informal sector (Dreher & Schneider, 2010). Furthermore, it is highly
plausible that the unofficial economy and the official economy are substitutes, rather than complements, in lower income coun-
tries causing the official economy to suffer when the unofficial economy is large (Dreher & Schneider, 2010; Johnson & Kaufmann,
1997). Therefore, it could be the case that we see this strong negative association between low-income countries and high cor-
ruption levels partly because corruption is driving individuals to participate in the informal sector, seemingly lowering income
∗
Corresponding author. Tel.: 724-882-0074.
E-mail address: jbologna@mix.wvu.edu, jlbologna@gmail.com
1
This finding is common in much of the empirical literature (e.g., Li et al., 20 0 0; Mo, 2001; Abed & Davoodi, 2002; Treisman, 2007; Aidt, 2009).
http://dx.doi.org/10.1016/j.jce.2015.12.001
0147-5967/© 2015 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved.
658 J. Bologna / Journal of Comparative Economics 44 (2016) 657–695
10
10
9
9
INFORMAL_INCOME
8
8
7
7
6
6
0 .2 .4 .6 .8 0 .2 .4 .6 .8
PROB_CORRUPT PROB_CORRUPT
Fig. 1. The relationship between corruption, as measured by the share of resources (R$) found to involve corruption, and formal and informal incomes per-
worker.
Source: Formal and informal incomes per-capita (FORMAL_INCOME; INFORMAL_INCOME) from IBGE Census 20 0 0 and the share of resources R$ involving corrup-
tion (PROB_CORRUPT) from Ferraz and Finan (2011).
and GDP per-capita due to its negative impact on the formal economy alone. The goal of this paper is to see if corruption impacts
income levels once variation in informal economic activity is taken into account. To the best of my knowledge, this is the first
paper to explicitly do so.
The negative relationship between corruption and income levels is well-documented, and numerous papers also link corrup-
tion to the size of the shadow economy (Johnson & Kaufmann, 1997; Hindriks et al., 1999; Hibbs & Piculescu, 2005; Dreher et al.,
2009; Dreher & Schneider, 2010). However, empirical studies that incorporate the effect of the informal sector when estimating
the relationship between corruption and economic outcomes at the aggregate level are scarce and there are virtually no studies
that incorporate this effect within a specific country.2 For example, broad surveys of the corruption literature rarely mention the
informal sector, other than the fact that its size may be influenced by corruption (e.g., Dreher & Herzfeld, 2005; Treisman, 2007).
If the informal sector has a significant impact on economic outcomes and if corruption and the informal sector are significantly
correlated, the previous empirical results are potentially biased. Therefore, I estimate the effect corruption has on economic out-
comes, while incorporating the effects of the informal sector. I do this using a unique data set on corruption and informal sector
employment across 476 Brazilian municipalities.
The size of the informal sector in Brazil is extremely large. Both employment and incomes from the informal sector are in-
cluded in each decennial Census, making this an ideal country for this specific type of analysis. In the sample of municipalities
used in this paper, the size of the Brazilian informal sector in 20 0 0, as measured by share of employees without a formal con-
tract, was over 50% of total employees. Furthermore, informal sector employees accounted for almost 40% of total income from
employees of all types in these municipalities. Thus, it is clear that the informal sector is an integral part of the Brazilian economy.
In addition, corruption in Brazil has long been a serious problem. Due to a number of scandals exposing specific politician’s
behavior in the latter half of the 20th century, Brazil is recognized as a country that is suffering from corrupt acts (Geddes & Neto,
1992; da Silva, 1999). In order to combat corruption, in May of 2003, the government of Luiz Inácio Lula da Silva implemented
a program based on random auditing of municipal governments’ expenditures (Ferraz & Finan, 2008; Ferraz & Finan, 2011).
This program randomly selects about one percent of the total number of municipalities with a population less than 50 0,0 0 0
(Controladoria-Geral da União (CGU)). The CGU auditors collect information on all federal funds transferred to the municipal
government, inspects public work construction, and consults the population through community councils on any complaints of
misconduct (CGU). This collected information is then organized into a report and made available to the public.
In a series of papers, Claudio Ferraz and Frederico Finan used these reports to construct measures of corruption in the audited
municipalities (2008; 2011). This paper uses the corruption indicators developed in their 2011 paper, specifically. Ferraz and Finan
(2011) focus on three different types of corruption that could be found in these audit reports: fraud in the public procurement of
goods and services, diversion of funds, and the over invoicing of goods and services. They estimate that corruption in these local
governments alone amounts to be about $550 million per year.
The main corruption variable of interest in this paper is the share of total federal resources (R$) associated with any of these
three types of corruption.3 As can be seen in Fig. 1, this corruption variable is negatively related to average incomes of both formal
2
Dreher and Gassebner, (2013) conduct a cross-sectional analysis and include the size of the informal sector in addition to corruption measures when ex-
amining how corruption impacts entrepreneurship, finding no change in their results. Wiseman, (2015) argues that corruption only impacts entrepreneurship
through its effect on the shadow economy within the United States, but does not include both effects in a single regression.
3
Several other measures of corruption are discussed and used as robustness checks in the results section of this paper.
J. Bologna / Journal of Comparative Economics 44 (2016) 657–695 659
and informal workers.4 This figure also shows that, on average, formal employees have higher incomes than informal employees
and that corruption has more of a negative impact on the informal sector than the formal. Furthermore, as Fig. 3 shows, a higher
level of corruption is associated with an increase in the size of the informal sector, although this positive relationship is relatively
small in magnitude. Taken together, these figures suggest that even if corruption and the informal sector are complementary
in the sense that corruption induces individuals to participate in the informal sector as in Dreher and Schneider, (2010), these
individuals are worse off regardless of being formally or informally employed in the presence of more corruption.
As a preview of the results, this paper finds that while corruption tends to have a negative association with GDP per-capita
and income levels, this relationship is largely statistically insignificant. This is especially true in regressions that include state-
level fixed effects. Corruption is a product of its environment (Svensson, 2005). The negative effect found in regressions that do
not include state-fixed effects may simply be driven by the (unobserved) cultural environment of the area, not because of the
effect of corruption itself. Thus, perhaps there is something more fundamental, specific to these areas, causing corruption to be
higher and economic outcomes to be lower.
This paper does find that the size of the informal sector has a negative and robustly significant relationship with GDP per-
capita and average income from all sources of employment (both formal and informal). This effect is robust to the inclusion of
state-fixed effects and an instrumental variable analysis. This is therefore suggesting that the size of the informal sector may be
more important in determining income levels than corruption. Furthermore, even if informal employment is a relevant alterna-
tive to formal employment and results in positive spillovers into the formal economy, society as a whole is worse off in the face
of a larger informal sector since both GDP per-capita and total incomes are lower.
Theoretical considerations of how corruption and the informal economy may impact economic outcomes are presented in
in Section 2. Section 3 presents a review of the literature measuring corruption and the informal sector. Section 4 describes the
data. Section 5 summarizes the results. Section 6 concludes, discusses potential policy implications, and suggests areas of future
research.
2. Theoretical considerations & methodology
The simple notion that corruption is ‘bad’, in the sense that it acts as a deterrent to development, is widely held. Corruption
not only poses direct costs on society through the act of bribing, it also results in the misallocation of productive resources
(Svensson, 2005). For example, even if corruption ‘greases the wheels’ of an economy by allowing more firms to operate (Dreher
& Gassebner, 2013), it is possible that the firms allowed to operate are the least efficient. A more specific example of this is
when the government is looking to contract the production of a good to a private company, as is often the case in these Brazilian
municipalities, a firm may gain the government’s contract through bribes. However, the firm may pay to gain this contract at
the expense of the quality of their product (Rose-Ackerman, 1996). Fraud in the procurement of goods and services is one of
the major ways local politicians extract resources in Brazil, and much of this fraud appears in the bidding process for contracts
(Ferraz & Finan, 2011). Thus, corruption is likely negatively related to overall income through these various effects.
The informal sector and corruption are likely complementary, in the sense that they are positively related, in Brazil. In addi-
tion, it is likely that the size of the informal sector also has a direct impact on income potentially biasing the relationship between
corruption and income levels if excluded. Thus, the size of the informal sector is an important variable that needs to be controlled
for, though researchers are still undecided as to its effect on incomes.
Some researchers argue that the official economy and the unofficial economy are complementary, and therefore argue that
an increase in the informal sector is not necessarily bad for economic growth (Choi & Thum, 2005). Schneider and Enste (2000)
estimate that about two-thirds of the income earned in the informal sector is immediately spent in the formal economy, therefore
causing there to be positive spillovers. However, others argue that reducing the shadow economy would lead to a significant
increase in the tax base leading to higher quality public services and economic growth (Loayza, 1996). Researchers also argue
that the informal sector distorts competition and prevents more efficient formal firms from gaining market share, consequently
decreasing economic growth (Farrell, 2004).
La Porta and Shleifer (2008) summarize the above views of researchers into three broad categories: (1) a romantic view, (2)
a parasitic view, and (3) a dual view. Each view makes a prediction regarding the effect the informal sector has on economic
outcomes. The romantic view predicts that the informal economy has a positive effect on income levels overall. According
to this view, the informal sector consists of firms and individuals that are extremely similar in productivity (i.e., have similar
characteristics) to the formal sector, however are only held back by governmental policies and practices. Individuals seek refuge
in the informal sector, possibly increasing production beyond what it otherwise would have been. In this case, wages in both
the formal sector and the informal sector would be fairly similar, since individuals in the formal and informal sector are similar
in productivity. However, as shown in Figs. 1 and 2, formal employees earn more income on average.5 In addition, as shown in
Fig. 2 municipalities with larger informal economies have a lower income per-worker in general. Thus, it is unlikely the case that
the romantic view is the reality in Brazil.
The parasitic and dual views both argue that informal firms are far less productive than formal firms. Both of these views are
reflected in the income per-worker estimates provided in Figs. 1 and 2. However, the parasitic view argues further that these
4
Data will be explained in further detail in the data section. Descriptions and summary statistics of variables presented in Fig. 1 through Fig. 3 are presented
in Table 1.
5
This is true even after accounting for income taxes.
10
10
9
9
8
8
7
7
6
6
.2 .4 .6 .8 1 .2 .4 .6 .8 1
SIZE_INFORMAL SIZE_INFORMAL
Fig. 2. The relationship between the size of the informal economy in 20 0 0 and formal and informal incomes per-worker in 20 0 0.
Source: Formal and informal incomes per-capita (FORMAL_INCOME; INFORMAL_INCOME) as well as the size of the informal economy, as measured by
SIZE_INFORMAL from IBGE Census 20 0 0.
1
.8
.6
.4
.2
0 .2 .4 .6 .8
PROB_CORRUPT
Fig. 3. The relationship between corruption, as measured by the share of resources (R$) found to involve corruption, and the size of the informal economy is
20 0 0.
Source: Size of informal economy (SIZE_INFORMAL), as measured by share of employees without a formal contract, from IBGE Census 20 0 0 and the share of
resources R$ involving corruption (PROB_CORRUPT) from Ferraz and Finan (2011).
informal firms actually have negative effects on income levels overall, while the dual view contends that these informal firms are
far too small to have a negative impact on overall income. La Porta and Shleifer (2008) find the most evidence in support of the
dual view when looking at a cross-section of countries.
As can be seen in Fig. 2, the size of the informal sector is negatively related to average incomes of both formal and informal
workers, but has a stronger effect on the latter. The increase in the relative number of informal employees may be driving the
wage down in the informal sector; however, all else equal this should also be increasing or leaving the wages in the formal sector
unchanged, however the opposite is true. This is suggesting that the parasitic view may be the most relevant in Brazil. In the
case of the Brazilian economy these informal, less productive, firms may take up a large enough share of the market to have a
negative impact on the overall economy. Thus, I expect that municipalities with a larger informal sector will have lower income
levels overall.
I estimate how the level of corruption and the size of the informal sector impact GDP per-capita and both formal and informal
income per-worker, while controlling for an additional 18 variables. Therefore, this paper will estimate the following:
yT = α0 + α1CORRU P T _MEASU RE + α2 SIZE _INF ORMAL + θ X + ε , (2.1)
where yT is a measure of the municipal income level, CORRUPT_MEASURE is one of the two measures of corruption described in
detail in the following sections, SIZE_INFORMAL is the relative share of total employees without a formal contract, X is a matrix
of additional controls described in the following sections (see Table 2), and ε is the error term. I expect that both α 1 and α 2 are
less than zero when looking at GDP per-capita and total (both formal and informal) income per-worker. In other words, I expect
that corruption negatively impacts overall income, holding the size of the informal economy constant. Similarly, I expect that
the size of the informal economy negatively impacts overall income, holding corruption constant. I estimate these results with
and without state and time fixed effects. In addition, since there is always the possibility of omitted variables, I supplement the
results with an instrumental variable analysis.6 In general, though, I find that only the size of the informal sector has a robust
negative relationship with both GDP per-capita and total income per-worker.
It is possible that corruption has differing effects depending on the size of the informal sector. This could make its overall
average effect insignificant. More specifically, it is possible that a large informal sector is indicative of a well-established informal
sector. In this case, it may be relatively easy for an individual to move their business underground if desired, making any growth
enhancing, or ‘grease the wheels’, type of corruption unnecessary. Thus, corruption will likely have more of a negative effect in
areas with large informal sectors.
Similarly, a small informal sector may imply that the informal economy is not as well established. This may make it less
appealing for individuals to move their business underground if desired, consequently making individuals more willing to work
with corrupt officials in the formal economy. This makes growth enhancing corruption more likely. Thus, corruption will likely
have less of a negative effect in areas with small informal sectors.
To test these additional hypotheses I re-estimate Eq. (1) with an interaction between CORRUPT_MEASURE and SIZE_INFORMAL
included. In other words I estimate the following:
yT = α0 + α1CORRU P T _MEASU RE + α2 SIZE _INF ORMAL + α3CORRU P T _MEASU RESIZE _INF ORMAL + θ X + ε , (2.2)
where all is defined as above. I expect that α 3 < 0. In other words, I expect that the effect of corruption worsens as the size of
the informal sector increases. I then present the properly calculated marginal effect of corruption at specified levels of informal
sector size.
3. Measuring corruption and the informal sector
As emphasized in Olken and Pande (2012), the ideal way to measure corruption is by direct observation. However, due to
corruption’s secretive nature, this is obviously difficult. Until recently, much of the literature on corruption has focused on cross-
country comparisons between corruption perceptions and economic growth (e.g., Mauro, 1995). Nevertheless, these perception
indices have been shown to be biased at both the macro level (Donchev & Ujhelyi, 2014) and at the micro level (Olken, 2009).
Perceptions of corruption can easily be influenced by factors other than corruption, such as the level of incomes or education in
each country. Furthermore, corruption perceptions are rather difficult to compare across countries. Given this, the use of more
experience based measures of corruption has been emphasized at the aggregate level, the most common of which is an index of
bribe payers (Treisman, 2007; Bologna, 2015).
However, while bribes represent a more direct measure of corruption, these aggregate measures are largely survey based and
could still be biased. Given this, there have been several micro based studies that attempt to measure corruption through direct
observation of bribes (e.g., Mcmillion & Zoido, 2004; Olken & Barron, 2009; Sequeria & Djankov, 2010) or by estimating theft of
public resources (e.g., Reinikka and Svensson, 2005; Olken, 2007). However each of these micro based studies typically focuses
on specific categories of public services such as education (Reinikka & Svensson, 2005) or public infrastructure (Olken, 2007).7
The measure of corruption used in this paper is based on instances of corruption in Brazilian municipalities using audit reports
that are made available to the public. This measure of corruption is unlikely to suffer from the biases of the perception indices,
as it is estimating corruption from actual incidences of corruption. Furthermore, it is a much broader measure of corruption
than bribes or theft of a specific type of government expenditure alone as it includes corruption in all types of government
expenditures, rather than focusing on exclusive categories.
As explained in Ferraz and Finan (2011), Brazil is one of the most decentralized countries in the world. The country is divided
into over 5,500 municipalities, the smallest political and administrative units in the country. These municipalities receive about
$35 billion per year from the federal government to provide public services. Federal transfers are allocated based on the munici-
pality’s population. Importantly, these transfers are the major source of revenue for local governments (Bragga, et al., 2015) and
thus comprise a majority of government action within each municipality.
The mayor and the elected council are relatively free to decide how to spend federal transfers, allowing a great deal of room
for corruption. The three most common ways in which local politicians can engage in corruption are fraud in the procurement
in public services, diversion of funds, and over invoicing public services (Ferraz & Finan, 2011). As an example, Ferraz and Finan
(2011) explain that mayors often divert funds intended for public service projects, such as education, toward the purchases of
personal items for either themselves or for their friends and family.8
Given the discretion allowed in municipal governments, it is not surprising that there is a vast amount of corruption in
Brazilian municipalities. In order to combat this corruption, in May 2003, Luiz Inácio Lula da Silva implemented an anticorruption
6
Instruments described in detail in the results section.
7
In the U.S. specifically, researchers often use corruption convictions as a measure of corruption across the 50 states (e.g., Glaeser & Saks, 2006). However, this
includes acts of corruption in which the individual was actually caught and convicted. In addition, relying on convictions as a measure of corruption requires
there to be no political bias in the judicial system. Research has shown that this requirement fails to hold in U.S. courts (e.g., Gordon, 2009). It is unlikely to hold
in Brazil courts, given Brazil experiences significantly more corruption than the U.S.
8
See Ferraz and Finan (2011) for more examples of corruption occurring in these municipalities.
program based on random auditing of municipal expenditures. Out of a total of about 5,500 municipalities, 50–60 are chosen to
be audited on a monthly basis (CGU). Municipalities with more than 50 0,0 0 0 inhabitants cannot be chosen, therefore excluding
approximately 8% of the municipalities (Ferraz & Finan, 2011). Furthermore, to ensure a fair and truly random process, individuals
are invited to witness each lottery.
Once a municipality is chosen, 10–15 CGU auditors gather information on all federal funds transferred to the municipal gov-
ernment from 2001 onward. These auditors also meet with members of the local community and municipal councils in order to
hear their direct complaints of corruption. The results of each audit are made publicly available through CGU reports and press
releases on the CGU’s website. While the audit program is still in existence today, the dataset used in Ferraz and Finan (2011)
utilizes audits only for the first eleven months of the program, or the first 11 lottery draws. Therefore, although Olken and Pande
(2012) argue that the results of these audit reports need to be interpreted with caution as it represents both actual corruption
and the inability to hide corruption from auditors, it unlikely that this would be a large issue using only the earliest audits, rather
than the most recent ones.
Another potential concern with the use of these reports is the potential bias of the auditors. If the auditors are encouraged to
under-report or over-report, estimates of corruption using audits may be politically biased. Ferraz and Finan (2011) shows this is
unlikely to be the case. In their paper, they find that first term mayors are less corrupt than second term mayors due to reelection
incentives. If auditors are politically biased, or easily swayed, we would expect that this result is driven by first-term mayors
having more of an incentive to bribe auditors for a favorable report. In the data, this would show up as more favorable reports
for first-term mayors during election years. They find no evidence of this effect. In addition, Ferraz and Finan (2011) test if audit
reports are more favorable when the mayor is more closely affiliated with the federal and state governments. Again, they find no
evidence of this effect.9
In addition to lack in evidence, Ferraz and Finan (2011) argue more generally that these auditors are likely unbiased. Auditors
are federal employees that are selected based on a highly competitive process. They are also highly paid.10 Auditors work as a
team with each auditor reporting to a regional supervisor. This makes it difficult to bias the reports without the other auditors,
or the supervisor, knowing. As noted in Ferraz and Finan (2011), there has never been a single incidence of auditors receiving
bribes.11 Given the unlikeliness of political bias in the reports themselves, and the extensive training auditors receive prior to
being sent to municipalities (Ferraz & Finan, 2011), the measurement of corruption using these audit reports is likely consistent
across municipalities.
As mentioned above, corruption is not the whole story when it comes to determining its effect on GDP and income levels.
Although the literature is somewhat mixed as to the exact direction of the relationship, corruption and the informal (shadow)
economy are clearly related. Some argue that since the option of entering into the shadow economy can act as a constraint
on politicians’ behavior, corruption and the informal economy are substitutes for one another (Choi & Thum, 2005; Dreher et
al., 2009). However, others argue that if the option of participating in the informal sector is plausible then corruption may
drive entrepreneurs underground. In this case, corruption is often needed to expand the informal sector and therefore we may
actually see that corruption and the informal sector are complementary (Johnson & Kaufmann, 1997; Hindriks et al., 1999; Hibbs
& Piculescu, 2005).
Dreher and Schneider (2010) argue that the relationship between corruption and the informal sector depends on the level
of income or development of the specific country in question. Using the World Bank’s definition of income classes, Dreher and
Schneider (2010) break their sample up into two groups: low income and high income. Specifically, countries are classified as
low income if their 2004 GNI per-capita is less than US $3,255. Using this classification, Dreher and Schneider (2010) find that
corruption and the informal sector are complements for countries in the low income group and substitutes for countries in the
high income group.
Given that Brazil’s 2004 GNI per-capita is approximately US $3,310 (World Bank), Brazil is a country that just missed the low
income cutoff point in the Dreher and Schneider (2010) analysis. From 20 01–20 03 the GNI per-capita in Brazil was actually less
than the low income cutoff point of US $3,255. Since the corruption data is measuring the level of corruption from 20 01–20 03,
it is likely that Brazil falls into the low income group where corruption and the informal sector are complementary. Thus, in the
case of Brazil it is likely that an increase in corruption corresponds to an increase in the size of the informal sector.
Since the informal sector is likely to have its own impact on development (see, e.g., Schneider & Estne, 20 0 0; La Porta &
Shleifer, 2008), it is important to incorporate the informal sector into the analysis as well when looking at corruption-income
level relationships. This leads to the question of how to properly measure the size of the informal sector. As noted in Schneider
and Estne (20 0 0), Schneider (20 05), and Schneider (20 07), a precise definition of the shadow, or informal, economy is still lacking.
The informal sector includes both legal activities, such as unreported income related to the production of legal goods and services,
as well as illegal activities, such as trade in stolen goods. A common working definition is to include all market-based production
of legal goods and services that are deliberately concealed from public authorities (see, e.g., Ulyssea, 2010; Schneider & Williams,
2013).
9
Ferraz and Finan (2011) also test to see if auditors favor first term mayors or political allies by auditing less valued projects. They find no evidence of such
effects.
10
Higher wages are commonly associated with less corruption (e.g., Rijckeghem and Weder, 2001). Since auditors are highly paid employees they are less likely
going to be politically biased.
11
This is noted at the time of their study (2011). Any incidence of bribery with auditors past 2011 is irrelevant for the audits used in the Ferraz and Finan (2011)
paper.
Table 1
Summary statistics and brief descriptions of dependent variables and main independent variables.
Variable Description Year Source Obs. Mean Std. dev. Min Max
Dependent variables
GDP_PC_04 GDP per-capita for years 2004. 2004 IPEADATA 476 4819 7951 829 124638
GDP_PC_10 GDP per-capita for years 2010. 2010 IPEADATA 476 5904 9232 1055 129629
Types of income
TOTAL_00 Average yearly income per employees from 20 0 0 Census 476 3497 1308 1311 14843
main job.
FORMAL_00 Average yearly income per employees with 20 0 0 Census 476 4369 1457 1257 16466
formal contracts only.
INFORMAL_00 Average yearly income per employees 20 0 0 Census 476 2454 885 838 8927
without formal contracts only.
TOTAL_10 Average yearly income per employees from 2010 Census 476 4014 1157 2058 12804
main job.
FORMAL_10 Average yearly income per employees with 2010 Census 476 4612 1026 2718 13788
formal contracts only.
INFORMAL_10 Average yearly income per employees 2010 Census 476 2768 766 1363 7593
without formal contracts only.
Main independent variables
PROB_CORRUPT Share of audited resources (R$) found to 2001–04 F&F(2011) 476 0.063 0.102 0.0 0 0 0.794
involve corruption.
PROB_CORRUPT_2 Share of audited items found to involve 2001–04 F&F(2011) 476 0.053 0.052 0.0 0 0 0.323
corruption.
MISMANAGE Proportion of irregularities associated with 2001–04 F&F(2011) 366 1.624 1.624 0.0 0 0 8.0 0 0
mismanagement.
SIZE_INFORMAL Share of total employees without a formal 20 0 0 Census 476 0.639 0.186 0.129 0.977
contract.
Notes: All dependent variables enter regressions in logged form. 79% of municipalities experienced at least one instance of corruption. 99% of munici-
palities experienced at least one instance of mismanagement.
Given the availability of data in Brazil, this paper will use a slightly different, yet simpler, definition of the size of the informal
sector. Each Brazilian Census includes data on all “types” of employees in the production of goods and services in all legal sectors
of the economy– both formal and informal. Consistent with the definition used in the methodology of constructing GDP in
Brazilian municipalities, this paper will define informal employees as the difference between the total number of employees and
employees with a formal contract (Instituto Brasileiro de Geografia e Estatística - IBGE). Therefore, these informal employees
are engaged in legal productive activities that contribute to the calculation of GDP; however, they are mostly non-tax payers.12
Specifically, only nine percent of employees without a formal contract are taxpayers of pension (IBGE). Thus, rather than using
value-added estimates of these employees as the size of the informal sector, this paper will simply use the relative share of total
employment allocated to the informal sector. This gives a simple idea of how large the informal sector is in the Brazilian economy.
4. Data
The data for this paper comes from three different sources: Instituto Brasileiro de Geografia e Estatística (Brazilian Institute
of Geography and Statistics - IBGE), Instituto de Pesquisa Econômica Aplicada (Institute of Applied Economic Research – IPEA),
and Ferraz and Finan (2011). All dependent variables and controls, other than the measure of corruption, come from either IBGE
or IPEA. The corruption measures come from Ferraz and Finan (2011). Brief descriptions and summary statistics for all variables
are presented in Tables 1 and 2.
4.1. Dependent variables
The first dependent variable of interest is simply GDP per-capita for each municipality put into R$20 0 0 dollars by the IPEA.
Municipal GDP is calculated by the IBGE and made available by IPEADATA. This data is given annually from 20 0 0 until 2010.
Thus, since the corruption incidences that form the data of this paper take place mainly in 20 01–20 03, it makes sense to look at
subsequent GDP per-capita levels in 2004. Furthermore, as corruption is often argued to have differing short-term and long-term
relationships checking robustness with the most recent year available (2010) is also necessary (Aidt, 2009).
Looking at the relationships between GDP per-capita, corruption, and the informal sector will give some indication of how
corruption and the informal sector influence official estimates of GDP. The estimate of GDP in each municipality explicitly in-
cludes value added in the construction industry from employees employed informally, accounting for approximately 70% of
12
This definition excludes the production of illegal, yet potentially productive, goods and services. To try to capture these activities I use an alternative measure
of informal sector size that includes self-employed workers (SIZE_INFORMAL_1). Self-employed individuals face significantly less oversight and are more likely
going to participate in informal activity (Wiseman, 2013). This informal activity involves the production of legal or illegal goods and services. A brief descrip-
tion of this alternative variable, and results using this alternative variable are shown in Appendix A Table A1 and Table A4 panel a, respectively. This variable
remains negatively related to total income, however, it is not significantly related to GDP per-capita. This is likely because the self-employed workers do typically
contribute to GDP officially, offsetting the negative effect of any official activity they engage in. Hence the main results are focused on employees that explicitly
identify as being informal.
Table 2
Summary statistics and brief descriptions of remaining independent variables and instrumental variables.
Variable Description Year Source Obs. Mean Std. dev. Min Max
POPULATION Total population, enters regressions in 20 0 0 Census 476 25412 40598 1270 449476
logged form.
URBAN Share of urban population. 20 0 0 Census 476 0.601 0.231 0.119 1.0 0 0
DENSITY People per square mile, enters regressions in 20 0 0 Census 476 320 2072 0.662 33,072
logged form.
TEEN Share of population between ages 10–19. 20 0 0 Census 476 0.223 0.027 0.144 0.284
WORK Share of population between ages 20–59. 20 0 0 Census 476 0.473 0.052 0.336 0.579
ELDER Share of population age 60 or over. 20 0 0 Census 476 0.088 0.026 0.031 0.214
SCHOOL Average # of years of schooling population 20 0 0 Census 476 4.332 1.188 1.178 8.504
10 years or older.
AGRICULTURE Share of employment in agricultural sector. 20 0 0 Census 476 0.398 0.208 0.0 0 0 0.899
CONSTRUCTION Share of employment in construction sector. 20 0 0 Census 475 0.055 0.031 0.003 0.284
MANUFACTURING Share of employment in manufacturing 20 0 0 Census 472 0.080 0.076 0.002 0.507
sector.
TRADE Share of employment in trade sector. 20 0 0 Census 475 0.091 0.045 0.009 0.27
TRANSPORT Share of employment in transportation 20 0 0 Census 475 0.029 0.015 0.001 0.086
sector.
SERVICES Share of employment in services sector. 20 0 0 Census 434 0.015 0.019 0.0 0 0 0.227
PROVISION Share of employment in the provision of 20 0 0 Census 476 0.134 0.066 0.008 0.441
services sector.
SOCIAL Share of employment in the social services 20 0 0 Census 475 0.077 0.030 0.030 0.203
sector.
GOVERNMENT Share of employment in the government 20 0 0 Census 476 0.054 0.035 0.035 0.278
sector.
OTHER Share of employment in other activities 20 0 0 Census 476 0.058 0.028 0.028 0.182
sector.
DISTANCE_SP (Log) min. cost of transport to São Paulo 1995 IPEADATA 476 1913 1311 68.063 8582
Municipal Headquarters index
Instrumental Variables
RAIN Quarterly averages of rainfall of summer 1961–90 IPEADATA 433 119.652 41.491 36.723 240.050
months Dec.–Feb. (mm).
TEMP Quarterly averages of temperature of 1961–90 IPEADATA 433 24.983 1.913 17.330 28.590
summer months Dec.–Feb. (°C).
employment in that sector (IBGE). Thus, indicating that the IBGE is attempting to incorporate the informal sector into its official
estimates. However, as mentioned in Pederson (2003), while official estimates of GDP are able to capture a portion of the infor-
mal sector, they are likely going to under-estimate “total” output if the size of the informal sector is significant. Since, on average,
over 60% of employees fit the definition of informal as defined in this paper and in the GDP methodology, the GDP estimate is
likely biased.
Therefore, these results are supplemented with average total income per-worker estimates that come from the IBGE Census in
20 0 0 and in 2010. Total income is defined as the average yearly income employees, of any type, receive from their main job. Thus,
this is a per-worker measure, rather than a strict per-capita measure. Furthermore, both the Census in 20 0 0 and 2010 split this
average income into several different types of employees. The incomes of the two types of employees that this paper is focused
on are employees hired with a formal contract and those hired without a formal contract. The former are referred to as formal
employees, and the latter are referred to as informal employees. As can be seen in Table 1, in both 20 0 0 and 2010, employees
hired with a formal contract make almost double what those employees hired without a formal contract make. This paper
will first estimate the relationship between corruption, the informal sector, and incomes in 20 0 0, and then supplement these
results with incomes in 2010. The latter estimates, as well as a supplemental IV analysis, will allow more causal arguments to be
made.
4.2. Independent variables
The corruption variables come from Ferraz and Finan (2011). Given that not all municipalities receive the same amount of
federal funds, as these transfers are based on population brackets, the number of corrupt instances or value of corrupt instances
found in these reports may not be a comparable measure of corruption across municipalities.13 Areas that receive a larger
amount of funds likely have a larger number of corruption instances. However, this doesn’t necessarily mean they have more
corruption. Therefore, Ferraz and Finan (2011) develop two measures to get around this.14 The first, and the main corruption
13
I present results in Appendix A Table A2 where I regress economic outcomes on the raw number of instances of corruption finding a negative, though
statistically insignificant relationship. I also look at the number of corrupt instances per-capita and the value of corruption per-capita find a negative, mostly
insignificant relationship. It does appear that the value of corruption matters more so than the frequency. These findings fit with the main results.
14
All measures developed by Ferraz and Finan (2011) and employed in this paper examine the actual incidence of corruption rather than the number of
individuals involved. This corruption is reflective of both government corruption, as well as the willingness of private individuals to engage in corruption with
these officials. This does not involve, however, the illegal acts of private individuals alone. By the nature of this study, I am only interested in acts of misconduct
that involve government officials, as I am focusing on public, or political, corruption.
variable used in this paper, is defined as the share of audited resources (R$) found to involve corruption (PROB_CORRUPT). This
gives an idea of the proportion of federal resources that are involved in corrupt acts. However, since a dollar value was not
available for every single instance of corrupt acts, Ferraz and Finan (2011) also use the share of audited items found to involve
corruption (PROB_CORRUPT_2) as an alternative measure of corruption. This second measure gives an idea of the frequency of
corrupt transactions, whereas the first measure gives more of an idea of how serious these transactions were, in terms of their
value.
As can be seen in Table 1, there is a great amount of variation across municipalities in their levels of corruption. Beginning with
PROB_CORRUPT, we see that approximately 6% of resources are found to involve corruption on average. However, this variable
has a fairly large standard deviation at 10%. Further, in some municipalities we see that 80% of resources are associated with
corruption, while other municipalities experience no corruption at all. Though, experiencing no corruption is not the norm as
nearly 80% of municipalities experienced at least one instance of corruption.
The summary statistics of PROB_CORRUPT_2 indicate that when corruption does occur, it typically deals with more valuable
resources. The frequency of corruption also seems to be more consistent across municipalities. On average, approximately 5%
of audited items are associated with corruption, which is fairly similar to PROB_CORRUPT. However, whereas the maximum for
PROB_CORRUPT was 80%, the maximum for PROB_CORRUPT_2 is only 32%. In addition, PROB_CORRUPT_2 is far less variable with
a standard deviation of only 5%. Thus, information revealed in PROB_CORRUPT is slightly different than information revealed in
PROB_CORRUPT_2.
In addition to these main corruption variables, Ferraz and Finan (2011) develop an alternative measure of corruption associ-
ated with less blatant acts of corruption and more broadly inclusive of acts of mismanagement in general. They argue that these
types of violations are less visible, yet may still be associated with acts of corruption. Essentially, any violation of a standard
rule or procedure is recorded as an act of mismanagement. An example Ferraz and Finan (2011) give is in the procurement of
public goods, at least three firms have to bid for the contract. When less than three firms bid, yet the good is still provided, this is
coded as an act of mismanagement. To get a relative measure of mismanagement across municipalities, Ferraz and Finan (2011)
divide the total number of violations by the total number of service items audited (MISMANAGE).15 As shown in Table 1, acts of
mismanagement seem to be much more common than acts of corruption. In fact, 99% of municipalities experience at least one
act of mismanagement.
MISMANAGE will be included in the regressions as a robustness check. If politicians do attempt to hide their blatant acts of
corruption from the auditors, these acts are more likely to be uncovered using this variable. In addition, since not all munici-
palities in the sample were selected in the same lottery, lottery fixed effects are included in the regressions to control for time
effects.
As mentioned in the previous section the size of the informal sector will be defined as the difference between the total
number of employees and the number of employees with a formal contract.16 This amount is then divided by the total num-
ber of employees so that we are measuring the relative size of the informal sector in each municipality (SIZE_INFORMAL).
This definition is used because it is consistent with the construction of the GDP in each municipality (IBGE). Thus, this part
of the informal sector is at least somewhat included in the official GDP estimates. As shown in Table 1, the informal employ-
ees account for over 60% of total employees on average. In addition, Table 1 shows that every municipality has some degree of
informality.
Recall, the corruption variables represent corruption for the years 20 01–20 03. Therefore, given that the Census is available
for years 20 0 0 and 2010, this paper uses the size of the informal sector in 20 0 0 as the control variable of interest. Thus, making
SIZE_INFORMAL a predetermined variable in this model. Similarly, this paper includes an additional 18 variables to control for a
variety of characteristics at or prior to the year of 20 0 0 that are common in the literature (see, e.g., Higgins et al., 2006; Resende,
2013; Cravo and Resende, 2013). These variables include population and demographic controls (e.g., Resende, 2013), average
years of schooling (e.g., Cravo and Resende, 2013), and industrial employment structure (e.g., Higgins et al., 2006). An additional
control is included to measure the cost of transport from each municipal headquarters to the city of São Paulo (Resende, 2013).
This variable is included as São Paulo is essentially the center of economic activity in Brazil, and therefore is what most areas of
Brazil depend on for development. All variable names, brief descriptions, and sources of these additional controls are given in
Table 2.
5. Results
For brevity, results are presented only for the main variables of interest (PROB_CORRUPT, PROB_CORRUPT_2, MISMANAGE, and
SIZE_INFORMAL). Full results for the main estimations are available in Appendix B. I first discuss the results using Eq. (2.1) as the
equation of interest. These are the main results. I check these results for robustness through the inclusion of state and lottery
fixed effects and through an instrumental variable analysis. I then discuss results using Eq. (2.2) as the equation of interest. This
equation tests if the effect corruption has on income levels depends on the size of the informal sector.
15
Ferraz and Finan (2011) calculate this additional variable for only a subset of audited municipalities, dropping the total number of observations down to 331.
16
This definition of the informal sector includes military and public service employees, accounting for only about 17% of informal employment as defined in
this paper (IBGE). Results do not change when excluding military and public service employees from the definition. For brevity, these results are not presented
in this paper, but are available upon request.
Table 3
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have
on GDP per-capita in 2004 and income per-capita from different sources in 20 0 0.
Types of income in 20 0 0 Types of income in 20 0 0
GDP_PC_04 TOTAL FORMAL INFORMAL GDP_PC_04 TOTAL FORMAL INFORMAL
(1) (2) (3) (4) (5) (6) (7) (8)
PROB_CORRUPT −0.453∗∗ −0.050 −0.053 −0.191∗ −0.353 0.012 −0.029 −0.123

(0.227) (0.080) (0.088) (0.106) (0.248) (0.075) (0.084) (0.113)
SIZE_INFORMAL −0.948∗∗∗ −0.399∗∗∗ 0.225∗∗ 0.002 −0.719∗∗∗ −0.441∗∗∗ 0.281∗∗ −0.144
(0.208) (0.078) (0.108) (0.105) (0.268) (0.081) (0.111) (0.124)
Observations 434 434 434 434 434 434 434 434
R-squared 0.634 0.806 0.503 0.626 0.679 0.852 0.623 0.702
State fixed effects? No No No No Yes Yes Yes Yes
Lottery fixed effects? No No No No Yes Yes Yes Yes
PROB_CORRUPT −0.460∗ −0.105 −0.100 −0.224∗ −0.475 0.015 −0.072 −0.062
(0.265) (0.098) (0.108) (0.130) (0.312) (0.093) (0.107) (0.133)
MISMANAGE 0.072 0.001 0.002 −0.006 0.046 0.002 0.002 −0.004
(0.045) (0.008) (0.012) (0.010) (0.050) (0.010) (0.013) (0.012)
SIZE_INFORMAL −0.890∗∗∗ −0.444∗∗∗ 0.170 0.002 −0.484 −0.448∗∗∗ 0.304∗∗ −0.155
(0.240) (0.099) (0.131) (0.135) (0.294) (0.105) (0.131) (0.161)
Observations 331 331 331 331 331 331 331 331
R-squared 0.656 0.798 0.504 0.616 0.701 0.847 0.628 0.700
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically significant at 5% level, ∗
indicates statistically
significant at 10% level. Standard errors are in parentheses. Constant included in all regressions.
Table 4
The effect corruption, as measured by the share of items found to involve corruption, and the informal economy have on GDP
per-capita and income per-capita from different sources in 20 0 0.
Types of income in 20 0 0 Types of income in 20 0 0
(1) (2) (3) (4) (5) (6) (7) (8)
PROB_CORRUPT_2 −0.044 −0.206 −0.246 −0.213 −0.043 −0.104 −0.226 −0.022

(0.477) (0.177) (0.226) (0.239) (0.497) (0.204) (0.237) (0.264)
SIZE_INFORMAL −0.925∗∗∗ −0.399∗∗∗ 0.225∗∗ 0.009 −0.706∗∗∗ −0.444∗∗∗ 0.276∗∗ −0.140
(0.205) (0.078) (0.108) (0.104) (0.268) (0.080) (0.111) (0.124)
Observations 434 434 434 434 434 434 434 434
R-squared 0.630 0.806 0.505 0.623 0.677 0.852 0.624 0.700
PROB_CORRUPT_2 −0.189 −0.262 −0.348 −0.007 0.299 −0.023 −0.240 0.217
(0.701) (0.259) (0.324) (0.339) (0.729) (0.260) (0.333) (0.326)
MISMANAGE 0.072 0.005 0.007 −0.008 0.036 0.003 0.005 −0.008
(0.050) (0.009) (0.013) (0.011) (0.054) (0.009) (0.014) (0.012)
SIZE_INFORMAL −0.869∗∗∗ −0.447∗∗∗ 0.165 0.016 −0.449 −0.449∗∗∗ 0.296∗∗ −0.141
(0.237) (0.099) (0.133) (0.137) (0.290) (0.106) (0.133) (0.163)
Observations 331 331 331 331 331 331 331 331
R-squared 0.653 0.799 0.506 0.612 0.698 0.847 0.628 0.701
5.1. Main results
Tables 3 and 4 present the results for the first period of interest – the effect corruption and the informal sector have on GDP
per-capita in 2004 as well as the relationship between corruption, the informal sector, and incomes in 20 0 0. Specifications (1)
through (4) include results without any fixed effects, while specifications (5) through (8) include state-level fixed effects and
lottery number fixed effects.
As can be seen in these two tables, while the effect of corruption, as measured by either PROB_CORRUPT or PROB_CORRUPT_2,
is negative in a majority of the specifications, it is largely statistically insignificant. Only in the case where corruption is mea-
sured by PROB_CORRUPT does corruption seem to have a negative and statistically significant association with GDP per-capita,
however, this result is not robust to the inclusion of fixed effects. Furthermore, it appears that the inclusion of less blatant forms
of corruption, MISMANAGE, has very little impact on the results. Furthermore, MISMANAGE is never statistically significant.
However, the size of the informal sector (SIZE_INFORMAL) has a fairly robust, negative and statistically significant relationship
with both GDP per-capita and total income levels. This is especially interesting since SIZE_INFORMAL also seems to have a positive
Table 5
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have
on GDP per-capita and income per-capita from different sources in 2010.
Types of income in 2010 Types of income in 2010
(1) (2) (3) (4) (5) (6) (7) (8)
PROB_CORRUPT −0.499∗∗ 0.006 −0.061 −0.049 −0.311 −0.003 −0.085 0.131

(0.211) (0.054) (0.069) (0.148) (0.214) (0.051) (0.065) (0.103)
SIZE_INFORMAL −0.861∗∗∗ −0.324∗∗∗ −0.025 −0.102 −0.638∗∗ −0.263∗∗∗ 0.096 −0.256∗∗
(0.207) (0.060) (0.069) (0.131) (0.265) (0.065) (0.074) (0.126)
Observations 434 434 434 434 434 434 434 434
R-squared 0.615 0.803 0.522 0.552 0.665 0.854 0.633 0.801
PROB_CORRUPT −0.421∗ −0.017 −0.100 −0.118 −0.303 −0.012 −0.129∗ 0.230
(0.235) (0.063) (0.080) (0.169) (0.266) (0.058) (0.077) (0.141)
MISMANAGE 0.055 −0.001 0.013∗ −0.043∗∗∗ 0.029 0.005 0.024∗∗∗ −0.026∗
(0.035) (0.006) (0.007) (0.014) (0.039) (0.008) (0.008) (0.014)
SIZE_INFORMAL −0.839∗∗∗ −0.379∗∗∗ −0.089 −0.003 −0.441 −0.266∗∗∗ 0.082 −0.190
(0.235) (0.071) (0.083) (0.147) (0.286) (0.080) (0.089) (0.137)
Observations 331 331 331 331 331 331 331 331
R-squared 0.656 0.800 0.512 0.555 0.707 0.855 0.655 0.803
relationship with formal income per-worker. Given its negative relationship with total income, this is suggesting that even if
SIZE_INFORMAL results in positive spillovers to the formal sector, it has a negative overall impact on incomes.
In addition, the effects of SIZE_INFORMAL on GDP per-capita are large in magnitude. Considering only the statistically signif-
icant effects, a one standard deviation increase in the size of the informal sector is associated with a decrease in GDP per-capita
of a maximum of nearly 18% to a minimum of about 13%, depending on the specification. The effects of the informal sector on
total income levels are smaller in magnitude than their effects on GDP per-capita, however the size of the effects are relatively
stable across all specifications. A one standard deviation increase in the size of the informal sector is associated with a decrease
in income per-worker of a maximum of about 8% to a minimum of about 7%.
Magnitudes of these effects are broken down in Tables 7 and 8 based on municipal population in 20 0 0.17 Tables 7 and 8 use
coefficient estimates from specification (1) and specification (2), respectively, in Table 3 to calculate these magnitudes. This gives
the reader an idea of the dollar value of these effects. Looking at the smallest municipality in Table 7 we see that a standard
deviation increase in SIZE_INFORMAL corresponds to a R$ 692.66 decrease in GDP per-capita. This is nearly equivalent to the
entire amount of GDP per-capita in certain municipalities (see Table 1). GDP per-capita tends to be higher in larger municipalities,
making these effects larger, however the correlation is relatively weak (0.178). In comparison, the correlation between income
per-worker and population is stronger (0.374) making the effects of both corruption and the size of the informal sector on income
per-worker more significant in large municipalities. Thus, changes to corruption and the informal sector in these larger areas will
have more meaningful effects.
The negative relationship between the size of the informal sector and incomes, however, could be picking up the fact that
as GDP falls, informal employment acts as a safety net in the short run even if its effects are negative in the long run therefore
causing the share of informal employment to rise (Loayza & Rigolini, 2011). However, since we are looking at the effect the size
of the informal sector in 20 0 0 has on GDP per-capita in 2004, it is unlikely that this effect is exclusively picking up the business
cycle effect. However, these results will be checked for robustness looking at the effect the size of the informal sector has on GDP
per-capita after 10 years as well.
A similar story emerges when looking at the results from the second time period of interest, GDP per-capita in 2010 and
income levels in 2010, presented in Tables 5 and 6. The measures of corruption, PROB_CORRUPT and PROB_CORRUPT_2 have little
relationship to GDP per-capita and income per-worker. As before, PROB_CORRUPT has a negative relationship with GDP per-capita
that is not robust to the inclusion of state and lottery fixed effects. The major difference between the first time period and the
time period given in Tables 5 and 6 is that now when state and lottery fixed effects are included, PROB_CORRUPT_2 actually
has a positive association with total incomes. Since PROP_CORRUPT_2 measures the frequency of corrupt transactions, this could
be indicating that a “grease the wheels” situation is present (e.g., Leff, 1964; Huntington, 1968; Heckelman & Powell, 2010).
In this case a higher frequency of corrupt transactions leads to higher income levels. However this effect is only positive and
statistically significant in four of the 16 specifications using PROB_CORRUPT_2 in Table 6. Thus in general, corruption remains
largely insignificant.18
17
Tables also present the dollar value of the effect of corruption, though the effect is statistically insignificant in some of the specifications.
18
It is possible that a combination of the severity of corruption and the frequency of corruption is more important to economic outcomes. If corruption
occurs frequently, but in small magnitudes, it may not have as large of an effect as if it occurred frequently in large magnitudes. I explore this possibility using
Table 6
The effect corruption, as measured by the share of items found to involve corruption, and the informal economy have on GDP
per-capita and income per-capita from different sources in 2010.
Types of income in 2010 Types of income in 2010
(1) (2) (3) (4) (5) (6) (7) (8)
PROB_CORRUPT_2 −0.167 0.124 0.144 −0.191 0.061 0.240∗∗ 0.245∗ 0.332

(0.404) (0.121) (0.128) (0.264) (0.419) (0.117) (0.143) (0.220)
SIZE_INFORMAL −0.837∗∗∗ −0.323∗∗∗ −0.020 −0.101 −0.624∗∗ −0.257∗∗∗ 0.105 −0.252∗∗
(0.204) (0.060) (0.069) (0.131) (0.264) (0.064) (0.073) (0.126)
Observations 434 434 434 434 434 434 434 434
R-squared 0.610 0.803 0.522 0.552 0.664 0.855 0.634 0.801
PROB_CORRUPT_2 −0.102 0.175 0.133 0.125 0.539 0.323∗ 0.170 0.649∗∗
(0.617) (0.182) (0.202) (0.335) (0.615) (0.164) (0.189) (0.317)
MISMANAGE 0.053 −0.004 0.010 −0.046∗∗∗ 0.018 0.0 0 0 0.020∗∗ −0.034∗∗
(0.041) (0.007) (0.008) (0.015) (0.043) (0.008) (0.009) (0.015)
SIZE_INFORMAL −0.817∗∗∗ −0.372∗∗∗ −0.079 0.008 −0.402 −0.250∗∗∗ 0.096 −0.169
(0.232) (0.072) (0.084) (0.149) (0.283) (0.080) (0.089) (0.136)
Observations 331 331 331 331 331 331 331 331
R-squared 0.653 0.801 0.510 0.554 0.706 0.856 0.653 0.804
Table 7
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have on GDP per-capita in 2004
estimated for the three smallest cities, three cities closest to median, and three largest cities.
Smallest municipalities POPULATION y = GDP_PC_04 Effect of std. dev. increase in Effect of std. dev. Increase in
No. of People Value (R$) PROB_CORRUPT SIZE_INFORMAL
4.62% decrease in y (R$) 17.63% decrease in y (R$)
(1) Chapada de Areia, TO 1270 3924.41 181.31 692.66

(2) Viçosa, RN 1521 2005.79 92.63 354.02
(3) Lajaedo Grande, SC 1572 7628.76 352.44 1346.48
Median municipalities y = GDP_PC_04 Effect of std. dev. increase in Effect of std. dev. increase in
PROB_CORRUPT SIZE_INFORMAL
Value (R$) 4.62% decrease in y (R$) 17.63% decrease in y (R$)
(237) Apuí, AM 13864 2681.14 123.87 472.69
(238) Maraial, PE 14017 1807.59 83.511 318.68
(239) Cruz do Espírito Santo, PB 14081 2403.47 111.04 423.73
Largest municipalities y = GDP_PC_04 Effect of std. dev. increase in Effect of std. dev. increase in
(474) Betim, MG 306675 23522.87 1086.76 4151.79
(475) Mauá, SP 363392 7988.66 369.08 1407.60
(476) São João de Meriti, RJ 449476 3762.39 173.82 664
Notes: These estimates are calculated using coefficient estimates in Table 3, specification (1) without MISMANAGE.
Another major difference between the first period and the period presented in Tables 5 and 6 is that MISMANAGE seems to
have more of an impact in the longer run. It seems that MISMANAGE has a statistically significant positive effect on formal income
per-worker and a statistically significant negative effect on informal income per-worker. MISMANAGE still has no statistically
significant relationship to total income per-worker or GDP per-capita in any case.
However, just as in Tables 3 and 4, the size of the informal sector (SIZE_INFORMAL) exhibits a negative and statistically signif-
icant relationship to both GDP per-capita and total income levels, although this effect is slightly smaller in magnitude. The dollar
values of this effect, again broken down based on 20 0 0 municipal population, are presented in Tables 9 and 10 for specification
(1) and (2), respectively, of Table 5.19 Thus, across both time periods, the size of the informal sector has a negative relationship
with both GDP per-capita and total income levels. This is robust to the inclusion of a variety of different corruption measures,
several alternative measures of corruption that combine these two individual variables (Table A1 describes these measures and Table A3 presents results using
these measures, both found in Appendix A). Specifically, I use the weighted average between PROB_CORRUPT and PROB_CORRUPT_2. I explore several different
weighting schemes, where I continually give PROB_CORRUPT more weight as it seems that the value of the corrupt resources is slightly more important than
the frequency of corruption (see Table A2). Note that all regressions in Appendix A include state fixed effects, as these have been shown to be important in
determining corruption’s impact. Corruption is never statistically significant using these alternative measures.
19
Tables also present the dollar value of the effect of corruption, though the effect is statistically insignificant in some of the specifications. In addition, the
marginal effect of corruption in specification (2) is positive, though not statistically different from zero.
Table 8
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have on total income
per-worker in 20 0 0 estimated for the three smallest cities, three cities closest to median, and three largest cities.
Smallest municipalities POPULATION y = TOTAL_00 Effect of std. dev. increase in Effect of std. dev. increase in
Value (R$) PROB_CORRUPT SIZE_INFORMAL
.5% decrease in y (R$) 7.42% decrease in y (R$)

(2) Viçosa, RN 1521 2572.32 13.12 190.87
Median municipalities y = TOTAL_00 Effect of std. dev. increase in Effect of std. dev. increase in
Value (R$) .5% decrease in y (R$) 7.42% decrease in y (R$)
(237) Apuí, AM 13864 3877.56 19.78 287.71
(238) Maraial, PE 14017 2375.28 12.11 176.25
Largest municipalities y = TOTAL_00 Effect of std. dev. increase in Effect of std. dev. increase in
Value (R$) .5% decrease in y (R$) 7.42% decrease in y (R$)
(474) Betim, MG 306675 5008.68 25.54 371.64
(475) Mauá, SP 363392 7040.04 35.90 522.37
(476) São João de Meriti, RJ 449476 5395.2 27.52 400.32
Table 9
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have on GDP per-
capita in 2010 estimated for the three smallest cities, three cities closest to median, and three largest cities.
Smallest municipalities POPULATION y = GDP_PC_10 Effect of std. dev. increase Effect of std. dev. increase in
Value (R$) in PROB_CORRUPT SIZE_INFORMAL
5.10% decrease in y (R$) 16.01% decrease in y (R$)

(2) Viçosa, RN 1521 2482.00 126.58 397.37
Median municipalities y = GDP_PC_10 Effect of std. dev. increase Effect of std. dev. increase in
in PROB_CORRUPT SIZE_INFORMAL
(237) Apuí, AM 13864 3962.53 202.09 634.40
(238) Maraial, PE 14017 3341.77 170.43 535.02
Largest municipalities y = GDP_PC_10 Effect of std. dev. increase Effect of std. dev. increase in
(474) Betim, MG 306675 33394.93 1703.14 5346.53
(475) Mauá, SP 363392 7865.69 401.15 1259
Table 10
The effect corruption, as measured by the share of resources (R$) found to involve corruption, and the informal economy have on total income
per-worker in 2010 estimated for the three smallest cities, three cities closest to median, and three largest cities.
Smallest municipalities POPULATION y = TOTAL_10 Effect of std. dev. increase Effect of std. dev. increase in
Value (R$) in PROB_CORRUPT SIZE_INFORMAL
.06% increase in y (R$) 6.03% decrease in y (R$)

(2) Viçosa, RN 1521 3243.09 1.95 204.31
Median municipalities y = TOTAL_10 Effect of std. dev. increase Effect of std. dev. increase in
Value (R$) .06% increase in y (R$) 6.03% decrease in y (R$)
(237) Apuí, AM 13864 4054.22 2.43 255.40
(238) Maraial, PE 14017 3141.41 1.88 197.91
Largest municipalities y = TOTAL_10 Effect of std. dev. increase Effect of std. dev. increase in
Value (R$) .06% increase in y (R$) 6.03% decrease in y (R$)
(474) Betim, MG 306675 5357.53 3.21 337.52
(475) Mauá, SP 363392 6079.67 3.65 383.02
Table 11
The effect corruption and the informal economy have on GDP per-capita and total income per-capita in 20 0 0 and 2010 IV robustness
checks when only the size of the informal sector is instrumented.
GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10
PROB_CORRUPT −0.415∗∗ 0.037 −0.461∗∗ 0.053

(0.205) (0.093) (0.200) (0.064)
PROB_CORRUPT_2 −0.106 −0.094 −0.304 0.154
(0.503) (0.209) (0.421) (0.135)
∗∗∗
SIZE_INFORMAL −1.543 0.273 −1.554∗∗∗ 0.284 −1.152 ∗∗
−0.014 −1.145∗∗ −0.027
(0.545) (0.248) (0.542) (0.247) (0.519) (0.192) (0.517) (0.191)
Observations 404 404 404 404 404 404 404 404
R-squared 0.639 0.766 0.636 0.764 0.628 0.789 0.625 0.790
J-statistic 0.321 0.394 0.369 0.332 0.173 0.672 0.192 0.755
F-statistic 22.437∗∗∗ 22.437∗∗∗ 22.206∗∗∗ 22.206∗∗∗ 22.437∗∗∗ 22.437∗∗∗ 22.206∗∗∗ 22.206∗∗∗
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically significant at 5% level, ∗ indicates statistically significant at
10% level. Standard errors are in parentheses. SIZE_INFORMAL is instrumented for with RAIN and TEMP. Constant included in all regressions.
Table 12
The effect corruption and the informal economy have
on GDP per-capita in 2004 and 2010 additional IV ro-
bustness checks where both corruption and the size
of the informal sector are instrumented.
GDP_PC_04 GDP_PC_10
PROB_CORRUPT −0.024 0.192

(0.445) (0.511)
SIZE_INFORMAL −1.566∗∗∗ −1.187∗∗
(0.546) (0.522)
Observations 404 404
R-squared 0.636 0.619
J-statistic 0.373 0.240
F-statistic 18.656∗∗∗ 18.656∗∗∗
Notes: ∗∗∗ indicates statistically significant at 1% level,

∗∗
indicates statistically significant at 5% level, ∗ in-
dicates statistically significant at 10% level. Standard
errors are in parentheses. PROB_CORRUPT is instru-
mented for with RESOURCE; SIZE_INFORMAL is instru-
mented for with RAIN and TEMP. Constant included in
all regressions.
as well as state and lottery fixed effects.20 It is possible, however, that the results are still suffering from omitted variable bias
as some unobserved factor may be influencing both the size of the informal sector as well as GDP and income levels. Therefore,
these results will be supplemented with an instrumental variable analysis.
A significant portion of informal employment in general fluctuates seasonally (Hussmanns, 2004). This is likely due to the
types of industries that tend to comprise the informal sector. For example, tourism employment, where the informal sector
flourishes (Cukier & Wall, 1994), fluctuates seasonally with tourists. Agricultural employment, which makes up over a quarter
of informal employment in this dataset, and construction employment are largely informal, with nearly 70% of its employees
working without a formal contract. Both are also both highly dependent on weather patterns. This is in contrast with non-
weather dependent sectors that contain predominantly formal sector employment, such as finance. These examples suggest that
the size of the informal sector is determined ultimately by weather patterns in each municipality. Therefore, it seems reasonable
to expect that the size of the informal sector is going to be dependent upon seasonal rainfall and temperature patterns. I choose
the look at summer months (December–February) specifically, as this is Brazil’s peak tourism period and part of its agricultural
growing season.
IPEADATA provides quarterly measurements of both rainfall and temperature patterns across each municipality averaged
over a thirty year period. This data summarizes the weather patterns that present in the municipalities for each season from
1961–1990. This paper uses the summer season specifically.
It is reasonable to assume that average weather patterns from 1961 to 1990 are otherwise unrelated to GDP per-capita and
total income levels in 20 0 0 onwards. Historical rain and temperature patterns may impact the overall industrial makeup of a
20
Though industrial composition is controlled for, I also check that these results are not driven by one major industry that takes up a major portion of informal
employment. Specifically, as shown in Table 1 the agricultural industry makes up a significant portion of overall employment in Brazil (40%). Importantly, 30%
of this employment is in the informal sector. Therefore, SIZE_INFORMAL could be simply picking up the effect of agriculture alone and not necessarily informal
employment in general. Therefore I reconstruct SIZE_INFORMAL to exclude agricultural employment (Table A1 in Appendix A) and re-estimate the results (Table
A4 panel b in Appendix A) and show that estimates remain negative and statistically significant, yet smaller in magnitude as would be expected.
Table 13
The marginal effect of corruption on economics outcomes conditional on specified levels of informal sector size.
Panel a: The marginal effect of corruption on GDP per-capita in 2004 and total income per-worker in 20 0 0; state and lottery fixed effects not included.
SIZE_INFORMAL d(GDP_PC_04)/d(PROB_CORRUPT) d(TOTAL_00)/d(PROB_CORRUPT)
Value Value meaning dy/dx p-value dy/dx p-value
0.129 Minimum 0.136 0.769 0.339 0.026

0.657 Median −0.525 0.017 −.097 0.208
0.977 Maximum −0.925 0.014 −0.362 0.018
Panel b: The marginal effect of corruption on GDP per-capita in 2010 and total income per-worker in 2010; state and lottery fixed effects not included.
0.129 Minimum −0.270 0.616 0.163 0.137

0.657 Median −0.527 0.009 −0.013 0.813
0.977 Maximum −0.683 0.069 −0.119 0.266
Panel c: The marginal effect of corruption on GDP per-capita in 2004 and total income per-worker in 20 0 0; state and lottery fixed effects included.
0.129 Minimum 0.884 0.064 0.304 0.033

0.657 Median −0.516 0.030 −0.027 0.722
0.977 Maximum −1.365 0.003 −0.228 0.130
Panel d: The marginal effect of corruption on GDP per-capita in 2010 and total income per-worker in 2010; state and lottery fixed effects included.
0.129 Minimum 0.408 0.407 0.092 0.408

0.657 Median −0.406 0.044 −0.016 0.765
0.977 Maximum −0.899 0.024 −0.081 0.438
Notes: Regressions correspond to equation (3.2) in text. Constant included in all regressions. Full results available upon request.
municipality by altering the allocation of formal versus informal sector jobs.21 It is through this effect alone that these historical
weather patterns are able to influence current income levels. Therefore, I argue that historical rain and temperature patterns
(RAIN; TEMP) are valid instruments for SIZE_INFORMAL.
As can be seen in Table 11 in all cases the first stage F-statistics always reject the weak instrument null hypothesis ac-
cording to the Stock and Yogo (2005) critical values and the J-statistics indicate no evidence to reject the exogeneity of the
instruments. Furthermore, the effect of SIZE_INFORMAL on GDP per-capita remains negative and statistically significant in every
specification. However, the negative and statistically significant effect of SIZE_INFORMAL on total income per-worker disappears.
In addition, PROB_CORRUPT still has a negative and statistically significant impact on GDP per-capita in both periods. Though
PROB_CORRUPT_2 remains statistically insignificant in all specifications.
Though corruption is found to have little effect overall, it is important to address potential endogeneity with this variable as
well. If there is an omitted variable driving the relation between corruption and income, the minimal number of negative effects
I do find in this paper may be overstated. Therefore, in Table 12 I re-estimate the effect PROB_CORRUPT has on GDP per-capita
after instrumenting for both the size of the informal sector and corrupt.
As in Table 11, I instrument SIZE_INFORMAL using RAIN and TEMP. To instrument for corruption, I use variables related to the
amount of federal funds each municipality receives. The more funds a municipality receives, the more appealing it may be to
engage in corruption. As Ferraz and Finan (2011) shows, mayors in these municipalities are less likely to engage in corruption
when they face reelection. This shows that these mayors calculate the costs and benefits of corruption before acting. When facing
reelection, the costs of corruption are much higher. Similarly, when the resources available for extraction are more valuable, the
benefits of corruption are higher, making it more likely that mayors will act corruptly.
Importantly, the amount of funding a municipality receives depends exclusively on the municipality’s population bracket, not
its level of development (Braga, et al., 2015). Given this, it is unlikely that these funds have any impact on income per-person other
than their effect on corruption. Therefore, I argue that the value of federal resources allocated to each municipality (RESOURCE)
is an exogenous instrument for PROB_CORRUPT.22
As can be seen in Table 12, the first stage F-statistics always reject the weak instrument null hypothesis and the J-statistics
indicate no evidence to reject the exogeneity of the instruments as before. Importantly, all statistical significance regarding
corruption disappears, while the negative effect of the informal sector remains.
21
The industrial composition in general is controlled for in the analysis; see Table 2 for a list of industries.
22
The correlation coefficient between these two variables is 0.285.
Panel a: Dependent variable is GDP per-capita in 2004. Panel b: Dependent variable is total income per-worker in 2000.
Marginal Effect of Corruption on GDP_PC_04 Marginal Effect of Corruption on TOTAL_00

2
1
Marginal Effect of PROB_CORRUPT
.5
1
0
0
-.5
-1
-2
-1
0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1
Panel c: Dependent variable is GDP per-capita in 2004. Panel b: Dependent variable is total income per-worker in 2010.
Marginal Effect of Corruption on GDP_PC_10 Marginal Effect of Corruption on TOTAL_10

1
.4
Marginal Effect of PROB_CORRUPT
.5
.2
0
0
-.5
-.2
-1
-1.5
-.4
0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1
Fig. 4. The marginal effect of corruption (PROB_CORRUPT) contingent upon the size of the informal economy.
Notes: Figures summarize effects presented in Table 13 panels a and b.
Thus, it appears that the size of the informal sector is negatively related to officially calculated GDP per-capita. This effect is
robust to the inclusion of alternative corruption measures, a measurement of nonvisible corruption, and various fixed effects,
as well as an instrumental variable analysis. Although this affect may be partially attributed to the fact that officially calculated
GDP imperfectly includes the value added from the informal sector, the fact that the size of the informal sector also seems to be
negatively related to total income per-worker leads to the conclusion that a larger informal sector leads to lower levels of actual
output and incomes.
5.2. The effect of corruption contingent upon informal sector size
It is possible that corruption has differing effects depending on the size of the informal sector. If this is the case then we may
be seeing little significance overall because there are off-setting effects. In order to test this I re-estimate the results using Eq.
(2.2) with the main corruption variable of interest, PROB_CORRUPT, only using GDP per-capita and total income per-worker as
the dependent variables. Table 13 presents the properly calculated marginal effects of PROB_CORRUPT at the minimum, median,
and maximum levels of SIZE_INFORMAL. Fig. 4 summarizes these effects in four separate graphs. Full results of this estimation
are available upon request.
As outlined in Section 2 of this paper, I expect that corruption has a positive effect in areas with a small informal sector
and a negative effect in areas with a large informal sector. As can be seen in Table 13, this is exactly what is occurring. At the
minimum value of SIZE_INFORMAL, PROB_CORRUPT tends to be positively related to both GDP per-capita and total income per-
worker. The median effect of corruption does tend to be negative. This aligns with the main results finding an overall negative
effect of corruption.
Thus, even if the overall effect of corruption is statistically insignificant when looking at municipalities as a whole, corruption
may still be important in determining economic outcomes. This is especially true when the informal sector is large. Though
the marginal effects of SIZE_INFORMAL are not reported in Table 13, SIZE_INFORMAL has a consistently negative and statistically
significant effect on both GDP per-capita and total income per-worker. The negative effect is present regardless of the amount of
corruption. Thus, as with the main results, the size of the informal sector seems to be the most important overall in determining
income levels.
6. Conclusion
The negative association between average income levels and corruption is well documented. In addition, although the litera-
ture is mixed as to the exact direction of the relationship, the idea that corruption and the informal economy are highly connected
is widely accepted as well. In particular, a common argument is that the informal economy and corruption are complementary
to each other and therefore a higher amount of corruption will be associated with a larger informal economy. Moreover, a larger
informal economy is also associated with lower levels of incomes. If this is the case, then it is plausible that the negative as-
sociation documented between corruption and income levels may be partly due to the effect corruption has on the informal
economy.
However, the empirical literature analyzing the links between corruption and income levels almost entirely ignores the effect
of the informal economy. This paper exploits a unique data set on corruption and informal sector employment in 476 Brazilian
municipalities to estimate whether corruption impacts GDP or income levels once variation in informal economic activity is
taken into account. Overall, I find that both corruption and the informal economy are generally associated with poor economic
outcomes. However, only the size of the informal economy has a robustly statistically significant effect.
Further, these effects are large in magnitude. For example, one standard deviation increase (0.186) in the share of total em-
ployees that are informally employed is associated with a decrease in GDP per-capita of a maximum of about 18% to a minimum
of about 13%, depending on the specification. Further, a one standard deviation increase in the size of the informal sector is asso-
ciated with a decrease in income (both formal and informal) per-worker of a maximum of about 8% to a minimum of about 7%.
Given that the size of the informal economy varies from a minimum of about 13% to a maximum of 98% across municipalities in
Brazil, these effects can be extremely large. Thus, the results indicate that the size of the informal economy has a robust negative
effect on income levels in Brazil.
The findings of this paper suggest that policies aimed at increasing income should be focused on reducing the size of the
informal sector. To do so, one needs to better understand the underlying causes of the informal economy. In developed countries,
there are many benefits of partaking in the formal sector. An example of a major benefit is the use of the formal legal system.
There are also costs of being a part of the formal sector. These costs include, but are not limited to, paying taxes and abiding by
formal rules and regulations. If the costs of engaging in the formal economy outweigh the benefits, individuals will choose to go
underground. Future research should be focused on revealing these costs and benefits.
If policy makers find that the benefits firms receive from participating in the formal sector are relatively low, they may think
about ways to improve the quality of their public service. Similarly, if policy makers find that the costs to firms for participating
in the formal sector are relatively high, they need to think about how to reduce these costs, without sacrificing quality. Perhaps
some of the regulations in place are inefficient, causing firms to move underground. In this case, these regulations may not be
serving their intended purpose since they are not being followed. Thus, the benefits of these regulations are likely outweighed
by the costs.
Relatedly, policy makers need to think more about enforcement. An informal sector that accounts for over half of the work-
force as in Brazil indicates that individuals who participate in the informal sector are not being effectively punished. As noted in
La Porta and Shleifer (2014), reducing the size of the informal sector in countries where the informal sector is well established
takes a significant amount of time. This is likely because informality is so well accepted in these countries. Changing this view re-
quires policy makers to be more stringent in their enforcement of rules and regulations. Policy makers can quicken the reduction
in informal sector size by encouraging work in the formal sector, and punishing of those who are caught working in the informal
sector.
Acknowledgments
I would like to thank Claudio Ferraz and Frederico Finan for kindly sharing their data.
Appendix A. Alternative measures of corruption and the size of the informal sector
Tables A1 and A4.

Table A1
Summary statistics and brief descriptions of alternative corruption and informal sector size measures.
Variable Description Mean Std. dev. Min Max
Alternative measures of corruption
NUM_CORRUPT Number of violations found to involve corrupt. 1.931 1.707 0 10

NUM_CORRUPT_PC Number of violations found to involve corrupt divided by 2003 population. 0.0 0 0 0.0 0 0 0 0.002
VALUE_CORRUPT_PC Value of corrupt resources divided by 2003 population. 14.985 30.529 0 344
PROB_CORRUPT_AV The average of PROB_CORRUPT and PROB_CORRUPT_2. 0.058 0.067 0 0.430
PROB_CORRUPT_AV_2 (0.4 x PROB_CORRUPT_2) + (0.6 x PROB_CORRUPT) 0.058 0.073 0 0.502
Alternative measures of informal sector
SIZE_INFORMAL_2 Share of total employment comprised of self-employed or non-formal contract workers. 0.603 0.124 0.167 0.940
SIZE_INFORMAL_3 Share of total, non-agricultural, employees without a formal contract. 0.621 0.184 0.124 0.985
Notes: All corruption data comes from Ferraz and Finan (2011) and all informal sector data comes from the Brazilian Census Bureau (IBGE).
Table A2
The effect of corruption, measured using the alternative variables described in Table A1, and the informal econ-
omy on GDP per-capita and total income per-worker over both periods.
Panel a: Effects in the first period.
GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00
NUM_CORRUPT −0.013 −0.003

(0.014) (0.005)
NUM_CORRUPT_PC −75.389 −76.735
(170.074) (54.179)
VALUE_CORRUPT_PC −0.001∗ 0.0 0 0
(0.001) (0.0 0 0)
∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗
SIZE_INFORMAL −0.710 −0.443 −0.705 −0.442 −0.696∗∗ −0.442∗∗∗
(0.267) (0.080) (0.268) (0.080) (0.269) (0.080)
Observations 434 434 434 434 434 434
R-squared 0.677 0.852 0.677 0.853 0.679 0.852
Panel b: Effects in the second period.
GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10
NUM_CORRUPT −0.005 0.005

(0.013) (0.003)
NUM_CORRUPT_PC −7.784 40.621
(146.661) (38.078)
VALUE_CORRUPT_PC −0.001∗ 0.0 0 0
(0.001) (0.0 0 0)
SIZE_INFORMAL −0.628∗∗ −0.261∗∗∗ −0.626∗∗ −0.263∗∗∗ −0.619∗∗ −0.263∗∗∗
(0.263) (0.064) (0.264) (0.064) (0.264) (0.065)
Observations 434 434 434 434 434 434
R-squared 0.664 0.854 0.664 0.854 0.666 0.854
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically significant at 5% level, ∗ indicates
statistically significant at 10% level. Standard errors are in parentheses. State and lottery fixed effects, as well as
a constant, are included in all regressions. Full results available upon request.
Table A3
The effect of corruption, measured using the alternative variables described in Table A1, and the informal economy on GDP per-capita and total
income per-worker over both periods.
Panel a: Effects in the first period.
PROB_CORRUPT_AV −0.482 −0.010

(0.378) (0.123)
PROB_CORRUPT_AV_2 −0.462 −0.001
(0.346) (0.110)
PROB_CORRUPT_AV_3 −0.436 0.004
(0.317) (0.099)
(0.291) (0.089)
SIZE_INFORMAL −0.721∗∗∗ −0.442∗∗∗ −0.721∗∗∗ −0.441∗∗∗ −0.720∗∗∗ −0.441∗∗∗ −0.720∗∗∗ −0.441∗∗∗
(0.268) (0.081) (0.268) (0.081) (0.268) (0.081) (0.268) (0.081)
Observations 434 434 434 434 434 434 434 434
R-squared 0.678 0.852 0.678 0.852 0.678 0.852 0.678 0.852
Panel b: Effects in the second period.
PROB_CORRUPT_AV −0.400 0.054

(0.331) (0.081)
(0.303) (0.073)
(0.277) (0.067)
(0.253) (0.061)
SIZE_INFORMAL −0.639∗∗ −0.261∗∗∗ −0.639∗∗ −0.262∗∗∗ −0.639∗∗ −0.262∗∗∗ −0.639∗∗ −0.262∗∗∗
(0.265) (0.064) (0.265) (0.064) (0.265) (0.064) (0.265) (0.064)
Observations 434 434 434 434 434 434 434 434
R-squared 0.665 0.854 0.665 0.854 0.665 0.854 0.665 0.854
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically significant at 5% level, ∗ indicates statistically significant at 10%
level. Standard errors are in parentheses. State and lottery fixed effects, as well as a constant, are included in all regressions. Full results available
upon request.
Table A4
The effect of corruption, measured as the share of resources (R$) found to involve
corruption, and the informal economy, measured using the alternative variables
described in Table A1, on GDP per-capita and total income per-worker over both
periods.
GDP_PC_04 TOTAL_00 GDP_PC_10 TOTAL_10
PROB_CORRUPT −0.326 0.032 −0.286 0.008

(0.257) (0.079) (0.218) (0.052)
SIZE_INFORMAL_2 0.017 −0.383∗∗∗ −0.089 −0.133∗∗
(0.291) (0.084) (0.271) (0.061)
Observations 434 434 434 434
R-squared 0.672 0.848 0.659 0.849
PROB_CORRUPT −0.327 0.042 −0.286 0.001
(0.257) (0.076) (0.218) (0.050)
SIZE_INFORMAL_3 −0.465∗ −0.368∗∗∗ −0.306 −0.222∗∗∗
(0.265) (0.086) (0.251) (0.069)
R-squared 0.672 0.845 0.657 0.850
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically

significant at 5% level, ∗ indicates statistically significant at 10% level. Standard
errors are in parentheses. State and lottery fixed effects, as well as a constant, are
included in all regressions. Full results available upon request.
Appendix B. Full tables of main results presented in Table 3 through Table 6 and Table 11 through Table 12
Tables B1 and B19.
Table B1
The effect corruption, as measured by the share of audited resources (R$) found
to involve corruption, and the informal economy has on GDP per-capita and
income per-capita from different sources.
Types of income in 20 0 0
GDP_PC_04 TOTAL FORMAL INFORMAL
PROB_CORRUPT −0.453∗∗ −0.050 −0.053 −0.191∗

(0.227) (0.080) (0.088) (0.106)
SIZE_INFORMAL −0.948∗∗∗ −0.399∗∗∗ 0.225∗∗ 0.002
(0.208) (0.078) (0.108) (0.105)
POPULATION 0.017 0.050∗∗∗ 0.086∗∗∗ 0.016
(0.042) (0.013) (0.017) (0.019)
URBAN −0.068 −0.152∗∗ −0.116 −0.082
(0.208) (0.067) (0.091) (0.088)
SCHOOL 0.196∗∗∗ 0.202∗∗∗ 0.173∗∗∗ 0.131∗∗∗
(0.052) (0.020) (0.026) (0.025)
DENSITY −0.118∗∗∗ −0.059∗∗∗ −0.057∗∗∗ −0.048∗∗∗
(0.029) (0.009) (0.012) (0.013)
DISTANCE_SP 0.009 −0.048∗∗∗ −0.038∗ −0.061∗∗∗
(0.050) (0.016) (0.019) (0.021)
TEEN −4.563∗ −4.189∗∗∗ −2.121 −5.383∗∗∗
(2.400) (0.882) (1.344) (1.045)
WORK 4.431∗∗∗ −1.635∗∗∗ −0.905 −1.267∗∗
(1.253) (0.485) (0.725) (0.613)
ELDER −6.061∗∗∗ −4.090∗∗∗ −2.695∗∗∗ −5.766∗∗∗
(1.270) (0.429) (0.613) (0.540)
AGRICULTURE −0.404 −0.078 −0.723 0.653
(1.349) (0.534) (1.047) (0.510)
CONSTRUCTION 1.380 0.412 0.123 1.052
(1.848) (0.680) (1.217) (0.786)
MANUFACTURING −0.041 −0.192 −0.575 0.894∗
(1.381) (0.532) (1.049) (0.509)
TRADE −1.744 0.217 −0.570 1.121∗
(1.563) (0.571) (1.020) (0.618)
TRANSPORT 0.405 1.304 0.532 2.140∗
(2.500) (0.917) (1.417) (1.138)
SERVICES 0.573 0.377 −0.351 0.984
(2.218) (0.747) (1.289) (0.764)
PROVISION −0.694 −0.733 −1.602 0.080
(1.498) (0.585) (1.090) (0.600)
SOCIAL 0.211 0.702 0.008 0.957
(1.748) (0.664) (1.254) (0.712)
GOVERNMENT 0.508 0.007 −0.715 0.576
(1.677) (0.583) (1.118) (0.618)
OTHER −2.168 −0.030 −0.524 1.027
(1.657) (0.642) (1.111) (0.669)
R-squared 0.634 0.806 0.503 0.626

errors are in parentheses. Constant included in all regressions.
Table B2
∗
PROB_CORRUPT −0.460 −0.105 −0.100 −0.224∗
(0.265) (0.098) (0.108) (0.130)
MISMANAGE 0.072 0.001 0.002 −0.006
(0.045) (0.008) (0.012) (0.010)
SIZE_INFORMAL −0.890∗∗∗ −0.444∗∗∗ 0.170 0.002
(0.240) (0.099) (0.131) (0.135)
POPULATION 0.012 0.058∗∗∗ 0.100∗∗∗ 0.019
(0.041) (0.016) (0.021) (0.022)
URBAN −0.046 −0.165∗∗ −0.186∗ −0.043
(0.234) (0.076) (0.101) (0.108)
SCHOOL 0.199∗∗∗ 0.191∗∗∗ 0.146∗∗∗ 0.125∗∗∗
(0.055) (0.024) (0.031) (0.030)
DENSITY −0.129∗∗∗ −0.060∗∗∗ −0.055∗∗∗ −0.051∗∗∗
(0.033) (0.011) (0.014) (0.015)
DISTANCE_SP −0.019 −0.054∗∗∗ −0.048∗∗ −0.070∗∗∗
(0.060) (0.019) (0.023) (0.025)
TEEN −4.771∗ −4.660∗∗∗ −2.385 −5.450∗∗∗
(2.650) (1.038) (1.616) (1.256)
WORK 4.489∗∗∗ −1.647∗∗∗ −0.526 −1.362∗
(1.376) (0.558) (0.860) (0.724)
ELDER −5.862∗∗∗ −4.249∗∗∗ −2.921∗∗∗ −5.827∗∗∗
(1.350) (0.542) (0.777) (0.682)
AGRICULTURE −1.413 −0.666 −2.590∗∗ 1.066
(2.028) (0.888) (1.013) (0.739)
CONSTRUCTION −1.173 −0.294 −2.093 1.841
(2.752) (1.136) (1.284) (1.137)
MANUFACTURING −1.196 −0.855 −2.448∗∗ 1.230∗
(2.067) (0.879) (1.016) (0.719)
TRADE −1.869 −0.346 −2.205∗∗ 1.518∗
(2.134) (0.891) (1.014) (0.825)
TRANSPORT −2.154 0.804 −1.703 2.925∗∗
(3.145) (1.273) (1.480) (1.436)
SERVICES 1.438 −0.265 −2.476∗ 1.505
(3.857) (1.198) (1.414) (1.229)
PROVISION −1.784 −1.415 −3.547∗∗∗ 0.307
(2.159) (0.935) (1.086) (0.857)
SOCIAL −0.474 0.028 −1.783 1.087
(2.492) (1.051) (1.263) (0.975)
GOVERNMENT −0.621 −0.490 −2.689∗∗ 1.066
(2.282) (0.922) (1.107) (0.798)
OTHER −2.576 −0.522 −2.274∗∗ 1.391
(2.354) (1.009) (1.156) (0.872)
R-squared 0.656 0.798 0.504 0.616

Table B3
income per-capita from different sources with fixed effects.
PROB_CORRUPT −0.353 0.012 −0.029 −0.123

(0.248) (0.075) (0.084) (0.113)
SIZE_INFORMAL −0.719∗∗∗ −0.441∗∗∗ 0.281∗∗ −0.144
(0.268) (0.081) (0.111) (0.124)
POPULATION 0.024 0.042∗∗∗ 0.074∗∗∗ 0.011
(0.052) (0.012) (0.015) (0.019)
URBAN 0.021 −0.077 −0.030 −0.017
(0.196) (0.068) (0.088) (0.090)
SCHOOL 0.218∗∗∗ 0.133∗∗∗ 0.131∗∗∗ 0.055∗
(0.065) (0.024) (0.031) (0.030)
DENSITY −0.152∗∗∗ −0.034∗∗ −0.034∗∗ −0.018
(0.042) (0.013) (0.017) (0.015)
DISTANCE_SP 0.017 −0.100∗∗∗ −0.051∗ −0.157∗∗∗
(0.064) (0.021) (0.026) (0.026)
TEEN −5.724∗∗ −2.209∗∗ −0.517 −2.906∗∗
(2.878) (1.006) (1.487) (1.315)
WORK 2.540 −0.615 −0.171 −0.279
(1.594) (0.571) (0.825) (0.709)
ELDER −8.576∗∗∗ −2.078∗∗∗ −0.866 −2.990∗∗∗
(1.710) (0.546) (0.734) (0.730)
AGRICULTURE −0.764 −0.362 −0.892 0.303
(1.392) (0.567) (1.123) (0.527)
CONSTRUCTION 0.786 0.403 0.341 0.906
(1.980) (0.717) (1.331) (0.849)
MANUFACTURING −0.064 −0.512 −0.809 0.494
(1.450) (0.562) (1.128) (0.520)
TRADE −1.513 −0.081 −0.849 0.643
(1.540) (0.620) (1.117) (0.618)
TRANSPORT 0.242 0.862 0.410 1.686
(2.637) (0.903) (1.490) (1.058)
SERVICES 1.074 −0.102 −0.799 0.531
(2.301) (0.727) (1.267) (0.720)
PROVISION −1.219 −0.932 −1.694 −0.211
(1.512) (0.610) (1.148) (0.620)
SOCIAL 0.023 0.417 −0.529 0.371
(1.902) (0.645) (1.302) (0.691)
GOVERNMENT −0.055 −0.587 −1.208 −0.173
(1.883) (0.634) (1.161) (0.683)
OTHER −2.824 −0.718 −1.158 0.515
(1.765) (0.675) (1.174) (0.695)
R-squared 0.679 0.852 0.623 0.702

Table B4
income per-capita from different sources with fixed effects.
PROB_CORRUPT −0.475 0.015 −0.072 −0.062

(0.312) (0.093) (0.107) (0.133)
MISMANAGE 0.046 0.002 0.002 −0.004
(0.050) (0.010) (0.013) (0.012)
SIZE_INFORMAL −0.484 −0.448∗∗∗ 0.304∗∗ −0.155
(0.294) (0.105) (0.131) (0.161)
POPULATION 0.040 0.052∗∗∗ 0.088∗∗∗ 0.014
(0.051) (0.016) (0.020) (0.023)
URBAN 0.071 −0.098 −0.109 −0.007
(0.234) (0.084) (0.104) (0.115)
SCHOOL 0.225∗∗∗ 0.137∗∗∗ 0.131∗∗∗ 0.054
(0.078) (0.030) (0.039) (0.036)
DENSITY −0.156∗∗∗ −0.033∗∗ −0.028 −0.022
(0.048) (0.016) (0.019) (0.020)
DISTANCE_SP 0.028 −0.101∗∗∗ −0.051 −0.167∗∗∗
(0.089) (0.029) (0.036) (0.035)
TEEN −8.184∗∗ −3.105∗∗ −1.664 −2.936∗
(3.498) (1.283) (1.924) (1.703)
WORK 1.519 −0.851 −0.337 −0.275
(1.832) (0.689) (1.012) (0.838)
ELDER −9.137∗∗∗ −2.480∗∗∗ −1.502 −3.079∗∗∗
(2.168) (0.691) (0.949) (0.910)
AGRICULTURE −2.363 −0.799 −2.760∗∗ 0.766
(2.598) (0.989) (1.137) (0.726)
CONSTRUCTION −2.945 −0.030 −2.103 2.189∗
(3.454) (1.255) (1.434) (1.146)
MANUFACTURING −2.139 −1.036 −2.744∗∗ 0.892
(2.633) (0.983) (1.140) (0.695)
TRADE −2.279 −0.604 −2.617∗∗ 0.996
(2.558) (1.0 0 0) (1.147) (0.795)
TRANSPORT −2.981 0.269 −2.064 1.958
(3.735) (1.327) (1.658) (1.333)
SERVICES 1.789 −0.712 −2.918∗∗ 1.006
(4.340) (1.200) (1.407) (1.132)
PROVISION −2.622 −1.546 −3.706∗∗∗ 0.011
(2.633) (1.029) (1.179) (0.841)
SOCIAL −0.998 −0.045 −2.270∗ 0.688
(3.126) (1.110) (1.326) (0.910)
GOVERNMENT −1.201 −1.128 −3.183∗∗∗ 0.038
(2.881) (1.034) (1.214) (0.834)
OTHER −3.387 −1.093 −2.965∗∗ 0.875
(2.865) (1.086) (1.242) (0.874)
R-squared 0.701 0.847 0.628 0.700

Table B5
The effect corruption, as measured by the share of audited items found to in-
volve corruption, and the informal economy has on GDP per-capita and income
per-capita from different sources.
PROB_CORRUPT_2 −0.044 −0.206 −0.246 −0.213

(0.477) (0.177) (0.226) (0.239)
SIZE_INFORMAL −0.925∗∗∗ −0.399∗∗∗ 0.225∗∗ 0.009
(0.205) (0.078) (0.108) (0.104)
POPULATION 0.023 0.049∗∗∗ 0.086∗∗∗ 0.017
(0.043) (0.013) (0.017) (0.019)
URBAN −0.096 −0.151∗∗ −0.114 −0.090
(0.209) (0.067) (0.091) (0.089)
SCHOOL 0.200∗∗∗ 0.199∗∗∗ 0.170∗∗∗ 0.129∗∗∗
(0.053) (0.020) (0.027) (0.025)
DENSITY −0.119∗∗∗ −0.059∗∗∗ −0.056∗∗∗ −0.049∗∗∗
(0.029) (0.010) (0.012) (0.013)
DISTANCE_SP 0.006 −0.050∗∗∗ −0.041∗∗ −0.064∗∗∗
(0.051) (0.016) (0.020) (0.021)
TEEN −4.954∗∗ −4.100∗∗∗ −2.008 −5.420∗∗∗
(2.411) (0.882) (1.338) (1.073)
WORK 4.372∗∗∗ −1.642∗∗∗ −0.912 −1.293∗∗
(1.259) (0.482) (0.721) (0.609)
ELDER −6.282∗∗∗ −4.060∗∗∗ −2.655∗∗∗ −5.805∗∗∗
(1.262) (0.424) (0.607) (0.538)
AGRICULTURE −0.502 −0.095 −0.742 0.605
(1.368) (0.548) (1.052) (0.512)
CONSTRUCTION 0.871 0.337 0.043 0.820
(1.794) (0.672) (1.198) (0.831)
MANUFACTURING −0.133 −0.214 −0.601 0.843∗
(1.397) (0.546) (1.053) (0.510)
TRADE −1.736 0.231 −0.554 1.137∗
(1.585) (0.588) (1.031) (0.623)
TRANSPORT 0.277 1.340 0.577 2.134∗
(2.523) (0.919) (1.426) (1.121)
SERVICES 0.526 0.390 −0.335 0.981
(2.233) (0.773) (1.313) (0.773)
PROVISION −0.759 −0.750 −1.622 0.044
(1.522) (0.599) (1.095) (0.604)
SOCIAL 0.164 0.704 0.011 0.943
(1.771) (0.678) (1.261) (0.713)
GOVERNMENT 0.532 −0.031 −0.761 0.546
(1.696) (0.596) (1.121) (0.626)
OTHER −2.272 −0.035 −0.528 0.990
(1.675) (0.656) (1.118) (0.665)
R-squared 0.630 0.806 0.505 0.623

Table B6
PROB_CORRUPT_2 −0.189 −0.262 −0.348 −0.007

(0.701) (0.259) (0.324) (0.339)
MISMANAGE 0.072 0.005 0.007 −0.008
(0.050) (0.009) (0.013) (0.011)
SIZE_INFORMAL −0.869∗∗∗ −0.447∗∗∗ 0.165 0.016
(0.237) (0.099) (0.133) (0.137)
POPULATION 0.018 0.058∗∗∗ 0.100∗∗∗ 0.023
(0.041) (0.016) (0.020) (0.023)
URBAN −0.085 −0.168∗∗ −0.187∗ −0.063
(0.230) (0.076) (0.101) (0.110)
SCHOOL 0.205∗∗∗ 0.189∗∗∗ 0.142∗∗∗ 0.129∗∗∗
(0.056) (0.025) (0.032) (0.031)
DENSITY −0.131∗∗∗ −0.060∗∗∗ −0.056∗∗∗ −0.052∗∗∗
(0.033) (0.011) (0.014) (0.015)
DISTANCE_SP −0.020 −0.056∗∗∗ −0.051∗∗ −0.069∗∗∗
(0.061) (0.019) (0.023) (0.025)
TEEN −5.062∗ −4.604∗∗∗ −2.276 −5.640∗∗∗
(2.645) (1.052) (1.623) (1.309)
WORK 4.451∗∗∗ −1.651∗∗∗ −0.528 −1.382∗
(1.391) (0.557) (0.859) (0.729)
ELDER −5.984∗∗∗ −4.270∗∗∗ −2.938∗∗∗ −5.888∗∗∗
(1.342) (0.536) (0.771) (0.678)
AGRICULTURE −1.537 −0.691 −2.612∗∗ 1.005
(2.015) (0.913) (1.045) (0.723)
CONSTRUCTION −1.365 −0.366 −2.174∗ 1.759
(2.759) (1.154) (1.303) (1.130)
MANUFACTURING −1.329 −0.890 −2.484∗∗ 1.167∗
(2.050) (0.903) (1.045) (0.704)
TRADE −1.876 −0.335 −2.189∗∗ 1.510∗
(2.130) (0.917) (1.051) (0.823)
TRANSPORT −2.273 0.808 −1.685 2.854∗∗
(3.164) (1.295) (1.512) (1.414)
SERVICES 1.255 −0.289 −2.490∗ 1.409
(3.850) (1.224) (1.449) (1.208)
PROVISION −1.961 −1.441 −3.565∗∗∗ 0.215
(2.147) (0.958) (1.115) (0.839)
SOCIAL −0.540 0.035 −1.766 1.046
(2.478) (1.076) (1.298) (0.951)
GOVERNMENT −0.694 −0.527 −2.734∗∗ 1.038
(2.266) (0.948) (1.134) (0.798)
OTHER −2.656 −0.550 −2.305∗ 1.356
(2.353) (1.035) (1.190) (0.856)
R-squared 0.653 0.799 0.506 0.612

Table B7
per-capita from different sources with fixed effects.
PROB_CORRUPT_2 −0.043 −0.104 −0.226 −0.022

(0.497) (0.204) (0.237) (0.264)
SIZE_INFORMAL −0.706∗∗∗ −0.444∗∗∗ 0.276∗∗ −0.140
(0.268) (0.080) (0.111) (0.124)
POPULATION 0.027 0.042∗∗∗ 0.074∗∗∗ 0.012
(0.052) (0.012) (0.015) (0.019)
URBAN 0.010 −0.074 −0.025 −0.021
(0.195) (0.068) (0.088) (0.092)
SCHOOL 0.215∗∗∗ 0.131∗∗∗ 0.128∗∗∗ 0.054∗
(0.066) (0.024) (0.032) (0.030)
DENSITY −0.152∗∗∗ −0.034∗∗ −0.033∗ −0.019
(0.042) (0.013) (0.017) (0.015)
DISTANCE_SP 0.007 −0.100∗∗∗ −0.052∗∗ −0.161∗∗∗
(0.065) (0.021) (0.026) (0.027)
TEEN −5.593∗ −2.249∗∗ −0.582 −2.863∗∗
(2.890) (0.995) (1.491) (1.285)
WORK 2.652∗ −0.650 −0.228 −0.242
(1.582) (0.561) (0.826) (0.683)
ELDER −8.730∗∗∗ −2.062∗∗∗ −0.855 −3.043∗∗∗
(1.722) (0.546) (0.727) (0.724)
AGRICULTURE −0.871 −0.345 −0.873 0.266
(1.386) (0.578) (1.136) (0.523)
CONSTRUCTION 0.344 0.419 0.308 0.752
(1.884) (0.712) (1.335) (0.907)
MANUFACTURING −0.173 −0.500 −0.799 0.457
(1.441) (0.572) (1.141) (0.514)
TRADE −1.497 −0.069 −0.823 0.650
(1.546) (0.632) (1.132) (0.621)
TRANSPORT 0.091 0.907 0.484 1.636
(2.663) (0.900) (1.499) (1.029)
SERVICES 1.030 −0.089 −0.778 0.517
(2.309) (0.742) (1.289) (0.720)
PROVISION −1.324 −0.909 −1.662 −0.247
(1.507) (0.618) (1.161) (0.607)
SOCIAL −0.047 0.445 −0.480 0.348
(1.901) (0.656) (1.316) (0.681)
GOVERNMENT −0.064 −0.578 −1.192 −0.176
(1.901) (0.645) (1.174) (0.686)
OTHER −2.947∗ −0.703 −1.145 0.473
(1.760) (0.684) (1.185) (0.686)
R-squared 0.677 0.852 0.624 0.700

Table B8
PROB_CORRUPT_2 0.299 −0.023 −0.240 0.217

(0.729) (0.260) (0.333) (0.326)
MISMANAGE 0.036 0.003 0.005 −0.008
(0.054) (0.009) (0.014) (0.012)
SIZE_INFORMAL −0.449 −0.449∗∗∗ 0.296∗∗ −0.141
(0.290) (0.106) (0.133) (0.163)
POPULATION 0.046 0.051∗∗∗ 0.089∗∗∗ 0.015
(0.052) (0.016) (0.020) (0.023)
URBAN 0.029 −0.097 −0.110 −0.015
(0.230) (0.084) (0.103) (0.117)
SCHOOL 0.226∗∗∗ 0.136∗∗∗ 0.130∗∗∗ 0.055
(0.079) (0.031) (0.039) (0.037)
DENSITY −0.161∗∗∗ −0.033∗∗ −0.027 −0.023
(0.049) (0.016) (0.019) (0.019)
DISTANCE_SP 0.010 −0.100∗∗∗ −0.051 −0.171∗∗∗
(0.089) (0.029) (0.035) (0.035)
TEEN −7.801∗∗ −3.123∗∗ −1.747 −2.798∗
(3.551) (1.260) (1.923) (1.653)
WORK 1.863 −0.866 −0.376 −0.173
(1.826) (0.679) (1.014) (0.808)
ELDER −9.192∗∗∗ −2.479∗∗∗ −1.527 −3.076∗∗∗
(2.165) (0.694) (0.943) (0.905)
AGRICULTURE −2.563 −0.792 −2.760∗∗ 0.721
(2.500) (0.992) (1.162) (0.701)
CONSTRUCTION −3.071 −0.027 −2.138 2.183∗
(3.360) (1.258) (1.455) (1.131)
MANUFACTURING −2.299 −1.031 −2.762∗∗ 0.867
(2.537) (0.985) (1.164) (0.669)
TRADE −2.294 −0.603 −2.620∗∗ 0.995
(2.487) (1.004) (1.170) (0.780)
TRANSPORT −3.432 0.288 −2.035 1.838
(3.688) (1.318) (1.673) (1.296)
SERVICES 1.571 −0.705 −2.956∗∗ 0.982
(4.293) (1.204) (1.421) (1.119)
PROVISION −2.938 −1.533 −3.699∗∗∗ −0.064
(2.543) (1.030) (1.204) (0.812)
SOCIAL −1.166 −0.037 −2.242∗ 0.633
(3.036) (1.113) (1.354) (0.882)
GOVERNMENT −1.453 −1.118 −3.196∗∗ −0.011
(2.808) (1.036) (1.233) (0.824)
OTHER −3.542 −1.088 −2.979∗∗ 0.849
(2.777) (1.090) (1.264) (0.848)
R-squared 0.698 0.847 0.628 0.701

Table B9
Types of income in 2010
∗∗
PROB_CORRUPT −0.499 0.006 −0.061 −0.049
(0.211) (0.054) (0.069) (0.148)
SIZE_INFORMAL −0.861∗∗∗ −0.324∗∗∗ −0.025 −0.102
(0.207) (0.060) (0.069) (0.131)
POPULATION 0.021 0.040∗∗∗ 0.046∗∗∗ 0.077∗∗∗
(0.041) (0.009) (0.010) (0.023)
URBAN −0.145 −0.110∗∗ −0.057 −0.401∗∗∗
(0.186) (0.055) (0.065) (0.120)
SCHOOL 0.195∗∗∗ 0.154∗∗∗ 0.109∗∗∗ 0.240∗∗∗
(0.051) (0.017) (0.018) (0.029)
DENSITY −0.097∗∗∗ −0.048∗∗∗ −0.033∗∗∗ −0.040∗∗∗
(0.026) (0.008) (0.009) (0.013)
DISTANCE_SP −0.007 −0.017 −0.018 0.054∗
(0.046) (0.012) (0.014) (0.029)
TEEN −2.577 −1.583∗∗ −0.966 −8.411∗∗∗
(2.325) (0.672) (0.812) (1.500)
WORK 4.459∗∗∗ 0.096 −0.046 −3.552∗∗∗
(1.234) (0.370) (0.439) (0.834)
ELDER −4.925∗∗∗ −3.008∗∗∗ −1.693∗∗∗ −8.544∗∗∗
(1.196) (0.312) (0.382) (0.788)
AGRICULTURE −0.176 0.723 −0.011 0.966
(1.412) (0.541) (0.545) (0.691)
CONSTRUCTION 2.726 0.854 0.288 0.878
(1.949) (0.662) (0.681) (1.023)
MANUFACTURING −0.177 0.749 0.094 1.150
(1.458) (0.538) (0.544) (0.719)
TRADE −1.851 1.081∗∗ 0.300 0.917
(1.564) (0.545) (0.575) (0.824)
TRANSPORT 1.016 0.410 −0.366 2.035
(2.384) (0.814) (0.819) (1.476)
SERVICES 0.293 0.596 −0.282 1.111
(2.034) (0.709) (0.759) (0.851)
PROVISION −0.425 0.525 −0.584 0.881
(1.545) (0.563) (0.583) (0.783)
SOCIAL −0.204 1.025 0.075 0.257
(1.749) (0.658) (0.671) (1.003)
GOVERNMENT 0.205 0.960∗ 0.230 1.470∗
(1.704) (0.578) (0.627) (0.841)
OTHER −1.238 0.693 0.337 1.410
(1.745) (0.605) (0.664) (0.872)
R-squared 0.615 0.803 0.522 0.552

Table B10
PROB_CORRUPT −0.421∗ −0.017 −0.100 −0.118

(0.235) (0.063) (0.080) (0.169)
MISMANAGE 0.055 −0.001 0.013∗ −0.043∗∗∗
(0.035) (0.006) (0.007) (0.014)
SIZE_INFORMAL −0.839∗∗∗ −0.379∗∗∗ −0.089 −0.003
(0.235) (0.071) (0.083) (0.147)
POPULATION 0.011 0.049∗∗∗ 0.052∗∗∗ 0.083∗∗∗
(0.039) (0.011) (0.012) (0.027)
URBAN −0.140 −0.134∗∗ −0.048 −0.423∗∗∗
(0.202) (0.063) (0.076) (0.139)
SCHOOL 0.205∗∗∗ 0.137∗∗∗ 0.099∗∗∗ 0.229∗∗∗
(0.052) (0.019) (0.022) (0.033)
DENSITY −0.100∗∗∗ −0.049∗∗∗ −0.035∗∗∗ −0.030∗∗
(0.027) (0.009) (0.010) (0.014)
DISTANCE_SP −0.024 −0.020 −0.012 0.047
(0.054) (0.015) (0.016) (0.035)
TEEN −2.257 −1.974∗∗∗ −0.905 −8.294∗∗∗
(2.539) (0.760) (0.965) (1.629)
WORK 4.683∗∗∗ 0.192 0.244 −3.476∗∗∗
(1.362) (0.411) (0.505) (0.933)
ELDER −4.490∗∗∗ −2.992∗∗∗ −1.735∗∗∗ −8.243∗∗∗
(1.263) (0.393) (0.502) (0.912)
AGRICULTURE −1.503 0.081 −0.316 −0.218
(2.064) (0.824) (1.006) (1.112)
CONSTRUCTION −0.414 0.062 −0.082 −1.160
(2.834) (1.061) (1.263) (1.485)
MANUFACTURING −1.872 0.107 −0.207 −0.071
(2.102) (0.811) (0.999) (1.127)
TRADE −2.480 0.473 0.002 −0.285
(2.113) (0.799) (0.976) (1.188)
TRANSPORT −1.767 −0.521 −1.220 1.584
(3.011) (1.110) (1.279) (1.930)
SERVICES 0.596 −0.163 −0.604 0.071
(3.499) (1.055) (1.251) (1.347)
PROVISION −1.745 −0.169 −0.894 −0.331
(2.152) (0.827) (1.035) (1.167)
SOCIAL −1.035 0.486 −0.267 −1.272
(2.412) (0.958) (1.180) (1.404)
GOVERNMENT −1.334 0.474 −0.010 0.639
(2.243) (0.867) (1.080) (1.269)
OTHER −1.674 0.056 −0.154 0.725
(2.414) (0.898) (1.087) (1.278)
R-squared 0.656 0.800 0.512 0.555

Table B11
The effect corruption, as measured by the share of audited resources (R$)
found to involve corruption, and the informal economy has on GDP per-capita
and income per-capita from different sources with fixed effects.
PROB_CORRUPT −0.311 −0.003 −0.085 0.131

(0.214) (0.051) (0.065) (0.103)
SIZE_INFORMAL −0.638∗∗ −0.263∗∗∗ 0.096 −0.256∗∗
(0.265) (0.065) (0.074) (0.126)
POPULATION 0.016 0.047∗∗∗ 0.049∗∗∗ 0.056∗∗∗
(0.050) (0.009) (0.011) (0.016)
URBAN −0.075 −0.088 −0.034 −0.220∗∗
(0.177) (0.056) (0.069) (0.093)
SCHOOL 0.203∗∗∗ 0.119∗∗∗ 0.097∗∗∗ 0.051∗
(0.064) (0.019) (0.021) (0.026)
DENSITY −0.117∗∗∗ −0.026∗∗ −0.018 0.013
(0.035) (0.011) (0.011) (0.013)
DISTANCE_SP −0.027 −0.005 0.004 −0.084∗∗∗
(0.062) (0.017) (0.020) (0.025)
TEEN −3.383 −1.260∗ −1.104 −2.617∗
(2.856) (0.734) (0.854) (1.380)
WORK 2.962∗ 0.218 0.018 0.717
(1.546) (0.405) (0.481) (0.720)
ELDER −6.956∗∗∗ −1.719∗∗∗ −0.899∗ −2.633∗∗∗
(1.531) (0.430) (0.489) (0.749)
AGRICULTURE −0.486 0.410 −0.270 0.448
(1.452) (0.585) (0.616) (0.631)
CONSTRUCTION 1.856 0.890 0.388 0.608
(2.031) (0.708) (0.739) (0.807)
MANUFACTURING 0.004 0.475 −0.153 0.632
(1.522) (0.585) (0.615) (0.639)
TRADE −1.279 0.659 −0.199 0.229
(1.565) (0.596) (0.641) (0.660)
TRANSPORT 0.401 −0.184 −0.862 1.011
(2.563) (0.804) (0.855) (1.183)
SERVICES 0.735 0.284 −0.603 0.464
(2.093) (0.680) (0.708) (0.716)
PROVISION −0.887 0.244 −0.777 0.848
(1.562) (0.604) (0.648) (0.692)
SOCIAL −0.690 0.975 −0.140 −0.281
(1.865) (0.675) (0.704) (0.868)
GOVERNMENT −0.410 0.482 −0.284 −0.073
(1.906) (0.622) (0.700) (0.724)
OTHER −1.726 0.229 −0.167 0.411
(1.862) (0.663) (0.729) (0.727)
R-squared 0.665 0.854 0.633 0.801

Table B12
The effect corruption, as measured by the share of audited resources (R$) found to involve
corruption, and the informal economy has on GDP per-capita and income per-capita from
different sources with fixed effects.
PROB_CORRUPT −0.303 −0.012 −0.129∗ 0.230

(0.266) (0.058) (0.077) (0.141)
MISMANAGE 0.029 0.005 0.024∗∗∗ −0.026∗
(0.039) (0.008) (0.008) (0.014)
SIZE_INFORMAL −0.441 −0.266∗∗∗ 0.082 −0.190
(0.286) (0.080) (0.089) (0.137)
POPULATION 0.033 0.056∗∗∗ 0.055∗∗∗ 0.055∗∗∗
(0.047) (0.011) (0.012) (0.019)
URBAN −0.065 −0.094 −0.017 −0.206∗
(0.195) (0.069) (0.084) (0.109)
SCHOOL 0.210∗∗∗ 0.122∗∗∗ 0.118∗∗∗ 0.063∗∗
(0.073) (0.024) (0.026) (0.032)
DENSITY −0.116∗∗∗ −0.028∗∗ −0.017 0.019
(0.039) (0.012) (0.013) (0.015)
DISTANCE_SP −0.025 0.008 0.035 −0.075∗∗
(0.084) (0.022) (0.024) (0.034)
TEEN −4.747 −2.208∗∗ −1.739∗ −2.601
(3.462) (0.893) (1.016) (1.595)
WORK 2.537 0.004 −0.047 0.656
(1.768) (0.462) (0.531) (0.799)
ELDER −7.154∗∗∗ −2.020∗∗∗ −1.085∗ −2.687∗∗∗
(1.872) (0.526) (0.577) (0.882)
AGRICULTURE −2.432 −0.170 −0.624 −0.159
(2.536) (0.933) (1.122) (0.887)
CONSTRUCTION −2.038 0.285 0.132 0.364
(3.381) (1.156) (1.375) (1.192)
MANUFACTURING −2.486 −0.146 −0.583 0.016
(2.597) (0.925) (1.117) (0.896)
TRADE −2.686 0.076 −0.646 −0.630
(2.477) (0.899) (1.081) (0.878)
TRANSPORT −3.261 −1.062 −1.703 0.857
(3.559) (1.185) (1.389) (1.494)
SERVICES 0.593 −0.344 −0.875 −0.271
(3.927) (1.051) (1.225) (1.034)
PROVISION −2.566 −0.502 −1.337 −0.110
(2.566) (0.933) (1.131) (0.926)
SOCIAL −1.871 0.507 −0.593 −1.101
(2.925) (1.031) (1.252) (1.136)
GOVERNMENT −2.236 −0.089 −0.807 −0.702
(2.761) (0.964) (1.188) (0.954)
OTHER −2.320 −0.363 −0.754 −0.053
(2.849) (1.007) (1.190) (0.980)
R-squared 0.707 0.855 0.655 0.803
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statistically significant at

5% level, ∗ indicates statistically significant at 10% level. Standard errors are in parentheses.
Constant included in all regressions.
Table B13
PROB_CORRUPT_2 −0.167 0.124 0.144 −0.191

(0.404) (0.121) (0.128) (0.264)
SIZE_INFORMAL −0.837∗∗∗ −0.323∗∗∗ −0.020 −0.101
(0.204) (0.060) (0.069) (0.131)
POPULATION 0.027 0.041∗∗∗ 0.048∗∗∗ 0.077∗∗∗
(0.042) (0.009) (0.010) (0.023)
URBAN −0.173 −0.113∗∗ −0.064 −0.400∗∗∗
(0.186) (0.055) (0.064) (0.120)
SCHOOL 0.198∗∗∗ 0.156∗∗∗ 0.112∗∗∗ 0.237∗∗∗
(0.052) (0.017) (0.019) (0.030)
DENSITY −0.098∗∗∗ −0.048∗∗∗ −0.034∗∗∗ −0.040∗∗∗
(0.026) (0.008) (0.008) (0.013)
DISTANCE_SP −0.012 −0.015 −0.017 0.051∗
(0.047) (0.012) (0.014) (0.029)
TEEN −2.930 −1.658∗∗ −1.117 −8.331∗∗∗
(2.371) (0.668) (0.817) (1.518)
WORK 4.393∗∗∗ 0.097 −0.054 −3.558∗∗∗
(1.242) (0.369) (0.440) (0.832)
ELDER −5.136∗∗∗ −3.039∗∗∗ −1.764∗∗∗ −8.517∗∗∗
(1.208) (0.312) (0.378) (0.795)
AGRICULTURE −0.289 0.729 −0.020 0.949
(1.447) (0.529) (0.526) (0.700)
CONSTRUCTION 2.154 0.872 0.232 0.807
(1.895) (0.635) (0.647) (1.031)
MANUFACTURING −0.285 0.758 0.091 1.129
(1.490) (0.527) (0.525) (0.731)
TRADE −1.835 1.073∗∗ 0.291 0.930
(1.597) (0.536) (0.559) (0.832)
TRANSPORT 0.905 0.381 −0.420 2.068
(2.425) (0.809) (0.809) (1.479)
SERVICES 0.252 0.585 −0.302 1.123
(2.068) (0.694) (0.737) (0.863)
PROVISION −0.502 0.532 −0.586 0.865
(1.582) (0.552) (0.566) (0.791)
SOCIAL −0.252 1.022 0.063 0.258
(1.792) (0.648) (0.655) (1.013)
GOVERNMENT 0.208 0.984∗ 0.264 1.435∗
(1.736) (0.569) (0.611) (0.850)
OTHER −1.349 0.690 0.318 1.405
(1.779) (0.594) (0.644) (0.880)
R-squared 0.610 0.803 0.522 0.552

Table B14
PROB_CORRUPT_2 −0.102 0.175 0.133 0.125

(0.617) (0.182) (0.202) (0.335)
MISMANAGE 0.053 −0.004 0.010 −0.046∗∗∗
(0.041) (0.007) (0.008) (0.015)
SIZE_INFORMAL −0.817∗∗∗ −0.372∗∗∗ −0.079 0.008
(0.232) (0.072) (0.084) (0.149)
POPULATION 0.017 0.050∗∗∗ 0.054∗∗∗ 0.086∗∗∗
(0.039) (0.011) (0.012) (0.027)
URBAN −0.177 −0.140∗∗ −0.061 −0.438∗∗∗
(0.198) (0.062) (0.075) (0.138)
SCHOOL 0.211∗∗∗ 0.140∗∗∗ 0.103∗∗∗ 0.233∗∗∗
(0.054) (0.020) (0.023) (0.033)
DENSITY −0.102∗∗∗ −0.049∗∗∗ −0.035∗∗∗ −0.031∗∗
(0.027) (0.009) (0.010) (0.014)
DISTANCE_SP −0.024 −0.019 −0.010 0.049
(0.055) (0.015) (0.016) (0.036)
TEEN −2.564 −2.087∗∗∗ −1.067 −8.467∗∗∗
(2.556) (0.770) (0.978) (1.669)
WORK 4.647∗∗∗ 0.187 0.232 −3.490∗∗∗
(1.373) (0.411) (0.510) (0.937)
ELDER −4.604∗∗∗ −3.002∗∗∗ −1.767∗∗∗ −8.280∗∗∗
(1.261) (0.394) (0.500) (0.919)
AGRICULTURE −1.617 0.074 −0.345 −0.252
(2.052) (0.791) (0.969) (1.080)
CONSTRUCTION −0.580 0.078 −0.101 −1.187
(2.837) (1.021) (1.220) (1.454)
MANUFACTURING −1.991 0.107 −0.232 −0.101
(2.089) (0.779) (0.962) (1.102)
TRADE −2.490 0.463 −0.009 −0.297
(2.108) (0.769) (0.943) (1.166)
TRANSPORT −1.887 −0.552 −1.271 1.528
(3.035) (1.086) (1.250) (1.902)
SERVICES 0.423 −0.185 −0.658 0.010
(3.492) (1.021) (1.212) (1.310)
PROVISION −1.912 −0.187 −0.944 −0.388
(2.141) (0.795) (1.001) (1.136)
SOCIAL −1.103 0.465 −0.299 −1.307
(2.403) (0.921) (1.141) (1.374)
GOVERNMENT −1.394 0.489 −0.009 0.637
(2.226) (0.836) (1.047) (1.242)
OTHER −1.743 0.061 −0.163 0.712
(2.413) (0.868) (1.052) (1.253)
R-squared 0.653 0.801 0.510 0.554

Table B15
∗∗ ∗
PROB_CORRUPT_2 0.061 0.240 0.245 0.332
(0.419) (0.117) (0.143) (0.220)
SIZE_INFORMAL −0.624∗∗ −0.257∗∗∗ 0.105 −0.252∗∗
(0.264) (0.064) (0.073) (0.126)
POPULATION 0.020 0.048∗∗∗ 0.051∗∗∗ 0.056∗∗∗
(0.050) (0.009) (0.011) (0.016)
URBAN −0.087 −0.094∗ −0.043 −0.224∗∗
(0.177) (0.056) (0.068) (0.094)
SCHOOL 0.202∗∗∗ 0.122∗∗∗ 0.100∗∗∗ 0.056∗∗
(0.065) (0.019) (0.022) (0.026)
DENSITY −0.118∗∗∗ −0.027∗∗ −0.019∗ 0.012
(0.035) (0.010) (0.011) (0.013)
DISTANCE_SP −0.035 −0.005 0.002 −0.081∗∗∗
(0.062) (0.017) (0.020) (0.025)
TEEN −3.234 −1.177 −0.985 −2.558∗
(2.876) (0.733) (0.852) (1.378)
WORK 3.090∗∗ 0.290 0.121 0.769
(1.543) (0.404) (0.478) (0.719)
ELDER −7.102∗∗∗ −1.746∗∗∗ −0.962∗∗ −2.609∗∗∗
(1.546) (0.426) (0.482) (0.756)
AGRICULTURE −0.593 0.379 −0.328 0.448
(1.446) (0.559) (0.589) (0.603)
CONSTRUCTION 1.466 0.883 0.279 0.769
(1.949) (0.678) (0.709) (0.758)
MANUFACTURING −0.101 0.453 −0.202 0.645
(1.515) (0.560) (0.591) (0.612)
TRADE −1.275 0.631 −0.224 0.186
(1.564) (0.573) (0.618) (0.638)
TRANSPORT 0.231 −0.277 −0.997 0.945
(2.597) (0.788) (0.838) (1.179)
SERVICES 0.686 0.257 −0.642 0.445
(2.101) (0.649) (0.677) (0.694)
PROVISION −0.998 0.200 −0.849 0.830
(1.556) (0.580) (0.625) (0.668)
SOCIAL −0.776 0.916 −0.220 −0.332
(1.867) (0.647) (0.677) (0.842)
GOVERNMENT −0.426 0.463 −0.306 −0.094
(1.921) (0.599) (0.679) (0.701)
OTHER −1.844 0.203 −0.224 0.424
(1.860) (0.638) (0.703) (0.705)
R-squared 0.664 0.855 0.634 0.801

Table B16
volve corruption, and the informal economy has on GDP per-capita and in-
come per-capita from different sources.
PROB_CORRUPT_2 0.539 0.323∗ 0.170 0.649∗∗

(0.615) (0.164) (0.189) (0.317)
MISMANAGE 0.018 0.0 0 0 0.020∗∗ −0.034∗∗
(0.043) (0.008) (0.009) (0.015)
SIZE_INFORMAL −0.402 −0.250∗∗∗ 0.096 −0.169
(0.283) (0.080) (0.089) (0.136)
POPULATION 0.037 0.057∗∗∗ 0.056∗∗∗ 0.054∗∗∗
(0.047) (0.011) (0.012) (0.019)
URBAN −0.097 −0.101 −0.030 −0.199∗
(0.193) (0.069) (0.084) (0.110)
SCHOOL 0.213∗∗∗ 0.124∗∗∗ 0.119∗∗∗ 0.067∗∗
(0.074) (0.023) (0.026) (0.031)
DENSITY −0.120∗∗∗ −0.030∗∗ −0.019 0.018
(0.039) (0.012) (0.013) (0.015)
DISTANCE_SP −0.040 0.004 0.029 −0.075∗∗
(0.083) (0.022) (0.024) (0.034)
TEEN −4.330 −2.042∗∗ −1.591 −2.393
(3.498) (0.886) (1.012) (1.603)
WORK 2.868 0.114 0.075 0.745
(1.752) (0.456) (0.526) (0.798)
ELDER −7.168∗∗∗ −2.002∗∗∗ −1.094∗ −2.613∗∗∗
(1.869) (0.521) (0.575) (0.895)
AGRICULTURE −2.597 −0.208 −0.688 −0.146
(2.428) (0.879) (1.078) (0.825)
CONSTRUCTION −2.099 0.300 0.102 0.470
(3.276) (1.097) (1.331) (1.124)
MANUFACTURING −2.596 −0.157 −0.628 0.075
(2.495) (0.874) (1.078) (0.842)
TRADE −2.693 0.077 −0.649 −0.619
(2.388) (0.853) (1.046) (0.826)
TRANSPORT −3.668 −1.181 −1.856 0.805
(3.513) (1.138) (1.341) (1.465)
SERVICES 0.461 −0.343 −0.933 −0.149
(3.879) (1.007) (1.187) (0.980)
PROVISION −2.833 −0.569 −1.440 −0.108
(2.463) (0.882) (1.089) (0.869)
SOCIAL −2.043 0.444 −0.655 −1.168
(2.820) (0.969) (1.204) (1.074)
GOVERNMENT −2.427 −0.123 −0.883 −0.649
(2.678) (0.914) (1.150) (0.896)
OTHER −2.430 −0.378 −0.799 −0.005
(2.757) (0.958) (1.150) (0.924)
R-squared 0.706 0.856 0.653 0.804

Table B17
The effect corruption and the informal economy have on GDP per-capita and in-
come per-capita; IV robustness checks.
∗∗
(0.205) (0.093)
PROB_CORRUPT_2 −0.106 −0.094
(0.503) (0.209)
SIZE_INFORMAL −1.543∗∗∗ 0.273 −1.554∗∗∗ 0.284
(0.545) (0.248) (0.542) (0.247)
POPULATION 0.033 0.066∗∗∗ 0.040 0.065∗∗∗
(0.039) (0.015) (0.040) (0.015)
URBAN 0.013 −0.218∗∗∗ −0.013 −0.215∗∗
(0.214) (0.084) (0.214) (0.084)
SCHOOL 0.164∗∗∗ 0.244∗∗∗ 0.167∗∗∗ 0.242∗∗∗
(0.057) (0.026) (0.058) (0.027)
DENSITY −0.124∗∗∗ −0.054∗∗∗ −0.126∗∗∗ −0.053∗∗∗
(0.030) (0.010) (0.031) (0.010)
DISTANCE_SP 0.028 −0.069∗∗∗ 0.025 −0.070∗∗∗
(0.053) (0.018) (0.052) (0.017)
TEEN −3.602 −4.438∗∗∗ −3.746∗ −4.370∗∗∗
(2.208) (1.083) (2.203) (1.090)
WORK 4.727∗∗∗ −1.493∗∗∗ 4.728∗∗∗ −1.490∗∗∗
(1.240) (0.556) (1.251) (0.557)
ELDER −5.750∗∗∗ −4.493∗∗∗ −5.848∗∗∗ −4.473∗∗∗
(1.281) (0.533) (1.273) (0.531)
AGRICULTURE −0.118 −0.263 −0.197 −0.261
(1.191) (0.661) (1.207) (0.670)
CONSTRUCTION 0.947 0.064 0.437 0.107
(1.603) (0.818) (1.546) (0.810)
MANUFACTURING −0.324 0.173 −0.433 0.185
(1.261) (0.680) (1.280) (0.690)
TRADE −1.589 −0.138 −1.560 −0.136
(1.450) (0.676) (1.471) (0.689)
TRANSPORT 0.831 2.320∗∗ 0.767 2.359∗∗
(2.351) (1.070) (2.392) (1.072)
SERVICES 0.617 0.253 0.594 0.265
(2.039) (0.836) (2.052) (0.852)
PROVISION −0.586 −1.025 −0.649 −1.031
(1.355) (0.718) (1.375) (0.725)
SOCIAL 0.684 0.558 0.679 0.566
(1.596) (0.808) (1.616) (0.818)
GOVERNMENT 0.992 0.220 1.013 0.206
(1.533) (0.737) (1.554) (0.747)
OTHER −2.017 0.043 −2.090 0.056
(1.464) (0.793) (1.474) (0.807)
R-squared 0.639 0.766 0.636 0.764
J-statistic 0.321 0.394 0.369 0.332

errors are in parentheses. Constant included in all regressions. F-statistic 22.437
with PROB_CORRUPT and 22.206 with PROB_CORRUPT_2.
Table B18
The effect corruption and the informal economy have on GDP per-capita and
income per-capita; IV robustness checks.
∗∗
(0.200) (0.064)
PROB_CORRUPT_2 −0.304 0.154
(0.421) (0.135)
SIZE_INFORMAL −1.152∗∗ −0.014 −1.145∗∗ −0.027
(0.519) (0.192) (0.517) (0.191)
POPULATION 0.035 0.048∗∗∗ 0.042 0.048∗∗∗
(0.039) (0.011) (0.040) (0.010)
URBAN −0.096 −0.123∗∗ −0.123 −0.122∗∗
(0.191) (0.061) (0.192) (0.060)
SCHOOL 0.188∗∗∗ 0.176∗∗∗ 0.188∗∗∗ 0.177∗∗∗
(0.056) (0.021) (0.057) (0.022)
DENSITY −0.099∗∗∗ −0.047∗∗∗ −0.101∗∗∗ −0.048∗∗∗
(0.026) (0.009) (0.027) (0.009)
DISTANCE_SP 0.006 −0.027∗ −0.0 0 0 −0.024∗
(0.048) (0.014) (0.048) (0.014)
TEEN −0.931 −1.804∗∗ −0.993 −1.860∗∗
(2.070) (0.826) (2.075) (0.828)
WORK 4.934∗∗∗ 0.142 4.940∗∗∗ 0.139
(1.160) (0.394) (1.167) (0.394)
ELDER −4.842∗∗∗ −3.374∗∗∗ −4.932∗∗∗ −3.376∗∗∗
(1.189) (0.375) (1.192) (0.374)
AGRICULTURE 0.094 0.711 −0.0 0 0 0.727
(1.308) (0.568) (1.340) (0.552)
CONSTRUCTION 2.478 0.764 1.908 0.832
(1.778) (0.699) (1.714) (0.664)
MANUFACTURING −0.149 0.924 −0.265 0.934∗
(1.356) (0.573) (1.391) (0.559)
TRADE −1.931 0.944∗ −1.892 0.935∗
(1.473) (0.574) (1.506) (0.561)
TRANSPORT 1.975 0.883 1.964 0.845
(2.228) (0.868) (2.283) (0.860)
SERVICES 0.337 0.631 0.329 0.621
(1.868) (0.723) (1.902) (0.703)
PROVISION −0.321 0.396 −0.410 0.419
(1.471) (0.604) (1.501) (0.588)
SOCIAL 0.304 1.208∗ 0.312 1.199∗
(1.603) (0.684) (1.641) (0.669)
GOVERNMENT 0.720 1.067∗ 0.723 1.081∗
(1.579) (0.610) (1.612) (0.595)
OTHER −0.961 0.865 −1.030 0.865
(1.580) (0.638) (1.604) (0.623)
R-squared 0.628 0.789 0.625 0.790
J-statistic 0.173 0.672 0.192 0.755

errors are in parentheses. Constant included in all regressions. F-statistic 22.437
with PROB_CORRUPT and 22.206 with PROB_CORRUPT_2.
Table B19
The effect corruption and the informal economy have on GDP per-capita
in 2004 and 2010 additional IV robustness checks where both corruption
and the size of the informal sector are instrumented.
GDP_PC_04 GDP_PC_10

(0.445) (0.511)
PROB_CORRUPT_2
SIZE_INFORMAL −1.566∗∗∗ −1.187∗∗

(0.546) (0.522)
POPULATION 0.040 0.046
(0.040) (0.042)
URBAN −0.013 −0.140
(0.215) (0.195)
SCHOOL 0.168∗∗∗ 0.193∗∗∗
(0.057) (0.055)
DENSITY −0.126∗∗∗ −0.103∗∗∗
(0.030) (0.027)
DISTANCE_SP 0.027 0.004
(0.053) (0.049)
TEEN −3.788∗ −1.243
(2.236) (2.115)
WORK 4.726∗∗∗ 4.933∗∗∗
(1.249) (1.170)
ELDER −5.850∗∗∗ −5.013∗∗∗
(1.291) (1.201)
AGRICULTURE −0.188 −0.022
(1.211) (1.357)
CONSTRUCTION 0.468 1.680
(1.863) (2.114)
MANUFACTURING −0.431 −0.326
(1.280) (1.400)
TRADE −1.565 −1.893
(1.467) (1.507)
TRANSPORT 0.736 1.818
(2.374) (2.293)
SERVICES 0.587 0.285
(2.042) (1.892)
PROVISION −0.634 −0.402
(1.377) (1.513)
SOCIAL 0.673 0.286
(1.614) (1.647)
GOVERNMENT 1.023 0.773
(1.556) (1.628)
OTHER −2.093 −1.088
(1.466) (1.611)
Observations 404 404
R-squared 0.636 0.619
J-statistic 0.373 0.24
Notes: ∗∗∗ indicates statistically significant at 1% level, ∗∗ indicates statisti-

cally significant at 5% level, ∗ indicates statistically significant at 10% level.
Standard errors are in parentheses. Constant included in all regressions.
F-statistic is equal to 18.656.
References
Abed, G.T., Davoodi, H.R., 2002. Corruption, structural reforms, and economic performance in the transition economies. In: Abed, Gupta (Eds.), Governance,
Corruption, and Economic Performance. International Monetary Fund, Washington D.C.
Aidt, T., 2009. Corruption, institutions, and economic development. Oxford Review of Economic Policy 25, 271–291.
Bologna, J. 2015. Corruption: the good, the bad, and the uncertain. SSRN Working Paper (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2379061) (accessed
17.09.15).
Braga, B., Guillén, D., Thompson, B., 2015. Local Government Spending and Employment in Brazil: Evidence from a Natural Regression Discontinuity.
Campos, J., Lien, D., Pradhan, S., 1999. The impact of corruption on investment: predictability matters. World Development 27 (6), 1059–1067.
Choi, J.P., Thum, M., 2005. Corruption and the shadow economy. International Economic Review 46 (3), 817–836.
Cravo, T.A., Resende, G.M., 2013. Economic growth in Brazil: a spatial filtering approach. The Annals of Regional Science 50 (2), 555–575.
Cukier, J., Wall, G., 1994. Informal tourism employment: vendors in Bali. Indonesia. Tourism Management 15 (6), 464–467.
da, Silva, M., F., 1999. The political economy of corruption in Brazil. Revista de Administração de Empresas 39 (3), 26–41.
Donchev, D., Ujhelyi, G., 2014. What do corruption indices measure? Economics and Politics 26 (2), 309–331.
Dreher, A., Gassebner, M., 2013. Greasing the wheels? The impact of regulations and corruption on firm entry. Public Choice 155 (3–4), 413–432.
Dreher, A., Herzfeld, T., 2005. The Economic Costs of Corruption: A Survey and New Evidence. SSRN Working Paper (http://papers.ssrn.com/sol3/papers.cfm?
abstract_id=734184) (accessed 7.07.14).
Dreher, A., Kotsogiannis, C., McCorriston, S., 2009. How do institutions affect corruption and the shadow economy? International Tax and Public Finance 16 (6),
773–796.
Dreher, A., Schneider, F., 2010. Corruption and the shadow economy: an empirical analysis. Public Choice 144 (1), 215–238.
Farrell, D., 2004. The hidden dangers of the informal economy. McKinsey Quarterly 2004, 26–37.
Ferraz, C., Finan, F., 2008. Exposing corrupt politicians: the effects of Brazil’s publicly released audits on electoral outcomes. The Quarterly Journal of Economics
123 (2), 703–745.
Ferraz, C., Finan, F., 2011. Electoral accountability and corruption: evidence from the audits of local governments. American Economic Review 101 (4), 1274–1311.
Geddes, B., Neto, A.R., 1992. Institutional sources of corruption in Brazil. Third World Quarterly 13 (4), 641–661.
Glaeser, E.L., Saks, R.E., 2006. Corruption in America. Journal of Public Economics 90 (6–7), 1053–1072.
Gordon, S.C., 2009. Assessing partisan bias in federal public corruption prosecutions. 103(4), 534–554.
Heckelman, J.C., Powell, B., 2010. Corruption and the institutional environment for growth. Comparative Economic Studies 52 (3), 351–378.
Hibbs Jr., D.A., Piculescu, V., 2005. Institutions, Corruption and Tax Evasion in the Unofficial Economy. University, Göteborg Working Paper.
Higgins, M.J., Levy, D., Young, A.T., 2006. Growth and convergence across the US: Evidence from county-level data. Review of Economics and Statistics 88 (4),
671–681.
Hindriks, J., Keen, M., Muthoo, A., 1999. Corruption, extortion and evasion. Journal of Public Economics 74 (3), 395–430.
Huntington, S.P., 1968. Political Order in Changing Societies. Oxford University Press, New York.
Hussmanns, R., 2004. Measuring the Informal Economy: From Employment in the Informal Sector to Informal Employment. Bureau of Statistics International
Labour Office, Geneva Policy Integration Department Working Paper No. 53.
Johnson, S., Kaufmann, D., Shleifer, A., 1997. The unofficial economy in transition. Brookings Papers on Economic Activity 1997 (2), 159–239.
Porta, R.L., Shleifer, A., 2014. Informality and development. Journal of Economic Perspectives 28 (3), 109–126.
La Porta, R.L., Shleifer, A., 2008. The Unofficial Economy and Economic Development. NBER Working Paper Series No. 14520.
Leff, N.H., 1964. Economic development through bureaucratic corruption. The American Behavioral Scientist 8 (3), 8–14.
Li, H., Xu, L.C., Zou, H., 20 0 0. Corruption, income distribution, and growth. Economics and Politics 12 (2), 155–182.
Loayza, N.A., 1996. The economics of the informal sector: a simple model and some empirical evidence from Latin America. Carnegie-Rochester Conference Series
on Public Policy 45, 129–162.
Loayza, N., Rigolini, J., 2011. Informal employment: safety net or growth engine? World Development 39 (9), 1503–1515.
Mauro, P., 1995. Corruption and growth. Quarterly Journal of Economics 110 (3), 681–712.
Mcmillan, J., Zoido, P., 2004. How to subvert democracy: Montesinos in Peru. Journal of Economic Perspectives 18 (4), 69–92.
Mo, P.H., 2001. Corruption and economic growth. Journal of Comparative Economics 29, 66–79.
Olken, B.A., 2009. Corruption perceptions vs. corruption reality. Journal of Public Economics 93 (7–8), 950–964.
Olken, B.A., 2007. Monitoring corruption: evidence from a field experiment in Indonesia. Journal of Political Economy 115 (2), 200–249.
Olken, B.A., Barron, P., 2009. The simple economics of extortion: evidence from trucking in Aceh. Journal of Political Economy 117 (3), 417–452.
Olken, B.A., Pande, R., 2012. Corruption in developing countries. Annual Review of Economics 4, 479–509.
Pederson, S., 2003. The Shadow Economy in Germany, Great Britain, and Scandinavia: A Measurement Based on Questionnaire Surveys. The Rockwool Foundation
Research Unit, Copenhagen Research Unit Study No. 5.
Reinikka, R., Svensson, J., 2005. Fighting corruption to improve schooling: evidence from a newspaper campaign in Uganda. Journal of the European Economic
Association 3 (2–3), 259–267.
Resende, G.M., 2013. Spatial Dimensions of Economic Growth in Brazil. ISRN Economics 2013. http://dx.doi.org/10.1155/2013/398021.
Rijckeghem, C.V., Weder, B., 2001. Bureaucratic corruption and the rate of temptation: do wages in the civil service affect corruption, and by how much. Journal
of Development Economics 65 (2) 307–331.
Rose-Ackerman, S. 1996. The Political Economy of Corruption: Causes and Consequences. The World Bank Note No. 74.
Schneider, F., 2005. Shadow economies around the world: what do we really know? European Journal of Political Economy 21 (3), 598–642.
Schneider, F., 2007. Shadow economies and corruption all over the world: new estimates for 145 countries. Economics: The Open-Access, Open-Assessment
E-Journal 2007 9 (2), 1–53 http://dx.doi.org/10.5018/economics- ejournal.ja.2007- 9.
Schneider, F., Enste, D.H., 20 0 0. Shadow economies: size, causes, and consequences. Journal of Economic Literature 38 (1), 77–114.
Schneider, F., Williams, C.C., 2013. The Shadow Economy. The Institute of Economic Affairs, London.
Sequeira, S., Djankov, S., 2010. An Empirical Study of Corruption in Ports. MPRA Paper 21791, University Library of Munich, Germany.
Shleifer, A., Vishny, R.W., 1993. Corruption. Quarterly Journal of Economics 108 (3), 599–617.
Stock, J.H., Yogo, M., 2005. Testing for Weak Instruments in Linear IV Regression. NBER Technical Working Paper Series No. 284.
Svensson, J., 2005. Eight questions about corruption. Journal of Economic Perspecives 19 (3), 19–42.
Treisman, D., 2007. What have we learned about the causes of corruption from ten years of cross-national empirical research? Annual Review of Political Science
10, 211–244.
Ulyssea, G., 2010. Regulation of entry, labor market institutions and the informal sector. Journal of Development Economics. 91, 87–99.
Wei, S., 1997. Why is Corruption so Much More Taxing than Tax? Arbitrariness kills. NBER Working Paper No. 6255.
Wiseman, T., 2013. U.S. shadow economies: a state-level study. Constitutional Political Economy 24 (4), 310–335.
Wiseman, T., 2015. Entrepreneurship, Corruption, and the Size of the U.S. Underground Economies. Journal of Entrepreneurship and Public Policy 4 (3), 313–330.

Bologna 2016, The Effect of Informal Employment and Corruption On Income Levels in Brazil

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Bologna 2016, The Effect of Informal Employment and Corruption On Income Levels in Brazil

Enviado por

Direitos autorais:

Formatos disponíveis

Journal of Comparative Economics 44 (2016) 657–695

Contents lists available at ScienceDirect

Journal of Comparative Economics

The effect of informal employment and corruption on income

2. Theoretical considerations & methodology

3. Measuring corruption and the informal sector

4.1. Dependent variables

4.2. Independent variables

PROB_CORRUPT −0.453∗∗ −0.050 −0.053 −0.191∗ −0.353 0.012 −0.029 −0.123

PROB_CORRUPT_2 −0.044 −0.206 −0.246 −0.213 −0.043 −0.104 −0.226 −0.022

5.1. Main results

PROB_CORRUPT −0.499∗∗ 0.006 −0.061 −0.049 −0.311 −0.003 −0.085 0.131

PROB_CORRUPT_2 −0.167 0.124 0.144 −0.191 0.061 0.240∗∗ 0.245∗ 0.332

(1) Chapada de Areia, TO 1270 3924.41 181.31 692.66

(1) Chapada de Areia, TO 1270 3059.76 15.30 227.03

(1) Chapada de Areia, TO 1270 4432.28 226.05 709.61

(1) Chapada de Areia, TO 1270 3767.23 2.26 237.34

GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10

PROB_CORRUPT −0.415∗∗ 0.037 −0.461∗∗ 0.053

PROB_CORRUPT −0.024 0.192

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level,

0.129 Minimum 0.136 0.769 0.339 0.026

0.129 Minimum −0.270 0.616 0.163 0.137

0.129 Minimum 0.884 0.064 0.304 0.033

0.129 Minimum 0.408 0.407 0.092 0.408

Marginal Effect of Corruption on GDP_PC_04 Marginal Effect of Corruption on TOTAL_00

Marginal Effect of Corruption on GDP_PC_10 Marginal Effect of Corruption on TOTAL_10

5.2. The effect of corruption contingent upon informal sector size

Tables A1 and A4.

Variable Description Mean Std. dev. Min Max

Alternative measures of corruption

NUM_CORRUPT Number of violations found to involve corrupt. 1.931 1.707 0 10

Panel a: Effects in the ﬁrst period.

GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00

NUM_CORRUPT −0.013 −0.003

Panel b: Effects in the second period.

GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10

NUM_CORRUPT −0.005 0.005

Panel a: Effects in the ﬁrst period.

GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00 GDP_PC_04 TOTAL_00

PROB_CORRUPT_AV −0.482 −0.010

Panel b: Effects in the second period.

GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10 GDP_PC_10 TOTAL_10

PROB_CORRUPT_AV −0.400 0.054

GDP_PC_04 TOTAL_00 GDP_PC_10 TOTAL_10

PROB_CORRUPT −0.326 0.032 −0.286 0.008

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

Tables B1 and B19.

GDP_PC_04 TOTAL FORMAL INFORMAL

PROB_CORRUPT −0.453∗∗ −0.050 −0.053 −0.191∗

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

GDP_PC_04 TOTAL FORMAL INFORMAL

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

GDP_PC_04 TOTAL FORMAL INFORMAL

PROB_CORRUPT −0.353 0.012 −0.029 −0.123

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

GDP_PC_04 TOTAL FORMAL INFORMAL

PROB_CORRUPT −0.475 0.015 −0.072 −0.062

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

GDP_PC_04 TOTAL FORMAL INFORMAL

PROB_CORRUPT_2 −0.044 −0.206 −0.246 −0.213

Notes: ∗∗∗ indicates statistically signiﬁcant at 1% level, ∗∗ indicates statistically

GDP_PC_04 TOTAL FORMAL INFORMAL