tax evasion, disclosure, and

World Development Vol. 36, No. 11, pp. 2512–2525, 2008� 2008 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matter

www.elsevier.com/locate/worlddevdoi:10.1016/j.worlddev.2007.11.010

Tax Evasion, Disclosure, and Participation

in Financial Markets: Evidence from Brazilian Firms

THOMAS KENYON *

United Nations Industrial Development Organization, Vienna, Austria

Summary. — This paper draws on survey data and qualitative evidence from Brazilian manufactur-ing firms to examine the scale and consequences of tax evasion at the enterprise level. It discussesthe costs and benefits of under-reporting from the entrepreneur’s perspective and provides evidencethat evasion of sales tax is only weakly correlated with firm size. The paper then shows that med-ium-sized and large manufacturing firms that evade taxes are less likely to undergo an externalaudit and more likely to be asked for informal payments by tax officials. It also argues that theymay be less likely to participate in markets for equity finance.� 2008 Elsevier Ltd. All rights reserved.

Key words — tax evasion, equity markets, Brazil, Latin America

* I would like to acknowledge the helpful comments

of Katrina Burgess, Daniel Geffen, Maddalena Hon-

orati, Robert Kaufman, Richard Locke, Helen Milner,

John Nasir, Pablo Pinto, Graeme Robertson and two

anonymous reviewers for World Development. Part of

the work on which this paper is based was financed by

the Investment Climate Unit of the World Bank. Finalrevision accepted: November 13, 2007.

1. INTRODUCTION

This paper addresses the scale and conse-quences of tax evasion in Brazil at the firmlevel. It is concerned with what has been calledthe ‘‘unreported’’ economy—activity con-ducted by firms that are known to and regis-tered with the authorities, but nonethelesskeep a fraction of their activities undeclaredfor tax purposes (Feige, 1990). While mostfirms engaged in such unreported activity tendto be small, in some countries the sub-sectorcan also include medium-sized and even largecompanies, as this paper shows. Drawing ondata from the World Bank’s Investment Cli-mate Survey (ICS) of Brazilian manufacturingfirms, the paper analyzes the costs and benefitsof tax evasion from the firm’s perspective andmakes two empirical points. The first is that lar-ger Brazilian manufacturing firms declare agreater proportion of their activities to the taxand labor authorities, but that the differenceis small: a doubling of firm size is associatedwith an increase of just 4 percentage points inthe fraction of sales reported. The second isthat firms that are evading taxes are less likelyto undergo an independent audit and also morelikely to be asked for informal payments by thetax authorities. As a consequence of the former,they may also be less likely to participate inmodern capital markets.

251

Tax evasion matters for several reasons. It hasconsequences for resource allocation, enablingnon-compliant firms to draw labor away fromthose that do pay taxes, and potentially shiftingresources into sectors more amenable to eva-sion, such as trade, services and construction.To the extent that evading firms are constrainedin their access to sources of finance, as thispaper argues, they may also operate morelabor-intensively than would be optimal (Eilat& Zinnes, 2002). They may also choose to oper-ate at below optimal size, in order to avoid theattention of the authorities (McKinsey, 2004,p. 6). 1 More generally, tax evasion erodes thestate’s capacity to raise revenues and providenecessary public services, placing an unfair bur-den on those individuals and businesses that dopay their taxes. These consequences are of par-ticular concern when evading firms competewith compliant ones, or when tax and otherforms of regulatory evasion, such as smuggling,

2

TAX EVASION, DISCLOSURE, AND PARTICIPATION IN FINANCIAL MARKETS 2513

become a source of competitive advantage. 2

Under these circumstances, the presence of alarge shadow economy can undermine the socialcontract that exists between the state and com-pliant economic actors, potentially leading tothe disintegration of norms that encouragecompliance (Gerxhani, 2004).

The distorting influence of tax evasion onfactor markets indicates the need for a moreholistic approach to microeconomic reform indeveloping countries. As an influential studyhas put it, ‘‘Another set of big think questionscenters on the proposition. . . that the benefitsfrom any one set of reforms are significantly en-hanced when other markets are well function-ing. . . Comparative analysis of policy changesin countries where polices toward other sectorsvary may yield valuable insights’’ (Krueger,2000). This is of particular interest to a countrylike Brazil, where access to finance is one of themost severe firm-level constraints on economicgrowth (World Bank, 2005). It should also beof interest to policy-makers trying to stimulateequity market development. The practicalimplication is that reforms to capital marketregulation are unlikely to yield significantgrowth in financial intermediation unless pol-icy-makers also address the causes of tax eva-sion. This in turn requires a careful approachthat induces firms to comply with their taxand regulatory obligations while avoiding driv-ing them out of business altogether.

2. STATES, MARKETS, AND TAXATION

In the last few years, policy-makers, andscholars have turned their attention to the mi-cro-institutions of capitalism outside thedeveloped world. Following a wave of macro-economic reforms in the 1990s, many develop-ment practitioners came to recognize theimportance of institutions and incentive struc-tures in sustaining economic activity. Attentiontherefore focused on what has come to be calledthe ‘‘investment climate’’—the ‘‘opportunitiesand incentives for firms to invest productively,create jobs, and expand’’ (World Bank, 2005,p. 19). Among the most important determi-nants of the investment climate is the scopeand quality of government regulation. Thereis a long and distinguished literature on theways in which state institutions support marketactivity: by channeling information about mar-ket conditions and participants, by defining andenforcing property rights and contracts, and by

ensuring an appropriate level of competition(World Bank, 2002, pp. 5–6). Unfortunately,efficient states are a rarity in the developingworld. Most governments in these countriesdispose of the ability to issue regulations andcommands—what Mann called ‘‘despoticpower’’—but have little if any capacity to en-force them (Mann, 1984). The most extrememanifestation of this failure is organized pri-vate thuggery of the sort that prevailed inSomalia in the early 1990s. But a more com-mon outcome is the growth of the unrecordedor ‘‘gray’’ economy which characterizes a num-ber of middle-income countries. In these coun-tries much economic activity is unregulated andor untaxed. As a recent analysis put it, ‘‘Therelationship between the informal economyand the state is, by definition, one of inevitableconflict. The whole point of the state is to assertthe monopoly of its authority within a territory,but the whole point of informal entrepreneursis to avoid or subvert that authority. . . an infor-mal economy will develop when and where itcan’’ (Centeno and Portes, 2003, p. 7).

The obvious question is what allows this sit-uation to persist. One explanation is a simplelack of capacity. To be effective, a tax adminis-tration must both make contact with potentialtaxpayers and obtain information about thescale of their activities. Doing so requires a highdegree of coordination and information-shar-ing among government departments, particu-larly the tax authorities and police. Evenadvanced industrial countries have difficultyin ensuring full compliance (OECD, 2004).Achieving anything comparable is beyond adeveloping country like Brazil, which spends amuch smaller proportion of national incomeon tax collection and which lacks lawyers andjudges with tax expertise (McKinsey, 2004, p.30). The identification problem is particularlyacute for smaller businesses. Survey data fromtransition and from some African economiesshow that smaller firms generally report a lowerfraction of their revenues to the authorities,presumably because their size makes it easierfor them to deal in cash (Gauthier and Gerso-vitz, 1997; Gauthier and Reinikka, 2001; Gehl-bach, 2004, 22). It is also more problematic forlabor-intensive businesses because they havefewer visible fixed assets than capital-intensiveones. Much unrecorded activity in developedcountries consists of labor-intensive servicessuch as construction, road haulage, contractcleaning and catering (OECD, 2004, p. 262).When enforcement capacity is limited, the tax

2514 WORLD DEVELOPMENT

authorities may pursue a triage strategy—pre-ferring not to pursue some firms because theadministrative costs of doing so outweigh therevenue benefits.

But a second explanation is that firms areable to evade taxation through political connec-tions, bribery and other forms of manipulation.In developing as in developed countries, largefirms often engage in sophisticated tax planningand avoidance schemes. Studies in Cameroonand Uganda have found that larger firms aremore likely to be granted exemptions and takeadvantage of tax incentive schemes than smalland medium-sized enterprises (Gauthier &Gersovitz, 1997; Gauthier & Reinikka, 2001).But tax administrations often run into troublebecause they suffer from corruption within theirown ranks. This can take various forms—fromproviding exemptions to firms and individualswho would otherwise not qualify, to closing atax audit without any adjustment being madeor penalties being imposed, to deleting taxpayerrecords altogether (Tanzi, 2003, p. 5). If the va-lue of the tax foregone exceeds that of thebribe, the arrangement can be to the benefitof both parties—though not to the rest of soci-ety. If not, it can amount to an extortion racketwith low-level officials blackmailing firms intopaying bribes to avoid prosecution for evasion.

Whichever side of the debate one takes, thescale of evasion is impressive. By one estimate,unregulated activity accounted for more than30% of all economic activity in the developingworld in 2003. Furthermore, the proportion ofoutput generated by firms that are wholly orpartly outside the regulatory system has in-creased over the past decade and a half: from29% in 1990 to 38% in 2003 in Latin Americanand the Caribbean, and from 28% to 36%among developing countries as a whole, exclud-ing China (Schneider & Klinglmair, 2004). Moststudies of regulatory non-compliance have fo-cused on micro-enterprises and self-employedindividuals. But this focus may be misleading,since the phenomenon extends well beyondthese categories of firms and workers. As theWorld Bank has put it, ‘‘The informal economyis diverse, ranging from subsistence farmers andthose engaging in entrepreneurship out of neces-sity, to more affluent firms that find it feasible toevade tax and regulatory obligations, and oth-ers in the middle’’ (World Bank, 2005, p. 61).

In what follows I analyze the benefits andcosts of tax evasion from the perspective ofthe firm, concentrating in particular on its con-sequences for the ability of these firms to partic-

ipate in the arms-length transactions thatcharacterize many markets, including thosefor finance. The data on which the analysis isbased are drawn from Brazilian manufacturingenterprises with 10 or more employees. Theytherefore exclude the service sector, in whichthe majority of Brazilian firms are located andwhere we might expect tax evasion to be morewidespread, given their generally higher shareof labor in value added, lesser degree of relianceon fixed assets, and consequent lower visibil-ity. 3 Experience from the OECD suggests thattax evasion tends to be much higher in thesesectors than in most manufacturing industries.And anecdotal evidence from Brazil indicatesthat the evasion of customs duties is prevalentin retail trade and especially in the distributionof beverages, cigarettes, fuel, pharmaceuticalsand processed food (ETCO, 2003).

3. A FIRM-LEVEL PERSPECTIVE

Government regulation imposes costs onfirms and there are benefits to be derived fromevading it. Governments regulate companies inall sorts of ways: to restrict who may enter amarket, to control where firms may locate orto ensure minimum health, safety, and productstandards (World Bank, 2005, p. 95). The regu-latory burden associated with doing businessvaries significantly across countries, with workby the World Bank suggesting that it is largeron average in developing than in developedcountries. 4 In some countries firm managersreport spending over 10% of their time dealingwith government regulations and inspections(World Bank, 2005, p. 100). The cost and timeinvolved in starting a business is also muchhigher in low and middle income than indeveloped countries—up to 100 times higherin some cases according to the World Bank’sDoing Business Survey (World Bank, 2005,p. 99).

Governments in developing countries alsoraise revenues from companies, generallythrough sales or payroll taxes (World Bank,2005, p. 106). Indeed, the fraction of revenuederived from corporate taxes and levies oncommercial transactions tends to be higher indeveloping than in developed countries, duelargely to inefficiencies in tax administrationof the type discussed above (World Bank,2005, p. 107). From the firm’s point of view,the problem is often not so much the level oftaxation—which is little different from that in


the developed world—but the difficulties andinconvenience associated with its collection. Inseveral of the countries covered by the WorldBank’s Investment Climate Surveys, includingBrazil, over 50% of firms listed tax administra-tion as a very severe or major obstacle to thegrowth of their business (World Bank, 2005,p. 109). The practical consequence, accordingto some Brazilian entrepreneurs, is that evasionhas become a necessary condition for survival(Lieberman, 2003, p. 231).

(a) What are the consequences of tax evasion forfirms?

But if tax evasion confers important benefitson firms it also imposes significant costs. Themost obvious are those associated with discov-ery and prosecution: penalties in the form offines or confiscated assets (Loayza, 1997, p.1). But there are other potentially more seriousconsequences. These follow from two observa-tions. The first is that evading firms do not typ-ically have access to the courts or other disputeresolution mechanisms, most of which are atthe disposal only of companies that operate leg-ally (World Bank, 2005, p. 8). The second isthat evading firms also generally prefer to keepa low public profile to avoid detection.

Lack of access to the courts can renderevading firms vulnerable to crime. It may alsorestrict them to doing business with a circum-scribed set of partners whom they considertrustworthy, limiting the use of arms lengthtransactions or those based on disclosure andtransparency rather than personal acquain-tance. This can have consequences for theirability to enter into licensing agreements orrelationships with suppliers of capital equip-ment and technology. Firms that are evadingtaxes and other government regulations mayalso subject to extortion by agents of the state.Macro level data suggest a strong correlationbetween perceptions of corruption and the ex-tent of the unreported economy, even after con-trolling for the level of income across countries(World Bank, 1999, 20). This may be, as the lit-erature on transition economies has empha-sized, because firms pay bribes to officials inorder to avoid paying taxes. But it may alsobe because evading enterprises are subject topredation on the part of corrupt officials. Sev-eral studies have found a strong firm-level cor-relation between hidden activity and thepropensity to pay bribes (Johnson et al., 2000,p. 513). 5

A further consequence of tax evasion is thatit may make it harder for firms to access exter-nal finance. Banks and other financial institu-tions are generally unwilling to grant credit tofirms that lack proper documentation, includ-ing that relating to government registrationand licensing. Meanwhile, participation in cap-ital markets requires a level of disclosure thatmany less modern and less well-organized firmsfind difficult to meet, accustomed as they gener-ally are to poor or non-existent book-keeping.Moreover, listing on the stock market requirescompanies to submit their accounts for externalauditing; in some countries it may also requirethem to engage the services of outside directors.By making their financial information publiclyavailable to investors, under-reporting firmsnot only incur a substantial burden of regula-tion, they also are forced to state their earningstransparently and open themselves to a morerigorous assessment of their tax liabilities. 6

As a consequence, they may also be more likelyto finance themselves through retained earningsor through informal sources, such as money-lenders, friends, and family.

4. THE IMPACT OF TAX EVASION INBRAZIL

Brazil makes a particularly good case forexamining the consequences of tax evasion atthe firm level. First, the scale of unreportedactivity is larger than in most other countries.According to one estimate, it accounts foraround 40% of Brazilian GDP, compared to13% in China, 23% in India, and 30% in Mex-ico (Capp, Elstrodt, & Bebb Jones, 2005). Ofthe countries covered by the World Bank’sDoing Business Survey, only Russia had a lar-ger unofficial sector—at just over 46% of grossnational income. And anecdotal evidence sug-gests that unrecorded and untaxed activity isprevalent even among large firms. Lieberman(2003), for instance, quotes an accountant froman American-based Big-Six accounting firm inthe late 1990s as saying that most of his corpo-rate clients paid ‘‘no income tax at all’’ (Lieber-man, 2003, p. 229). There is also a particularlylarge productivity differential between firmsthat comply with government regulations,including tax, and those that do not. Accordingto McKinsey, output per worker among theformer is on average 46% that of the latter(McKinsey, 2004, p. 6).


Second, the retarding impact of evasion oncapital markets is probably more evident no-where than in Brazil. According to the marketresearch firm IDC, there are about 6,000 com-panies in Brazil with over 250 employees, only120 of which are listed and traded on the SaoPaulo Stock Exchange (Viegas, 2005, p. 1). In2000, the ratio of stocks traded in relation toGDP stood at 19%, compared to 40% inMalaysia and 43% in South Korea (WorldBank, 2007). Anecdotal evidence points to taxevasion as being one of the chief reasons: com-panies that do not pay taxes do not generallykeep reliable accounts and have an understand-able aversion to outside scrutiny by banks orother financial intermediaries. This is particu-larly troublesome for capital market develop-ment since the relationship between the issuersand buyers of equity securities depends on ahigh level of disclosure. Interviews with fundmanagers in Sao Paulo confirm the relevanceof these considerations. Many controllingshareholders balk at undertaking a stock mar-ket listing or at selling control to an outsideinvestor when they realize that doing so wouldrequire them to obtain an audit and settle theirtax liabilities—thus eroding their profitabilityand the value of their stake in the companyby up to 30 or 40%. 7

Historically Brazilian companies have re-sponded to the challenge of greater opennessby carrying multiple sets of books (known lo-cally as ‘‘caixa dois’’). 8 They may also bribeauditors to overlook tax evasion or other irreg-ularities. But the tightening of reporting andother requirements for participation in publiccapital markets since the financial crises of thelate 1990s has raised the cost of these subter-fuges. Companies wishing to list on the SaoPaulo stock exchange must now meet a set ofcriteria explicitly modeled on and as rigorousas those required by the New York and Lon-don stock exchanges. These include the manda-tory use of independent directors, thepresentation of certified accounts according toUS GAAP and stringent minority shareholderprotections. One consequence of this evolutionin corporate governance standards for listedcompanies is that, as a recent Brazilian govern-ment report put it, the scale of tax evasion hasbecome ‘‘an obstacle to financial intermedia-tion: it undermines the evaluation of credit riskby banks, raises the cost of borrowing, and cur-tails the use of capital market instruments,where transparency is a necessary conditionfor efficiency’’ (Rocca, 2004, p. 24).

5. MEASURING THE IMPACT OF TAXEVASION

In what follows, I use the World Bank’sInvestment Climate Survey database to test aseries of firm-level hypotheses concerning therelationship between tax evasion, access to for-mal markets for finance and business govern-ment relations. This database containsresponses from a random sample of firms onvarious aspects of the investment climate,including corporate governance, governmentregulation, access to finance, physical infra-structure and labor relations (World Bank,2005). Launched in 2001, the ICS data coverboth objective and perception-based indicators.The objective indicators measure the time re-quired to complete various processes and themonetary costs of disruptions and regulations.The subjective indicators cover firms’ percep-tions of the constraints on doing business andtheir assessments of various risks, includingthose associated with tax and labor laws(World Bank, 2005, p. 245).

The Brazilian survey was conducted in 2003for a random sample of 1642 business entitiesstratified by sector, location and size. Asalready mentioned, all were manufacturingfirms and spanned the following activities:food processing, textiles, garments, leatherand footwear, chemicals, metals and machin-ery, electronics, auto-parts and furniture.They were also chosen to be the representa-tive of the country’s five principal regions.Consistent with the geographical distributionof economic activity, 70% of firms weredrawn from the country’s relatively developedsouth and south-east, while 17% were drawnfrom the Amazon and center-west regionsand 13 from the poorer north-east. Entitiesbelonging to companies employing fewer thanten workers were excluded. Of these 1642businesses, 69 are listed publicly—represent-ing approximately a fifth of the total listedon the Sao Paulo stock exchange. These areoverwhelmingly privately owned companies,most of them controlled by domestic Brazil-ian shareholders. Only four are majority for-eign-owned and only one is whollygovernment-owned. Of these domestic Brazil-ian shareholders, the majority are individualsand families—not banks or institutionalinvestors, as might be the case in the UnitedStates or some European countries. The listedcompanies themselves range in size from 11employees to over 6,000.


As the sample frame was based on officialstatistics, the survey tended to over-sampleregistered and, possibly by extension, taxcompliant firms. Since in Brazil, as in manyother countries, the size distribution of firmsis skewed toward micro and small enterprises,the survey also over-sampled large firms toensure a minimum number of respondentsin each size category to allow for meaningfulanalysis. This means that we should be care-ful about extrapolating the results to otherfirm populations, including those in serviceindustries as already mentioned, but alsothe micro-enterprise sector or unregisteredfirms. 9

(a) Hypotheses

First, I am interested in the degree of re-ported tax evasion among respondent firms.What is the distribution and how does it varyby size and sector? Is it true that smallerfirms are more likely to evade than largerones? Second, I am interested in whetherfirms that are evading taxes are less likelyto have their accounts certified by externalauditors and whether, as a consequence, theyare less likely access markets for equity fi-nance and more likely to rely on informalsources of funds. Third, I am interested inthe relationship between firms and the stateagencies responsible for tax and administra-tion. Are firms that are evading taxes moreor less likely to face demands for informalpayments from government officials and dothey have more or less confidence in theircapacity to influence regulations that affecttheir business? These questions are capturedin the following hypotheses:

H1: The distribution of tax evasion across firms

is a continuous not a binary variable; smaller

firms should report higher levels of evasionH2: Firms that are evading taxes and labor reg-

ulations will be less likely to have their accounts

audited by an external auditor

H3: Firms that are under-reporting sales will be

less likely to issue new shares and more likely to

rely on informal sources of finance

H4: Firms that are under-reporting sales will be

more likely to receive demands for informal pay-

ments from government inspectors

H5: Firms that are under-reporting sales will re-

port less influence over the content of regulations

that affect their business

(b) Variables and specification

The variables I use in the analysis are definedbelow. Survey questions are reported in Table1. In the regression analysis that follows, I em-ploy a simple specification of the type:

Y ¼ aþ b1½tax compliance�þ b2½control variables� þ e:

I use OLS estimation where the dependent vari-ables are continuous, that is, for hypotheses (1)and (3), probit where the dependent variable isbinary, that is, for hypotheses (2) and (4) andordered probit where the dependent variableis categorical and orderable, that is, hypothesis(5). The data are cross-sectional and unstan-dardized.

(c) Independent variable

(i) Tax evasionObtaining reliable data on the extent of tax

evasion is notoriously difficult, since firms havean understandable aversion to acknowledgingit. Instead, the survey asks respondents a lead-ing question designed to elicit informationabout their own behavior while purporting tobe about that of their competitors. The ques-tion about tax compliance is worded as follows:‘‘Recognizing the difficulties many enterprisesface in fully complying with taxes and regula-tions, what percentage of total sales wouldyou estimate the typical establishment in yourarea of activity reports for tax purposes?’’ Itake this as a proxy for the level of compliancepracticed by the respondent firm itself, antic-ipating that the firm will take its own practiceas a baseline. Naturally, this assumption isopen to question. It might be, for instance, thatrespondent firms deliberately overstate the de-gree of compliance out of fear of the authori-ties. To counter this, the survey enumeratorsmade clear to surveyed firms that the firm con-ducting the survey was private, unconnectedwith the tax authorities and that the data wouldbe disseminated only in such a way as to makeit impossible to identify firms from theirresponses. 10

(d) Dependent variables

(i) Aversion to disclosureI am interested in the extent to which the firm

is willing to grant access to outsiders, the

Table 1. World Bank investment climate survey questions

Tax evasion:

Recognizing the difficulties many enterprises face in fully complying with tax and other regulations, what percent-age of total annual sales would you estimate the typical establishment in your sector reports?

Aversion to disclosure:

Does your establishment have its annual financial statement checked and certified by an external auditor? [Yes,No]

Sources of finance:

What percentage of your new investment and working capital requirements over the past year was financed fromnew equity or the sale of stock?What percentage of your new investment and working capital requirements over the past year was financed frominformal sources (e.g., moneylender)?

Corruption and influence:

During the last year, did anyone [from the federal/state/municipal tax authorities] suggest that you make an infor-mal payment (gratuity or tip)?How much influence do you think your firm had on recently enacted national laws and regulations that have asubstantial impact on your business? [1 = none, 2 = minor, 3 = moderate, 4 = major, 5 = decisive]

Source: World Bank investment climate surveys, available at www.enterprisesurveys.org.


assumption being that firms that are evadingtaxes will be wary of disclosure. The indicatorI use is whether the firm’s accounts were au-dited by an external auditor.

(ii) Use of external financeI am interested in the degree to which firms

are able to tap formal markets for finance,particularly equity finance. My variable isthe proportion of the firm’s working capitaland investment requirements that is met bythe issuance of new shares. Unfortunately,the ICS dataset does not distinguish betweenequity raised on public capital markets andthat raised through the sale of shares to indi-viduals. To further assess the reluctance ofevading firms to draw on formal sources offinance, I also include the percentage ofworking capital and investment requirementsmet through ‘‘informal sources (such as amoneylender).’’

(iii) Experience of corruptionThe ICS dataset includes several questions

concerning relations between firms and govern-ment inspectors. My variable indicates whetheror not the firm was, within the previous year,asked for a gift or informal payment by officialsfrom the federal, state or municipal tax author-ities. (Specifically, the question was worded asfollows, ‘‘Did anyone suggest you make infor-mal payments (gratuities or tips)?’’). It doesnot indicate whether such a payment was infact made or not. 11

(iv) Influence over policyI am also interested in the extent to which

firms have a say in policy-making. I use thefirm’s own assessment of the extent to whichit was able to influence ‘‘recently enacted na-tional laws and regulations that have a substan-tial impact on [its] business.’’ This is coded on ascale from 1 to 5, with 1 representing ‘‘no influ-ence’’ and 5 representing ‘‘decisive influence.’’

(e) Controls

Probably, the single largest difficulty inassessing the impact of tax evasion on firms’participation in formal markets is endogeneity.The same factors that lead firms to evade taxmay also lead them to eschew external auditsand markets for finance. Firms in labor-inten-sive industries, for instance, might be both bet-ter equipped to hide from the authorities and atthe same time less in need of external financethan more capital-intensive firms. Capital-intensive firms, or those with a higher propor-tion of fixed assets, might also find it easier toprovide collateral (and therefore be less depen-dent on informal sources of finance). Similarly,firms in areas of the country with relativelypoorly developed formal institutions might besubject to less rigorous tax enforcement andat the same time lack information about or ac-cess to formal markets for finance. Another twopotential sources of endogeneity are size andownership structure. Other things being equalI would expect larger firms both to have greater

http://www.enterprisesurveys.org


access to capital markets and to be more visibleto the tax authorities. The same may be true ofexporters and foreign-owned companies.

I address these issues through control vari-ables. For labor intensity, I include the shareof wage costs—defined as salaries plus socialsecurity and other contributions—in total pro-duction costs, measured at the firm level andobtained through the ICS survey instrument. 12

To control for differences in the availability ofcollateral, I include the proportion of fixed as-sets (machinery, equipment and land) in totalassets. I also control for variations in theoverall institutional environment by includingstate-level per capita output, as reported bythe Brazilian Institute for Geography andStatistics (IBGE). Finally, I include firm size,as measured by the log of the number ofemployees, the proportion of the firm’s outputthat is exported (either directly or through adistributor) and the proportion of its capitalthat is foreign-owned. The latter three variablesare self-reported by firms in the ICS. 13 Table 2reports the correlations among these controlvariables.

6. RESULTS

(a) Descriptive statistics

Several preliminary results emerge from thedescriptive statistics. The first is that there isa substantial variance in the level of reportedcompliance, consistent with the view that eva-sion is a matter of degree not kind. As Figure1 shows, the likelihood of a firm reporting50% or 70% of its sales is almost as great asthat of it reporting 100% (around 16% com-

Table 2. Control variables, pai

Size(log employees)

Expor(% out

Exporter (% output) 0.33(0.00)**

Foreign ownership 0.20 0.14(0.00)** (0.00

Labor intensity �0.01 �0.0(0.83) (0.51

Fixed assets (%total) �0.07 �0.0(0.01)** (0.06

State GDP per capita 0.09 0.15(0.00)** (0.00

Source: As above P-values in parentheses.* denotes significant at 5% level.** denotes significant at 1% level.

pared to 19%). Meanwhile, Figure 2 showsthat the reported tax evasion occurs amongall sizes of firm, even large enterprises. Whilethe degree of under-reporting is highest amongfirms with fewer than 20 employees, even largefirms—with more than 100 employees—typi-cally only declare around 75% of their salesto the authorities. The regression results re-ported in the first column of Table 5 provideadditional support for this finding, which isrobust to the inclusion of appropriate controlvariables. They imply that the hypothetical ef-fect of doubling firm size is to increase thefraction of sales declared by only 3.8 percent-age points (note that the size variable isexpressed as the natural log of the numberof employees). In short, larger firms are morelikely to report their peers as complying withtax requirements than smaller firms but thedifference between them is not great. 14

The level of reported tax evasion varies quitesubstantially across sectors, as shown in Table4. It is highest in the garments, leather and fur-niture industries, lowest among electronics,chemicals, and manufacturers of auto compo-nents. This is broadly consistent with thehypothesis that there are greater opportunitiesfor evasion in labor intensive as opposed tocapital-intensive sectors—and is supported bythe negative coefficient on labor intensity inthe OLS regression results reported in Table5, column 1, which implies that a 5 percentagepoint increase in the share of labor costs in to-tal production costs is associated with a 1 per-centage point decrease in the fraction of salesdeclared to the tax authorities. The incidenceof informal payments is lower than might beexpected: just under 10% of firms, on average,report having been asked for such a payment.

rwise correlation coefficients

terput)

Foreignownership

Laborintensity

Fixed assets(% total)

–)**

2 �0.01 –) (0.67)5 �0.02 �0.03 –) (0.50) (0.19)

�0.01 �0.01 �0.02)** (0.63) (0.61) (0.37)

020

4060

8010

0

taxc

ompl

ianc

e

0 5 10 15 20

Percent

Figure 1. Reported tax compliance, percentage of respondents by fraction of sales declared.

0 20 40 60 80

mean of taxcompliance

>100

51-99

21-50

11-20

Figure 2. Reported tax compliance, by firm size (number of employees).


This may be because the survey question onwhich this analysis is based refers to interac-tions only with the tax and not other publicauthorities, such as health and safety, labor,or environmental inspectors. 15

(b) Regression analysis

(i) Financial disclosure and sources of financeThe first set of findings from the regression

analysis is that there is also a significant

Table 3. Dependent variables, pairwise correlation coefficients

External audit(Yes/No)

Equity finance(%)

Informal finance(%)

Bribe asked(Yes/No)

Equity finance (%) 0.12 –(0.00)**

Informal finance (%) �0.05 �0.01 –(0.05)* (0.71)

Bribe asked (Yes/No) �0.07 0.05 �0.02 –(0.02)* (0.14) (0.57)

Influence over policy 0.01 0.04 �0.03 �0.01(1 = none, 5 = decisive) (0.72) (0.11) (0.31) (0.71)

Source: As above P-values in parentheses.* denotes significant at 5% level.** denotes significant at 1% level.

Table 4. Descriptive statistics by firm sector (means)

Taxcompliance

(%)

External audit(No = 0,Yes = 1)

Equityfinance

(%)

Informalfinance (%)

Bribe asked(No = 0,Yes = 1)

Policyinfluence

(0–4)

Laborintensity

(%)

Leather 63.2 0.12 3.4 2.7 0.10 0.62 25.2Garments 59.7 0.10 2.0 1.9 0.12 0.85 83.6Textiles 72.8 0.27 5.7 0.3 0.11 0.93 17.2Food 70.8 0.33 4.7 1.0 0.10 0.91 13.8Metals and machinery 73.1 0.19 2.3 2.6 0.09 0.72 95.0Electronics 76.2 0.28 9.6 1.3 0.10 0.92 24.1Chemicals 81.2 0.37 4.9 1.5 0.09 0.94 12.1Wood and furniture 63.5 0.15 2.2 1.4 0.09 0.70 25.2Auto components 78.1 0.32 9.1 0.9 0.06 0.78 21.4

Source: As above.


positive association between a firm’s self-re-ported degree of tax compliance and its willing-ness to submit to an external audit. 16 Aftercalculating marginal effects, the coefficient ontax compliance in column 2 of Table 5 impliesthat the probability of a firm having its ac-counts audited is 15% higher among fully taxcompliant firms than among non-compliantfirms, holding other explanatory variables con-stant at their means.17 I also find the expectedassociation between evasion and firms’ financ-ing decisions, though the effects for both equityand informal finance are very small. The coeffi-cient on compliance implies that, holding otherfactors constant, an increase of 1 percentagepoint in the proportion of sales declared is asso-ciated with an increase of 0.04 percentagepoints in the proportion of the firm’s financingneeds met by equity and a decrease of 0.03 per-centage points in the proportion met throughinformal sources. The effect is robust to firmsize and to the other control variables discussed

above. Foreign ownership also has a significantpositive—and much larger—effect on equityfinancing, as might be expected. But exportprofile and labor intensity appear not to matter,while size has the opposite effect of what onemight expect—larger firms seem to meet a low-er proportion of their financing requirementsvia equity than smaller ones. This may be dueto ambiguity in the phrasing of the survey ques-tion and in particular its failure to distinguishbetween equity issued on public markets andshares sold privately. The phrasing of the sur-vey question may also explain the very low(0.12), though significant, correlation betweenequity finance and the likelihood of a firm’shaving an external audit, as shown in Table 3.

(ii) Relations with governmentThe second set of findings to emerge from the

regression analysis is that evading firms aremore likely to face demands for informal pay-ments by officials from the tax authorities.

Table 5. Regression results—tax evasion, disclosure, sources of finance and corruption/influence

Taxcompliance

Externalaudit

Equityfinance

Informalfinance

Paymentrequested

Policyinfluence

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

Tax Compliance – 0.007 0.039 �0.026 �0.005 0.003(3.58)** (2.80)** (3.22)** (2.81)** (2.10)*

Employees (log) 3.845 0.400 �1.049 �0.557 0.006 0.047(6.36)** (9.85)** (2.57)** (2.77)* (0.12) (1.63)

Exports 0.108 0.001 0.044 �0.000 �0.002 �0.001(3.79)** (0.68) (1.53) (0.01) (0.75) (0.68)

Foreign ownership 0.135 0.014 0.248 0.003 �0.001 �0.000(4.35)** (6.86)** (4.54)** (0.35) (0.45) (0.05)

Labor intensity �0.221 �0.479 �0.008 �0.012 0.007 �0.005(3.08)** (1.77)* (1.20) (4.68)** (1.98)* (6.87)**

Percent fixed assets �3.047 �0.188 �2.016 1.100 �0.056 0.114(1.21) (1.07) (1.35) (1.20) (0.24) (0.94)

Per capita income 0.000 �0.000 �0.000 0.000 �0.000 �0.000(1.26) (0.43) (2.36)* (2.59)** (2.68)* (4.60)**

Constant 51.55 �2.96 5.12 5.27 �0.72 –(18.65)** (12.59)** (2.53) (4.58)** (2.83)**

R2 0.08 0.21 0.14 0.02 0.04 0.01Observations 1413 1412 1177 1177 876 1399

OLS (1, 3, 4); probit (2, 5); ordered probit (6).* Significant at 5% level;** Significant at 1% level.Absolute values of t-statistics in parentheses. Variables are not standardized.


Again, after calculating marginal effects, thecoefficient on tax compliance in column 5 ofTable 5 implies that the probability of a firmbeing asked for an informal payment is 9% low-er among fully tax compliant firms than amongnon-compliant firms, holding other variablesconstant at their means. This is consistent withthe second explanation for tax evasion outlinedabove—that it persists because of corruptionwithin the system of tax administration. It isalso robust to the inclusion of the relevant con-trol variables.

As it stands, however, it does not necessarilymean that firms are paying bribes in order toavoid being taxed. It might also be that thesefirms are targeted by officials because they donot have legal or other forms of redress. Thesurvey question in itself does not provide anyinformation as to whether the relationship be-tween entrepreneur and tax official is collusiveor exploitative. However, the regression analy-sis also provides us with another finding: firmsthat report lower levels of tax compliance alsosay they have less influence over regulationsthat affect the operation of their business. Thisis not conclusive, but it does suggest that therelationship between officials and under-report-

ing firms is more one of extortion by the latterthan capture by the former.

7. CONCLUSIONS

What have we learnt from this analysis? Thedata for manufacturing firms from Brazil showthat tax evasion is a matter of degree and that itis not limited to small and medium-sized enter-prises. Even quite large firms acknowledge con-cealing around a quarter of their sales from thetax authorities—assuming that the level ofcompliance reported by the firm as typical forits area of activity corresponds to the level actu-ally practiced by the firm itself. The regressionanalysis also illustrates the firm-level economiccosts of tax evasion. Enterprises that are evad-ing taxes are less likely to obtain an externalaudit, possibly because they are worried thatdisclosure will alert the tax authorities. Theyare also somewhat less likely to issue newshares to finance their working capital andinvestment requirements, and more likely torely on informal sources of finance such asmoneylenders, though the effects observed inthis dataset are small.


The analysis also shows that manufactur-ing firms operating in an environment inwhich tax evasion is more prevalent aremore likely to suffer demands for bribesfrom corrupt officials and are less likely tohave a say in regulations that affect theirbusiness. I interpret this as evidence thatthe firms’ decision to evade renders themvulnerable to blackmail and excludes themfrom participation in formal decision-makinginstitutions. But an alternative explanation,also consistent with the data, is that evadingfirms are engaged in a voluntary transactionby paying government inspectors to look theother way. Either way, the results suggestthat the problem is more one of corruptionthan identification per se. It is not so muchthat tax evasion is impossible to detect,rather that a dysfunctional system of taxadministration prevents tax laws from beingenforced properly.

These findings do not necessarily meanthat these firms themselves feel constrained

by their lack of access to capital markets.It is possible, for instance, that the entrepre-neurs in question might be quite willing toforfeit these benefits as a condition of pay-ing less tax, consistent with the voluntaristview of evasion outlined above. But thisdoes not imply that such an outcome isdesirable. After all, what is ‘‘voluntary’’ onthe part of the firm—in the sense that itrepresents a rational response to economiccircumstances—may not be socially optimal.There is little payoff from strengtheningminority shareholder protections, for in-stance, if firms are not going to access cap-ital markets because they are afraid ofalerting the tax authorities to their existence.Furthermore, if tax evasion has this effecton capital market participation, it may havea similar impact on other markets—includ-ing those for skills. Indeed, it is likely toundermine the viability of any arms-lengthtransaction that is based on disclosure andtransparency.

NOTES

1. As the OECD has put it, ‘‘Firms that operate byhiding all or part of their activities, employment, andrevenues do not grow naturally and hit a ‘‘glass ceiling’’constraining their development’’ (OECD, 2004, p. 191).In Turkey, for instance, there is a strong correlationbetween firm size, regulatory evasion, and the level oflabor productivity (OECD, 2004, p. 189).

2. A series of industry-level studies by McKinsey &Company in the late 1990s emphasized the competitiveaspects of relations between compliant and non-compli-ant entrepreneurs. Because they are outside the law, it isargued, non tax-compliant firms enjoy a substantial costadvantage over their competitors. This means they canafford to be less productive. According to McKinsey, thegap typically varies from 5–10% of the final productprice to around 40% (McKinsey, 1998, p. 515).

3. According to the Brazilian Institute for Geographyand Statistics (www.ibge.gov.br), the share of services inBrazilian GDP in 2005 was 64%, while that of manu-facturing industry—from which the sample analyzed inthis paper was drawn—was 18.4%.

4. According to the Doing Business database, theaverage number of procedures required to open a newbusiness is 7 in high-income countries, 10 in upper-middle-income countries and 12 in lower-middle-income

countries, and 11 in low-income countries (World Bank,2004, p. 84). Similar differences exist for other areas ofregulation.

5. There is a long-standing debate on the implicationsof corruption for firm-level efficiency and economicgrowth more generally. One view, associated withHuntington (1968) and others, holds that corruptioncan ‘‘grease the wheels of commerce’’ in economies withinefficient or poorly designed regulation. Under thesecircumstances some level of bribery may be optimal,since it allows more productive firms to buy lowereffective red tape (Lui, 1985). Another view, noworthodox at the World Bank and other internationalfinancial institutions, is that corruption constitutes anegative equilibrium which is detrimental to firmperformance and growth. Many scholars in this veinhave argued that low-level public officials may createregulatory obstacles and generate uncertainty in order toextort payment from entrepreneurs. For an overview ofthe debate and some firm-level evidence see inter alia

Bardhan (1997), Kaufmann and Wei (1999), and Fismanand Svensson (2000).

6. Of course there are many other reasons why owner-entrepreneurs might not want to tap equity markets,chief among them that doing so requires ceding control

http://www.ibge.gov.br


to outsiders. For a discussion of the issues involved andthe firm-level costs and benefits of participating in publicequity markets see Roell (1996).

7. Interview with Edwyn Neves, Analyst, AdventCapital, Sao Paulo.

8. ‘‘Firms, companies, etc. usually have two accounts.Some companies have one very correct one with income,output, and profit. But, there is another one with thesame information, reported differently. . .. This happensmore in small and medium enterprises, but the big onesare also experienced in taxes’’ (Brazilian financialanalyst, quoted in Lieberman, 2003, pp. 230–231).

9. For a detailed description of the sample design seeDe Farias (2003).

10. There are also empirical grounds for thinking thatthe survey responses accurately mirror the respondents’own experience. De Mel, McKenzie, and Woodruff, forexample, find a very tight correspondence between whatfirms reported as typical behavior for enterprises ‘‘sim-ilar in all respects’’ and their own behavior, as verified bythe subsequent reports of independent enumerators (DeMel, McKenzie, and Woodruff, 2007, p. 17). Rama-chandran et al. similarly find a high degree of correlationbetween data from the World Bank’s InvestmentClimate Surveys and independent measures of thebusiness environment at the country level, even forpotentially sensitive or incriminating questions such asthe tax compliance variable used here (Ramachandran,Gelb, Shah, & Turner, 2007).

11. It might be that the fiscal authorities deliberatelychoose to visit firms they know are under-reporting inorder to extract payment. But this does not appear to bethe case. The relationship between the firm havingreceived a visit from the tax authorities in the previous

year and its self-reported level of tax compliance is eitherweak (0.06 in the case of the state-level authorities) orstatistically insignificant (in the case of the federal andmunicipal authorities).

12. As a robustness check, I include sector dummies inplace of the labor intensity and fixed assets variables,and find that it makes no substantive difference to theresults. The relevant regression output is available fromthe author on request.

13. As with the tax compliance variable, it is possible,though unlikely, that firms deliberately misreported theirlevels of employment to the survey enumerators.

14. A simple linear regression of tax compliance onfirm size measured by employees explains less than 3% ofthe variance.

15. Response rates to this question were also lowerthan for the other variables employed in this analysis,though this is because around half the firms surveyedhad not received a visit from the tax authorities in thepreceding year, and their responses to subsequentinquiries about the nature of the interaction weretherefore recorded as missing.

16. Contrary to what one might expect, however,several firms reported having a public listing but noexternal audit. Of the 68 publicly listed firms in thesample, 49 reported having had their financial statementreviewed by an external auditor while 19 did not.

17. Since I am using a probit analysis, the parameterestimates cannot be interpreted in the same way as in alinear regression. Instead, I calculate marginal effects foreach independent variable, using STATA’s mfx com-mand.

REFERENCES

Bardhan, P. (1997). Corruption and development: Areview of issues. Journal of Economic Literature,35(3), 1320–1346.

Capp, J., Elstrodt, H., & Bebb Jones, W. Jr., (2005).Reining in Brazil’s informal economy. Sao Paulo:McKinsey Global Institute.

Centeno, M. A., & Portes, A. (2003). The informaleconomy in the shadow of the state. In P. Fernandez-Kelly, & J. Sheffner (Eds.), Out of the shadows:Political action & the informal economy in LatinAmerica (pp. 23–49). University Park: PennsylvaniaState University Press.

De Farias, A. M. L. (2003). Desenho Amostral da Pesquisasobre Clima de Investimento. Unpublished manuscript,World Bank, Washington, DC.

De Mel, S., McKenzie, D. J., & Woodruff, C. (2007).Measuring microenterprise profits: Don’t ask how thesausage is made. World Bank Policy ResearchWorking Paper 4229. Washington, DC: WorldBank.

Eilat, Y., & Zinnes, C. (2002). The shadow economy intransition countries: Friend or foe? A policy per-spective. World Development, 30(7), 1233–1254.

ETCO (Instituto Brasileiro de Etica Concorrencial)(2003). Comercio Illegal: Implicacoes e Alternativas.Sao Paulo, Brazil.

Feige, E. L. (1990). Defining and estimating undergroundand informal economies: The new institutionaleconomics approach. World Development, 18(7),989–1002.


Fisman, R., & Svensson, J. (2000). Are corruption andtaxation really harmful to growth? Firm level evidence.World Bank Policy Research Working Paper2485.Washington, DC: World Bank.

Gauthier, B., & Gersovitz, M. (1997). Revenue erosionthrough exemption and evasion in Cameroon. Jour-nal of Public Economics, 64, 407–424.

Gauthier, B., & Reinikka, R. (2001). Shifting tax burdensthrough exemptions and evasion, an empirical investi-gation of Uganda. World Bank Policy ResearchWorking Paper 2735. Washington, DC: World Bank.

Gehlbach, S. (2004). Taxability and low-productivitytraps. CEFIR Working Paper.

Gerxhani, K. (2004). The informal sector in developedand less developed countries: A literature survey.Public Choice, 120, 267–300.

Huntington, S. (1968). Political order in changing soci-eties. New Haven, CT: Yale University Press.

Johnson, S., Kaufman, D., McMillan, J., & Woodruff,C. (2000). Why do firms hide? Bribes and unofficialactivity after communism. Journal of Public Eco-nomics, 76(3), 495–520.

Kaufmann, D., & Wei, S. (1999). Does ‘Grease Money’speed up the wheels of commerce? NBER WorkingPaper 7093. Boston, MA: Harvard University Press.

Krueger, A. O. (2000). Agenda for future research: Whatwe need to know. In A. O. Krueger (Ed.), Economicpolicy reform: The second stage (pp. 3–20). Princeton,NJ: Princeton University Press.

Lieberman, E. S. (2003). Race and regionalism in thepolitics of taxation in Brazil and South Africa.Cambridge: Cambridge University Press.

Loayza, N. V. (1997). The economics of the informalsector: A simple model and some empirical evidencefrom Latin America. World Bank Policy ResearchWorking Paper 1727. Washington, DC: World Bank.

Lui, F. (1985). An equilibrium queuing model of bribery.Journal of Political Economy, 93(4), 760–781.

Mann, M. (1984). The autonomous power of the state:Its origins, mechanisms and results. Archives Europe-ennes de Sociologie, 25, 185–213.

McKinsey Global Institute (1998). Country studies onBrazil and Russia. Available at <www.mckin-sey.com/mgi>.

McKinsey & Company (2004). Eliminando as Barreirasao Crescimento Economico e a Economia Formal noBrasil. Sao Paulo: McKinsey & Company.

Organization for Economic Cooperation and Develop-ment (OECD) (2004). Establishing an open and non-discriminatory business environment. In OECD,Turkey: Economic Survey. Paris: OECD.

Ramachandran, V., Gelb, A., Shah, M. K., & Turner, G.(2007). What matters to African firms? The relevanceof perceptions data. Washington, DC: World Bank.

Rocca, C. A. (2004). Mercado de Capitais Eficiente:Condicao Para o Crescimento Sustentado. Availableat <http://www.capitalderisco.gov.br/vcn/pdf/EP0067.pdf>.

Roell, A. (1996). The decision to go public: An overview.European Economic Review, 40, 1071–1081.

Schneider, F., & Klinglmair, R. (2004). Shadow econo-mies around the world: What do we know? CESIFOWorking Paper 0403. Johannes Kepler University,Linz, Austria.

Tanzi, V. (2003). Corruption in public finance. In 8thInternational anti-corruption conference.

Viegas, L. (2005). Corporate governance in Brazilian non-listed companies: Challenges and opportunities.OECD meeting on corporate governance of non-listed companies, Istanbul.

World Bank (2002). World development report 2002:Building institutions for markets. Washington, DC:World Bank.

World Bank (2004). Doing business in 2004: Understand-ing regulation. Washington, DC: World Bank.

World Bank, H. (2005). World development report: Abetter investment climate for everyone. Washington,DC: World Bank.

World Bank (2007). World development indicators.Washington, DC: World Bank.

Available online at www.sciencedirect.com

tax evasion, disclosure, and

Documents