loayza mimic 1996

7/30/2019 Loayza Mimic 1996

1/60

__VP I7 2 `LPOLICY RESEARCH WORKING PAPER 17-27

The Economics An increase n the size of theinformal sector hurts growthof the Informal Sector by reducing the availability forpublic services or everyone in. . . -- ~~~~~~~~~~~~theconomy and increasingA SimpleModel and SomeEmpirical tnthe number of activities hatEvidence from Latin America use spime existing publicservicesess efficiently or not

Norman A. Loayza at all.

The World BankPolicyResearchDepartmentMacroeconomicsand Growth DivisionFebruary 1997


2/60

POLICY RESEARCH WORKING PAPER 1727

Summary indingsLoayzapresents the view that informal economies arise countries - identifying the size of the informal sector towhen governments impose excessive taxes and latent variable for which multiple causes and indicatorsregulations that they are unable to enforce. exisr.

Loayza studies the determinants and effectsof the 'The results suggestthat:informal sector usingan endogenous growth model * The size of the informal sector depends positivelywhose production technology depends essentiallyon on proxies for tax burden and restrictions on the laborcongestablepublic services. In this model, changes (in market.both policy parameters and the quality of government * It depends negatively on a proxy for the quality ofinstitutions) that promote an increase in the relative size governimentinstitutions.of the informal economy will also generate a reduction in * An increase in the size of the informal sector hurtsthe rate of economic growth. growtlh by reducing the availability of public services forUsing data from Latin American countries in the early everyone in tiheeconomy and by increasing the number1990s, Loayza tests some of the model's implications and of activities that use some existing public services lessestimates the size of the informal sector in these efficientlyor not at all.

This paper - a product of the Macroeconomics and Growth Division, PolicyResearch Department - is part of a largereffort in the department to examine the determinants of economic growth. Copies of this paper are available free from theWorld Bank, 1818 H Street NW, Washington, DC,20433. Pleasecontact RebeccaMartin, room NI 1-059, telephone 202-473-9026, fax 202-522-3518, Internet address rmartinl(@'worldhank.org. February 1997. (52 pages)

The Policy ResearchWorking Paper Seriesdisseminates the findings of work in progress o encourage the exchange of ideas aboutdevelopment issues.An objective of the series s to get the findings out quickly, even if the presentationsare less han fully polished. Thepapers carrythe names of the authors and should be cited accordingly. T7hcindings, interpretations, and conclusions expressed n thispaper are entirely those of the authors. They do not necessarilyretnesent the Euiw of the World Bank, its Executive Directors,or thecountries they represent.

Produced by the Policy Rese;rch D)ssenminationC:enter


3/60

THE ECONOMICS OF THE INFORMAL SECTORA Simple Model and Some Empirical Evidence from Latin America

Norman V. Loayza'The World Bank, Washington,D.C. 20433

'Excellent research assistance by Daniel Wolfenzon and Wanhong Hu is gratefullyacknowledged.I am grateful for feedbackprovidedby participantsat the Carnegie-RochesterConferenceon Public Policy in Pittsburgh, in November 1995. 1 speciallyacknowledgethecomments from Patrick Asea, my discussantat the Carnegie-RochesterConference,and AllanMeltzer, the Conference'sorganizer. I also benefited greatly from the commentsof Pierre-RichardAgenor, AlbertoAlesina, CevdetDenizer, DavidDollar, Edward Glaeser, Aart Kray,Philip Lane, Luis Serven, Klaus Schmidt-Hebbel, and participants at the World BankMacroeconomicsnd Growth Seminar.


4/60


5/60

INTRODUCTIONThepresenceof large informalsectors is one of the most importantcharacteristics

of developing ountries. The study of the informal economy is importantinasmuchas itsheds light on how the state's regulatory and enforcementsystems affect economicperformance.

There are a few differentdefinitionsand approaches o the study of the informalsector n the development conomicsliterature.' This paper's approach follows he workby Hernandode Soto (1989)2, ccording o which the informalsector is defined as the setof economicunits thatdo not complywith government-imposedaxes and regulations. AsPortes, Castells, and Benton (1989, p. 12) put it, "The informal economy is ... a processof income-generationcharacterized by one central feature: it is unregulated by theinstitutions of society, in a legal and social environment n which similar activities areregulated.

The informal sector ariseswhen excessive axes and regulationsare imposedbygovernment hat lackthe capability o enforcecompliance.An excessive egulatory systemmakes heformal economyunattractiveby imposinghigh entry costs to legality -throughlicense ees and registrationrequirements- nd high costs to remain legal -through axes,red tape, and, among others, labor and environmental egulations.

However,escaping axes and regulations s notcostless. An informal status entailsmanydisadvantages.Informalactivitiesare subject o stiff penaltiesin the form of finesor capital confiscation. Furthermore,because of their illegalstatus, informal agentsdonot fullyenjoy publicservices,particularly hose hatallow hem full, enforceableproperty

-1-


6/60

rightsover their capitaland output. This has a number of negativeconsequences: First,informal roducers re poorly protectedby the police and the judicial courts from crimescommitted gainst heir property. Second,since hey lack the capacityto enter into legallybinding contractualobligations, their access to capital markets, for financial, insurance,and corporativepurposes, is seriouslyimpaired. Lastly, they find obstacles to use otherpublicservicessuch as socialwelfare, skill training programs, and government-sponsoredcredit facilities.

The state, as the institution hat both monitorsthe regulatory and enforcementsystems nd administerspublic services,plays a crucial role in the formationof informaleconomies. If state officials,or interestgroups related to them, profit in someway fromthe presenceof the informal ector, hey will createan environment hat makes informalityattractive or simply unavoidable. To the extent that excessive regulationsare created tobenefitparticular interestgroups and not society in general,the presence of the informalsector s a result of the failure of political institutions o protect and promote an efficientmarket economy.

Thispaper attempts o organize he current informationon the determinantsof theinformal sector and its effect on economicgrowth through, first, a simple endogenousgrowth model and, second, empirical evidencefrom Latin America. The next sectiondescribes the main determinantsand featuresof the informal sector. Then, the papermodels the presenceof the informal sector in the economy and its relationship o outputgrowth within he framework f the endogenousgrowth literature, specifically, he workin whichgovernment'sprovisionof public goodsand services s consideredexplicitly, as

-2-


7/60

in Barro (1990) and Barro and Sala-i-Martin 1992). The empirical section of the paperuses data from Latin American ountries o test some of the implicationsof the model andto provideestimatesof the relative size of the informalsector throughout hese countries.THE RATIONALITYOF BEING INFORMAL

Economic nitschoose o be partiallyor completelynformalby weighingthe costsand benefitsa legalstatusentailsand considering heir particular institutionaland resourceconstraints. In this sense, the choice to be informal s a rational one, a fact which doesnot imply that some firms are not forced by their constraints to be either formal orinformal.

The costs and benefits of legality are now analyzed in detail. The presentationadoptsde Soto's analytical rameworkas well as some of his reported evidence. De Sotostudied the informal sector in Peru. Other researchershave applied his methodology oa variety of different countries. Chickeringand Salahdine(1991) present evidencefromselected underdevelopedAsian countries. Tokman (1992) offers evidencefrom LatinAmerica and the Caribbean. These empirical findings also contribute to the followingpresentation.TheCosts of Formality

The costs of formality can be divided into costs to access the formal sector andcosts to remain in the formal sector.

Costs to access the formal sector. De Soto and his research group conductedvarious experiments o quantify the access costs. In one of them, they set up a smallclothingfactory in the outskirts of Limaand tried to register it legally. Four university

-3-


8/60

students were hired to do the necessary paperwork under the supervisionof a lawyerspecializedn Administrative aw. It was convened hat no bribes would be offered unlessthere was no other means to proceed with the registration. During the months theexperiment lasted, governmentofficials asked for bribes ten times; in two cases, it wasabsolutely necessary to pay them. It took ten months to complete the registrationprocedure. Licensesand other requirements cost 195dollars, and the loss in utilitiescausedby the ten-monthwaitingperiod was estimatedat 1,037 dollars. In fact, the totalcostof legalregistrationwas equivalent o 32 times he minimum monthlysalary. By wayof comparison, Chickering and Salahdine (1991) report that a similar procedure takesthree-and-a-halfhours in Florida and four hours in New York.

Tokman (1992) also finds high access costs to legality in other Latin Americancountries. Financialcosts of entry, excluding required modifications n the business'premises,are estimated t an averageof ten percentof annualprofits; when modificationsin thepremisesare requiredas part of the registrationprocedure, these financialcosts risesignificantly.Tokmanfinds that the average ime to register a small firm in his group ofLatin American ountries s ten months, going from aboutone month (in Bolivia, Brazil,and Chile) to two years (in Guatemala). Given the long time involved in fulfillingbusiness registration requirements, Tokman (1992, p.12) concludes that "... thecombination of regulation inadequacy and bureaucratic inefficiency ... applies to mostcountriesand particularly o [Ecuador, Guatemala,and Peru]."

Costs to remainin the formalsector. Stayingformal can also be very costly.De Soto finds that, in a sample of 50 small manufacturing irms, the costs of staying

-4-


9/60

formal representan average of 348 percent of after-taxprofits. De Soto reports that 22percentof such costsare due to taxes, 5 percent due to higherpublic utility rates, and 73percent due to regulations(mainly laborrelated) and bureaucraticrequirements.

The costs of staying formal can be divided into three broad categories: taxes,regulations,and bureaucraticrequirements.

Taxes on formal firms constitute a major source of government revenue,particular!y in developing countries. Since formal firms are registered with stateorganizationsand, thus, can be audited with relative ease, they are attractive targets fortaxation. This is especially true in countries that lack strong, well-equipped taxadministrations o properly monitor individualsand informal firms. Burgess and Stern(1993) eport that in developingcountries,corporate income axes represent17.8 percentof total tax revenues,while individual ncome axes, only 10.6 percent. In contrast, forindustrial countries corporate income taxes account for only 7.6 percent of total taxrevenues, while individual income taxes carry most of the weight with 27.7 percent.Confrontedwith a narrow base of formal firms, most underdevelopedcountrieshaveimposed, at leastuntil recently, very high marginal ax rates.

Regulations control the use of resources, the manner of production, and thedistribution of output and profits. Common types of regulations are those related toenvironmentalrotection,allocationof importedinputs, consumer protectionand qualitycontrol, financial capital availability,and workers' welfare. Of all types of regulations,those related to workers' welfare are the most restrictive and costly in underdevelopedcountries and in manydevelopedcountries as well). Regulationsdesigned, in theory, to

-5-


10/60

improve workers' welfare appear in various forms, namely, minimum wages, fringebenefits (paid vacations and sick leave, indemnities, health insurance), social security,constraints on free hiring and dismissal, and protection to unions. Portes, Castells, andBenton (1989, p. 30) argue that "the best-known economic effect of the informalizationprocess is to reduce the costs of labor substantially," costs which are mostly due to"indirect wages" such as benefits and social security contributions. Tokman writes thatfor small firms in Latin America, the additional costs related to labor regulations are themost importantcomponent of the permanency costs in the formal sector. Tokman (1992)reports that such regulations increase labor costs by an average of around 20 percent,which is about equally divided between benefits and social security contributions. Nipon(1991) estimates that informal firms in Thailand, by ignoring labor-protection laws, saveabout 13 to 22 percent of labor wages.

Mazumdar (1976) cites the interestingexample of the labor practices in the Bombaytextile industry: Unions, enforcing government regulations, protected formal workersagainst salary decreases or dismissals and ensured social benefits for them. In response,textile companies kept the number of formal workers as low as possible; and when theirbusiness required a higher labor input, they would supplement the stable core of formalworkers with casual day laborers, who were abundant in Bombay. Not surprisingly, thesecasual day workers, who constitute the informal labor force for the industry, were paidlower wages, had no job security, and enjoyed no fringe benefits. This example illustrateslabor market conditions in many cities of the developing world.

In most developing countries, capital is scarce relative to labor. This should-6-


11/60

induce irms to choose abor-intensiveechnologies.However, becauselabor-employmentregulations aise labor costs, formal firms in developingcountries tend to be abnormallycapital intensive.5

Bureaucraticequirements red tapeand paper work) are also a significant cost ofremaining formal. Alonzo (1991, pp. 48-49) reports that in the Philippines, "no matterhow small he business,an ownerneeds an accountantand a lawyer to comply with all ofthe requirements." Chickeringand Salahdine 1991, p. 191) write that in Egypt, "muchof the country's entrepreneurial talent is consumed in circumventing the country'snightmare bureaucraticregulatory system." De Soto (1989) surveyed 37 formal firmsoperating in areas in which informal firms abound. He found that 40 percent of theadministrativepersonnel working time is spent fulfilling the bureaucratic paper workimposedby the state.The Costs of Informality

Informal enterprisesface two kinds of costs: First, penaltieswhen the informalactivity is detected, and second, the inability to take full advantage of government-provided goods. Penaltiesfor informal activities are usually stiff; very often, detectedfirms have to surrender a considerablepart of their output or physicalcapital stock. DeSoto (1989) indsthat informalentrepreneurspay between 10 to 15 percent of their grossincome in bribes to corrupt governmentofficials, whereas formal entrepreneurs pay anaverage of only 1 percent of gross income in bribes (without counting bribes used tobecome formal). In order to avoid being caught, firms scale down the size of theirinformal operations. In the case of purely informalfirms, the efforts to avoid detection

-7-


12/60

prevent them from achievingeconomiesof scale and from choosing an optimal capital-labor mix; this is so because larger and morephysical capital-intensive irms are easierto detect.

Thesecondcostof informalitys the inability o take full advantageof government-providedgoods,in particular he legal and udicial system andthe police. Since informalactivities re illegal, informalbusinessmencannot exercisefull property rightsover theircapitaland product. Therefore,contracts elated o informalactivitiescan not be enforcedthroughthe judicial system and, thus, their value and usefulnessare greatly diminished.The inability to sign contracts enforceable through the courts creates uncertaintyandincreases the transaction and monitoring costs in all business dealings conductedbyinformal companies. This reduces investment hat comes both from internal sources(retained earnings) and from capital markets. De Soto (1989) provides an interestingexampleof low investment rom internal sources; he studies informal housing, that is,constructionon land over which families have not yet secured a property title. De Sotofinds hat beforeobtaining heir land titles, families nvest as little as possible in buildingtheir housesand prefer investing n other durables; however, once the property titlesareissued by the state, families shift their investment o build and improve their houses.

Investment from capital markets is also severely affected by the lack of propercontracts. In fact, the inefficiencyof capital marketsis manifested n several ways: thehighborrowing ates paidby informal irms, the relatively low value of informal physicalcapital,and the difficulty o transfer property and createcommon-stockcorporations. DeSoto (1989)points out that in Lima, in June 1985, he nominal orrowingrate for informal

-8-


13/60

firms was 22 percent monthly, whereasfor formal firms of comparablesize, it was 4.9percent. Huq and Sultan(1991) eport hat in Bangladesh,n 1988, firms which dependedon noninstitutionalsources to meet their financial needs paid rates between 48 to 100percent annually, whereas the borrowingrate from commercialbanks was around 12percent. The difficulty to transfer and mortgage property and create common-stockcorporationsimits the informal irms' abilityto expand, manage their risk, and use moreadvanced echnologies.

The high cost of capital faced by informal firms coupled with their low cost oflabor (due to their non-compliancewith labor laws) induces them to be more laborintensive han their formal counterparts.

As the costsof informality ise, the incentives or an enterprise o becomeformalbecomestronger. On the other hand, paradoxically, he higher these costs are, the moredifficult it is for an informalenterprise o accumulate he wealth that would enable it toenter the formal sector. Credit constraints and the inability to form common-stockcorporationsprevent informalcompanies from growing and, thus, from being able toafford the access costs to the formal economy.A SIMPLE ENDOGENOUS GROWTH MODEL

The followingmodelattempts o organize omeof the informationpresented in theprevious section on the determinantsof the informalsector and its effect on economicgrowth. To keep the model manageable, he paper ignores some aspects of informalitythat, although important, have been well studied in other papers, namely, the issues ofcapital-labor intensity,labor-marketsegmentation,and size of firms. These aspects of

-9-


14/60

informality have been examined by, among others, Chaudhuri (1989), Gupta (1993),Rauch (1991), and Loayza (1994).Setup6

The economy is populatedby agents endowedwith a (possiblydifferent) startinglevelof a broad measureof capital, which is meant o includephysical as well as humancapital. They operate a technology hat exhibits constantreturns to capital (a la Rebelo1991) o producea singlegood, which can be either consumedor invested. Raw labor isnot an input of production. We followBarro and Sala-i-Martin 1992) in assuming hatthe capital rate of return depends on the availableamount of public services relative toaggregateproduction. The basic production function is then given by

Y. = A G k O


15/60

captures he factthat some regulationsnd taxes imposedby governmentare unproductiveand/ormisused. Informalagentspay a proportionalfractionof their incomeas penalties.We assume hat theproceeds rom penalties re usedboth as bribes o governmentofficialsand to partiallyfinance government'senforcementsystem(the tax administration n thissimplecase); therefore, nformal enalty revenuesare not used to pay for public servicescontained in G. Becauseof their illegal status, informal agents have access only to afraction of availablepublicservices. Therefore,according o the sector agent i belongsto, his net-of-tax/penaltyncomeis given by

Yi (1-t)A( ) k 0


16/60

G = n(q,X) ( Y F), 0'( < I(4)

in>o, @n0aq ' ax axaqwhere ri(.) is the fraction of tax revenuesavailable for the provisionof public services.The fraction I-i(.) of tax revenues is in part wasted and in part used to finance theenforcement system. We assume that rj(.) is a positive function of the quality ofgovernment nstitutions proxiedby the parameterq). This capturesthe fact that higher-qualitygovernment nstitutions mpose fewer wastefulregulationsand administerfiscalresources more efficiently. Also, given that increasingenforcementeffectiveness akesaway fiscal resources, ri(.) is a negative function of enforcementstrength (which, asexplainedbelow, is represented y the parameter A). The positive partial cross-derivativeimplies hat the amount of resourcesit takes to raise enforcementstrength decreaseswiththe qualityof governmentnstitutions. (Theefficiency of governemnt nstitutions tself isassumed o be exogenous.)

The ratio of public services to total production s then given byG 5

For given rl(.) and t, an increase in the relative size of the informal sector, I, lowerscapitalproductivity or all agents n the economy. This is so because informalproductioncongestspublic services but does not contribute o financing hem.

-12-


17/60

We assume hat the effectivepenalty rate, i, dependson both the strengthof theenforcementsystem and the extent of public dissatisfactionwith the informal sector (asarguedbelow, this dissatisfactions due to the fact that an increasein the relativesize ofthe informal sector is most likely associatedwith a decreasein everyone's productivity.)The strength of the enforcement system affects the effective penalty rate because itdetermines the ability to detect and punish informalactivities. As explained above,increasingenforcement effectiveness akes away fiscal resources, the more so the lowerthe quality of government nstitutions.

The penalty rate is thengiven byit = T(X,I), O0ax aiwhere X measures the strengthof the enforcement system, and the relative size of theinformal sector, I, measures public dissatisfactionwith the effects of informalityoncapital's rate of return. Having the effectivepenalty rate partially depend on the size ofthe informal ector is a simple way to endogenizepublic policy in the face of informality(although here is no claim that such policy is optimal.)

A simple unctional orm that conformswith equation(6), additionallypresentingpositive interactionbetween the parametersX and I, is the following:

11 = XI (7)

-13-


18/60

EquilibriumWe restrict ourselves to the study of an interior solution, that is, one where the

model's parameters are such that both sectors, formal and informal, coexist in theeconomy. Then,given hat there is free mobility crosssectors, in equilibrium the formalandinformal atesof returnmustbe equalizedat all times. This conditiondetermines herelative size of the informalsector in equilibrium. From equations(2) and (7),

C(1))8a =(1-T) (8)

Therefore,

(9)

where he sign aboveeach parameter ndicates he sign of the partial derivativeof I withrespect to this parameter. The followingresults concerningthe determinationof theinformal ector's relativesize are obtained romequation 9). When he tax rate and, thus,the incentive to evade taxes rise, the informalsector expands. If in the compositionofpublic services, a larger share corresponds o those services not available to informalagents(e.g., police, udicial,and legalsystems),hen the equilibriumsize of I drops. Therelative ize of the informal ectoralso decreaseswhen enforcementstrengthrises. Giventhat an improvement in the quality of government institutions allows strengthening

-14-


19/60

enforcement without higher drain on fiscal resources, an increase in q is also likely to leadto a smaller informal sector. Finally, when public services are more productive relativeto private services, making public services forgone by informal activities more importantin production, the relative size of the informal sector is smaller.

Given the equilibrium value for 1, the economy's net capital rate of return, r, isgiven by,

r = [A I1 i)T'] IXf oq) I - xj,ki6 a)!] (10)

The expression in the first set of brackets corresponds to the case when there is noinformal sector. The rate of return in such case is first increasing and then decreasing inthe tax rate. The informal sector, through its detrimental impact on the availability ofpublic services, creates an additional negative effect of the tax rate on the capital rate ofreturn.Utility Optimization

Agents in the economy maximize the value of discounted utility subject to theirbudget constraint:

Max U =f e Pt dt0(11)

subject to k1(t) = y1(t) ci(t)= rki(t) - c,(t)

-15-


20/60

where p is the constant rate of time preference. We follow the common practice ofassuming that the instantaneous utility function has a constant intertemporal elasticity ofsubstitution, which is equal to 1/0 in equation (11). As in other Ak growth models, therate of return, r, is independent of the path of capital accumulation. The first-order andtransversality conditions imply the following constant rate of consumption growth:

t ) = y = !(r-p) (12)

GrowthAs in the Rebelo (1991)Ak growth model, there are no transitionaldynamics in this

model. The growth rates of aggregate capital (K), aggregate production (1), as well asformal (Y F) and informal (Y' ) production, are constant and equal to the consumptiongrowth rate, y.8 The economy's long-run growth rate depends on the technology,preference, and policy parameters; in this sense, the model is one of "endogenous"growth. From equations(10) and (12), we obtain an expression for the economy's growthrate:

y = > i[A(l_T)a] |rl|,q|1-(P (13)

As the quality of government institutions mproves(higher q), the growth rate risesbecause a larger share of tax revenues is allocated to finance public services. As explained

-16-


21/60

above, an improvement in the quality of government institutions can also lead to astronger, less costly enforcementsystem and, therefore, to a smaller informal sector; thisresults in less congestion of public services and, thus, higher growth. A decrease in theamount of public servicesavailable to informal agents (lower 6) also improves growth byreducing congestion of public services due to informality.

The effect of strengthening enforcement on the growth rate is ambiguous. On theone hand, it decreases congestion of public services due to informality; but on the otherhand, it takes away fiscal resources that could be used to finance public services.However, given a standard cost function for enforcement (e.g., U-shaped average costcurve), strengthening enforcement when it is initially at a small level will produce acongestion-reduction effect that outweighs the resource-cost effect, thus increasing theeconomy's growth rate.

The effect of an increase irithe tax rate on growth, discussed in the next section,can also be positive or negative: even though an increase in the tax rate both induces morepublic-service congestion due to informality and reduces the private net rate of return, itmay also lead to a larger amount of public services.Optimal Tax Policy

To examine optimal tax policy, we compare this model with the Barro and Sala-i-Martin, 1992 (henceforth B-SM) model, in which the informal sector is not present. Theoptimal tax rate when public servicesare subject to congestion in the B-SM model is givenby

-17-


22/60

* aUB-S =I +a (14)

It can be shown that in the present model, under certain parameter conditions9 , theoptimal tax rate involves some informalityand is related to the other parameters of interestas follows,'

+ +r= (a,X,5)(15)

t


23/60

The effects of enforcement strength (X) and the share of public services available toinformal agents (6) on the optimal tax rate are explained as follows: When informalitybecomes less attractive (becauseof higher A, and/or lower 6), a sufficiently small increasein the tax rate does not lead to more public-service congestion due to informality.

In Figure 1, the solid line graphs the growth rate, y, as a function of the tax rate,r, in our model; the dotted-line graphs this relationship in the B-SM model, which doesnot allow informality and where all tax revenues are used to finance public services. As6 decreases, X increasesfrom a sufficiently low level, and q rises, the growth function inour model shifts upwards and to the right; therefore, the growth rate becomes higher forall values of the tax rate, and the optimum tax rate becomes larger.Growth and Informality

From the model presented above, the following conclusion can be drawn. Incountries where the tax burden is larger than optimal and where the enforcement systemis too weak, the relative size of the informal sector is negatively correlated with the rateof economicgrowth; in other words, changes, both in policy parameters and in the qualityof government institutions, that promote an increase in the relative size of the informaleconomy will also generate a reduction in economic growth.THE INFORMAL SECTOR IN LATIN AMERICAN COUNTRIES

This section presents some evidenceon both the determinants of the informal sectorand its effect on the provision of public services and economic growth in Latin Americancountries. The estimation period is the early 1990's.51

-19-


24/60

Determinantsand Size of the Informal SectorEstimating the size of the informalsector is difficult, particularly when it is defined

as the sector that evades regulations and taxes. Several approaches have been used in theliterature, all with limited success." This paper uses a statistical model that considers therelative size of the informal sector (informal production as a percentage of totalproduction)as a latent variable that potentially has multiple causes and for which multipleindicators can be found. This Multiple-IndicatorMultlple-Cause (MIMIC) model was firstused for the estimation of the size of the informal sector by Frey and Weck-Hanneman(1984); they concentrated their study on OECD countries. We have chosen this approachfor the following reasons. First, in the process of estimating the size of the informalsector, we obtain and test the significanceof the estimated effects of some causal variableson the informal sector. Second, given that we estimate the size of the informal sectorjointly for all countries in the sample, we can make cross-country comparisons and usethose estimates to assess their correlation with other variables of interest. Third, we canuse the information contained in different alternative or complementary indicators of theinformal sector in a single estimation process.

The variables serving as either causes or indicatorsof the informal sector have beenselected in accordance with the theory and discussion presented in the previous sectionsof the paper. The causal variables considered are the following: First, the tax burden,proxied by the highest statutory corporate income tax rate in the country."3 Second,government-imposed restrictions on labor markets, measured by the Rama (1995) indexdivided by per capita GDP to control for differences in labor productivity across countries.

-20-


25/60


26/60

The variables serving as indicators are, first, the rate of value-added tax (VAT)evasion and, second, the percentageof the non-agricultural labor force does not contribute

to social security. VAT evasion data are taken from an IMF study conducted in 1993 bySilvani and Brondolo. Data on the percentageof non-agricultural labor force that does notcontribute to social security are taken from the World Bank study Averting the Old AgeCrisis (1995). These indicators reflect the two most important aspects of the informalsector in Latin America, namely, non-compliancewith taxes and labor market regulations.Whereas the rate of VAT evasion proxies directly for the share of informal production, thepercentageof non-agriculturalworkers not covered by social security proxies for the shareof informal employment.The MIMIC Model

The MIMIC model of a latent variable (Joreskog and Goldberger, 1975) isspecified as follows,

I = 'X+ eZ 41 + i (16)

and E(qi') = 0', E(62 ) = oE, E(pt') = Q, a diagonal matrix

In the first equation, the latent variable I is linearly determined, subject to a disturbance,by a set of observable exogenous causes X. In the second equation, the latent variablelinearly determines, subject to disturbances, a set of observable endogenous indicators Z.

A reduced-form equation is obtained by substituting the first into the second-22-


27/60

equation,

Z =C X + e) +(17)='X + v

Estimationof the structural parameters is obtained through maximum likelihood, makinguse of the restrictions implied in both the coefficient matrix II and the covariance matrixof the error term v. The basic idea of the MIMIC model is that the latent variable Iaccounts completely for the intercorrelations of the indicators Z. Once the effects of thecausal variablesX and the disturbanceE on each of the indicators are removed, there is nocorrelation among the indicators.

Note that the reduced-form parameters remain unchanged when '1 s multiplied bya scalar and f and OE are divided by the same scalar. Therefore, in order to remove theindeterminacyin the structuralparameters, it is necessaryto adopt a normalization, setting,for instance, one of the coefficientsin4 equal to one. A normalized estimate of the latentvariable can be obtainedfrom the estimated values of the causal-variablecoefficients P (seethe first equation in (16)). We can then compare the differences in the latent-variablevaluesof any two units (countries in our case) and, thus, rank all units accordingly. Thelevelof the latentvariable, however, remains undetermined unless additional information

is obtained (or assumed.)Given that it is difficult to compare the effects of different explanatory variables

on the same dependent variable when they have different units of measurement (and

-23-


28/60

specially when the regression coefficientshave been normalizedby setting one of themequal to an arbitrary value), it is useful to standardize the regression coefficients asfollows,

px= p a (18)

where P represents n estimatedegressioncoefficient,a represents an estimatedstandarddeviation,and the subscripts and I denote,respectively, givenexplanatoryvariable andthe dependent ariable. The standardized oefficientis, therefore, the expected change nstandard-deviationunits of the dependentvariablethat is produced by a one standard-deviation change of a given explanatoryvariable when the other variables are heldconstant.Estination Results

Figure 2 reports the estimationresults. The standardized egressioncoefficientsand their respective -values (inparenthesis)are presentedby the arrow pointing in thedirectionof influencein the model.15

The coefficients n the threecausal variables in the modelhave the expectedsignsand are statisticallysignificantat the 10%level. Both the tax burden and labor-marketrestrictionsaffect positivelytherelative size of the informalsector,with a one standard-deviationncrease n each of themproducinga rise in the informal ector of 0.33 and 0.49standard deviations, respectively. The strength and efficiency of the governmentinstitutionshas a negativeimpact on the informalsector,with a one standard-deviation

-24-


29/60

increase in the former leading to a decrease of 0.42 standard deviations in the latter.Therefore, he theoreticalmodel's conclusions oncerninghe determinantsof the informalsector are validated by the statisticalevidenceof the MIMIC model.

The VAT evasion ateand the percentageof non-agriculturalworkers not coveredby social security perform well as indicatorsof the relative size of the informal sector.The estimated atent variable explains75 % of the variance of the first indicatorand 34%of the variance of the second one.

Although the small size of the sample merits caution on the evaluation of theresults, several goodness-of-fitstatistics, including those that adjust for degrees offreedom, support the underlyingmodel (see the bottomof Figure 2). The Chi-squarestatistic for the null hypothesis that the constraints on the residual covariancematriximpliedby the modelare valid has a p-valueof 0.78. TheJoreskog and Sorbom (1985),Bentler and Bonett (1980), and Bollen (1986) indices (which do not have a knowndistribution nd whosevaluesmayrange from 0 to 1) also support the statisticalmodel asthey are quiteclose to 1, a value which representsperfect fit.

Using the procedure outlined in the previous section, we can estimate thestandardizedvalues (z-values)of the relative size of the informal sector for all countriesin our sample. These estimatesare presented n Table 1. The countries with the largestinformal ectorare Bolivia, Panama, and Peru, countrieswhich in the early 1990's werenotorious for their restrictive labor-market regulations and poor quality of theirenforcementsystems. They are followedby two Central Americancountries, namely,Guatemala ndHonduras. In the middleof the list we find, Brazil, Colombia, Uruguay,

-25-


30/60

Venezuela, Ecuador, and Mexico. Finally, the countries with the smallest informal sectorare Costa Rica, Argentina, and Chile; in the early 1990's these three countries were the

most advanced in the region in terms of market-oriented economic reforms.The correlationof the size of the informal sector with real per capita GDP in 1990

is -0.70, and its correlation with the share of urban population in the same year is -0.53.This suggests that the size of the informal sector is related to the development level of thecountry. Governments in less developed countries may attempt to regulate and tax theireconomies according to the standards of the more developed ones without having thecapacity to enforce compliance. Enforcement of regulations and taxes is difficult in lessdeveloped countriesbecause they tend to have weak bureaucracies, frail legal institutions,and poorly educated, largely rural populations.

As mentionedabove, the standardized z-values of the latent variable enables us todetermine only the relative position of countries in the sample. In order to obtain theabsolutevalue of the size of the informal sector (as a percentage of GDP), two points mustbe fixed: one to set the overall scale, another to set the distance between the ranks. Weuse the following two assumptions: First, the average size of the informal sector is thesame as the average of the VAT evasion rate. Second, the size of the informal sector inChile is the same as its VAT evasion rate. These assumption are sensible because, first,

the VAT evasion rate is directly related to the share of informal production (whereas oursecond indicator is more closely linked to informal labor employment); second, a largefraction of the variance of the VAT evasion rate is explained by the size of the informalsector according to our statistical model. The last column of Table 1 show our estimated

-26-


31/60

values of the size of the informal sector, as a percentage of GDP.The VAT evasion rate is calculated as a ratio to official GDP. Therefore, given

that the VAT evasion rate has been used to pin down the overall scale and the distancebetween the ranks of the informal sector, the estimated values of the informal sectorreported in Table 1 are also given as a percentage of official GDP. The question remainswhether official GDP includes informal production. National account authorities in LatinAmerica are aware of the large presence and order of magnitude of the informal sector intheir respectivecountries. For this reason they rely not only on income reports to estimateGDP but, most importantly, on sampling and surveys of production and prices byeconomic sectors. Therefore, although actual GDP may be measured with error, thedirection of the bias is uncertain. At any rate, official GDP does not ignore thecontribution of the informal sector.The Relation Between the InformalSector, Public Infrastructure, and EconomicGrowth

The theoretical model presented n section III links the informalsector with theprovisionof publicservices nd the rate of economicgrowth. The model predicts that therelative size of the informal sector is negativelycorrelated with both the availabilityofpublic services and the economy's growth rate (except when the increasein the informalsector is due to a small rise in the tax rate from a sufficientlysmall level or a drop inenforcement trength rom a highand costly level. Nonetheless, he case of suboptimallylow statutory tax rates or excessiveenforcementstrength hardly applies to any LatinAmericancountry).

-27-


32/60

Table 2 reportssome regressionesults. The economy'sgrowthrate is representedby the averagegrowth rate of real per capita GDP for the period 1980-92. This periodis long enough o minimize he effectof business-cycleluctuationson the growth rate andshort enough to be able to assumethat the informal sector is basically stable throughoutthe period. The provision of public services is proxied by an index of publicinfrastructure, which consists of the average of standardized values of four indicators.These indicatorsare per capita electricityconsumption,per capita telephone mainlines,percentage f populationwith access o safe water,and per capita roads in good condition.There are some importantpublic services not included n these indicators, most notably,the police and the judicial system. Nevertheless, o the extent that these omitted publicservices are correlated with the four indicators considered, the proposed public-infrastructure ndex is a good proxy for all availablepublic services.

Columns 1 and 3 of Table 2 show that the size of the informal sector has asignificantlynegativeimpact on real per capita GDP growth, whetheror not the public-infrastructurendex s includedas an explanatoryvariable in the regression. The fact thatthe informal sector's size has a negative effect on growth when the index of publicinfrastructures held constantcan be explained n two ways, both probably correct. First,an increase n the informal sector's size means that more activitiesare using some of the

existing public services less efficiently or not at all; second, our public-infrastructureindex does not account for all productive public services. Comparingcolumns 2 and 3shows that the public-infrastructure ndex affects positive and significantly he rate ofeconomicgrowth when it is included as the only explanatoryvariable in the regression;

-28-


33/60

however,whenthe size of the informal ectorbecomesan additionalexplanatoryvariable,the coefficienton the public-infrastructure ndex becomesinsignificant. The informalsector, being stronglynegativelycorrelated to the public-infrastructurendex, capturesmost of the index's explanatorypower in the growth regression.

In regressionnumber4, the determinantsof the informalsectorare useddirectlyas explanatory ariables. Althoughheir respectivecoefficientspresent the expectedsign,they are not individuallyignificant. However,they are jointly significant(the p-valueofthe F-statisticis 0.0798), which, whencoupled with the lack of individualsignificance,indicates he presence of multicollinearity. As a "weightedaverage" of the tax burden,labor-market restrictions, and the quality of government institutions, the size of theinformal sector serves as an overall indicatorof regulatoryefficiency.

There is the possibility hat the negative orrelation etween he size of the informalsector and economic growthis spurious, in the sense that the size of the informal sectormay be proxying for something different from regulatory efficiency; for example, theinformal sector size may be proxyingfor fiscal and monetarydiscipline, lack of overalldevelopment, low human capital, and trade restrictiveness,variableswhich have beenshown o be correlatedwitheconomicgrowth. To address his issue, we ran "augmented"growth egressions, which consider severalexplanatoryvariables hat represent differentaspectsof policy and economicdevelopment. The resultsare presented n Table 3. Theconclusion s that the sizeof the informal ectorremains ignificantlynegativelycorrelatedwith economicgrowth, thus demonstrating o be an independentdeterminantof growth.

-29-


34/60

CONCLUSIONThis paper presents the view that the informal economyarises when excessive taxes

and regulationsare imposedby governmentsthat lack the capability to enforce compliance.The determinants and effects of the informal sector are studied in an endogenous

growth model whose production technology depends essentially on congestable publicservices. The model concludes that in economies where the statutory tax burden is largerthan optimal and where the enforcement system is too weak, the relative size of theinformal sector is negativelycorrelated with the rate of economic growth; in other words,changes, both in policy paraneters and the quality of government institutions, that promotean increase in the relative size of the informal economy will also generate a reduction inthe rate of economic growth.

Many Latin American countries have (or had until recently)a tradition of excessiveregulations and weak government institutions. The paper uses data from Latin Americancountries in the early 1990's to test some of the implications of the model and to provideestimates for the size of the informal sector throughout these countries. The empiricalapproach consists of identifying the size of the informal sector to a latent variable forwhich multiple causes and multiple indicators exist. The size of the informal sector isfound to depend positively on proxies for tax burden and labor-market restrictions, andnegatively so on a proxy for the quality of government institutions. Furthermore, theempirical results indicate that an increase in the size of the informal sector negativelyaffects growth by, first, reducing the availability of public services for everyone in theeconomy, and, second, increasing the number of activities that use some of the existing

-30-


35/60

public services less efficiently or not at all. These results hold even when additionalexplanatory variables are added to account for other aspects of economic conditions and

policies.Finally, it is estimated that the largest informal sectors in Latin America are found

in Bolivia, Panama, and Peru, and the smallest, in Chile, Argentina, and Costa Rica.

-31-


36/60

ENDNOTES1. The concept of informalitywas first introduced by the 1972 ILO Employment Mission

to Kenya. For an excellent survey regarding definitions, empirical studies, and policyrecommendations on the informal sector, see Lubell (1991).2. The work by Hernando de Soto, presented in his book The Other Path (1989), hascontributed enormously to the understanding of the role of political institutions and legalstructures in the rise of informal sectors. De Soto's analysis draws from the contributionsof Douglas North (1981) and Mancur Olson (1982).3. Readers familiar with the determinants of informal economies may want to go directlyto the theoretical model.4. See Romer (1986), Lucas (1988), Barro (1990), and Rebelo (1991).5. Given that the codes of labor legislation are also rather extensive in developedcountries, it is not surprising to find "clandestine," or informal, employment in those

countries as well. However, given that the productivity of labor is higher in developedthan in underdevelopedcountries, the distortionarypower of labor regulations in developedeconomies is less important. Kaltzmann (cited in De Grazia, 1982, p. 34), discussing theFrench experience, writes that "the main reason for clandestine employment is economic:one of the parties increases his income, the other reduces his costs. Two economic causesplay a dominant role: inadequate remuneration and the burden of taxation and socialsecurity costs." De Grazia (1982, p. 35) writes that "In Italy, the phenomenon [ofinformal employment] is ... more blatant, if not universal. Entire neighborhoods ofNaples have been transformed into secret workshops(specializing particularly in shoe- and

-32-


37/60

garment-making) which move on quickly or disappear the moment a visit by the laborinspectors seems likely." De Grazia (1982) also reports that in Milan only 5,000homeworkers, from a total of 100,000, and 1,000 small enterprises, from a total of50,000, are listed on the city's commercial register. Indeed, the practice of subcontractingas a means to avoid labor-protection legislation (including prohibitions to employ illegalimmigrants) has become particularly popular in developed countries, especially the UnitedStates, Canada, Italy, Spain, and France (Portes, Castells, and Benton, 1989.)6. The model presented here is a substantiallymodified version of the model in Braun andLoayza (1994).7. From equation (9), an interior solution for I requires the following parameterrestrictions:

850+t-1 > 0 (-I>0)(1-A)6a+t-l1< 0 (=Ir> p1-0

The second inequality ensures positive growth, and the first one ensures that attainableutility is bounded and the transversality condition holds (see Barro and Sala-i-Martin,1995).9. If 1-6 < a1/(a+X+aX), the growth-maximizing tax rate implies some informality.

-33-


38/60

Otherwise,hat is, if 1-8">aX/(a+Xl+a),he optimal ax rate brings about a fully formaleconomy. We assume that the first inequalityholds.10. In general, it is notpossible to obtainan explicit solutionfor t+. However, for thecase when X= 1 (maximumstrengthof the enforcementsystem),T* takesa simple form:t*=c/(1 +2a). The comparative tatics resultspresented n equation(15) for the generalcase are obtainedapplying the implicit function heorem.11. From 1993 to the present, a few countries (notably Peru) have undertaken anaccelerated pace of market-orientedreforms. Given that the period of estimationconsidered in this paper is the early 1990's, the estimated results may not apply to therecent experienceof such countries.12. For a good survey on different approaches o the estimationof the informal sector,see Gupta and Gupta (1984).13. It would have been preferable o use the average marginalcorporate tax rate as theproxy for tax burden; unfortunately, data on tax revenue by rates was unavailable.Nevertheless, he highest tax rate is not only a good measureof the level of tax rates butalso of their dispersion.14. The appendixpresents the data used in the paper and their sources.15. In order to remove he indeterminacy f the structuralcoefficients, he coefficientonthe VAT evasion rate wasset equal to one; therefore, a t-test can not be appliedto thiscoefficient. Setting he coefficient n the other indicatorequal to one producedessentiallythe same results.16. The average growthrate was calculatedusing the least-squaresmethod, in order to

-34-


39/60

make use of the observations or all years in the period.

-35-


40/60

REF'ERENCELISTAlonzo, R.

(1991) The Informal Sector in the Philippines. TheSilent Revolution. Eds. A.Chickeringand M. Salahdine. InternationalCenter for Economic Growth, San Francisco.

Barro, R.(1990) GovernmentSpending in a Simple Model of Endogenous Growth. Journal

of Political Economy, October, supplement.

Barro, R., and X. Sala-i-Martin(1992) Public Finance in Models of EconomicGrowth. The Review of Economic

Studies, October.

Barro, R., and X. Sala-i-Martin(1995) Economic Growth. McGraw-Hill, Inc.

Bentler, P.M., and D.G. Bonett(1980) Significance Tests and Goodness of Fit in the Analysis of Covariance

Structures. Psychological Bulletin, 88: 588-606.

Bollen, K.A.

-36-


41/60

(1986) Sample Size and Bentler and Bonett's Nonnormed Fit Index.Psychometrika,51: 375-377.

Braun, J., and N. Loayza(1994) Taxation, Public Services, and the Informal Sector in a Model of

EndogenousGrowth. Policy ResearchWorking Paper No. 1334, The World Bank, July.

Burgess, R., and N. Stern(1993) Taxation and Development. Journal of Economic Literature, Vol. 31,

June.

Chaudhury, T.(1989) A Theoretical Analysis of the Informal Sector. World Development,

March.

Chickering, A., and M. Salahdine(1991) Introduction and The Informal Sector Search for Self-Governance. The

Silent Revolution. Eds. A. Chickering and M. Salahdine. International Center forEconomic Growth, San Francisco.

De Grazia, R.

-37-


42/60

(1982) ClandestineEmployment:A Problemof Our Times. The UndergroundEconomy in the UnitedStatesandAbroad. Eds. V. Tanzi, D.C. Heath and Company.Lexington,Massachusetts.

De Soto, H.(1989) The OtherPath. Harper& Row, NewYork.

Frey, B., and H. Weck-Hanneman(1984) The HiddenEconomyas 'UnobservedVariable'. EuropeanEconomic

Review,26: 33-53.

Gupta, M.(1993) Rural-UrbanMigration, InformalSector,and DevelopmentPolicies:A

TheoreticalAnalysis. The Journalof DevelopmentEconomics,June.

Gupta, P., and S. Gupta(1984) Black Economy: A Review of Methodologies. The Unsanctioned

Economy.

Huq, M., and M. Sultan(1991) 'Informality'n Development:The Poor as Entrepreneurs n Bangladesh.

-38-


43/60

The SilentRevolution. Eds. A. Chickeringand M. Salahdine. InternationalCenterforEconomicGrowth, San Francisco.

Joreskog,K., and A. Goldberger(1975) Estimation f a ModelwithMultipleIndicatorsand MultipleCausesof a

Single LatentVariable. Journalof the AmericanStatisticalAssociation,September.

Joreskog, K., and D. Sorbom(1985) LISREL VI: Analysis of Linear StructuralRelationshipsby Maximum

Likelihood, nstrumentalVariables,andLeast Squares. Uppsala:Universityof Uppsala.

Knack, S., and P. Keefer(1994) Institutionsand EconomicPerformance:Cross-CountryTests Using

Alternative nstitutionalMeasure. IRISWorkingPaper#109 (Economicsand Politics,forthcoming).

Loayza, N.(1994) Labor Regulationsand the InformalSector. Policy ResearchWorking

Paper No. 1335,The World Bank, July.

Lucas,R.

-39-


44/60

(1988) On the Mechanics of Economic Development. Journal of MonetaryEconomics, July.

Lubell, H.(1991) The InfonnalSector n the 1980sand 1990s. DevelopmentCentre Studies,

OECD, Paris.

Mazumdar,D.(1976) The Urban InformalSector. WorldDevelopment,Vol. 4, No. 8.

Nipon, P.(1991) The Informal Sector in Thailand. The Silent Revolution. Eds. A.

Chickering nd M. Salahdine.InternationalCenterfor EconomicGrowth, San Francisco.

North, D.(1981) Structureand Change n EconomicHistory. Norton, New York.

Olson, M.(1982) The Rise and Decline of Nations: Economic Growth, Stagflation,and

Social Rigidities. Yale UniversityPress, New Haven.

-40-


45/60


46/60

Silvani,C., and J. Brondolo(1993) An Analysisof VAT Compliance. Mimeo, Fiscal Affairs Department,

InternationalMonetaryFund, 1993.

Shome, P.(1992) Trends and Future Directionsin Tax PolicyReform:A Latin American

Perspective. Bulletinof the InternationalBureauof Fiscal Documentation,September.

Tokman,V.(1992) The InformalSector in Latin America: From Underground o Legal.

Beyond Regulation: The Informal Economy in Latin America. Ed. V. Tokman.PREALC,Lynne Rienner,Boulder,Colorado.

TheWorldBank(1994) WorldDevelopmentReport. Washington,D.C.

TheWorldBank(1995) WorldDevelopmentReport. Washington,D.C.

The World Bank(1995) Averting he Old Age Crisis. Washington,D.C.

-42-


47/60

AppendixI Data SourcesCausal Variables:1. Corporate ncome ax rate. Data on corporateincome tax rates in 1991 are taken fromShome(1992). When he countryhas more han one rate,the highest ate is used. Boliviais a specialcase. In 1991, there was no tax on corporate incomeper se. A tax of 3 % ofnet worthwas levied nstead. In order to estimate the impliedcorporate income ax rate,the net-worthtax rate is dividedby the real interest rate. For the calculationof the realinterest rate, the neoclassical-growth-modelelation between growth, real interest anddepreciationrates is used. Using the parametervalues presented in Barro and Sala-i-Martin (1995), the real interest rate in Bolivia is estimated at 8.1 %.2. Labor marketrestrictions. The proxy for labor market restrictionsis the ratio of theindex of labor market egulationspresented in Rama(1995) to real per capita GDP. TheRama index s dividedby real per capita GDP in order to account for differences n laborproductivity crosscountries. Rama considerseight types of labor-marketregulations or31 countries in Latin Americaand the Caribbeanduring the period 1980-92,when theseregulations emained basicallyunchanged. The eight variables consideredby Rama arenumberof ILO conventionsatified, number of days of annual leave withpay, number ofdays of maternity eave,social ecuritycontributions a percentageof wages,governmentemploymentas a percentageof labor force, minimumwage as a percentageof averagewage, severancepay and percentageof labor force in unions. After normalizing eachvariable, he Rama indexis obtained by averagingover the eight variables. The real percapitaGDP data correspond o the year 1990and are taken from Summers-HestonPenn

-43-


48/60

World Table 5.6.3. The strength of the enforcementsystem. The strength of the enforcementsystem isproxiedby an averageof three subjectivendicators eported in InternationalCountryRiskGuide (ICRG) or the period 1990-92. The three indicatorsconsideredare quality of thebureaucracy, orruption n government,and rule of law. For the indicatorquality of thebureaucracy,high scores indicate "autonomyfrom politicalpressure" and "strength andexpertise to govern without drastic changes in policy or interruptionsin governmentservices",also existenceof an "establishedmechanism or recruiting and training." Forcorruption n government,lower scores indicate "high governmentofficials are likely todemand pecialpayments"and "illegalpaymentsare generallyexpected throughout owerlevelsof government" in the form of "bribes connectedwith import and export licenses,exchange ontrols, ax assessment, oliceprotection, or loans". Rule of law "reflects thedegree o whichthe citizens of a countryare willing to accept the established nstitutionsto make and implement aws and adjudicatedisputes." Higher scores indicate "soundpolitical institutions,a strong court system, and provisions for an orderly successionofpower." Lower scores indicate "a tradition of dependingon physical force or illegalmeans o settleclaims." ICRG s a publicationof Political Risk Servicesof Syracuse, NYand is cited in Knack and Keefer (1994).Indicatorg:1. Value-added ax evasion ate. Data for this variable correspond o one of the years inthe period 1990-93for each country in the sample. The VAT evasion rate is calculatedas one minus the tax compliance rate, where the tax compliance rate is calculated as

-44-


49/60

follows,VAT revenue

compliance rate Potential base of VATAverage VAT rate

The VAT evasion rate data are obtained from Silvani and Brondolo(1993) for Chile,Honduras,Guatemala,Panama,Uruguay,Argentina,Ecuador,Mexico,Bolivia, Colombiaand Peru. For Brazil, Costa Rica and Venezuela, he compliance ates are not available.However, herevenueproductivity the percentage f GDP collectedper point of the VATrate) can be obtained from the IMF. Data on countrieswith both compliance ate andrevenue productivityare used to fit a line relating hese two measures. The relationshipis then used to estimate the compliance ates for thosethree countries.2. Fractionof labor force not contributing o socialsecurity. This variable is obtainedfromAverting The Old Age Crisis, the World Bank, 1995. The variable measuresthenumber of workersnot contributing o social security as a percentageof the labor force.Data are available for all Latin Americancountries in our sample. The years for whichdataare available, however,are not the same. For most countries the figurescorrespondto yearsaround 1991,but for some he latestavailable ear is in the mid 80's (e.g. Bolivia1985,and Guatemala1986).

Infrastructure:1. ElectricityConsumption KWH per person) in 1992. This is calculated as electricityproduction x (1 - system losses). Electricityproduction (KWHper person)and system

-45-


50/60

losses (% of total output) are obtained from the World Development Report 1995, theWorld Bank.2. TelephoneMainlines per 1,000 persons) in 1992. This variable is obtained from theWorldDevelopmentReport 1995, the World Bank.3. Roads n goodcondition km per millionpersons) in 1988. This is calculatedas Roaddensity (km per million persons) x Roads in good condition (% of paved roads). Dataare from the WorldDevelopmentReport 1994, the World Bank.4. Populationwith access to safe water (% of total) in 1990. Data are from the WorldDevelopmentReport 1994, the World Bank.5. Public nfrastructure ndex. It is the simple average of the standardizedvalues of theabove fou; indicators. All indicatorsare standardizedby computing z-values, that is,dividing he deviations from the mean by the standarddeviation.Growth Rate:The growth ratesof real per capitaGDP from 1980 o 1992are calculatedusing the least-square method. Real per capita GDP data are from theWorld Bank National Accounts.

-46-


51/60

Table 1. The Size of the Informal Sector: Standardized and Absolute ValuesStandardizedValue Absolute Value

Country (% of GDP)Chile -1.342 18.2Argentina -1.107 21.8Costa Rica -1.012 23.3Mexico -0.762 27.1Venezuela -0.523 30.8Ecuador -0.494 31.2Colombia -0.240 35.1Uruguay -0.236 35.2Brazil -0.062 37.8Honduras 0.516 46.7Guatemala 0.754 50.4Peru 1.243 57.9Panama 1.518 62.1Bolivia 1.746 65.6Mean 0.000 38.8Standard deviation 1.000 15.3

-47-


52/60

Table 2. The Growth Effects of Public Infrastuctureand the Informal Sector(t-statistics in parenthesis)Dependent Variable Growth Rate of Real Per Capita GDP Public

(1980-92) InfrastructureIndex(1) (2) (3) (4) (5)

Size of the Informal Sector -0.8852 -0.8435 -0.5814(-2.61) (-2.16) (-2.98)Public Infrastructure Index 0.5622 0.0718

(1.69) (0.24)Corporate Income Tax Rate -0.4436(-1.09)Labor-Market Restrictions -0.4333(-0.84)Strength and efficiency of 0.3598government institutions (1.16)

P-value (F-Statistic) 0.0233 0.0798AdjustedR2 0.3584 0.1201 0.3068 0.2537 0.3381Number of Observations 14 14 14 14 14Note: The above t-statistics are computed using heteroskedasticity-correctedstandard errors.Regression coefficients are standardized so that they reflect the change in the growth rateproduced by a one-standard deviation of the explanatory variable.

-48-


53/60

Table 3. The Informal Sector and Other Determinants of Economic Growth(t-statistics in parenthesis)

Dependent Variable Growth Rate of Real Per Capita GDP (1980-92)(1) (2) (3) (4) (5)

Size of the Informal Sector -1.2482 -1.3165 -1.3303 -1.4555 -1.2245(-3.33) (4.30) (-3.79) (-3.45) (-3.44)Real Per Capita GDP, 1980 -0.6418 -0.7956 -0.8906 -0.1523 -1.9685(-2.04) (-2.92) (-2.80) (-3.84) (-4.36)Secondary School 0.4302 0.7341 0.8426 0.7118Attainment, 1980 (1.86) (2.32) (2.66) (2.34)Average Tariff for -0.5226 -0.6138 -0.3439Intermediate and Capital (-1.26) (-1.46) (-0.86)Goods, 1985Average InflationRate, -0.2497 -0.60271980-92 (-0.46) (-1.60)Public Infrastructure Index, 1.19661990 (2.04)

P-value (F-Statistic) 0.0046 0.0077 0.0305 0.0724 0.0365Adjusted R2 0.4595 0.5119 0.5222 0.4735 0.6490Number of Observations 14 14 12 12 12Note: The above t-statistics are computed using heteroskedasticity-corrected standard errors.Regression coefficients are standardized so that they reflect the change in the growth rateproduced by a one-standard deviation of the explanatory variable.

-49-


54/60

Figure 1. The Growth Rate in the Presence of the InformalSector

-50-

, I

, I

, I

i-al a s 1

-50-


55/60

Figure 2. Results of the MIMIC Model

Share of VarianceCauses Indicators Explained by the

Tax Burden 0.33 Informal SectorL j U ~~~.84)VAT Evasion

Labor-Marke- 0.49 Size of Rate 0.751Labor-Market 10.49 the 0___ ~ ~ 2-4) Inform alestrictions (2-0i)1' Inoma

| Sec tor 0.5o Percentage of(2..30) Non-Strength and agricultural 0.341efficiency of W orkers Notgovernment 0.42 Covered byinstitutions (-1.75) SocialSecurity

Period of Estimation: 1990-1993Number of Observations: 14Overall Model Fit:

-- I: Constraints on residual covariance matrix implied by the model are validX2(2) = 0.5095, p-value = 0.7751-- Joreskog and Sorbom 1986: Goodness of Fit (GFI) = 0.9849

GFI Adjusted for Degree of Freedom = 0.8864-- Bentler and Bonett 1980: Normed Index = 0.9810-- Bollen 1986: Normed Index (Accounting for Degrees of Freedom) = 0.9051

-51-


56/60

Figure 3. The Relation of Public Infrastructure and Growth with the InformalSector

(a)n~~~~~~~~~~~~~~~~~~~~~~~~~~' y =-0.8852x

(-2.61)3

2

.5 -1 1.5

-2

Size of the Informal Sector

(b)

.................................................................................................... ,.,... ......................................................y = -0.5814x(-2.98)

.......................................... - . ...................................................................................................... 1__Size of the Informal Sector

-52-


57/60


58/60


59/60

Policy Research Working Paper SeriesContact

Title Author Date for paperWPS1705 The Polish Expenence with Bank Fernando Montes-Negret January 1997 T Ishibe

and Enterprise Restructuring Luca Papi 38968WPS1706 Monetary Policy during Transition: Martha de Melo January 1997 C. Bernardo

An Overview Cevdet Denizer 37699WPS1707 Trade Reorientation and Productivity Simeon Djankov January 1997 J Ngaine

Growth in Bulgarian Enterprises Bernard Hoekman 37947WPS1708 Has Latin America's Post-Reform William Easterly January 1997 R. Martin

Growth Been Disappointing? Norman Loayza 31320Peter Montiel

WPS1709 Poverty Comparisons with Jeari Olson Lanjouw January 1997 A RamirezNoncompatible Data Theory and Peter Lanjouw 85734Illustrations

WPS1710 Why Paper Mills Clean Up: Raymond S. Hartman January 1997 D. WheelerDeterminants of Pollution Abatement Mainul Huq 33401in Four Asian Countries

WPS1711 Issues in Comparing Poverty Trends Christine Jones January 1997 R MartinOver Time in C6te d'lvoire Xiao Ye 31320

WPS1712 Demand Elasticities in International Arvind Panagariya December 1996 J BadamiTrade: Are They Really Low? Shekhar Shah 80425

Deepak Mishra

WPS1713 Why Did Colombian Private Savings Alejandro Lopez January 1997 E. KhineDecline in the Early 1990s? 37471WPS1714 Fiscal Federalism in Bosnia- William Fox January 1997 Y. Jiwa

Herzegovina: The Dayton Challenge Christine Wallich 34848WPS1715 The Evolution of Poverty and Welfare SudharsharnCanagaralan January 1997 B. Casely-Hayford

in Nigeria, 1985-92 John Ngwafon 34672Saji Thomas

WPS1716 Reforming Pensions in Zambia: An Monika Queisser January 1997 H ArbiAnalysis of Existing Schemes and Clive Bailey 34663Options for Reforrn John Woodall

WPS1717 Fiscal Federalism in Bosnia- William Fox January 1997 Y JiwaHerzegovina: The Dayton Challenge Christine Wallich 34848WPS1718 Does Environmental Regulation Muthukumara Mani February 1997 E. de CastroMatter? Determinants of the Location Sheoli Pargal 89121of New Manufacturing Plants in India Mainul Huqin 1994


60/60

PolicyResearchWorkingPaperSeriesContactTitle Author Date for paper

WPS1719 Shifting Responsibility for Social David Sewell February 1997 C LawrenceServices As Enterprises Privatize in 32216Belarus

WPS1720 The Distribution of Foreign Direct Harry G. Broadman February 1997 J. GrigsbyInvestment in China Xiaolun Sun 82423WPS1721 EU Accession of Central and Luca Barbone February 1997 L. Barbone

Eastern Europe: Bridging the Juan Zalduendo 32556Income GapVVPS1722 Uncertainty, Instability, and Luis Serven February 1997 E. KhineIrreversible Investment: Theory, 37471

Evidence, and Lessons for AfricaWPS1723 The Productivity Effects of Lixin Colin Xu February 1997 P. Sintim-Aboagyeof Decentralized Reforms: An 38526

Analysis of the ChineseIndustrial Reforms

WPS1724 Debt Maturity and Firm Performance Fabio Schiantarelli February 1997 B. MooreA Panel Study of Indian Companies Vivek Srivastava 38526WPS1725 Access to Long-Term Debt and Fidel Jaramillo February 1997 B. MooreEffects on Firms' Performance: Fabio Schiantarelli 38526

Lessons form EcuadorWPS1726 Roads, Population Pressures, and Maureen Cropper February 1997 A. MaranonDeforestation in Thailand, Charles Griffiths 39074

1976-89 Muthukumara ManiWPS1727 The Economics of the Informal Norman A. Loayza February 1997 R. MartinSector A Simple Model and Some 39026

Empirical Evidence from LatinAmerica

WPS1728 Regional Labor Markets during Chris Manning February 1997 J. IsraelDeregulation in Indonesia: Have the 85117Outer Islands Been Left Behind?

loayza mimic 1996

Documents