coming out the shadows - estimating the impact of bureaucracy simplification and tax cut on...

8/10/2019 Coming Out the Shadows - Estimating the Impact of Bureaucracy Simplification and Tax Cut on Formality in Brazili

1/11

Coming out of the shadows? Estimating the impact of bureaucracy simplication andtax cut on formality in Brazilian microenterprises

Joana C.M. Monteiro, Juliano J. Assuno Department of Economics, Pontifcia Universidade Catlica do Rio de Janeiro (PUC-Rio), Rua Marqus de So Vicente, 225-Gvea Rio de Janeiro, RJ, 22453-900, Brazil

a b s t r a c ta r t i c l e i n f o

Article history:

Received 20 September 2007

Revised 8 September 2011Accepted 16 October 2011

JEL classication:

D73E26K34

Keywords:

Informal economyTax legislationBureaucracy

This paper evaluates the impact of a program of bureaucracy simplication and tax reduction on formalityamong Brazilian microenterprises the SIMPLES program. We document an increase of 13 percentage pointsin formal licensing among retail rms created after the program when compared to rms in ineligible sectors.The impact on retailers is robust to a series of tests. We nd no impact on construction, transportation, ser-vices and manufacturing sectors.

2011 Elsevier B.V. All rights reserved.

1. Introduction

In most countries, a substantial portion of the GDP is produced bythe so-called shadow or underground economy. In Latin America, forexample, the size of the informal sector relative to ofcial GDP rangesfrom 25% to 50%. For OECD countries, underground activities account,on average, for 16% of GDP (Enste and Schneider (2000)).

A large body of literature addresses the measurement of the shad-ow economy. Many contributions are published in a special issue oftheEconomic Journal(109:456, June 1999) and a survey of the differ-ent methodologies and main estimates can be found inEnste andSchneider (2000). However, less attention has been devoted to thecauses and consequences of informality. Empirical evidence aboutkey determinants is still very scattered due, predominantly, to the ab-sence of data. As mentioned byEnste and Schneider (2000), gather-ing information about underground economic activity is difcult,because no one engaged in such activity wants to be identied.

This paper evaluates the effects of new bureaucracy simplicationand tax reduction legislation for micro and small rms in Brazil the

so-called SIMPLES system. The analysis of SIMPLES program offers agreat opportunity tocontribute to theliterature on determinants of infor-malitybecause the program promotes a sizeable reduction in tax burdenandreduces theredtape involved in taxpayments, andthereforebypass-ing cumbersome procedures that increase the costs of being formal.

Our evidence is based on a special cross-sectional survey of microand small rms conducted in 1997, less than one year after the imple-mentation of the program. This database along with the implementa-tion of this new taxation in Brazil provides an opportunity forinvestigating the informal economy at therm level.

Theidenticationstrategyisbasedonafewkeyaspectsoftheempir-ical environment. First, the new tax system is restricted to a subset ofsectors.We explore this characteristic using adifference-in-differenceap-proach, comparing the legal status ofrmsin sectors affected and not af-fected by the reform,created before andafter the program. By restrictingthe analysis to a single country, our empirical strategy is less subject toother changes in the legal environment than other studies based oncross-country comparisons. Second, we do have data on unofcialrms more than 75% of therms in our sample are unlicensed andwe investigate the variation in the ofcial registration ofrms.

We show that the SIMPLES program affects the formalization ofeconomic sectors differently. There is an increase of 13 percentagepoints in the licensing of retail rms, while the licensing of theother eligible sectors (construction, manufacturing, transportationand service) remains unaffected by the new legislation. Since only27% of the retailers which started-up before the program are licensed,this result represents a measurable reduction in unlicensedrms in

Journal of Development Economics 99 (2012) 105115

We would like to thank ureo de Paula, Srgio Firpo, Gustavo Gonzaga, NarcioMenezes Filho and Rodrigo Soares for useful comments and suggestions on thispaper. Financial support from CNPq and FINEP is gratefully acknowledged. Any remain-ing errors are our own. Corresponding author. Tel.: +55 21 35271078; fax :+55 21 35271084.

E-mail addresses:[email protected](J.C.M. Monteiro),[email protected](J.J. Assuno).

0304-3878/$ see front matter 2011 Elsevier B.V. All rights reserved.

doi:10.1016/j.jdeveco.2011.10.002

Contents lists available at SciVerse ScienceDirect

Journal of Development Economics

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / d e v e c
http://dx.doi.org/10.1016/j.jdeveco.2011.10.002http://dx.doi.org/10.1016/j.jdeveco.2011.10.002http://dx.doi.org/10.1016/j.jdeveco.2011.10.002mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.jdeveco.2011.10.002http://www.sciencedirect.com/science/journal/03043878http://www.sciencedirect.com/science/journal/03043878http://dx.doi.org/10.1016/j.jdeveco.2011.10.002mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.jdeveco.2011.10.002


2/11

the retail trade sector. The estimated impact on retailers holds after aseries of robustness checks, accounting for differentiated time effectsacross groups, the possibility of split-ups, rm age measurement er-rors, time to formalize and sector or occupational changes.

The result that SIMPLES effect varies by economic sector is in linewith previous indications that the requirements to enter and operatein the formal sector depend on the economic activity (seeCapp et al.(2005), Farrell (2004), Paula and Scheinkman (2010)). Since SIMPLES

promoted a partial reduction in the cost of being formal, it is possiblethat the incentives generated vary by sector. We discuss further in thepaper why wend positive results just for the retail sector. In partic-ular, we argue that the benets of SIMPLES reform are clearer for re-tailrms. Transportation and constructionrms face other importantbarriers to register and there is uncertainty over eligibility in the ser-vice and manufacturing sectors.

We are also aware that our empirical setting leaves open the pos-sibility that the effect on the retail sector could be generated by a spe-cic sectoral shock coincident with the SIMPLES reform. Therefore, inaddition to the interpretation described above, one can argue that theSIMPLES reform had little signicant impact, and that the effect onthe retail sector is a statistical artifact generated by another simulta-neous sectoral shock. Even after the robustness tests we provide, weare not able to rule out this possibility, especially because we cannotexplore possible variations of the reform within our sample. Only fu-ture research, with better data, can shed light on these issues.

This paper contributes to the literature on determinants of formal-ity. A rst branch of this literature is primarily based on cross-countrycomparisons.Johnson et al. (1997) and Johnson et al. (1998)presentevidence of close relationships between the size of the unofcialeconomy, taxes, quality of public goods, regulatory discretion, andcorruption. Based on a sample of 69 countries and using an instru-mental variable approach,Friedman et al. (2000)suggest that bu-reaucracy, corruption and a weaker legal environment are alldeterminants of the informal sector. However, theynd that the taxrate has no effect on informality.Djankov et al. (2002), studying 85countries, show that rms have signicant entry costs, both interms of time and monetary fees for registration and licensing, and

that stricter entry regulation is associated with higher levels of cor-ruption and the size of the unofcial economy.Auriol and Warlters(2005), using a sample of 53 countries, also show that the shadoweconomy diminishes when thexed cost of market entry is reduced.

Recently, there is an emerging literature exploringrm-level dataand within country variation to understand the causes and conse-quences of informality.McKenzie and Sakho (2010)argue that prox-imity to tax registration ofce increases the information arm hasabout registration and, by using the distance of a rm to the tax ofceas an instrument, shows that registering to pay taxes leads to signi-cantly higher prots among middle rms but a decrease in prots ofsmall and large rms.Paula and Scheinkman (2010)show that for-malization is affected by the tax structure in different value chains.Bruhn (2011) nds that an improvement in business entry regulation

stimulates formality. By analyzing the economic effects of a reformthat simplied business entry regulation in Mexico, she nds thatthe reform increased the number of registered businesses by 5%,which was driven mostly by former wage earners opening businesses.However, her results indicate that the program did not stimulate reg-istration among existing informalrms.De Mel et al. (2011) nd thatsimply providing information and reimbursing the cost of registrationis not enough to leadrms in Sri Lanka to register. By conducting aeld experiment which provided incentives to register to randomlyselected informal sectorrms, they show that only payments equiva-lent to one-half to one month prot leads to an increase in registra-tion. Among the rms not registering after receiving the largerincentive, authors show they face issues related to ownership of land.

Taking all together, these studies suggest that the effect of a reform

aiming to reduce to costs of formalization may be heterogeneous

among rms, depending on the economic and institutional environ-ment. The specic constraints that preventrms to become formal canvary across sectors and circumstances. Therefore, reforms that do nothave a sufcient large scope may fail to increase formalization in somesectors. In this context, our results contribute to a better understandingof the determinants of formality a joint effort of taxreduction andbu-reaucracy simplication is found to be effective in reducing informalityin retail sector, although having no effect on construction, transporta-

tion, services and manufacturing.The rest of the paper is organized as follows.Section 2providesthe institutional background of the SIMPLES reform. Data are pre-sented inSection 3. The empirical strategy is presented inSection 4.Main results are depicted inSection 5,while a series of robustnesschecks are provided inSection 6. InSection7, we discuss why the re-sults are expected to vary by sector. Finally, we summarize our mainndings in the conclusion.

2. Institutional background: the SIMPLES reform

The SIMPLES system (Sistema Integrado de Pagamento de Impos-tos e Contribuies das Microempresas e Empresas de Pequeno Porte)was enacted in December 1996 with the objective to reduce and sim-plify the tax system to micro and smallrms. According to the Law,microenterprises are rms with an annual revenue equal to orlower than R$ 120,000, while small rms are the ones with annualrevenue up to R$ 1,200,000.1

The system combines six different federal taxes and social contri-butions into one single and monthly-based rate. The taxes included inthe system are IRPJ (corporate income tax), PIS/PASEP (contributionto employees' savings programs), CSLL (contribution on net prot),COFINS (contribution for nancing the social security system), IPI (in-dustrialized products tax) and the employer's social security contri-bution. The system represents a partial simplication and reductionof the tax burden sincerms still need to pay for other federal, stateand municipal taxes.2

The Law also opens the possibility that states and municipalities col-lect their most important taxes, respectively, ICMS (value-added tax)

and ISS (servicetax), through the SIMPLES system. However, in October1997 when our survey was carried out, only 45 municipalities in thecountry (out of 5565) have signed an agreement to collect ISS throughthe SIMPLES system.3 No Brazilian state adhered to the system in1997. Apartfrom that, thesystem was setnationally andthereis no var-iation across states or municipalities. In 2006, the governmentadded themunicipal and state taxes SIMPLES system (instead of simply allowedthis possibility). With this modication, the incentives to adhere to thelaw can vary among states. Unfortunately, there is no survey that en-ables us to assess the impact of this second reform.

Table 1 presents a diagram that shows the tax and bureaucratic re-quirements that an entrepreneur needs to perform under the Brazil-ian regular tax system and the requirements of SIMPLES. Under theregular system, the burden ofve taxes covered by SIMPLES (IRPJ,

CSLL, COFINS, PIS and IPI) varies from 5% to 11% of gross revenue,depending on the economic activity. In addition,rms must contrib-ute with 20% of the payroll to the social security.

On the other hand,rms under the SIMPLES system pay a singlerate that varies from 3% to 5% of the total revenue for microenter-prises and from 5.4% to 8.6% for small rms. The SIMPLES rates formicroenterprises are the following: 3% of total revenue for rmswith annual gross revenue up to R$ 60,000; 4% forrms with annual

1 The exchange rate in December 1996 was US$ 1=R$ 1.0365.These limits were in-creased, respectively, to R$ 240,000 and R$ 2,400,000 in 2004.

2 Federal taxes not included in the system include FGTS, employees' social securitycontribution (INSS dos empregados), IOF (nancial operations tax), CPMF, ITR (landproperty tax) and II (Import tax).

3 Source: http://www.receita.fazenda.gov.br/PessoaJuridica/SIMPLES/Municipios

Conveniados.htm

106 J.C.M. Monteiro, J.J. Assuno / Journal of Development Economics 99 (2012) 105115
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://www.receita.fazenda.gov.br/PessoaJuridica/SIMPLES/MunicipiosConveniados.htmhttp://www.receita.fazenda.gov.br/PessoaJuridica/SIMPLES/MunicipiosConveniados.htmhttp://www.receita.fazenda.gov.br/PessoaJuridica/SIMPLES/MunicipiosConveniados.htmhttp://www.receita.fazenda.gov.br/PessoaJuridica/SIMPLES/MunicipiosConveniados.htmhttp://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


3/11

revenue between 60,000 and 90,000; 5% for the ones with annual rev-enue between 90,000 and 120,000. For smallrms, rates begin in 5.4%for the ones with annual revenue between 120,000 and 240,000 andincrease by 0.4% for every R$ 120,000 of additional annual revenue.

In our main sample, only two

rms (out of 6127) have annual reve-nue above 60,000. The SIMPLES rate for most of our sample is 3%.Table 1also emphasizes that SIMPLES promoted a reduction in the

red tape. Brazilian rms must report to the Internal Revenue Servicethat they paid specic taxes by lling specic forms on a frequentbasis. Under SIMPLES system,ve tax forms were substituted by one,which needs to be lled just once a year. The system did not changethe procedure to registerrms, which is specially cumbersome in Brazilas pointed out by Doing Business report (seeWorldBank (2004)).4

The SIMPLES system does not affect all Brazilian micro and smallenterprises. The economic activity performed by therm determineswhether it can apply to participate in the system or not. Real estatecompanies, developers, the nancial sector, trading companies, ad-vertising agencies, cleaning servicerms and outsourcing companiesare not eligible. In addition, SIMPLES system is not an option for rmsthat provide services which require professionals with regulated oc-cupations such as dentists, physicians, auditors, architects, engineers,

journalists, actors, sales representative, musicians and others.By the time the law was enacted and the data were collected the fol-

lowing activities were clearly included in the system: retail trade,manufacturing which does not require a professional with a regulatedoccupation, transportation, civil construction and other services whichdo not require a professional with a regulated occupation. Accordingto ofcial data,5 2/3 of formal Brazilian rms used the SIMPLES system

in 1997. This percentage remained barely unchanged in the followingyears. The appendix presents the economic activities in our databasethat areclearly eligible for the system, the sectors which are notcoveredby the legislation, and a third group of sectors about which the legisla-

tion is unclear. This latter group was dropped from the analysis.6

We should emphasize that this exercise to classify rms as eligibleand ineligible is not straightforward for the service and manufactur-ing sectors. The main problem is that the law does not state explicitlywhich activities require professionals with regulated occupations.Indeed, the Internal Revenue Service has disclosed several pieces oflegislation and normative acts since the Law was enacted to deter-mine whether specic activities are eligible or not. In practice, manyrms have the conrmation of the adhesion to the system only aftertheir processes have been analyzed and approved by the Internal Rev-enue Service. For an idea of how uncertain was the eligibility statusfor some activities, rms that provide services of maintenance and re-pair of vehicles, machines, computers and home appliances were ini-tially considered not eligible because they were associated with theengineering profession, although it is hard to believe that an engineerworks in a auto repair rm. In 2004, the government reviewed thatrestriction and allowed theserms to adhere SIMPLES.

In this paper, we use the classication criteria in effect by the timethe database was collected (October 1997). However, for many rmsin the service and manufacturing sectors the eligibility statusdepended on theinterpretation of the Law by therm's owner, his ac-countant (who suggests the tax system and does the application) andthe Internal Revenue Service. This can challenge the analysis for ser-vice and manufacturing sectors as discussed inSection 7.

4 Recently, several municipalities and states tried to simplify the procedures or/andcreate one-stop-shops to facilitate formalization. We are not aware of any initiative ofthis type which were in place in 1997 that could be used in our analysis.

5 Secretaria da Receita Federal (Internal Revenue Service), Coordenao-Geral de

Estudos Econmico-Tributrios.

6 Otherrm's characteristics also affect their eligibility. Firms listed in the stock mar-kets, which have a foreign partner that lives abroad or partly owned by another com-pany are not eligible to the system. We don't have information on whether therms inour sample have these characteristics or not. But we don't think it is a concern due to

the type of entrepreneur under analysis.

Table 1

Cost and time to pay taxes.

Construction Transport

ation

Service 1 2 3 4 5 6 7 8 9 10 11 12

Regular system

Taxes

Employer's

INSS

CSLL

COFINSPIS

IRPJ

IPI

Total

Red tape

DCTF

DACON

MonthsMonetary cost by sector

R$ 500 fine for not filling in the form

Once a

year


DIPJ

DIPI R$ 500 fine

DIRF

SIMPLES system

Taxes

SIMPLES

Tax base

Gross revenue

Gross revenueGross revenue

Gross revenue

Value-added

Gross revenue

Payroll

Gross revenue

Retail

trade

1.08%

2%0.65%

1.2%

0%

4.9%

20%

3%8.6%

1.08%

2%0.65%

4.8%

0%

8.5%

20%

3%8.6%

1.08%

2%0.65%

2.4%

0%

6.1%

20%

3%8.6%

1.08%

2%0.65%

2.4%

0%

6.1%

20%

3%8.6%

1.08%

2%0.65%

1.2%

0%20%

4.9%10.9%

20%

3.5%9.1%

Red tapeDAS



R$ 500 fine for not filling

Manufacturing

Note: This Table is based on several laws and instructions that regulate Brazilian tax system. The rates indicated are the ones in effect in October 1997. IPI is the tax that levies inindustrialized products and its rate varies according to the product manufactured by the rm. To calculate total taxation in manufacturing sector, we considered that the type ofproduct in our sample has an IPI rate that varies from 0% to 20% and assumed that 30% of the revenue of manufacturing rms is value added by the rm. Firms considered IPItaxpayers have a SIMPLES rate of 0.5% higher, even when the IPI rate is 0%. SIMPLES rate varies according to the following rule: 3% of the total revenue for rms with annual grossrevenue upto R$60,000;4% forrmswithannual revenue between 60,000and90,000;5% fortheoneswithannual revenue between 90,000and120,000.For smallrms,rates beginin5.4% for theoneswithannualrevenue between 120,000 and240,000andincreaseby 0.4% forevery additional R$ 120,000 ofannualrevenue.Red tape refersto the forms thatrmsneed to ll in to inform the Internal Revenue Service that they paid the taxes.

107J.C.M. Monteiro, J.J. Assuno / Journal of Development Economics 99 (2012) 105115
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


4/11

3. Data

We use data on rms gathered by the Urban and Informal Econo-my Survey (ECINF) conducted by the Brazilian census bureau in Octo-ber 1997. The survey comprises about 40,000 rms located inBrazilian state capitals and metropolitan areas.

The survey was conducted by using a probabilistic sample ofhouseholds, which were selected in two steps. In the rst round,

households were selected with probability proportional to the per-centage of households which declared that its head was employedin the 1991 Demographic Census. In the second round, heads ofhouseholds who were self-employed or small employers (with lessthan 5 employees) were stratied according to their economic activ-ity and selected with respect to uniform probability in each strata.

The key denition in our analysis is informality. According toGerxhani (2004), there is a full range of descriptions and concepts the most common denitions are based on the size of the rm,labor regularization, licensing, tax evasion, among others. Our analy-sis focuses on a crucial step for arm to becomelegalin Brazil, whichis the holding of an ofcial license, issued by a state or a municipalityauthority.Ofcial licensingis an essentialrequirement forrmstobeableto print anofcial invoice for taxpurposes.Firmsare subject to a

series of penalties and nes in case of not having these licenses. Only24% of the rms in our sample held ofcial licenses. Our study doesnot address labor informality. The registration of wage workers, forexample, is not our focus here and thus we restrict our sample torms without a payroll.

Another characteristic of our sample is that allrms that were cre-ated more than 20 months before the survey are excluded.7 Part ofour empirical strategy is based on the comparison betweenrms cre-ated before and after the SIMPLES system. Since the survey was col-lected 10 months after the enactment of the SIMPLES law, weconsider the same time window of 10 months to build the set ofrms created before the law.

Table 2presents the variables considered in our study. Basically,we have information about rms and their owners. The dataset hasinformation about the economic activity, location, sales, assets, non-paid workers, main customers (individuals, small and large rms, orthe government), origin of thenancial resources invested, compli-ance with different governmental registrations and others. Aboutowners, the dataset has information on gender, educational level,age, and time when s/he started the business.

About 40% of the

rms in the sample have not indicated their assetvalue. Thus, to avoid losing information, we created a new asset var-iable with missing points replaced by zero along with a dummy vari-able indicating rms without information on assets. All othervariables are considered exactly as they are available in the survey.

4. Identication strategy

Our empirical analysis aims to evaluate the effect of SIMPLES sys-tem onrms' formality. As mentioned before, the outcome variable isa binary one indicating the possession of an ofcial license. Ideally, wewould like to compare the probability of ineligible rms holding a li-cense after the program with the probability of theserms being for-mal in the absence of the program. However, we face a typical

missing data problem since

rms are observed as either facing theprogram or not, but not both. As a consequence, constructing thecounterfactual is the central issue in the analysis.

Our main strategy relies on the use of ineligiblermsasameansofbuilding counterfactuals. Since the SIMPLES program was designedfor a subset of sectors of the Brazilian economy, we userms from in-eligible sectors to build a comparison group. In the end, we contrastthe responses of eligiblerms, which constitute thetreatment group,with the responses of ineligiblerms in thecomparison group.

Another important issue in the analysis is the time dimension.Firms in the treatment group may exhibit structural and signicantdifferences when contrasted with rms in the comparison group,and that is a potential source of problems. These differences can be ei-ther observed or non-observed. Observable variables are introducedexplicitly into the analysis. Non-observable differences, however,are considered implicitly through time differences.

Although we have only a cross-sectional survey ofrms in October1997, we introduce a time dimension in our analysis by consideringrms created before and after the new legislation, which was imple-mented in December 1996. Since the survey was collected 10 monthsafter the implementation of the SIMPLES system, we consider thesame time frame of 10 months for building the set ofrms createdbefore the program. Therefore, any difference betweenrms in thetreatment and comparison groups that is constant with respect to arm's age is controlled in our strategy. Firms in the treatment andcontrol groups are assumed to share the same aggregate shocks af-fecting their decision to register.

An underlying assumption in this approach is that formalization ispredominantly decided at the creation of therm. This assumption is

not testable in our sample, due to lack of information. However, basedon the 2003 edition of the same survey,Table 3shows strong evi-dence in this direction.8 Almost 90% of the owners of unlicensedrms did not try to formalize their business at the startup. This per-centage is reduced to 24% in the case of licensedrms. For approxi-mately 3/4 of licensed rms, formalization occurred when theywere starting-up. This evidence suggests a strong correlation be-tween current legal status and formalization attempts at startup.

The possibility that formalization is not decided at the startup in-troduces a potential negative bias in our analysis. Eligiblerms creat-ed before the law still face a decision whether to formalize or not

7

Firms with more than 20 monthsare considered in therobustness checksofSection 5.

Table 2

Summary statistics of eligible and ineligible rms.

Total Ineligible Eligible

Mean Mean SE Mean SE

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

Dependent variable

Licensed rm 0.24 0.26 (0.02) 0.23 (0.01)

Characteristics of the ownerPrimary education 0.58 0.37 (0.03) 0.65 (0.01)Secondary education 0.25 0.32 (0.02) 0.23 (0.01)College degree 0.11 0.28 (0.02) 0.05 (0.01)Age 35.17 33.56 (0.53) 35.77 (0.34)Male 0.64 0.58 (0.03) 0.66 (0.01)Owns his house 0.72 0.68 (0.02) 0.74 (0.01)Has another job 0.11 0.15 (0.02) 0.09 (0.01)

Characteristics of therm

Total assets 3544.51 4638.38 (1968.74) 3135.00 (277.99)Did not declare assets 0.35 0.38 (0.02) 0.34 (0.01)Revenue 883.31 876.63 (79.50) 885.82 (66.97)Located out of owners' house 0.61 0.58 (0.03) 0.62 (0.01)Sell to other rms and

government0.14 0.21 (0.02) 0.12 (0.01)

Startup was nanced by

the owner

0.49 0.40 (0.03) 0.53 (0.01)

Has non-paid employees 0.05 0.01 (0.00) 0.06 (0.00)Employs owner's relatives 0.10 0.04 (0.01) 0.13 (0.01)

Note: This table presents the means and standard-errors of the main variables used inour analysis. Column 1 shows the meanfor the whole sample and columns 34 and 56present the means and standard errors for ineligible and eligible rms.

8 ECINF/2003 is a more recent edition of the survey conducted in 1997, with a morecomprehensive questionnaire. The newer edition was based on the same procedures of

the 1997 edition, with the same sample design.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/11

theirs activities afterwards. In other words, the SIMPLES law might in-crease the formalization of eligible rms created before December1996. Consequently, when we compare eligiblerms created after-wards with eligible rms created before the SIMPLES system, wecan get a negative bias in the effect of the new tax system on formal-ization. For example, if all eligible rms become licensed with thenew tax system, our strategy would lead us to estimate a wrong im-pact of 0, since we consider only thegure of October 1997. In thissense, our estimate can be interpreted as a lower bound for the effectof the SIMPLES system to the formalization.

In summary, our empirical approach is based on a difference-in-difference strategy. The impact of the new SIMPLES system on infor-mality is estimated through comparisons between rms in eligibleand ineligible sectors, created before and after the new legislation.In this sense, we face two usual and important assumptions of any

difference-in-differences analysis. First, we assume that there arecommon time trends across groups rm's age has the same effecton eligible and ineligiblerms. Second, we assume there are no sys-tematic changes within groups. We address these issues and otherconcerns inSection 6, which provides several robustness checks.

Descriptive statistics of observable characteristics of eligible andineligiblerms and their owners are also presented inTable 2.

5. Empirical results

Our starting point is an unconditional analysis with raw data.Table 4shows the percentage ofrms holding ofcial licenses in dif-ferent groups. Therst two lines refer to therms in the comparisonand in the treatment group, respectively. Then, the treatment group is

decomposed with respect to different sectors.Columns 1 and 4 indicate the number ofrms in the sample that

were created before and after SIMPLES reform. Therst group is com-posed byrms which are in operation for 11 up to 20 months and thesecond group is formed by young rms which had less than10 months of operation in October 1997. We observe that there is ahigher number of youngrms, which may reect the high mortalityrate among microenterprises in Brazil. The number of youngrms isapproximately 35% higher than the number of olderrms in the retailand manufacturing sectors and also in the comparison group. Trans-portation and service sectors are exceptions in this pattern. Whilethe number of youngrms is relative the same of the olderrms inthe transportation sector, it is almost 50% higher in the service sector.

Table 4presents a decrease in the proportion of licensed rms

both in the comparison and treatment groups (columns 2 and 5).

This is expected because the process of formalization is time consum-ing. Thus, it is natural to observe a reduction in licensing in the com-parison betweenrms created before and after the SIMPLES programdue to the time required to complete all the paperwork involved inthe process. Younger rms in the comparison group (ineligiblerms) have a lower formalization rate (24%) than older rms in thisgroup (28%). Firms in the treatment group (eligible rms), on theother hand, have a license rate 3 percentage points lower (25% versus22%). This rst approximation indicates that the new legislation in-creases formalization of eligible rms, on average, by 1 percentagepoint.

Formalization is also an issue of multiple aspects. Effective tax ratesand regulatory costs usually differ from sector to sector. Some sectorsare more prone to informality because of tax rates, while other sectors

are more informal because of strict operational obligations, such as san-itary and environmental requirements, quality control, safety measures,copyright rules and so on. Thus, the causes of informality can vary con-siderably among different sectors. Indeed, when the treatment groupis disaggregated into sectors, signicant differences are uncovered. Re-tailers, for example, expanded licensing by 9 percentage points, from27% to 36%, suggesting that the SIMPLES program increases by 13 per-centage points the formalization of suchrms.

However, the resultsofTable 4 might be driven by the characteristicsof the rms and their owners. In order to account for such variation, wecarry out a linear regression analysis. The basic difference-in-differences(DID) estimates of the introduction of SIMPLES on formality ofrms arebased on the following equation for armi:

Pr Zi 1Xij TI afterf gi GI eligiblef gi TGI afterf gi I eligiblef gi

Xi;

1

whereZiis a binary variable indicating whetherrmiis licensed;Ii{after}

denoteswhetherrm i wascreatedafter December1996;Ii{eligible} repre-

sents whether rmiis eligible for the SIMPLES; andXiis a vector of ob-served characteristics. Notice that Eq.(1)is a linear probability model,which provides easier interpretations for the marginal effects on theprobability of licensing.9 The parameter of interest is TG. Under the as-sumption that the selection into eligiblerms, conditional onXi, is inde-pendent of the age individual-specic effects ofrms, it measures theaverage effect of the SIMPLES on the licensing of eligiblerms.

Table 4

Number and percentage of licensed rms created before and after the program bysector.

Created beforeSIMPLES

Created after SIMPLES Allrms

N %licensed

SE N %licensed

SE N %licensed

(1) (2) (3) (4) (5) (6) (7) (8)

Compar ison 623 0.28 (0.04) 864 0.24 (0.03) 1487 0.26Treatment 2020 0.25 (0.02) 2648 0.22 (0.02) 4668 0.23Retail 513 0.27 (0.04) 689 0.36 (0.03) 1202 0.32Construction 275 0.11 (0.07) 329 0.05 (0.02) 604 0.08Manufacturing 237 0.28 (0.08) 329 0.11 (0.03) 556 0.18Transportation 398 0.37 (0.04) 418 0.39 (0.05) 816 0.38Services 597 0.25 (0.03) 883 0.20 (0.02) 1480 0.22Total 2643 0.26 (0.02) 3512 0.23 (0.01) 6155 0.24

Note: This table reports the number of rms in our sample and the percentage oflicensed rms before and after the program by sector. The rst two lines correspondto the comparison group (ineligible rms) and the treatment group (eligible rms),while the following ve lines disaggregate the treatment group into economicsectors. A licensed rm is the one which holds a municipal or state license. Thepercentage of licensed rms and the standard errors were computed using thesample weights.

9 As a matter of fact, our results do not change (quantitatively or qualitatively) if weuse aprobit model. Moreover, the percentage of predicted values outside the [0,1] in-

terval is not high (around 10% of the sample).

Table 3

Obstacles and attempt to formalization at the startup, 2003.

Unlicensedrms

Licensedrms

Total

Firms with obstacles to formalizeat the startup

316,610 449,728 766,3384.1% 17.9% 7.5%

Firms without obstacles toformalize at the startup

590,806 1,438,968 2,029,7747.7% 57.4% 19.9%

Firms which did not try to

formalize at the startup

6,771,162 607,988 7,379,150

88.0% 24.3% 72.3%Total 7,695,819 2,506,809 10,202,628

75.4% 24.6%

Note: This table is based on ECINF/2003 which is a more recent edition of the surveyconducted in 1997, with a more comprehensive questionnaire. The newer editionwas based on the same procedures and sample design of the 1997 edition, which isthe one used in the rest of the paper. Each cell in the table presents two numbers the number ofrms in that position and the percentage with respect to the total ofeach column (except the last line). All statistics are expanded through the sampleweights. The total sample used in the table comprises 47,196 rms. The table showsthat 88% of the owners of unlicensed rms did not try to formalize at the startup. Onthe other hand, 75.3% (17.9%+57.4%) of licensed rms engaged in the formalizationprocess at the startup.

http://-/?-http://-/?-http://-/?-http://-/?-


6/11

Table 5presents different estimates for Eq.(1). All regressionscontrols for the whole set of observed characteristics described inTable 2. In Panel A, we show the results when we consider justrms managed by self-employed entrepreneurs, which is our baselinesample. Column 1 corresponds to the full sample estimate with alleconomic sectors, while columns 2 to 6 consider theve sectors inthe treatment group taken separately. These columns consider thesame control group and the observations of the treatment groupfrom each sector. For example, in column 2, the sample comprisesthe whole comparison group plus retailers in the treatment group.Column 3 corresponds to a sample formed by the comparisongroup and eligible rms from the construction sector, and so on.The parameters G and T represent the differences in licensingamong eligible and ineligible rms and between rms created be-fore and after the new legislation, respectively.Table 5shows thatthe observable variables are capturing all the statistically signicantdifferences between groups, with respect to either eligibility for the

program or startup time. All estimated parametersGand Tare notsignicantly different from 0, with the only exception beingGforconstruction.

Considering the sectors altogether, we do not observe a statistical-ly signicant effect of the introduction of SIMPLES on the formality ofrms. However, the following columns reveal important heterogene-ity in the response ofrms belonging to different sectors. Retailrmspresent a positive and statistically signicant increase of 13 percent-age points in licensing. The other sectors do not exhibit any signi-cant response to the new tax program. Panel B shows that theresults are essentially the same when we include in the samplerms which have up to ve employees. Again, the results indicatethat formalization increased only among rms in the retail sector,though the effect is reduced to an increase in 9 percentage points in

licensing.

In addition to vary by sector, the incentives that SIMPLES providemay change according torms' prole. On the one hand, largerrmsand the ones located outside owner's house may be more keen to for-

malize since they are more visible and easier tobe monitored by thegovernment. On the other hand, smallerrms and the ones locatedin the owner's house compose a group that have few incentives to be-come formal in the regular system and can be specially stimulated bythe new system to enter formality.

Table 6analyzes the prole of the retailers at the margin of be-coming formal by showing the results of the following regression:

PrZi 1Xij TI afterf gi GI

eligiblef gi TGI

afterf gi I

eligiblef gi

TGWI afterf gi I

eligiblef gi Wi Wi

Xi;

2

where Wi is a interaction variable that indicates arm's characteristic.In the rst three columns, we consider alternative measures of

rm's size. In column 1, we analyze whether the effect of SIMPLESon formalization is larger for rms which have 1 to 5 paid em-ployees,10 while in columns 2 and 3 we investigate whether there isa differential effect for rms with larger annual revenue and assets,respectively.

The results in column 1 show that, althoughrms with paid em-ployees have a higher probability of holding a license, they are notspecially stimulated by SIMPLES to obtain a municipal or state li-cense.11 Results from columns 2 and 3 reinforce the idea that SIMPLES

Table 5

Difference-in-difference estimates.

All sectors Retail trade Construction Manufacturing Transportation Services

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

Panel A Self-employed entrepreneurs

Eligible created after SIMPLES (YTG) 0.032 0.130 0.004 0.116 0.048 0.020(0.041) (0.046) (0.034) (0.113) (0.070) (0.037)

Eligible (YG) 0.005 0.045 0.201 0.094 0.046 0.050(0.050) (0.049) (0.047) (0.094) (0.085) (0.057)

Created after SIMPLES (YT) 0.051 0.054 0.055 0.054 0.044 0.050(0.035) (0.036) (0.035) (0.036) (0.035) (0.036)

Observations 5911 2517 2033 1988 2242 2871R2 0.156 0.176 0.226 0.215 0.205 0.203

Panel B Self-employed entrepreneurs and rms with up to 5 employees

Eligible created after SIMPLES (YTG) 0.007 0.093 0.027 0.095 0.006 0.005(0.032) (0.043) (0.030) (0.088) (0.060) (0.031)

Eligible (YG) 0.014 0.071 0.200 0.045 0.044 0.048(0.049) (0.054) (0.042) (0.073) (0.091) (0.054)

Created after SIMPLES (YT) 0.016 0.019 0.022 0.019 0.014 0.019

Observations 8408 2892 2280 2352 2515 3285R2 0.203 0.231 0.256 0.242 0.217 0.239Owner's characteristics Yes Yes Yes Yes Yes YesFirm's characteristics Yes Yes Yes Yes Yes YesState dummies Yes Yes Yes Yes Yes Yes

Note: This table reports OLS coefcients of Eq.(1)where the dependent variable is a binary variable indicating whether the rm holds a state or municipal license. All regressionscontrol for rm's owner characteristics (dummies indicating primary education level, secondary education level, college degree, age, gender, lives on his/her own house, has other

job), characteristics of the rm (total assets, annual revenue, and dummies indicating non-declaration of assets, location out of owner's house, sales to other rms and government,startup was nanced by the owner, non-paid employees in the rm, relatives employed in the rm), and 27 state dummies. The sample used varies across the columns and panels.Panel A includes only rms without employees, while panel B includes both rms managed by self-employed entrepreneurs andrms with up to ve employees. Column 1 showsthe results when we include all economic sectors. Columns 2 to 6 depict the estimates for the ve economic sectors in the treatment group. The sample used in each of the columns2 to 6 comprises the whole set of ineligible rms and the set of eligible rms in the correspondent sector. Robust standard errors clustered at economic activity are reported inparentheses. Signicantly different than zero at 99%. Signicantly different than zero at 95%.

10 ECINF survey only interviews rms with up to 5 employees.11 Additional exercises also show that SIMPLES does not increase formalization of

workers among rms with paid employees (results not shown and available upon

request).

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


7/11

does not have a differential effect on larger rms. These results mightbe determined by the fact that almost all rms in our sample have an-nual revenue less than R$ 60,000 and, consequently, face the same(and lowest) tax rate. In column 4 we analyze whetherrms locatedoutside the owner's house are more affected by the SIMPLES system,since they have higher exposure to inspections and monitoring. The

coefcient of the interaction is not statistically signicant. Finally, col-umn 5 looks at whether SIMPLES has a differential effect on rms thatsell to otherrms and government, which are clients that usually de-mand a tax receipt. The results indicate that SIMPLES actually reducethe probability of this group ofrms to obtain a license. This result,however, should be interpreted with caution because it is identiedby a small number ofrms. Only 68 out of 1,086 rms, or 7% of thesample, sell to this type of clients and among these rms only 19have a license.

Taken together, the results fromTable 6do not indicate that SIM-PLES has a differential effect on rms that are more likely to demandformalization because of their customer base or their exposure toinspection.

We now decompose the impact according to the previous occupa-

tions and sectors. This exercise not only sheds light on the nature ofthe changes induced by the reform, but also contributes to a bettercharacterization of potential selection bias in our analysis. The SIM-PLES system improves the business environment of eligible sectorsrelative to ineligible ones. Thus, the new tax regime may stimulateentrepreneurs to switch from an ineligible sector to an eligible one.This switch would not be very difcult since the sectors which are af-fected by the reformdo notrequire special skills.12 Food retailers, gar-ment manufacturing, transportation and food services are highlyrepresented among eligiblerms in our sample. If this switching ishappening, the difference-in-differences estimator might not reectany improvement in formalization but rather a reallocation amongsectors. The same argument can also be applied to occupations.

InTable 7, we interact the binary variable which indicates that therm was created after SIMPLES law and is from the retail sector withowner's position in his previous occupation (column 1) and the eco-nomic sector the owner used to work (column 2). We follow thesame econometric specication of Eq.(2).

Column 1 inTable 7 analyzes how the estimated impact of the

SIMPLES reform on retailers varies according to the previous occupa-tion. Column 1 shows that rms whose owners were self-employersand employers in previous occupation are more affected by the SIM-PLES when compared to the other occupations. A test of equality ofcoefcients shows that the effects on self-employers and employersin previous occupation are statistically the same, but different fromthe effect on non-workers.

Sector transitions are considered in column 2 ofTable 7. There isno statistically signicant difference between the impact on rmswhose owners were retailers and impact on rms whose ownerswere non-retailers in their previous business. The point estimate ofthe two is quite similar, although only the coefcient of non-retail issignicantly different from zero.

Table 7suggests that the increased formalization induced by the

SIMPLES reform is not likely to be determined by transitions betweenoccupations or sectors. This evidence reinforces the numbers pre-sented inTable 4, which show that the change in the number of re-tailers created before and after the reform is similar to the changein the number ofrms in the comparison group.

6. Testing for pitfalls

The results of the previous section suggest that SIMPLES increasedthe proportion of licensed retailers by 13 percentage points. However,our empirical strategy is subject to caveats that are addressed below.The objective of the following exercises is to reduce the chance ofhaving our results generated by other reasons not related to the

changes in the tax legislation. The estimates are presented inTable 8.

Table 6

Decomposition of the effect on retailers.

Dependent variable Firm holds of cial license

Interaction variable Employer Annual revenue Assets Located outside owner's house Sell to rms or government

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

Eligible created after 0.100 0.144 0.126 0.189 0.165SIMPLES (YTG) (0.043) (0.044) (0.047) (0.062) (0.054)

Eligible created after SIMPLES 0.059 0.001 0.004 0.103 0.330

interaction variable (0.054) (0.001) (0.003) (0.076) (0.145)

Eligible (YG) 0.065 0.044 0.044 0.047 0.053(0.050) (0.047) (0.049) (0.046) (0.046)

Created after SIMPLES (YT) 0.028 0.048 0.050 0.050 0.051(0.030) (0.038) (0.037) (0.037) (0.037)

Interaction variable 0.299 0.001 0.001 0.193 0.009(0.035) (0.001) (0.000) (0.041) (0.040)

Owner's characteristics Yes Yes Yes Yes YesFirm's characteristics Yes Yes Yes Yes YesState dummies Yes Yes Yes Yes YesObservations 2892 2517 2517 2517 2517R2 0.254 0.175 0.175 0.177 0.183

Note: This table reports OLS coefcients of Eq.(2)where the dependent variable is a binary variable indicating whether the rm holds a state or municipal license. The second lineshows the coefcient of the interaction term of the dummy indicating the effect of SIMPLES and a variable indicated in the columns. Employer indicates whether the rm has from 1to 5 employees. Annual revenue is the total value of revenues accumulated in the last 12 months. Assets indicate the value of equipments, tools and buildings owned by the rm.Located outside owner's house indicates whether the rm does not operate in owner's house. Sell to rms or government indicates whether the rm has other rms and/or thegovernment in its client base. All regressions control for rm's owner characteristics (dummies indicating primary education level, secondary education level, college degree,

age, gender, lives on his/her own house, has other job), characteristics of the rm (total assets, annual revenue, and dummies indicating non-declaration of assets, location outof owner's house, sales to other rms and government, startup was nanced by the owner, non-paid employees in the rm, relatives employed in the rm), and 27 state dummies.In column 1, the sample comprises all retail rms and ineligible rms managed by self-employed entrepreneurs or employers. The sample in columns 2 to 5 includes only retailrms and ineligible rms managed by self-employed entrepreneurs. Robust standard errors clustered at economic activity are reported in parentheses.

Signicantly different than zero at 99%. Signicantly different than zero at 95%. Signicantly different than zero at 90%.

12

We thanks an anonymous referee to point out this issue.

http://-/?-http://-/?-http://-/?-http://-/?-


8/11

6.1. Accounting for different time effects across groups

An important weakness of the DID methodology is the assumptionthat aggregate shocks have no differential effect across the comparisonand treatment groups. For example, the two groups might have experi-enced distinct time trends, andthe impact estimated inTable 5 could beassociated with differences in rm age rather than with the SIMPLESprogram.

In order to take this possibility into consideration, we re-estimatethe equation using a different time window. Originally, our resultswere generated considering rms created 10 months before and10 months after December 1996. In column 2 ofTable 8, we reportthe same exercise considering rms created 10 months before and10 months after December 1995, a month with no signicant changein the tax legislation. If our results are due to differences in timetrends across groups, we also expect a positive and signicant resultin this case. The results related to 1995 are not signicantly differentfrom zero, with a p-value of 0.67.

Although this exercise corroborates the key identication hypoth-esis, it does not rule out the possibility that the retail sector experi-enced an idiosyncratic shock between December 1996 and October1997 that induced more formalization. This would explain why we

only measure an effect in the retail sector and would imply that

SIMPLES does not induce formalization. Even though we are notaware of any specic shock like this, we cannot rule out this hypoth-esis and a skeptical reader may read our results as an evidence thatSIMPLES does not have much impact on microenterprises' decisionto register.

6.2. Splitting-up

The SIMPLES system is restricted to small and microenterprises rms with an annual revenue below thresholds of R$120,000 and R$1,200,000, respectively. Since the tax reduction decreases with rmsrevenue, there might be an incentive for largerrms to split up. If li-censing is more frequent amongrms reacting this way, we could ob-serve an increase in the number of formalrms created after the newsystem due to a change in the composition ofrms.

We believe this sort of bias seems to be of a second order and neg-ligible in our sample because less than 1% ofrms created before SIM-PLES had an annual revenue above R$ 120,000. This proportionremained unchanged after SIMPLES. In any case, we econometricallyassess this issue by restricting the analysis to households whichown only one rm. As shown in column 3 ofTable 8,the result re-mains unaltered. The underlying assumption behind this exercise isthat, in case of new rms are created, they are owned by someonewho lives in the household. If this is the case, restricting the sampleto households with only onerm creates a reduction in the composi-tion bias. As the results remain the same, it seems that this effect isnegligible in our case.

6.3. Measurement error in the age of therm

A crucial variable in our analysis is the reported age of therm.Firms are classied as created before or after the SIMPLES system ifthe owner reports an age above or below 10 months, respectively.This can be a potential problem if multiples of 12 months are focal

answers.

Table 7

Occupational choice and sector transitions.

Dependent variable Firm holds of cial license

Interaction variable Previousoccupation

Previoussector

(1) (2)

Eligible created after SIMPLES (YTG)*Past job: nonworkers 0.04

(0.06)Past job: employee 0.11

(0.06)

Past job: self-employed 0.19(0.05)

Past job: employer 0.27(0.13)

Past sector: retail trade 0.11(0.08)

Past sector: non-retail 0.13(0.05)

Owner's characteristics Yes YesFirm's characteristics Yes YesState dummies Yes YesObservations 2517 2459R2 0.18 0.18P-value for test of equality between coefcients

Nonworkers vs employee 0.35Nonworkers vs self-employed 0.04Nonworkers vs employer 0.07Employer vs employee 0.27Employer vs self-employed 0.59Employee vs self-employed 0.22Retail vs non-retail 0.83

Note: This table reports OLS coefcients of Eq.(2)where the dependent variable is abinary variable indicating whether the rm holds a state or municipal license. Past

job indicates rm's owner position in his previous occupation. Nonworkers arepeople who were previously unemployed or out of the labor force. Past sectorindicates the sector that the rm's owner worked in his previous occupation. Allregressions control for rm's owner characteristics (dummies indicating primaryeducation level, secondary education level, college degree, age, gender, lives on his/her own house, has other job), characteristics of the rm (total assets, annualrevenue, and dummies indicating non-declaration of assets, location out of owner'shouse, sales to other rms and government, startup was nanced by the owner, non-

paid employees in the rm, relatives employed in the rm), and 27 state dummies.The sample comprises retail rms and all ineligible rms. Robust standard errors clus-tered at economic activity are reported in parentheses.

Signicantly different than zero at 99% condence. Signicantly different than zero at 95% condence. Signicantly different than zero at 90% condence.

Table 8

Robustness checking.

Retailtrade

Yearbefore

Splitting-up

Memory/focuson 12 months

Time toformalize

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

Eligible created afterSIMPLES (YTG)

0.13 0.01 0.14 0.15 0.14(0.05) (0.08) (0.05) (0.08) (0.06)

Eligible (YG) 0.04 0.04 0.01 0.09 0.02(0.05) (0.05) (0.05) (0.08) (0.06)

Created afterSIMPLES (YT)

0.05 0.07 0.06 0.06 0.04(0.04) (0.05) (0.04) (0.06) (0.05)

Owner'scharacteristics

Yes Yes Yes Yes Yes

Firm's characteristics Yes Yes Yes Yes YesState dummies Yes Yes Yes Yes YesObservations 2517 1493 2404 1397 1540R2 0.17 0.19 0.15 0.15 0.22

Note: Each column in the table represents the least square estimate of Eq. (1)in thetext considering different contexts for the sake of robustness. The regression for theretail trade sector is reproduced from Table 5in column 1. Columns 2 to 5 refer todifferent falsication tests that are explained inSection 6. All regressions control forrm's owner characteristics (dummies indicating primary education level, secondaryeducation level, college degree, age, gender, lives on his/her own house, has other

job), characteristics of the rm (total assets, annual revenue, and dummies indicatingnon-declaration of assets, location out of owner's house, sales to other rms and gov-

ernment, startup was

nanced by the owner, non-paid employees in the

rm, relativesemployed in the rm), and 27 state dummies. The sample comprises retail rms and allineligible rms. Robust standard errors clustered at economic activity are reported inparentheses.

Signicantly different than zero at 99% condence. Signicantly different than zero at 95% condence. Signicantly different than zero at 90% condence.



9/11

Fig. 1shows spikes in the percentage ofrms with reported agesof 12 and 18 months. In particular, 19% of rms in our samplereported exactly 12 months. However, notice that the reported ageofrms with less than 6 months is more uniformly distributed.

Therefore, in order to accountfor potential biasesrelatedto this mea-surement error, column 4 ofTable 8reports an estimate of Eq.(1)ex-cluding 1,120 rms (44% of the sample) aged between 7 months and14 months. Again, the result is virtually the same: SIMPLES increasesby 15 percentage points the probability of holding a formal license.

6.4. Time to formalize

Column 5 ofTable 8considers a potential negative bias due to thefact that the process of formalization is time-consuming. Firms mightstart to operate without holding formal licenses, processing the pa-perwork to regularize their status concomitantly. Therefore, the effectof SIMPLES on the formalization of youngrms could be underesti-mated. Actually, when we discard those rms aged below 4 months

(and

rms above 16 months to keep the symmetry), we get a tiny in-crease in the estimated effect, which becomes 14 percentage points.This suggests that the bias due to the time to formalize is negative,as expected, but not economically important.

6.5. Selection of owners

Another potential source of selection bias is a potential effect ofthe SIMPLES reform on the prole ofrms' owners. Since the SIMPLESreform has affected the business environment, it could change thewillingness of different types of entrepreneurs to start new busi-nesses. Therefore, our results could reect a new composition of en-trepreneurs (potentially more prone to open formal business)rather than the impact of the new tax regime on formalization.

Table 9analyzes whether the average prole of the entrepreneurwho opens a rm in each sector changed before and after SIMPLESlaw. The structure of the table is as following. The lines depict a setof owner's characteristic under analysis and the columns show thepercents of owners in each sector with that particular characteristic.

Except by the age, there is no systematic statistical difference inthe average owners' characteristic betweenrms opened before and

Fig. 1.Distribution ofrms' age.

Table 9

Firms' owners prole before and after SIMPLES reform.

Sectors in the treatment groupOwners' characteristics Created before or after SIMPLES Retail Construction Manufacturing Transportation Services Comparison group

Primary education Before 0.51 0.87 0.67 0.63 0.61 0.37After 0.57 0.80 0.65 0.59 0.66 0.38Diff (AB) 0.06 0.06 0.02 0.04 0.05 0.02

Secondary education Before 0.32 0.06 0.25 0.27 0.26 0.31After 0.30 0.09 0.22 0.26 0.22 0.33Diff (AB) 0.03 0.03 0.03 0.01 0.04 0.02

College degree Before 0.09 0.01 0.03 0.02 0.03 0.29After 0.05 0.00 0.09 0.08 0.05 0.26Diff (AB) 0.04 0.00 0.07 0.05 0.01 0.02

Age Before 37.95 34.07 38.41 35.46 39.10 34.69After 35.71 31.73 32.89 34.63 36.51 32.62Diff (AB) 2.24 2.34 5.52 0.83 2.59 2.07

Male Before 0.58 0.97 0.52 0.99 0.49 0.60After 0.52 1.00 0.44 0.97 0.47 0.56

Diff (A

B)

0.07 0.03

0.08

0.02

0.02

0.04Owns his house Before 0.80 0.81 0.72 0.77 0.66 0.69After 0.78 0.79 0.69 0.66 0.72 0.67Diff (AB) 0.02 0.02 0.03 0.10 0.06 0.02

Has another job Before 0.11 0.05 0.04 0.08 0.12 0.15After 0.10 0.07 0.12 0.10 0.10 0.15Diff (AB) 0.01 0.02 0.08* 0.01 0.03 0.01

Startup was nanced by the owner Before 0.69 0.34 0.66 0.60 0.62 0.42After 0.62 0.18 0.45 0.60 0.52 0.39Diff (AB) 0.07 0.16 0.21 0.00 0.09 0.03

Work with relative in the rm Before 0.18 0.03 0.06 0.01 0.18 0.03After 0.16 0.02 0.26 0.02 0.18 0.05Diff (AB) 0.02 0.01 0.20 0.01 0.00 0.02

Note: This tablecomparesthe average rms' owners characteristics before and after SIMPLES law in the different economic sectors under analysis. Each cell indicates the percentageofrms whose owner has the characteristics indicated in thelines. Diff (AB) indicates the difference in means among rms created after and before SIMPLES law.

Signicantly different than zero at 99% condence. Signicantly different than zero at 95% condence.

Signicantly different than zero at 90% condence.



10/11

after SIMPLES law. This is exactly true for the retail sector. There is noevidence that our results are determined by selection of owner pro-les. Notice that this pattern is true for the other sectors as well, al-though some characteristics are different across periods for themanufacturing sector.

7. Discussion: why the results vary by sector?

Our results indicate that SIMPLES effect on formality varies withthe economic sector. In this section we discuss the rationale of thisdifference. We argue that both informality levels and the incentivesprovided by SIMPLES vary by economic sector since the requirementsto enter and operate in the formal economy depend on the activity ofthe rm and SIMPLES promoted a partial and uniform reduction inthe cost of being formal.

The idea that informality varies by economic sector is supported bythe analysis of formality rates across sectors and by some studies doneabout the subject. In our dataset, 24% of therms have a municipal orstate license but this rate ranges from 8% in the construction sector to38% among transportationrms.Capp et al. (2005)argue that levelsandforms of informality varyaccording to thevalue chain of a given sec-tor, the way it is taxed and regulated, and sector-specic schemes forgetting past regulatory or tax-enforcement agents.

In the decision to become formal or not, entrepreneurs balancethebenets and costs of formality and both can vary by sector. The costsof formalization can be divided between the costs of becoming formaland the costs of operating in the formal sector. The rst group in-cludes the several requirements to register a rm. The registrationprocess in Brazil demands 15 steps and the visit to several institu-tions, reaching 152 days on average according to Doing Business esti-mates (WorldBank (2004)). This process is more cumbersome forsome activities than others. For instance,rms that process or handlefood face an additional step, the health inspection, which adds con-siderable time and cost to the process. In other sectors, formalizationdoes not depend only on rms' demand. Cities usually regulate thenumber of taxis and busses that circulate in their area. Therefore,

even if an entrepreneur who works in the transportation sector de-cides to formalize his vehicle, this option may not be available dueto limits on licenses imposed by city authorities.

The costs to operate in the formal sector also depend on the eco-nomic activity. As shown inTable 1, the Federal tax burden variesby sector and therefore the net benet that SIMPLES promotes. In ad-dition, the system implies just a partial reduction in the tax burdensince other important taxes are not included in the reform. The ser-vice sector, for instance, must pay ISS, a municipal tax that reaches5% of gross revenue, which is more than the SIMPLES rate for micro-enterprises with annual revenue up to R$ 60,000. Another importantcost is related to labor regulations. Labor taxes and fees representmore than 50% of the wages in Brazil so companies are tempted tounderreport employment.Farrell (2004)reports that informality in

manufacturing industries is more prevalent in labor-intensive sectorssuch as apparel and food processing than in capital-intensive onessuch as automotive assembly, cement, oil, steel, and telecommunica-tions. Finally, underreporting of revenue may be easier in some sec-tors, which can reduce rms' de facto tax burden. Paula andScheinkman (2010)show that the formality of a rm is correlatedto the formality of suppliers and purchasers whenrms are subjectto the credit system of value added tax, which does not levy in all eco-nomic activities. In addition, arm's ability and necessity to underre-port revenues depend on how much its competitors rely on informalpractices. Underreporting of revenues and the consequent reductionin tax burden can dramatically reduce costs in Brazil. For instance,an estimate from McKinsey Global Institute reported inCapp et al.(2005)indicates that evasion of taxes and social charges can more

than triple a Brazilian supermarket's income. Therefore, it is hard to

compete in prices with informalrms and underreporting may be arequirement to survival in some markets.

In addition, the decision to enter the informal sector depends cru-cially on law enforcement. Limited enforcement of legal obligationsstimulates informality and rms' ability to avoid detection is lowerin some sectors. Entrepreneurs that serves the open public (e.g. res-taurants) or circulate around the city (e.g. drivers) are more exposedto inspection, while sectors composed by small and geographically

dispersed companies that serve mainly individuals such as construc-tion make the task of auditors and tax collectors more difcult.These examples illustrate that both informality levels and the in-

centives provided by SIMPLES vary by sector. In our context, we be-lieve that construction and transportation rms have specially fewincentives to enter the formal sector induced by SIMPLES reform.The typical constructionrm in our database is a plumber or electri-cian that serves individuals. Theserms are particularly hard-hit byinspection and can expect few benets from formalization sincetheir customers are often keen to exchange an invoice for a reductionin prices. Indeed, only 11% of the construction rms created beforethe program were licensed at the moment of the survey.

The transportation sector follows a particular dynamic. This sectorhas the highest formalization rate in our dataset (38%) mostly due torms that operate taxis, school busses and urban busses, which repre-sent 53% of the rms in this sector and have a formalization rate of52%. This activity is heavily regulated. In most Brazilian cities, for in-stance, an entrepreneur cannot simply decide to register a car or abus to carry passengers. It is necessary availability of licenses andthere is usually a cap on the number of licenses issued by a city,meaning that formalization in this sector depends both on demandand supply of licenses. Operate informally is also more risky in thissector since the fact that these vehicles circulate around the city letthem more exposed to inspection.

Therefore, the three other eligible sectorsretail trade, manufactur-ing and service are potentially more affected by the reform. However,we believe that the design of SIMPLES law makes the eligibility in themanufacturing and service sectors highly uncertain. As discussed, activi-ties which require professionalswith regulatedoccupations are noteligi-

ble to the system. The problem isthat this is not a precise denition and,specially in therst years of the new system, there was much doubt onwhich activitiesrequire professionals with regulated occupations. For in-stance, rms that provide services of maintenance and repair of vehicles,machines, computers and home appliances were initiallyconsidered noteligiblebecause theywere associated with the engineering profession. In2004, the government reviewed that restriction and allowed thesermsto adhere to SIMPLES. Firms which provide internet services were alsoineligible in the beginning of the system due to the interpretation thatthey require a computer programming analyst and daycare facilitieswere only accepted in the system in 2000. Much of these uncertaintieswere resolved along the years after clarications and law addenda.Zanluca (2006)cites 14 pieces of legislation and normative acts thatwere enacted from 1998 to 2006, aiming, among other things, to clarify

whether specic activities were eligible or not to SIMPLES.This eligibility uncertainty brings two difculties to our analysis.

From the econometrician point of view, it is hard to determinewhich rms among the ones in the manufacturing and service sectorsare part of the treatment or control group. From the point of view ofthe rms, this uncertainty may reduce their willingness to formalizetheir ventures. This decision depends highly on the advice providedby accountants and therefore their assessment on whether an activityis eligible or not.

8. Conclusion

This paper evaluates the impact of bureaucracy simplication andtax reduction on the formality of microenterprises. Our source of

identication is the enactment of a new tax system designed for



11/11

micro and small rms, the SIMPLES system. We explore the fact thatthe new tax system is restricted to a subset of sectors and use adiffer-ence-in-differenceapproach, comparing the legal status of rms insectors affected and not affected by the reform, created before andafter the program.

Wend that the impact vary according to economic sector. Formal-ization increased by 13 percentage points in retailers with the new taxregime but we do not nd an effect on license rates among rms in

transportation, construction, services and manufacturing sectors. Thekey hypothesis behind this result is the assumption that aggregateshocks have no differential effect across the retail sector and the com-parisongroup. We perform a series of falsication checks and sensitivityanalysis andshow that our results are robust to all of them. In particular,we do a placebo test by replicating the same econometric exercise oneyear before the enactment of SIMPLES and show that there is no mea-sured effect on the retail sector for that period. We also discuss reasonsto expect that SIMPLES incentive varies with economic sector.

Our results are an indication that tax burden and bureaucracy,which are key elements in the SIMPLES reform, constitute importantobstacles to the regularization of rms. However, since we cannotrule out the possibility that an idiosyncratic shock to the retail sectorcaused the measuredincrease in formalization, a skeptical reader mayinterpret our results as an evidence that SIMPLES does not have muchimpact on microenterprises' decision to register. In any case, thispaper shows that this type of reform can produce limited effects onthe reduction of the informal sector because it does not induce alltypes ofrms to formalize. Economic activities such as transportation,manufacturing and construction require other initiatives to experi-ence lower levels of informality. We still need more research to iden-tify which initiatives are these.

Appendix A. Composition of treatment and comparison groups

Eligible economic activities (number ofrms in the sample)

Retail Trade (1202):retail trade of the following products: grocery(12), beverage, meat and food (576), garment and accessories (191),

decoration articles (18), books and magazines (34), construction mate-rial (31), homeappliance,machines and electric supplies (36), transportequipment (40), pharmaceutical and chemical products (52), oil andfuel (23), supermarkets (6), leisure articles (183).

Civil construction (604)

Manufacturing (566):manufacturing of wire, construction mate-rial and ceramics (27), metallic instruments (59), wood objects (28),bamboo, wicker and agave (4), furniture (42), paper goods (4), rub-ber goods (1), leather goods (1), plastic goods (3), textile goods(16), garment (173), shoes (20), food (76), printing and editing(30), medical material and hygiene products (82).

Transportation (816): cargo transportation (90), passenger trans-

portation (431), charter freight (292), maritime freight (2), air freight(1).

Other services (1480): lodging (10), bar and restaurant (944),furniture repair (29), plumber and electricity services (41), sewing(202), apparel rental (24), laundering, pressing and dying (65), gar-dening and housing maintenance and repair (24), entertainment(131), tourism (10).

Ineligible economic activities (number ofrms in the sample)

Comparison group (1487):machine manufacturing (18), home ap-pliance and electric supply manufacturing (23), transport equipment

manufacturing (5), chemical product manufacturing (1), cleaning andcosmetics manufacturing (5), cleaning companies (3), banks andnan-cial institutions (2), insurance companies (18), housing administration(31), exchange shop(1), state lotteries (1), credit cards androtating sav-ings companies (4), home appliance repair (72), auto repair (193),watches and precision article repair (25), gymnasium and beautyshops (303), housing and cleaning services (18), security services (31),law services (59), accounting and economics services (38), data proces-

sing and business consulting (69), services of architecture, engineeringandgeology (26), advertisement andevent promotion (27), writers, ser-vices of journalism, investigation and statistics (6), machines and rurallabor rental services (6), commercial representation and foreign tradeofces (90), tools and equipments rental; leasing and marketing ofces(35), lotteryshops (7), port services(6), employmentand trainingagen-cy; telecommunication services (24), services to hospitals, foundations,welfare and social securities (6),clubs and sport associations (6), clinics,hospitals and laboratories (45), odontological services (12), colleges,universities and educational courses (270).

Excluded sectors: mining,beverage manufacturing, tobacco and cig-arettes manufacturing, water distribution and supply, peddler, producefair, telecommunication companies, TV and radio stations, photography,lming and translation, ateliers of panting, decoration and design, carrental, parking, trafc engineering services and towing, social assistancecenters, cultural centers, museums and parks, religion centers, commu-nities associations, vets, notary, lottery shop, brothel and hunting, streetvendors, non-dened activities, non-declared activities.

Notes: The classication of economic activitiesas eligible and ineligi-ble wasmade by the authorsfollowing SIMPLES law, conversations withaccountants and lists provided by the Internal Revenue Service and ac-countancy consultingrms. This classication follows the rule in effectin October 1997. After that date, some activities were excluded fromthe system and others were accepted. Excluded activities are the oneswhose eligibility status in October 1997 was not clear to the authors.

References

Auriol, E., Warlters, M., 2005. Taxation base in developing countries. Journal of PublicEconomics 89 (4), 625646.Bruhn, M., 2011. License to sell: the effect of business registration reform on entrepre-

neurial activity in mexico. Review of Economics and Statistics 93 (1), 382386February.

Capp, J., Heinz-Peter, E., Willian, B., J. J., 2005. Reining in Brazil's informal economy. TheMcKinsey Quarterly 1, 17 January.

De Mel, S., McKenzie, D., Woodruff, C., 2011. What is the cost of formality? Experimen-tally estimating the demand for formalization. July.

De Paula, A., Scheinkman, J.A., 2010. Value-added taxes, chain effects, and informality.American Economic Journal: Macroeconomics 2 (4), 195221.

Djankov, S., Porta, R.L., Lopez-De-Silanes, F., Shleifer, A., 2002. The regulation of entry.The Quarterly Journal of Economics 117 (1), 137 February.

Farrell, D., 2004. The hidden dangers of the informal economy. The McKinsey Quarterly3, 2637 August.

Friedman, E., Johnson, S., Kaufmann, D., Zoido-Lobaton, P., 2000. Dodging the grabbinghand: the determinants of unofcial activity in 69 countries. Journal of Public Eco-nomics 76 (3), 459493 June.

Gerxhani, K., 2004. The informal sector in developed and less developed countries: a

literature survey. Public Choice 120 (34), 267300 09.Johnson, S., Kaufman, D., Shleifer, A., 1997. The unofcial economy in transition. Brook-

ings Papers on Economic Activity 28 (19972), 159240.Johnson, S., Kaufmann, D., Zoido-Lobaton, P., 1998. Regulatory discretion and the unof-

cial economy. American Economic Review 88 (2), 387392 May.McKenzie, D., Sakho, Y.S., 2010. Does it pay rms to register for taxes? The impact of

formality on rm protability. Journal of Development Economics 91 (1), 1524January.

Schneider, F., Enste, D.H., 2000. Shadow economies: size, causes, and consequences.Journal of Economic Literature 38 (1), 77114 March.

WorldBank, 2004. Doing Business Brazil. The International Bank for Reconstruction andDevelopment/The World Bank, Washington.

Zanluca, J.C., 2006. SIMPLES Federal Aspectos Tributrios. Portal Tributrio, Curitiba.


coming out the shadows - estimating the impact of bureaucracy simplification and tax cut on...

Documents