why is corruption less harmful to income inequality in latin america?

World Development Vol. 40, No. 8, pp. 1534–1545, 2012� 2012 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matter

www.elsevier.com/locate/worlddevhttp://dx.doi.org/10.1016/j.worlddev.2012.04.015

Why is Corruption Less Harmful to Income Inequality

in Latin America?

STEPHEN DOBSONUniversity of Hull, UK

and

CARLYN RAMLOGAN-DOBSON *

Nottingham Trent University, UK

Summary. — Conventional wisdom says corruption is bad for income inequality. But recent research on Latin America finds a trade-offbetween corruption and inequality and suggests this is due to the large informal sector in the region. Using data on a large sample ofcountries we find that the informal sector impacts the link between corruption and inequality. In particular, the marginal impact of cor-ruption becomes negative once the informal sector becomes large. This is true in Latin America and more generally. Corruption reducingpolicies should be accompanied by measures that help displaced informal sector workers.� 2012 Elsevier Ltd. All rights reserved.

Key words — corruption, Latin America, income inequality, informal sector, marginal impact

1. INTRODUCTION

Conventional wisdom says corruption is bad for incomeinequality. Numerous empirical studies report a positive rela-tionship between corruption and inequality: more corruptionleads to more inequality (e.g., Ades & Di Tella, 1997; Dincer& Gunlap, 2008; Gupta, Davoodi, & Alonso-Terme, 2002;Gyimah-Brempong & Munoz de Camacho, 2006; Li, Xu, &Zou, 2000). These studies rationalize the result by suggestingseveral avenues through which corruption is expected to havean adverse impact on inequality (for details see Andres andRamlogan-Dobson (2011)). Empirical research also points toa feedback relationship between corruption and inequality,which may help to create a corruption–inequality trap(Apergis, Dincer, & Payne, 2010; Chong & Gradstein,2007a; Uslander, 2007).

That corruption may be harmful to inequality has also beendemonstrated in theoretical work. Blackburn and Forgues-Puccio (2007) model the behavior of bureaucrats appointedto implement income redistribution programs to help thepoor. The model predicts that bureaucrats will conspire withthe rich in providing the authorities with false information.In the macroeconomic model of Foellmi and Oechslin (2007)corruption redistributes income toward the wealthy. This out-come arises because wealth serves as collateral in the determi-nation of how much can be borrowed on the (imperfectlyfunctioning) credit market. Payment of bribes reduces the nec-essary collateral so the lower the bribe the smaller is theamount of money that can be borrowed. Entrepreneurship isnot an option for the poorest in society since they can onlygenerate low levels of collateral.

Conventional wisdom has recently been questioned inempirical research on Latin America. Dobson and Ramlo-gan-Dobson (2010) and Andres and Ramlogan-Dobson(2011) provide evidence of a trade-off between corruption

1534

and income inequality. They explain this result with refer-ence to the (relatively large) informal sector in the region.The poorest individuals lack the personal characteristics re-quired to find work in the formal economy, while discrimi-nation and institutional barriers also restrict workopportunities. The informal sector therefore provides jobsand a source of income. Policies aimed at reducing corrup-tion impose labor market (and other) regulations and havean adverse impact on employment and welfare in the infor-mal sector. If this hypothesis is correct anti-corruption mea-sures introduced by governments and advocated byorganizations such as the World Bank may exacerbateinequality. 1 For this reason alone it is important to under-stand more about the corruption–inequality–informal sectorrelationship. 2

To the best of our knowledge this is the first paper to exam-ine empirically the link between inequality, corruption, andthe informal sector. Our key finding is that the informal sectorimpacts the trade-off between corruption and inequality. Inparticular, as the informal sector grows corruption is lessharmful to inequality. This result is important for three rea-sons: (i) it lends credence to the findings for Latin America;(ii) it applies to any country where the informal sector is large;and (iii) it suggests that anti-corruption policies need to beaccompanied by initiatives which help displaced workers fromthe informal sector.

The rest of the paper is structured as follows. Section 2considers the theory to support the hypothesis that the infor-mal sector impacts the relationship between corruption andinequality. Section 3 explores some empirical issues andSection 4 discusses the data and methodology. Section 5reports the results and Section 6 concludes.

* Final revision accepted: January 30, 2012

http://dx.doi.org/10.1016/j.worlddev.2012.04.015

WHY IS CORRUPTION LESS HARMFUL TO INCOME INEQUALITY IN LATIN AMERICA? 1535

2. CORRUPTION, INEQUALITY, AND THEINFORMAL ECONOMY

Poor institutions provide the impetus for the growth of aninformal sector. A weak legal system, restrictive labor regula-tions, excessive bureaucracy, social security contributions, anda heavy tax burden are all major disincentives for entrepre-neurs to start up a business in the formal economy (Dabla-Norris, Gradstein, & Inchauste, 2010; Feige, 1996; Schneider,2007; Tanzi, 1982). A business operating “informally” avoidstaxation and other contributions. Some would-be formal sec-tor entrepreneurs often find it necessary to bribe officials toobtain relevant documents. 3 De Soto (1989) provides anexample from Peru in the 1980s when entrepreneurs were ex-pected to pay a bribe for ten of the eleven requirements for set-ting up a business. In a cross-sectional study Chong andGradstein (2007b) find an inverse relationship between institu-tional quality and the informal sector: an improvement in thequality of institutions is associated with a decline in the infor-mal sector.

In developing countries 40–50% of the labor force andaround 50% of GDP originates in the informal sector (Schnei-der, Buehn, & Montenegro, 2010). The continued existence ofinformal firms comes from their ability to avoid detection or,if found out, to bribe officials to overlook their existence orobtain necessary documentation. Over time individuals andfirms build-up a network of contacts so as to enable continuedoperation. Furthermore, it is possible that the networks andchannels that facilitate corruption become stronger as moreindividuals engage in this activity and as a result it become lessexpensive to engage in corruption (Cule & Fulton, 2009). Theidea that corruption can assist in building an economy and bean effective substitute for poor institutions and missing law issummarized well by de Vaal and Ebben (2011, p. 112). Theysay “when a decent system of property rights is missing cor-ruption may become a crucial element of the economic system.In such an environment corruption could reduce uncertaintyand facilitate investment and productivity thus providing analternative system in which the indirect institutional effect ofcorruption more than compensates its negative direct effect.”But this notion is not new. Leff (1964) and Huntington(1968) speculate that corruption may be perceived as a usefulsubstitute for a weak rule of law, while Osterfeld (1992) notesthat corruption allows private citizens to evade bad laws.

If there are weak institutions the structure of productionadapts and mechanisms are developed that allow business towork in a corrupt environment. Corruption is not foughtagainst or regarded as immoral and for firms in the formaleconomy it may be the only way to circumvent cumbersomeand pervasive regulation that undermines efficiency andoutput. For example, in Latin America under protectionist re-gimes individuals bribed officials to obtain import licenses andto overlook foreign exchange controls (Franko, 2006). Arduz(2000) describes a system in Bolivia where goods are processedon the basis of customs officers setting their own tax ratesrather than using official ones. In many instances firms inthe formal sector may depend on firms in the informal sectorfor accessing essential inputs. Some “formal” firms may onlyregister part of their workforce or part of their sales or onlydeclare part of the salaries of their employees. Survey evidencefor Latin America suggests that medium to large-sized firmshave a substantial proportion of their operations off the books(Perry et al., 2007).

The nature of a region’s colonial heritage seems to be impor-tant in explaining why corruption is tolerated and why this tol-erance may benefit business. Engerman and Sokoloff (1997,

2000, 2002, 2005) and Acemoglu, Johnson, and Robinson(2002) trace the roots of current Latin American inequalityback into the colonial era when profitable activities were con-trolled by a ruling class (or “landed elite”) who set up institu-tions to serve their own ends. The concentration of land,mineral wealth, and political power in the hands of this eliteinduced a long run development path characterized by higheconomic and social inequality. Following independence, theCreole elite gained control of key institutions and exerted asignificant influence on the formation and implementation ofgovernment policies. The fact that corruption is embeddedinto the historical development of Latin America has giveneconomic agents time to develop ways of prospering withina corrupt environment. Hence, the adverse impact on inequal-ity is reduced.

Initial conditions are therefore crucial to the success of anti-corruption policies. If policy makers are in direct conflict withindividuals who have successfully developed businesses usingthe means at hand policy is unlikely to work in the shortrun. Firms and individuals operating partially or totally inthe informal sector face higher costs via improvements in taxcollection, new regulations and procedures, changes in person-nel, policing, and enforcement. 4 Entrepreneurs find it increas-ingly difficult to survive because the things that gave impetusto the growth of the informal sector are eroded and establishednetworks that once facilitated production begin to disappear.The effect of these measures is a reduction in the size of theinformal sector, a loss of jobs and income, and a subsequentrise in income inequality. In short, the impact of anti-corrup-tion policy in countries with weak institutions will not be thesame as in countries with more robust institutions.

Recent theoretical research corroborates this view. Albrecht,Navaroo, and Vroman (2009) show that labor market re-forms, via the imposition of certain taxes, lead to a rise in totalunemployment in an economy with a significant informal sec-tor. Defining corruption as a set of activities that smooth theprocess of transaction in the production sectors, Mandal andMarjit (2010) use a general equilibrium model to show thatlower corruption can produce a rise in wage inequality. Ulys-sea (2010) develops a two-sector matching model that incorpo-rates the main features of labor markets in Latin America.Simulations of the model using data for Brazil show that theenforcement of labor market regulations reduces the size ofthe informal sector, significantly increases unemployment,and produces substantial welfare losses. Chong and Calderon(2000) show empirically that the effect of corruption oninequality is nonlinear with an initial worsening followed byan improvement. They hypothesize that this is due to the pres-ence of the informal sector.

The argument that anti-corruption policies will have ad-verse consequences for business in the informal sector is chal-lenged by Maloney (2004). He argues that since the self-employed represent the core of the informal sector, labormarket regulation will have only minimal impact on theinformal sector. This may be the case if we think of anti-cor-ruption measures in terms of labor market regulation only.But if we see corruption more broadly, involving an ex-change between two parties which influences the allocationof resources either in the present or the future (Macrae,1982), a crackdown on corruption puts the squeeze on allparticipants (self-employed or not) in the informal sector.Moreover, even if the self-employed are the core of the infor-mal sector it is possible, as theoretical studies have shown,for labor market regulation to adversely affect the size ofthe informal sector. The implication of this is that the issueis an empirical one.

1536 WORLD DEVELOPMENT

A positive role for corruption has also been found in re-search on the impact of corruption on growth, efficiency,and trade. Sarte (2000) finds that the impact of corruptionon growth depends on the ability of the informal sector to sub-stitute for production in the formal economy. This in turn de-pends on the cost of operating informally. If the cost ofinformality is low, growth is higher. Houston (2007) finds thatin countries with poor institutions corruption helps to expandoutput and he concludes that corruption should not be indis-criminately attacked in countries that are poorly governed.Aidt, Dutta, and Sena (2008) show that corruption has a det-rimental impact on growth in countries with good governanceand the opposite effect where there is poor governance, whilede Vaal and Ebben (2011) show that in countries where insti-tutions are not well developed corruption may be conducive togrowth. Centralized corruption removes any uncertainty andproducers know whom to bribe in order to secure production,while in a decentralized system of corruption economic agentsdo not know whether their bribe will be effective. Meon andSekkat (2005) and Meon and Weil (2009) find no evidenceto support the idea that corruption has a detrimental impacton efficiency, rather the evidence supports the hypothesis thatcorruption “greases the wheels.” Blackburn and Forgues-Puc-cio (2009) show that if there is an organized corruption net-work innovation and growth is higher than when such anetwork is missing, and Dutt and Traca (2010) find that in ahigh tariff environment the marginal impact of corruption istrade-enhancing.

Initial conditions in a country also matter in work that ex-plores the impact of trade liberalization on poverty andinequality. According to Heckscher–Ohlin and the Stolper–Samuelson theorem trade liberalization should increase the de-mand for unskilled labor intensive activities in developingcountries and hence the demand for unskilled labor. Thisshould result in a fall in poverty and inequality. However,trade reforms in Latin America are associated with risinginequality and poverty. Perry and Olarreaga (2006) explainthis “puzzle” in terms of initial conditions. They argue thatthere is no puzzle because the initial conditions were such thata reduction in the demand for labor should have been ex-pected: liberalization reduced the level of protection affordedto unskilled labor intensive sectors and so reduced the demandfor labor. The authors conclude that the impact of trade re-forms will be different from that expected in perfectly func-tioning markets.

The literature on the relationship between corruption andthe informal sector is also relevant since the interaction ofthese two variables may be expected to influence inequality.A key question in this literature asks are corruption and theinformal sector substitutes or complements? Theoretical workshows that both types of relationship may hold. Choi andThum (2005) argue that the informal sector limits corruptionas it gives business an outside option of going underground.For example, if entrepreneurs are required to purchase alicense from a corrupt official in order to open a business inthe formal economy, the propensity of officials to extractbribes is reduced if the entrepreneur has the option of avoidingthe purchase and operating in the informal sector. Thus, eco-nomic activity in the formal sector is enhanced and corruptionand the informal sector are substitutes. The theoretical modelof Dreher, Kotsogiannis, and McCorriston (2009) shows thatcorruption and the informal sector are substitutes since theinformal sector reduces the propensity of officials to securebribes from firms. In contrast, Friedman, Johnson, Kauf-mann, and Zoido-Lobaton (2000) argue that corruption leads

to more informality because entrepreneurs have to operateunderground to avoid the predatory and corrupt behavior ofgovernment officials. Hindriks, Muthoo, and Keen (1999) alsosuggest the informal sector is a complement to corruption. Intheir model the tax payer colludes with the tax inspector insuch a way that the inspector is bribed to underreport the truetax liability.

Much of the empirical evidence supports the idea that cor-ruption and the informal sector are complements (e.g., Fried-man et al., 2000; Johnson, Kaufmann, & Zoido-Lobaton,1998). Dreher and Schneider (2010) suggest that the relation-ship between corruption and the informal economy may de-pend on the level of income. In a cross-sectional analysis of98 countries they show that corruption and the informal sectorare complements in low income countries and substitutes inhigh income countries. The authors also show that the rela-tionship is not robust when perception based indices of cor-ruption are used. Dreher et al. (2009) attempt to overcomethe problems associated with perception based measures ofcorruption by using structural equation modeling to capturethe link between institutional quality, corruption, and theinformal sector. Their empirical finding is that institutionalquality, under certain conditions, reduces both the size ofthe corruption market and the size of the informal sector.

In summary, weak institutions provide the basis for corrup-tion and give impetus to the growth of an informal sector. Theinformal sector substitutes for production in the formal econ-omy. Corrupt acts serve as a mechanism for overcoming insti-tutional barriers and play a positive role in facilitatinginvestment, employment, and production in the informal sec-tor. Anti-corruption policies may therefore do more harm toinequality than good.

The above discussion informs the three principal hypothesesthat we test in the empirical section. The hypotheses are: (i)higher corruption leads to higher inequality: the marginal im-pact of corruption is positive; (ii) higher corruption leads tolower inequality in Latin America: the marginal impact of cor-ruption is negative; (iii) the impact of corruption on inequalitydepends on the size of the informal sector: if there is no infor-mal sector the marginal impact of corruption is positive (sameas (i)); as the informal sector grows the marginal impactbecomes less positive and may become negative once the infor-mal sector becomes large (this explains (ii)).

3. EMPIRICAL ISSUES

We begin with a model that is designed to test hypotheses (i)and (ii). Initially we estimate the following equation:

Inequalityi ¼ a0 þ a1Ci þ a2DLA þ a3Ci � DLA

þ other determinants: ð1ÞIn (1) the dependent variable is a measure of inequality incountry i, C represents an index of corruption, with a highervalue implying more corruption and DLA is a dummy variablefor Latin America (equal to 1 for Latin America, 0 otherwise).The “other determinants” are explanatory variables usedwidely by researchers when studying inequality (see examplesin Andres & Ramlogan-Dobson, 2011; Dobson & Ramlo-gan-Dobson, 2010; Gupta et al., 2002; Gyimah-Brempong &Munoz de Camacho, 2006).

If hypothesis (i) holds a1 > 0 (higher corruption leads tohigher inequality). Differentiating (1) with respect to corrup-tion gives:


@Inequalityi

@Ci¼ a1 þ a3 � DLA: ð2Þ

Eqn. (2) shows the marginal impact of corruption. If LatinAmerica is different the marginal impact of corruption oninequality will be negative: more corruption reduces inequal-ity. For hypothesis (ii) to hold we require a3 < 0 in (2) andits absolute value must be larger than a1 (so overalla1 + a3 < 0). Estimating (1) can tell us whether Latin Americais different to the rest of the world but it does not give any spe-cific insight as to the reason for the difference since the inter-action term is a “catch all” for general differences.

To examine the role of the informal sector (and test hypoth-esis (iii)) we adapt (1) by incorporating informal sector vari-ables. We estimate the following:

Inequalityi ¼ b0 þ b1Ci þ b2DLA þ b3Ci � DLA þ b4Ci

� I i þ b5I i þ other determinants: ð3ÞIn (3) I is a measure of the size of the informal sector. The im-pact of the informal sector on inequality is uncertain. On theone hand a larger informal sector may have a positive impactif it provides an opportunity for otherwise jobless people toobtain work. On the other hand if it attracts people whowould have been employed in the formal sector the impacton inequality may be negligible. Furthermore, if the informalsector has an impact on other macroeconomic variables, forexample saving, investment, and growth, there will be knockon effects on inequality (positive and negative). There is alsothe possibility that the impact of the informal sector mayswitch from positive (inequality increases) to negative(inequality falls), consistent with a Kuznets-type relationship.

Differentiating (3) with respect to corruption:

@Inequalityi

@Ci¼ b1 þ b3 � DLA þ b4 � I i: ð4Þ

Eqn. (4) is the marginal impact of corruption on inequalitywhen the informal sector is in the model. It is comprised ofthe coefficients on both interaction terms in (3). If the informalsector explains why Latin America is different, as suggested inhypothesis (iii), two conditions in (4) must be met: (i) b3 < 0and its absolute value should not exceed b1 (b1 + b3 > 0);(ii) b4 < 0. Condition (i) says the marginal impact of corrup-tion on inequality in Latin America is positive but lower thanthe rest of the world. Condition (ii) says the marginal impactof corruption becomes less positive as the size of the informalsector increases (since b4 < 0). It follows from (4) that for anycountry in Latin America the marginal impact of corruptionwill be positive if there is no informal sector, less positive asthe size of the informal sector increases, and may become neg-ative as the informal sector becomes large.

For nonLatin American countries b3 in (4) is not relevant.For these countries we expect b1 > 0 (if hypothesis (i) holds).If this is so and if b4 < 0 then for a country with a small infor-mal sector the marginal impact of corruption will be positive.However, as the informal sector grows larger the positive mar-ginal impact will become smaller and may become negative ifthe informal sector becomes large.

4. DATA AND ESTIMATION METHODS

The empirical analysis is based on a large sample of devel-oped and developing countries. 5 Inequality is initially mea-sured with the Gini coefficient. Data are drawn from theUnited Nations World Income Inequality Database (WIID)

(UNU-WIDER, 2005). 6 The measure of corruption is thewidely used International Country Risk Guide (ICRG) cor-ruption index. The ICRG measure takes values from zero(most corrupt) to six (least corrupt). Other explanatory vari-ables are: real output per capita (lgdp), primary (primary)and secondary (secondary) gross school enrollment rates,domestic credit to the private sector as a ratio of GDP (dcps),openness of the economy measured as the ratio of export plusimport to GDP (openness), the share of agriculture in totaloutput (aggdp), and inflation (inflation). Data are obtainedfrom Penn World Tables 6.3 7 and World Development Indi-cators 2010. Data on the informal sector are taken fromSchneider (2004, 2007) and Buehn and Schneider (2007). Theinformal sector (shadow economy) is measured in terms ofpercentage of “official” GDP. 8 Data from Schneider havebeen used in other studies (e.g., Laoyza, Oviedo, & Serven,2005) as they are the most comprehensive estimates to havebeen obtained using a unified method. 9 In the cross-sectionestimation we use the average value of each variable overthe period 2000–2004/5.

As a robustness check we use the share of income in thelowest quintile (quintile 1) as an alternative measure ofinequality, and the corruption perception index (CPI) 10 asan alternative measure of corruption. The CPI takes valuesfrom 0 (most corrupt) to 10 (least corrupt). In the estima-tions, the values of both the ICRG and the CPI measuresof corruption are rebased so a larger value indicates a high-er level of corruption. Instrumental variable estimation isalso used to deal with potential endogeneity. The literaturepoints to several potential instruments for the corruptionvariable (or some measure of governance in a country)when using cross-section data. These include lagged vari-ables (Li et al., 2000); a black market premium, governmentspending on defense as a percentage of GDP, and the leg-islative tradition of the country (Chong & Calderon, 2000);democracy, initial real per capita income, latitude of acountry, ethnicity, initial corruption, the share of publicemployment in the labor force, and the ratio of governmentspending to GDP (Gupta et al., 2002); and ethno-linguisticfractionalization and mortality rate of colonial settlers (Gyi-mah-Brempong, 2002; Gyimah-Brempong & Munoz deCamacho, 2006).

An additional robustness test is provided by panel estima-tion over the period 2000–2004/5. 11 We do random effects(RE) estimation for several reasons. First, a key variableof interest (Latin America dummy) cannot be included ifwe use a fixed effects model; second, our sample containslimited observations for many of the cross-section units;and third, some explanatory variables change very slowlyfor countries and may be highly collinear with the fixed ef-fects. The RE model may be criticized for assuming that theunobserved effect is uncorrelated with the explanatory vari-ables. For this reason, we include a large number of time-constant controls (Woolridge, 2010) and we also includeyear dummies.

Summary statistics are provided in Table 1 for all coun-tries and separately for Latin American countries in thesample. Figure 1 illustrates the relationship between inequal-ity and corruption (using CPI in its original form: a highervalue indicates lower corruption). As the CPI increases, theGini coefficient falls: lower levels of corruption are associ-ated with lower levels of inequality. Figure 2 shows thatan increase in the size of the informal sector is associatedwith more corruption, while in Figure 3 we see that coun-tries with larger informal sectors are associated with higherlevels of inequality.

Table 1. Summary statistics

Gini coefficient Informal sector (% of GDP) Corruption (CPI) Corruption (ICRG) Quintile 1

All countries

Mean 41.2 34.91 4.05 2.70 6.11Standard deviation 10.57 13.09 2.16 1.14 2.38Maximum 73.9 67.74 9.77 6 12.60Minimum 21.9 8.38 1.22 0.21 1.07No. of countries 137 140 160 139 141

Latin America

Mean 53.7 45.3 3.33 2.50 3.83Standard deviation 4.7 13.21 1.34 0.56 0.99Maximum 61.8 67.74 7.42 3.62 5.60Minimum 44.2 20.14 1.75 1.31 2.38No. of countries 21 21 21 21 21

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Gin

i coe

ffic

ient

Corruption index (CPI)

Figure 1. Inequality and corruption.

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

0 10 20 30 40 50 60 70

Gin

i coe

ffic

ient

Size of the informal sector (% GDP)

Figure 3. The informal sector and inequality.

0.01.02.03.04.05.06.07.08.09.0

10.0

0 10 20 30 40 50 60 70 80

Cor

rupt

ion

inde

x (C

PI)


Figure 2. The informal sector and corruption.


Table 2. OLS estimation results (Eqn. (1))

Explanatory variables Dependent variable

(1) Gini coefficient (2) Gini coefficient (3) Quintile 1

lgdp �7.3284*** �3.3583 0.1200(0.002) (0.250) (0.783)

primary 0.0669 �0.0196*

(0.288) (0.101)secondary �0.1021** �0.1037** 0.0215*

(0.040) (0.044) (0.095)aggdp �0.4992** �0.2840* 0.0145

(0.012) (0.055) (0.262)dcps 0.0568*** 0.0425* �0.0202***

(0.006) (0.087) (0.004)openness �0.0254 0.0049

(0.112) (0.277)inflation 0.1762***

(0.000)corrupt (ICRG) 2.3160** �0.9488***

(0.046) (0.004)corrupt � ladum (ICRG) �4.0300** 1.0660**

(0.033) (0.021)corrupt (CPI) 1.3915**

(0.050)corrupt � ladum (CPI) �2.7047***

(0.007)ladum 6.2076** 5.6973** �1.4989**

(0.033) (0.052) (0.044)constant 114.0368*** 81.5739*** 6.2381***

(0.000) (0.001) (0.000)R2 0.601 0.571 0.451No. of observations 98 112 93

p values in parentheses are based on heteroskedastic-robust standard errors.* Significant at 10%.

** Significant at 5%.*** Significant at 1%.


5. RESULTS

Table 2 presents the OLS results of estimating Eqn. (1). TheICRG measure of corruption is used in columns (1) and (3)and the CPI is used in column (2). In column (3) the shareof income of the lowest quintile is the dependent variable. Ta-ble 2 shows that a rise in corruption results in a rise in the Ginicoefficient (corrupt is positive and significant in columns (1)and (2)) and a reduction in the share of income of the lowestquintile (corrupt is negative and significant in column (3)). Thissupports the conventional wisdom that corruption is bad forinequality. The dummy variable for Latin America (ladum)is positive (negative for quintile 1) and significant indicatingthat the constant in Latin America is significantly higher thanthe rest of the world (lower for quintile 1). 12 The interactionterm is significant and negative (positive for quintile 1) andits absolute value is larger than the corruption parameter. 13

For any Latin American country, since a1 + a3 < 0, the mar-ginal impact of corruption is negative.

The finding that the impact of corruption on inequality isdifferent in Latin America is consistent with the result inDobson and Ramlogan-Dobson (2010) and Andres andRamlogan-Dobson (2011), and provides support for hypothe-sis (ii). The results in column (1) of Table 2 show that themarginal impact of corruption in Latin America is2.3160 + (�4.0300 � 1) = �1.714. That is, a one unit rise incorruption (ICRG) leads to a fall in the Gini coefficient byapproximately 1.7 (higher corruption is associated with lower

inequality). Outside of Latin America a one unit rise in cor-ruption leads to a rise in the Gini coefficient by approximately2.3 (higher corruption leads to higher inequality). If the CPImeasure of corruption is used (column (2)) the marginal im-pacts are a little smaller: the Gini coefficient falls by 1.3 in La-tin America and rises by 1.4 in nonLatin American countries.

The results of estimating the model with the informal sectorincluded (Eqn. (3)) are shown in Table 3. The coefficient on thecorruption–Latin America interaction term now loses its sig-nificance. It may be the case in Table 2 that the interactionterm is acting as a “catch all” variable, not precise enoughto explain inequality. This result indicates that the informalsector alters the relationship between corruption and inequal-ity and is supportive of the intuitive argument in Dobson andRamlogan-Dobson (2010) and Andres and Ramlogan-Dob-son (2011). 14 Since the emergence of an informal sector isinextricably linked to corruption and since the informal sectorprovides a source of income and jobs for the poor, it is to beexpected that the informal sector has an influence on inequal-ity (see Section 2). This implies that models explaininginequality will be misspecified if the informal sector is ex-cluded.

The corruption–informal sector interaction term in Table 3 isnegative (b4 < 0) and significant (positive for quintile 1) there-fore indicating that the marginal impact of corruption oninequality falls as the informal sector becomes larger. We canuse the results in column (1) to illustrate. For a country withan informal sector around 12% of GDP the marginal impact

-8

-6

-4

-2

0

2

4

6

0 10 20 30 40 50 60 70 80 90 100

Gin

i coe

ffci

ent


Figure 4. Marginal impact of corruption on inequality.

Table 3. OLS estimation results (Eqn. (3))

Explanatory variables Dependent variable

(1) Gini coefficient (2) Gini coefficient (3) Quintile 1

lgdp �5.2629** �3.2906 0.1176(0.017) (0.282) (0.754)

primary 0.0474(0.447)

secondary �0.1054** �0.1238** 0.0121*

(0.038) (0.033) (0.078)aggdp �0.3595**

(0.050)dcps 0.0707*** 0.0532* �0.0258***

(0.001) (0.075) (0.001)openness �0.0305* 0.0055

(0.085) (0.180)inflation �0.0033* �0.0142*

(0.090) (0.064)informal �0.0645 �0.0416 0.0092

(0.479) (0.617) (0.687)corrupt � informal (ICRG) �0.1086** 0.0349*

(0.042) (0.069)corrupt (ICRG) 4.1059*** �1.2837***

(0.002) (0.002)corrupt � ladum (ICRG) �3.4715 0.5896

(0.107) (0.214)corrupt � informal (CPI) �0.0450**

(0.051)corrupt (CPI) 1.8216**

(0.029)corrupt � ladum (CPI) 1.21677

(0.471)ladum 9.0431*** 14.1867*** �1.8154***

(0.008) (0.003) (0.003)constant 44.8006*** 47.6357*** 5.5879***

(0.000) (0.002) (0.000)R2 0.568 0.501 0.574No. of observations 98 113 94


** Significant at 5%.*** Significant at 1%.


of corruption on inequality is 2.8027—a one unit rise in corrup-tion (ICRG) increases the Gini coefficient by almost three (high-er corruption is associated with higher inequality). For acountry where the average size of the informal sector is about

45% of GDP the marginal impact is �0.7811—a one unit risein ICRG reduces the Gini coefficient by close to one (higher cor-ruption is associated with lower inequality). 15 Clearly, the mar-ginal impact of corruption depends on the size of the informal

Table 4. IV and panel estimation results

Explanatory variables Dependent variable: Gini coefficient

(1) IV (2) LIML (3) LIML (4) RE (5) RE

lgdp �2.9276 �1.4757 �1.3579 �2.7319** �2.9609***

(0.171) (0.552) (0.551) (0.023) (0.000)primary 0.0881 0.0859 0.0045

(0.156) (0.207) (0.888)secondary �0.1331*** �0.1309*** �0.0511

(0.003) (0.006) (0.322)aggdp �0.2835** �0.1991

(0.047) (0.231)dcps 0.0789** 0.0944** 0.1217*** 0.0387*** 0.0462**

(0.024) (0.017) (0.002) (0.002) (0.000)openness �0.0301 �0.0169 �0.0794*

(0.132) (0.153) (0.080)inflation 0.1634***

(0.001)informal 0.0705 �0.1262 �0.1818 0.1538* 0.1303*

(0.239) (0.281) (0.275) (0.061) (0.070)corrupt � informal (ICRG) �0.2719** �0.0304**

(0.042) (0.041)corrupt (ICRG) 10.6505** 1.3617**

(0.053) (0.042)corrupt � ladum (ICRG) �1.3509 �0.4479

(0.821) (0.300)corrupt � informal (CPI) �0.0826** �0.1302** �0.0315**

(0.032) (0.031) (0.0511)corrupt (CPI) 3.4774** 5.0467** 1.2905**

(0.042) (0.022) (0.031)corrupt � ladum (CPI) �1.7938 0.0155 �0.0794

(0.286) (0.795) (0.911)ladum 7.6281*** 7.3596** 11.8893** 11.5218*** 12.38447***

(0.001) (0.020) (0.036) (0.000) (0.000)constant 67.5599*** 61.4746*** 58.7861** 56.4635*** 59.7713***

(0.001) (0.005) (0.016) (0.000) (0.000)Basmanb 0.3877 1.2889

(0.7622) (0.2799)Anderson–Rubin F test 1.83*

(p-value) (0.075)Sargan testa 2.1086(p-value) (0.8344)Cragg–Donald statistic 24.846R2 0.496 0.507 0.606 0.551 0.677No. of observations 93 109 90 304 344No. of cross-sections 91 89


** Significant at 5%.*** Significant at 1%.a Chi-sq(5): democracy, latitude, and ethno-linguistic fractionalization.b Chi-sq(3): democracy, latitude, military expenditure, and ethno-linguistic fractionalization. We also experimented with other instruments such asladum * democracy but the results were unchanged.


sector and the impact switches from positive to negative oncethe informal sector becomes large. A similar result is foundusing the numbers in columns (2) and (3) of Table 3. Figure 4illustrates the marginal impact using the results in column (1).The marginal impact of corruption on inequality declines asthe informal sector grows and ceases to be positive for countrieswith an informal sector that is greater than 37% of GDP. Thisrepresents 64 countries (all developing) in our study, almosttwo thirds of the overall sample.

Before considering the IV estimation results in Table 4 it isnecessary to mention a few technical issues. Instrumental vari-ables must satisfy two requirements for asymptotic consis-tency. They must be correlated with the endogenousvariables and orthogonal to the error term. In line with thesecriteria, researchers undertake an examination of the firststage F statistics and perform a test for over-identification.However, recent literature on weak instruments has shownthat these diagnostics may not be adequate. Therefore we also


use the Cragg–Donald (C–D) statistic among other diagnos-tics to decide whether or not the instruments are weak.

We instrument for corruption so that corrupt andcorrupt � ladum become endogenous regressors in column (1)of Table 4. Based on the Sargan statistic we are unable to re-ject the null hypothesis that the instruments are uncorrelatedwith the error term and so we can conclude that the instrumentset is valid. The C–D statistic suggests that the instruments areacceptable in the sense that they are not weak (Stock & Yogo,2005 provide a set of critical values for the C–D statistic).More specifically, the C–D statistic of 24.85 is greater than11.04 (critical value for 5% bias) and we conclude that themaximal bias is less than 5% of the OLS bias. The discrepancybetween the partial R2 and the Shea partial R2 is small (not re-ported). We also perform the Anderson–Rubin Wald F test(1949) test to check the null hypothesis that the coefficientsof the endogenous regressors in the structural equation arejointly equal to zero. The null hypothesis is rejected at the10% level. Given the diagnostics, we accept these results.

If we use informal as an exogenous regressor and cor-rupt � informal as an endogenous regressor there is a weakinstrument problem (result not presented). In the presence ofweak instruments, IV estimates can exhibit severe finite-sam-ple bias and the finite-sample distribution can be very differentfrom the asymptotic distribution, which distorts the size oftests and the coverage of confidence intervals (Stock & Yogo,2005). As an alternative we generate the point estimates of theparameters using Fuller’s (1977) modified limited informationmaximum likelihood (LIML) estimator. As a test for the jointvalidity of the instrument set we report the Basman statistic.The LIML results are presented in columns (2) and (3) of Ta-ble 4 for both the CPI and ICRG measures of corruption.

As well as adopting the LIML estimation procedure, onemay also deal with the weak instrument problem by usinglagged values of the endogenous variables. In this instancewe use lagged values for corrupt and informal 16 when specify-ing the corrupt � informal interaction term and the informalvariable. 17 We then proceed to treat the interaction term asan exogenous variable (see Table 4 column (1)). The relevantdiagnostics for the model are respectable.

The IV results in Table 4 support the results obtainedusing OLS. In column (1) the corruption parameter is po-sitive and significant (corruption is bad for inequality).The corruption–Latin America interaction term is insignifi-cant (as in Table 3) while the coefficient on the corrup-tion–informal sector interaction term is negative andsignificant (as in Table 3). The LIML results, presented incolumns (2) and (3), are comparable to the results in Table3. The corruption variable is positive and significant, thecorruption–Latin America interaction term is insignificant(although positive in column (2)) and the corruption–infor-mal sector interaction term is negative and significant. BothIV and LIML results support our hypothesis that the infor-mal sector impacts the relationship between corruption andinequality.

Panel estimation (RE) results are in columns (4) and (5) ofTable 4. These results support those from the cross-sectionestimates. In particular, the corruption–Latin America inter-action term is insignificant, the coefficient on corruption is po-sitive and significant, and the corruption–informal sectorinteraction term is negative and significant. This evidence sup-ports the view that the impact of corruption on inequality de-pends on the size of the informal sector. An interesting aspectof the panel data results is the finding that the informal sectorhas a significant direct effect on inequality, a result not foundin other estimations. 18

A key implication of our findings is that the effectiveness ofanti-corruption measures is likely to be reduced in countrieswith a large informal sector. In other words, where institutionsare weak the link between corruption and inequality may bealtered. The key role played by institutions is also seen inthe work of Chong and Gradstein (2007b). They use a cross-section of countries to examine the impact of inequality onthe informal sector. They show that where institutions are lessthan robust the impact of inequality on the informal sector isenhanced. Though the focus of their work is different it doesmake clear that the quality of institutions matter.

Our results also tie in with some of the arguments of Sarte(2000). He analyzes growth, corruption, and the informal sec-tor from a theoretical perspective and finds that large informalsectors are not necessarily detrimental to economic growth.The impact of corruption on growth depends on operatingcosts in the informal sector. When the cost of informality islow, a large number of firms choose to operate informallyand growth is higher. This is consistent with our finding thatwhen the informal sector is large (costs of informality arelow) corruption is less harmful to inequality.

Tables 2–4 also show the results for the other determinantsof inequality. There is some evidence of an inverse relationshipbetween income (lgdp) and inequality, though this result is notrobust. Secondary education is important in reducing inequal-ity in most of the estimates. There is some evidence to indicatethat a higher ratio of agriculture to GDP and more opennessare associated with lower inequality, while financial deepeningand higher inflation exacerbate inequality. In the panel estima-tions we tested the Kuznets hypothesis by adding a squared in-come term but the income parameters were insignificant(results not presented). However, when included on its own(columns (4) and (5) of Table 4) income is negative and signif-icant.

6. CONCLUSION

Perceived wisdom says corruption is bad for incomeinequality. Policy makers should therefore introduce anti-cor-ruption measures as a way of reducing inequality. But recentresearch on Latin America finds a trade-off between corrup-tion and inequality and explains the finding with regard tothe large informal sector in the region. If the informal sectorargument is correct anti-corruption policies may be misguided.This paper has examined the informal sector hypothesis usingdata for a large sample of countries and has established empir-ically that the informal sector impacts the relationship betweencorruption and inequality. In particular, when the informalsector is large corruption does less harm to inequality. Thisis true for Latin America and more generally. The findingsare robust to different estimation procedures and differentmeasures of inequality and corruption.

This research has important implications for policy. Oneobvious implication is that where institutions are weak (andthe informal sector is large) it may be beneficial to allow cor-ruption to grow. The problem with this is that corruptioncould become pervasive and later on countries may becomestuck in a downward spiral with a growing informal sectorand an ineffective institutional framework. Furthermore, theinformal sector is by nature unregulated with many workers(adults and children) facing exploitation and dangerous work-ing conditions. 19

A more suitable strategy is one that fights corruption withone hand and improves other aspects of governance with theother by building stronger institutions to allow improvements


in government efficiency and in the quality of the regulatoryframework. In short, policy makers in countries with a largeinformal sector must exercise caution in any attempt to reducecorruption as narrow reform programs may be counterpro-

ductive. Anti-corruption policies, institutional reform, andenforcement of laws which have an impact on the informalsector must be accompanied by measures which are able to ab-sorb workers displaced from the informal sector.

NOTES

1. There is some support for this in work on Brazil by Ulyssea (2010),though the author also notes that if anti-corruption policy aims atreducing the cost of entry into the formal sector the cost of creating newvacancies is reduced and so employment and welfare will rise.

2. One potential drawback with the Latin America research is itprecludes an accurate comparison with the rest of the world because theresults are based on a sample of Latin American countries only. By using adata set comprising a large sample of countries we are able to deal withthis issue.

3. Bribery is possible because officials have discretion with regard toallowing the start-up of a business. Also, the punishment if caught is oftenrelatively minor.

4. Regulation takes the form of labor regulation, environmental protec-tion, consumer protection, quality control, etc. There are also additionalcosts as requirements pertain to firms having accountants and lawyers.

5. The sample comprises Albania, Algeria, Argentina, Armenia, Austra-lia, Austria, Azerbaijan, Bangladesh, Barbados, Belgium, Benin, Bolivia,Botswana, Bosnia and Herzegovina, Brazil, Bulgaria, Burkina Faso,Burundi, Cambodia, Cameroon, Canada, Chile, China, Colombia, Congo(DR), Costa Rica, Cote d’Ivoire, Croatia, Czech Republic, Denmark,Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ethiopia,Fiji, Finland, France, Georgia, Germany, Ghana, Greece, Guatemala,Guyana, Haiti, Honduras, Hong Kong (China), Hungary, Iceland, India,Indonesia, Iran, Ireland, Israel, Italy, Jamaica, Japan, Jordan, Kazakh-stan, Kenya, Korea (Rep), Kuwait, Kyrgyz Republic, Laos, Latvia,Lebanon,, Lesotho, Lithuania, Luxembourg, Madagascar, MacedoniaFYR, Malawi, Mali, Malaysia, Malta, Mauritania, Mauritius, Mexico,Moldova, Mongolia, Morocco, Mozambique, Namibia, Nepal, Nether-lands, New Zealand, Nicaragua, Niger, Nigeria, Norway, Oman, Paki-stan, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Poland,Portugal, Puerto Rico, Romania, Russia Federation, Rwanda, SaudiArabia, Senegal, Serbia and Montenegro, Sierra Leone, Singapore, SlovakRepublic, Slovenia, South Africa, Spain, Sri Lanka, Swaziland, Sweden,Switzerland, Syria, Taiwan, Tajikistan, Tanzania, Thailand, Trinidad andTobago, Tunisia, Turkey, Uganda, Ukraine, United Arab Emirates,United Kingdom, United States of America, Uruguay, Uzbekistan,Venezuela, Vietnam, Yemen Republic, Zambia, and Zimbabwe.

6. Available at http://www.wider.unu.edu/wiid/wiid.htm.

7. Available at http://pwt.econ.upenn.edu/php_site/pwt_index.php.

8. This definition may be viewed as rather broad since it includes bothlegal and illegal activities. However, it provides the widest coverage ofcountries and there is no alternative data available for the large sampleused in this study.

9. The procedure used to generate these estimates may lead to measure-ment error because the estimates depend on the theoretical relationbetween the variable of interest (size of the shadow economy) and theindicators, which may be subject to debate.

10. http://www.transparency.org/policy_research/surveys_indices/cpi/2009.

11. Dynamic panel estimation is inappropriate for several reasons: first,there are missing observations; second, the variables of interest are dummyvariables; and third, variables such as the ICRG or CPI measure ofcorruption show little variation for some countries.

12. We tried other regional dummies (other than Latin America) butnone were significant. We also experimented without success with adummy variable to distinguish between developed and developingcountries and a location dummy (tropical versus non-tropical countries).

13. We also carried out estimations with only the corruption variable (nointeraction term). The corruption variable only becomes significant whenthe interaction term is included.

14. In order to check whether neo-liberal reforms may be responsible forthe ‘different’ result for Latin America, we experimented with severalvariables which may reflect neo-liberal reforms and interacted these withcorruption; for example, corrupt * openness and corrupt * dcps. Theseinteraction terms were not significant.

15. @Gini/@corruption = 4.1059 � (0.1086) � 12 = 2.8027; @Gini/@corruption = 4.1059 � (0.1086) � 45 = �0.7811; these marginal impactsare calculated without the coefficient on corrupt * ladum as it is insignificant.

16. So treating informal as an endogenous variable.

17. A possible alternative would be to specify the model with laggedexplanatory variables and apply OLS estimation.

18. The fact that the informal sector as an independent variable is notstatistically significant in cross section estimations but is significant inpanel estimations indicates that the results are sensitive to the type ofestimation. The panel results (columns (4) and (5) in Table 4) are pickingup within country effects arising out of extra data for each country and it islikely that this is the reason for a significant result in these estimations.However, it is not possible to confirm this without further research andthis requires a sufficiently large panel of data to become available. All wecan say for the moment is that it is an open empirical question as towhether the informal sector has a direct influence on inequality.

19. According to UN figures, one in five 10-14 year olds work in Brazil,Honduras and Haiti, and more than one in ten work in most LatinAmerican and Caribbean countries. The International Labor Organiza-tion puts the figure for the total number of working children (agedbetween 5 and 14 years) in Latin America and the Caribbean at 17.5million. See http://natlex.ilo.ch/wcmsp5/groups/public/—ed_norm/—dec-laration/documents/publication/wcms_decl_fs_51_en.pdf.

http://www.wider.unu.edu/wiid/wiid.htm

http://pwt.econ.upenn.edu/php_site/pwt_index.php

http://www.transparency.org/policy_research/surveys_indices/cpi/2009

http://www.transparency.org/policy_research/surveys_indices/cpi/2009

http://natlex.ilo.ch/wcmsp5/groups/public/---ed_norm/---declaration/documents/publication/wcms_decl_fs_51_en.pdf

http://natlex.ilo.ch/wcmsp5/groups/public/---ed_norm/---declaration/documents/publication/wcms_decl_fs_51_en.pdf


REFERENCES

Acemoglu, D., Johnson, S., & Robinson, J. A. (2002). Reversal offortune: Geography and institutions in the making of modern worldincome distribution. Quarterly Journal of Economics, 117(4),1231–1294.

Ades, A., & Di Tella, R. (1997). The new economics of corruption: Asurvey and some results. Political Studies, 45(3), 496–515.

Aidt, T. S., Dutta, J., & Sena, V. (2008). Governance regimes, corruptionand growth: Theory and evidence. Journal of Comparative Economics,36(2), 195–220.

Albrecht, J., Navaroo, L., & Vroman, S. (2009). The effects of labormarket policies in an economy with an informal sector. EconomicJournal, 119(539), 1105–1129.

Andres, A. R., & Ramlogan-Dobson, C. (2011). Is corruption really badfor inequality? Evidence from Latin America. Journal of DevelopmentStudies, 47(7), 959–976.

Apergis, N., Dincer, C. D., & Payne, J. E. (2010). The relationshipbetween corruption and income inequality in US states: Evidence froma panel co-integration and error correction model. Public Choice,145(1), 125–135.

Arduz, M. (2000). Descripcion y evaluacion de las iniciativas tomodascontra la corrupcion en el servicio nacional de aduanas en Bolivia.Unpublished case study, OECD Development Centre/UNDP.

Blackburn, K., & Forgues-Puccio, G. F. (2007). Distribution anddevelopment in a model of misgovernance. European Economic Review,51(6), 1534–1563.

Blackburn, K., & Forgues-Puccio, G. F. (2009). Why is corruption lessharmful in some countries than others?. Journal of Economic Behaviorand Organization, 72(3), 797–810.

Buehn, A., & Schneider, F. (2007). Shadow economies and corruption allover the world: Revised estimates for 120 countries. Economics: TheOpen-Access, Open-Assessment E-Journal, 1. http://dx.doi.org/10.5018/economics-ejournal.ja.2007-9.

Choi, J. P., & Thum, M. (2005). Corruption and the shadow economy.International Economic Review, 46(3), 817–836.

Chong, A., & Calderon, C. A. (2000). Institutional quality and incomedistribution. Economic Development and Cultural Change, 48(4),761–786.

Chong, A., & Gradstein, M. (2007a). Inequality and institutions. Reviewof Economics and Statistics, 89(3), 454–465.

Chong, A., & Gradstein, M. (2007b). Inequality and informality. Journalof Public Economics, 91(1–2), 159–179.

Cule, M., & Fulton, M. (2009). Business culture and tax evasion: Whycorruption and the unofficial economy can persist. Journal of EconomicBehavior and Organization, 72(3), 811–822.

Dabla-Norris, E., Gradstein, M., & Inchauste, G. (2010). What causesfirms to hide output? The determinants of informality. Journal ofDevelopment Economics, 85(1–2), 1–27.

De Soto, H. (1989). The other path. New York: Harper and Row.de Vaal, A., & Ebben, W. (2011). Institutions and the relation between

corruption and economic growth. Review of Development Economics,15(1), 108–123.

Dincer, O., & Gunlap, B. (2008). Corruption, income inequality andpoverty in the United States. Nota di Lavoro, No. 54.

Dobson, S., & Ramlogan-Dobson, C. (2010). Is there a trade-off betweeninequality and corruption? Evidence from Latin America. EconomicsLetters, 107(2), 102–104.

Dreher, A., Kotsogiannis, C., & McCorriston, S. (2009). How doinstitutions affect corruption and the shadow economy?. InternationalTax and Public Finance, 16(4), 773–796.

Dreher, A., & Schneider, F. (2010). Corruption and the shadow economy:An empirical analysis. Public Choice, 144(2), 215–238.

Dutt, P., & Traca, D. (2010). Corruption and bilateral trade flows:Extortion or evasion?. Review of Economics and Statistics, 92(4),843–860.

Engerman, S. L., & Sokoloff, K. L. (1997). Factor endowments,institutions and differential paths of growth among new worldeconomies: A view from economic historians of the United States. InS. Haber (Ed.), How Latin America fell behind (pp. 260–304). Stanford,CA: Stanford University Press.

Engerman, S. L., & Sokoloff, K. L. (2000). History lessons: Institutions,factor endowments, and paths of development in the new world. TheJournal of Economic Perspectives, 14(3), 217–232.

Engerman, S. L., & Sokoloff, K. L. (2002). Factor endowments,inequality and paths of development among new world economies.NBER working paper 9259, Cambridge, MA: National Bureau ofEconomic Research.

Engerman, S. L., & Sokoloff, K. L. (2005). Colonialism, inequality andlong-run paths of development. NBER working paper 11057, Cam-bridge, MA: National Bureau of Economic Research.

Feige, E. (1996). Overseas holding of US currency and the undergroundeconomy. In S. Pozo (Ed.), Exploring the underground economy:Studies of illegal and unreported activity (pp. 5–62). Kalamazoo, MI:W.E. Upjohn Institute for Employment Research.

Foellmi, R., & Oechslin, M. (2007). Who gains from non-collusive corruption?. Journal of Development Economics, 82(1),95–119.

Franko, P. (2006). The puzzle of Latin American development. Oxford:Rowman and Little Field Publishers Inc..

Friedman, E., Johnson, S., Kaufmann, D., & Zoido-Lobaton, P. (2000).Dodging the grabbing hand: the determinants of unofficial activity in69 countries. Journal of Public Economics, 76(3), 459–493.

Fuller, W. A. (1977). Some properties of a modification of the limitedinformation estimator. Econometrica, 45(4), 939–953.

Gupta, S., Davoodi, H., & Alonso-Terme, R. (2002). Does corruptionaffect income inequality and poverty?. Economics of Governance, 3(1),23–45.

Gyimah-Brempong, K. (2002). Corruption, economic growth and incomeinequality in Africa. Economics of Governance, 3(3), 183–209.

Gyimah-Brempong, K., & Munoz de Camacho, S. (2006). Corruption,growth, and income distribution: Are there regional differences?.Economics of Governance, 7(3), 245–269.

Hindriks, J., Muthoo, A., & Keen, M. (1999). Corruption, extortion andevasion. Journal of Public Economics, 74(3), 395–430.

Houston, D. A. (2007). Can corruption ever improve an economy?. CatoJournal, 27(3), 325–342.

Huntington, S. (1968). Political order in changing societies. New Haven:Yale University Press.

Johnson, S., Kaufmann, D., & Zoido-Lobaton, P. (1998). Regulatorydiscretion and corruption. American Economic Review, 88(2),387–392.

Laoyza, N. V., Oviedo, A. M., & Serven, L. (2005). The impact ofregulation on growth and informality. World Bank Staff Papers,3623.

Leff, N. H. (1964). Economic development through bureaucratic corrup-tion. American Behavioral Scientist, 82(3), 337–341.

Li, H., Xu, L. C., & Zou, H. F. (2000). Corruption, income distribution,and growth. Economics and Politics, 12(2), 155–181.

Macrae, J. (1982). Underdevelopment and the economics of corruption: Agame theory approach. World Development, 10(8), 677–687.

Maloney, W. (2004). Informality revisited. World Development, 32(7),1159–1178.

Mandal, B., & Marjit, S. (2010). Corruption and wage inequality.International Review of Economics and Finance, 19(1), 162–172.

Meon, P., & Sekkat, K. (2005). Does corruption grease or sand the wheelsof economic growth?. Public Choice, 122(1–2), 66–97.

Meon, P., & Weil, L. (2009). Is corruption an efficient grease?. WorldDevelopment, 38(3), 244–259.

Osterfeld, D. (1992). Prosperity versus planning: How government stifleseconomic growth. New York: Oxford University Press.

Perry, G., & Olarreaga, M. (2006). Trade liberalization, inequality andpoverty reduction in Latin America. Paper presented at ABCDE, SanPetersburg, January.

Perry, E. G., Maloney, W., Arias, O., Fajnzylber, P., Mason, A., &Saavedra-Chanduvi, J. (2007). Informality: Exit and exclusion. Wash-ington, DC: The World Bank.

Sarte, P. D. G. (2000). Informality and rent seeking in a model of long rungrowth. Journal of Monetary Economics, 46(1), 173–197.

Schneider, F. (2004). The size of the shadow economies of 145 countries allover the world: first results over the period 1999 to 2003. IZA DP No.1431.

Schneider, F. (2007). Shadow economies and corruption all over the world:new estimates for 145 countries. <http://www.lawrence.edu/fast/fin-klerm/ShadEconomyCorruption_July2007.pdf>.

http://dx.doi.org/10.5018/economics-ejournal.ja.2007-9


Schneider, F., Buehn, A., & Montenegro, C. (2010). Shadow economies allover the world: new estimates for 162 countries from 1999 to 2007.Policy research working paper 5356, World Bank.

Stock, J. H., & Yogo, M. (2005). Testing for weak instruments in linear IVregression. In D. W. K. Andrews, & J. H. Stock (Eds.), Identificationand inference for econometric models: Essays in honor of ThomasRothenberg (pp. 80–108). Cambridge: Cambridge University Press.

Tanzi, V. (1982). The underground economy in the United States andabroad. Lexington, MA: Lexington Books.

Ulyssea, G. (2010). Regulation of entry, labor market institutionsand the informal sector. Journal of Development Economics,91(1), 87–99.

Uslander, E. (2007). Corruption and the inequality trap in Africa. AfroBarometer working paper No. 69.

Woolridge, J. M. (2010). Introductory econometrics: a modern approach(4th ed.). Mason: Thomson South-Western.

why is corruption less harmful to income inequality in latin america?

Documents