db methodology regulation of entry - doing business

THEQUARTERLY JOURNAL

OF ECONOMICSVol. CXVII February 2002 Issue 1

THE REGULATION OF ENTRY*SIMEON DJANKOVRAFAEL LA PORTA

FLORENCIO LOPEZ-DE-SILANESANDREI SHLEIFER

We present new data on the regulation of entry of start-up rms in 85�countries. The data cover the number of procedures, ofcial time, and ofcial cost� �that a start-up must bear before it can operate legally. The ofcial costs of entry�are extremely high in most countries. Countries with heavier regulation of entryhave higher corruption and larger unofcial economies, but not better quality of�public or private goods. Countries with more democratic and limited governmentshave lighter regulation of entry. The evidence is inconsistent with public interesttheories of regulation, but supports the public choice view that entry regulationbenets politicians and bureaucrats.�

I. INTRODUCTION

Countries differ signicantly in the way in which they regu-�late the entry of new businesses.To meet government require-ments for starting to operate a business in Mozambique, anentrepreneur must complete 19 procedures taking at least 149business days and pay US$256 in fees. To do the same, anentrepreneur in Italy needs to follow 16 different procedures, payUS$3946 in fees, and wait at least 62 business days to acquire the

* We thank Tatiana Nenova, Ekaterina Trizlova, and Lihong Wang for ableresearch assistance,and three anonymous referees,Abhijit Banerjee, RichardCaves, Edward Glaeser, Roumeen Islam, Simon Johnson, Lawrence Katz, DavidLaibson, Guy Pfeffermann, and seminar participants at George Mason Universityand the University of Maryland at College Park for helpful comments.Thecollection of data for this paper was nanced by the World Bank’s Research�Advisory Group and the World Development Report 2002: Building Institutionsfor Markets. An appendix describing country data is available from the authors onrequest.

© 2002 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnology.The Quarterly Journal of Economics, February 2002

1

necessary permits.In contrast, an entrepreneur in Canada cannish the process in two days by paying US$280 in fees and�completing only two procedures.

In this paper we describe the required procedures governingentry regulation, as well as the time and the cost of followingthese procedures, in 85 countries. We focus on legal requirementsthat need to be met before a business can ofcially open its doors,�the ofcial cost of meeting these requirements, and the minimum�time it takes to meet them if the government does not delay theprocess. We then use these data to evaluate economic theories ofregulation. Our work owes a great deal to De Soto’s [1990] path-breaking study of entry regulation in Peru. Unlike De Soto,welook at the ofcial requirements, ofcial cost, and ofcial time—� � �and do not measure corruption and bureaucratic delays thatfurther raise the cost of entry.

Pigou’s [1938] public interest theory of regulation holds thatunregulated markets exhibit frequent failures, ranging from mo-nopoly power to externalities. A government that pursues socialefciency counters these failures and protects the public through�regulation. As applied to entry, this view holds that the govern-ment screens new entrants to make sure that consumers buy highquality products from “desirable” sellers. Such regulation reducesmarket failures such as low quality products from y-by-night�operators and externalities such as pollution. It is “done to ensurethat new companies meet minimum standards to provide a goodor service. By being registered, new companies acquire a type ofofcial approval, which makes them reputable enough to engage�in transactions with the general public and other businesses”[SRI 1999, p. 14]. The public interest theory predicts that stricterregulation of entry, as measured by a higher number ofproce-dures in particular, should be associated with socially superioroutcomes.

The public choice theory [Tullock 1967;Stigler 1971;Peltz-man 1976] sees the government as less benign and regulation associally inefcient.� It comes in two avors.� In Stigler’s [1971]theory of regulatory capture,“regulation is acquired by the in-dustry and is designed and operated primarily for its benet.”�Industry incumbents are able to acquire regulations that createrents for themselves, since they typically face lower informationand organization costs than do the dispersed consumers. In thistheory the regulation of entry keeps out the competitors andraises incumbents’ prots. Because stricter regulation raises bar-�

2 QUARTERLY JOURNAL OF ECONOMICS

riers to entry, it should lead to greater market power and prots�rather than benets to consumers.�

A second strand of the public choice theory, which we call thetollboothview, holds that regulation is pursued for the benet of�politicians and bureaucrats [McChesney 1987; De Soto 1990;Shleifer and Vishny 1998]. Politicians use regulation both tocreate rents and to extract them through campaign contributions,votes, and bribes. “An important reason why many of these per-mits and regulations exist is probably to give ofcials the power to�deny them and to collect bribes in return for providing the per-mits” [Shleifer and Vishny 1993,p. 601].The capture and toll-booth theories are closely related, in that they both address rentcreation and extraction through the political process. The capturetheory emphasizes the benets to the industry,� while the toll-booth theory stresses those to the politicians even when theindustry is left worse off by regulation.

In principle, the collection of bribes in exchange for releasefrom regulation can be efcient.� In effect, the government canbecome an equity holder in a regulated rm. In practice, however,�the creation of rents for the bureaucrats and politicians throughregulation is often inefcient, in part because the regulators are�disorganized,and in part because the policies they pursue toincrease the rents from corruption are distortionary. The analogyto tollbooths on a highway is useful. Efcient regulation may call�for one toll for the use of a road, or even no tolls if the operationof the road is most efciently nanced through general tax reve-� �nues. In a political equilibrium, however,each town throughwhich the road passes might be able to erect its own tollbooth.Toll collectors may also block alternative routes so as to force thetrafc onto the toll road.� For both of these reasons, political tollcollection is inefcient.�

In the tollbooth theory the regulation ofentry enables theregulators to collect bribes from the potential entrants and servesno social purpose. “When someone has nally made the decision�to invest, he then is subjected to some ofthe worst treatmentimaginable. . . In a few cases this treatment consists of outrightextortion: presenting the investor with insurmountable delays orrepeated obstacles unless he makes a large payoff . . .” [WorldBank 1999, p. 10]. More extensive regulation should be associatedwith socially inferior outcomes, particularly corruption.

We assess the regulation of entry around the world from theperspective ofthese theories by addressing two broad sets of

3THE REGULATION OF ENTRY

questions. First,what are the consequences of the regulation ofentry, and in particular,who gets the rents? If the regulation ofentry serves the public interest, it should be associated withhigher quality of goods, fewer damaging externalities, andgreater competition.Public choice theory, in contrast, predictsthat stricter regulation is most clearly associated with less com-petition and higher corruption.

A second question we examine to distinguish the alternativetheories of regulation is which governments regulate entry? Thepublic interest model predicts that governments whose interestsare more closely aligned with those of the consumers, which wethink of as the more representative and more limited govern-ments, should ceteris paribus regulate entry more strictly. Incontrast,the public choice model predicts that the governmentsleast subject to popular oversight should pursue the strictestregulations, to benet themselves and possibly the incumbent�rms. Knowing who regulates thus helps to discriminate among�the theories.

Our analysis of exhaustive data on entry regulation in 85countries leads to the following conclusions. The number of pro-cedures required to start up a rm varies from the low of 2 in�Canada to the high of 21 in the Dominican Republic, with theworld average of around 10. The minimum ofcial time for such a�start-up varies from the low of 2 business days in Australia andCanada to the high of 152 in Madagascar,assuming that thereare no delays by either the applicant or the regulators, with theworld average of 47 business days.The ofcial cost of following�these procedures for a simple rm ranges from under 0.5 percent�of per capita GDP in the United States to over 4.6 times per capitaGDP in the Dominican Republic,with the worldwide average of47 percent of annual per capita income.For an entrepreneur,legal entry is extremely cumbersome,time-consuming,and ex-pensive in most countries in the world.

In a cross section of countries,we do not nd that stricter�regulation of entry is associated with higher quality products,better pollution records or health outcomes, or keener competi-tion. But stricter regulation of entry is associated with sharplyhigher levels of corruption, and a greater relative size of theunofcial� economy.This evidence favors public choice over thepublic interest theories of regulation.

In response,a public interest theorist could perhaps arguethat heavy regulation in some countries is a reection of� both


signicant market failures and the unavailability of alternative�mechanisms of addressing them, such as good courts or freepress. In addition, corruption and a large unofcial economy may�be inadvertent consequences of benevolent regulation, and hencecannot be used as evidence against the public interest view. Suchinadvertent consequences might obtain as a side effect of screen-ing out bad entrants [Banerjee 1997; Acemoglu and Verdier2000],or simply as a result of a well-intended but misguidedtransplant of rich-country regulations into poor countries.Be-cause of this logic, the question of which countries regulate entrymore heavily may be better suited conceptually to distinguish thealternative theories.

We nd that the countries with more open access to political�power, greater constraints on the executive, and greater politicalrights have less burdensome regulation of entry— even control-ling for per capita income—than do the countries with less rep-resentative, less limited, and less free governments. The percapita income control is crucial for this analysis because it couldbe argued that richer countries have both better governments anda lower need for the regulation ofentry, perhaps because theyhave fewer market failures or better alternative ways of dealingwith them. The fact that better governments regulate entry less,along with the straightforward interpretation of the evidence oncorruption and the unofcial� economy,point to the tollbooththeory: entry is regulated because doing so benets� theregulators.

The next section describes the sample.Section III presentsour basic results on the extent of entry regulation around theworld. Section IV asks who gets the rents from regulation. SectionV presents the main results on which governments regulate.Section VI concludes.

II. DATA

A. Construction of the DatabaseThis paper is based on a new data set,which describes the

regulation of entry by start-up companies in 85 countries in 1999.We are interested in all the procedures that an entrepreneurneeds to carry out to begin legally operating a rm involved in�industrial or commercial activity.Specically, we record all pro-�cedures that are ofcially required of an entrepreneur in order to�


obtain all necessary permits and to notify and le with all requi-�site authorities. We also calculate the ofcial� costs and timenecessary for the completion ofeach procedure under normalcircumstances. The study assumes that the information is readilyavailable and that all governmentalbodies function efciently�and without corruption.

We collect data on entry regulation using all available writ-ten information on start-up procedures from government publi-cations, reports of development agencies such as the World Bankand USAID, and government web pages on the Internet. We thencontact the relevant government agencies to check the accuracy ofthe data. Finally, for each country we commission at least oneindependent report on entry regulation from a local law rm, and�work with that rm and government ofcials to eliminate dis-� �agreements among them.

We use ofcial� sources for the number of procedures,time,and cost. If ofcial sources are conicting or the laws are ambigu-� �ous, we follow the most authoritative source.In the absence ofexpress legal denitions, we take the government ofcial’s report� �as the source. If several ofcial sources have different estimates�of time and cost, we take the median. Absent ofcial estimates of�time and cost,we take the estimates of local incorporation law-yers. If several unofcial� (e.g.,a private lawyer) sources havedifferent estimates, we again take the median.

Our countries span a wide range of income levels and politi-cal systems. The sample includes fourteen African countries, nineEast Asian countries including China and Vietnam, three SouthAsian countries (India, Pakistan, and Sri Lanka), all Central andEastern European countries except for Albania and some of theformer Yugoslav republics,eight former Soviet Union republicsand Mongolia, ten Latin American countries, two Caribbeancountries (Dominican Republic and Jamaica), six Middle Easterncountries (Egypt, Israel, Jordan, Lebanon, Morocco, and Tunisia),and all major developed countries.

We record the procedures related to obtaining all the neces-sary permits and licenses,and completing all the required in-scriptions,verications,� and notications for the company to be�legally in operation. When there are multiple ways to beginoperating legally, we choose the fastest in terms of time. In somecountries, entrepreneurs may not bother to follow ofcial proce-�dures or bypass them by paying bribes or hiring the services of“facilitators.” An entrepreneur in Georgia can start up a company


after going through 13 procedures in 69 business days and paying$375 in fees. Alternatively, he may hire a legal advisory rm that�completes the start-up process for $610 in three business days. Inthe analysis, we use the rst set of numbers. We do so because we�are primarily interested in understanding the structure of ofcial�regulation.

Regulations of start-up companies vary across regions withina country, across industries, and across rm sizes. For concrete-�ness, we focus on a “standardized” rm, which has the following�characteristics:it performs general industrial or commercial ac-tivities, it operates in the largest city1 (by population),it is ex-empt from industry-specic requirements (including environmen-�tal ones),it does not participate in foreign trade and does nottrade in goods that are subject to excise taxes (e.g.,liquor, to-bacco, gas), it is a domestically owned limited liability company,2

its capital is subscribed in cash (not in-kind contributions) and isthe higher of (i) 10 times GDP per capita in 1999 or (ii) theminimum capital requirement for the particular type of businessentity, it rents (i.e., does not own) land and business premises, ithas between 5 and 50 employees one month after the commence-ment of operations all of whom are nationals, it has turnover of upto 10 times its start-up capital, and it does not qualify for invest-ment incentives.Although different legal forms are used in dif-ferent countries to set up the simplest rm, to make comparisons�we need to look at the same form.

Our data almost surely underestimate the cost and complex-ity of entry.3 Start-up procedures in the provinces are oftenslower than in the capital. Industry-specic requirements add�procedures.Foreign ownership frequently involves additionalverications and procedures. Contributions in kind often require�assessment ofvalue, a complex procedure that depends on thequality of property registries.Finally, purchasing land can bequite difcult and even impossible in some of the countries of the�sample (for example, in the Kyrgyz Republic).

1. In practice, the largest city coincides with the capital city except in Aus-tralia (Melbourne), Brazil (Sao Paulo), Canada (Toronto), Germany (Frankfurt),Kazakhstan (Almaty), the Netherlands (Amsterdam),South Africa (Johannes-burg), Turkey (Istanbul), and the United States (New York).

2. If the Company Law allows for more than one privately owned businessform with limited liability, we choose the more popular business form amongsmall companies in the country.

3. The World Economic Forum [2001] surveys business people on how impor-tant administrative regulations are as an obstacle to new business. Our threemeasures are strongly positively correlated with these subjective assessments.


B. Denitions of Variables�We use three measures ofentry regulation:the number of

procedures that rms must go through, the ofcial time required� �to complete the process, and its ofcial cost. In the public interest�theory, a more thorough screening process requires more proce-dures and demands more time. In the public choice theory, moreprocedures and longer delays facilitate bribe extraction (tollboothview) or make entry less attractive to potential competitors (cap-ture view).

Theoretical predictions regarding our measure ofcost areambiguous.A benevolent social planner who wants to spendsignicant resources on screening new entrants may choose to�nance such activity with broad taxes rather than with the direct�fees that we measure, leading to low costs as we measure them. Acorrupt regulator may also want to set fees low in order to raisehis own bribe income if, for example, fees are veriable and�cannot be expropriated by the regulator.4 In contrast, higher feesare unambiguously desirable as a tool to deter entry under thecapture theory.Because of these ambiguities,we present statis-tics on cost mainly to describe an important attribute of regula-tion and not to discriminate among theories.

We keep track of all the procedures required by law to starta business. A separate activity in the start-up process is a “pro-cedure” only if it requires the entrepreneur to interact with out-side entities:state and local government ofces,� lawyers, audi-tors, company seal manufacturers, notaries, etc. For example, alllimited liability companies need to hold an inaugural meeting ofshareholders to formally adopt the Company Articles and Bylaws.Since this activity involves only the entrepreneurs,we do notcount it as a procedure. Similarly, most companies hire a lawyerto draft their Articles of Association.However,we do not countthat as a procedure unless the law requires that a lawyer beinvolved. In the same vein, we ignore procedures that the entre-preneur can avoid altogether (e.g., reserving exclusive rights overa proposed company name until registration is completed) or thatcan be performed after business commences.5 Finally, when ob-

4. Shleifer and Vishny [1993] distinguish corruption with theft from corrup-tion without theft. In the latter case, the regulator must remit the ofcial fee to�the Treasury, and therefore has no interest in that fee being high.

5. In several countries, our consultants advised us that certain procedures,while not required, are highly recommended, because failure to follow them mayresult in signicant� delays and additional costs. We collected data on these


taining a document requires several separate procedures involv-ing different ofcials, we count each as a procedure. For example,�a Bulgarian entrepreneur receives her registration certicate�from the Company Registry in Soa,� and then has to pay theassociated fee at an ofcially designated bank. Even though both�activities are related to “obtaining the registration certicate,”�they count as two separate procedures in the data.

To measure time, we collect information on the sequence inwhich procedures are to be completed and rely on ofcial gures� �as to how many business days it takes to complete each proce-dure. We ignore the time spent to gather information,and as-sume that all procedures are known from the very beginning. Wealso assume that procedures are taken simultaneously wheneverpossible, for maximum efciency. Since entrepreneurs may have�trouble visiting several different institutions within the same day(especially if they come from out-of-town),we set the minimumtime required to visit an institution to be one day. 6 Anotherjustication for this approach is that the relevant ofces some-� �times open for business only briey:� both the Ministry of Econ-omy and the Ministry of Justice in Cairo open for business onlybetween 11 a.m. and 2 p.m.

We estimate the cost of entry regulation based on all identi-able ofcial expenses: fees, costs of procedures and forms, pho-� �tocopies,scal� stamps, legal and notary charges, etc. All costgures are ofcial and do not include bribes, which De Soto [1990]� �has shown to be signicant for registration. Setup fees often vary�with the level of start-up capital. As indicated, we report the costsassociated with starting to operate legally a rm with capital�equivalent to the larger of (i) ten times per capita GDP in 1999 or(ii) the minimum capital requirement stipulated in the law.Wehave experimented with other capital levels and found our resultsto be robust.

Theoretical predictions for the cost ofentry regulation areambiguous.As an alternative measure, we consider only the

procedures, but did not include them in the variables presented here because wewanted to stick to the mandatory criterion.We have rerun the regressions dis-cussed below including these highly recommended procedures. The inclusion doesnot have a material impact on the results.

6. In the calculation of time, when two procedures can be completed on thesame day in the same building, we count that as one day rather than two(following the urgings of ofcials in several countries, where several ofces are� �located in the same building). Our results are not affected by this particular wayof computing time.


component of the cost that goes to the government, which in thesample averages about half the total cost. The results for this costvariable are generally weaker than for the total out-of-pocketcost, but go in the same direction.Our basic cost estimates alsoignore the opportunity cost ofthe entrepreneur’s time and theforgone prots associated with bureaucratic delay.� To addressthis concern, we calculate a “full cost” measure, which adds up theofcial� expenses and an estimate ofthe value of the entrepre-neur’s time,valuing his time at the country’s per capita incomeper working day.We report this number below,and have repli-cated the analysis using it as a measure of cost. The resultsobtained using this cost measure are very similar to those usingthe raw data on time and cost, and hence are not presented.

Table I lists typical procedures associated with setting up arm in our sample. The procedures are further divided by their�function: screening (a residual category, which generally aims tokeep out “unattractive” projects or entrepreneurs),health andsafety, labor, taxes, and environment. The basic procedure instarting up a business,present everywhere,is registering withthe Companies’ Registry. This can take more than one procedure;sometimes there is a “preliminary license” and a “nal” license.�Combined with that procedure, or as a separate procedure, is thecheck for uniqueness of the proposed company name.Add-onprocedures comprise the requirements to notarize the CompanyDeeds, to open a bank account and deposit of start-up capital, andto publish a notication of� the company’s establishment in anofcial� or business paper.Additional screening procedures thatinclude obtaining different certicates and ling with agencies� �other than the Registry may add up to 97 days in delays, as is thecase in Madagascar.Another set of basic screening procedures,present in almost every country in the data set,covers certainmandatory municipalprocedures,registrations with statisticalofces and with Chambers of Commerce and Industry (or respec-�tive Ministries). In the Dominican Republic these procedures takeseven procedures and fourteen days. There is large cross-countryvariation in terms of the number, time, and cost of screeningprocedures as the Company Registry performs many of thesetasks automatically in the most efcient countries but the entre-�preneur does much of the legwork in the less efcient ones.�

Additional procedures appear in four areas. The rst covers�tax-related procedures, which require seven procedures andtwenty days in Madagascar. The second is labor regulations,


TABLE IL IST OF PROCEDURES FOR STARTING UP A COMPANY

This table provides a list of common proceduresrequired to start up acompany in the 85 countries of the sample.

1. Screening procedures- Certify business competence- Certify a clean criminal record- Certify marital status- Check the name for uniqueness- Notarize company deeds- Notarize registration certicate�- File with the Statistical Bureau- File with the Ministry of Industry and Trade, Ministry of the Economy, or the

respective ministries by line of business- Notify municipality of start-up date- Obtain certicate of compliance with the company law�- Obtain business license (operations permit)- Obtain permit to play music to the public (irrespective of line of business)- Open a bank account and deposit start-up capital- Perform an ofcial audit at start-up�- Publish notice of company foundation- Register at the Companies Registry- Sign up for membership in the Chamber of Commerce or Industry or the Regional

Trade Association2. Tax-related requirements

- Arrange automatic withdrawal of the employees’income tax from the company payrollfunds

- Designate a bondsman for tax purposes- File with the Ministry of Finance- Issue notice of start of activity to the Tax Authorities- Register for corporate income tax- Register for VAT- Register for state taxes- Register the company bylaws with the Tax Authorities- Seal, validate, rubricate accounting books

3. Labor/social security-related requirements- File with the Ministry of Labor- Issue employment declarations for all employees- Notarize the labor contract- Pass inspections by social security ofcials�- Register for accident and labor risk insurance- Register for health and medical insurance- Register with pension funds- Register for Social Security- Register for unemployment insurance- Register with the housing fund

4. Safety and health requirements- Notify the health and safety authorities and obtain authorization to operate from the

Health Ministry- Pass inspections and obtain certicates related to work safety, building,� re,�

sanitation, and hygiene5. Environment-related requirements

- Issue environmental declaration- Obtain environment certicate�- Obtain sewer approval- Obtain zoning approval- Pass inspections from environmental ofcials�- Register with the water management and water discharge authorities


which require seven procedures and 21 days in Bolivia. The thirdarea is health and safety regulations, which demand ve proce-�dures and 21 business days in Malawi. The nal� area coverscompliance with environmental regulations, which take two pro-cedures and ten days in Malawi if all goes well.

Figures I and II describe the number, time,and cost of theprocedures needed to begin operating legally in New Zealand andFrance, respectively. New Zealand’s streamlined start-up processtakes only three procedures and three days.The entrepreneurmust rst obtain approval for the company name from the web-�site of the Registrar of Companies,and then apply online forregistration with both the Registrar of Companies and the taxauthorities.

In contrast, the process in France takes 15 procedures and 53days.To begin, the founder needs to check the chosen companyname for uniqueness at the Institut National de la Proprie teIndustrielle (INPI). He then needs the mayor’s permit to use hishome as an ofce. (If the ofce is to be rented, the founder must� �secure a notarized lease agreement.) The following documentsmust then be obtained, each from a different authority: proof of a

F IGURE IStart-up Procedures in New Zealand

Procedures are lined up sequentially on the horizontal axis and described in thetext box. The time required to complete each procedure is described by the heightof the bar and measured against the left scale. Cumulative costs (as a percentageof per capita (GDP) are plotted using a line and measured against the right scale.


clean criminal record,an original extract of the entrepreneur’scerticate of marital status from the City Hall,� and a power ofattorney. The start-up capital is then deposited with a notarybank or Caisse des Depots, and is blocked there until proof ofregistration is provided.Notarization of the Articles of Associa-tion follows. A notice stating the location of the headquartersofce is published in a journal approved for legal announcements,�and evidence ofthe publication is obtained.Next, the founderregisters four copies of the articles of association at the local taxcollection ofce. He then les a request for registration with the� �Centre de Formalites des Entreprises (CFE) which handles dec-larations of existence and other registration-related formalities.The CFE must process the documents or return them in case therequest is incomplete.The CFE automatically enters the com-pany information in the Registre Nationale des Entreprises(RNE) and obtains from the RNE identication numbers: numero�SIRENE (Syste`me Informatique pour le Repertoire des Entre-prises), numero SIRET (Syste` me Informatique pour le Re´pertoiredes Etablissements), and numero NAF (Nomenclature des Activi-tees Francaises).The SIRET is used by, among others,the tax

F IGURE IIStart-up Procedures in France

Procedures are lined up sequentially on the horizontal axis and described in thetext box. The time required to complete each procedure is described by the heightof the bar and measured against the left scale. Cumulative costs (as a percentageof per capita GDP) are plotted using a line and measured against the right scale.


authorities. The RNE also publishes a notice of the companyformation in the ofcial� bulletin of civil and commercial an-nouncements. The rm then obtains a proof of registration form�“K-bis,” which is effectively its identity card. To start legal opera-tions, the entrepreneur completes ve additional procedures: in-�form the post ofce of the new enterprise, designate a bondsman�or guarantee payment of taxes with a cash deposit,unblock thecompany’s capital by ling with the bank a proof of registration�(K-bis), have the rm’s ledgers and registers initialed, and le for� �social security. The magazine L’Entreprise comments: “To be surethat the le for the Company Registry is complete, many promot-�ers check it with a counselor’s service, which costs FF200 in Paris(about US$30). But there’s always something missing, and mostentrepreneurs end up using a lawyer to complete the procedure.”

III. BASIC RESULTS

Table II describes all the variables used in this study. TableIII presents the basic information from our sample. Countries areranked in ascending order rst� by the total number of entryprocedures,then by the time it takes to complete them, andnally by the cost of entry. We classify each procedure as one of�ve types:� safety and health, environmental,tax, labor, and aresidual category which we label “screening,” whose purpose un-der the public interest theory is to weed out the undesirableentrepreneurs. We then compute and report the total number ofprocedures and their breakdown into our ve categories for each�country.We also report the minimum number of business daysthat are ofcially required to comply with entry regulations, the�costs arising from the ofcial� fees, and the total costs whichimpute the entrepreneur’s time (as a fraction of GDP per capita).Finally, we take averages by income level and report t-testscomparing the regulation of entry across income groups.

The data show enormous variation in entry regulation acrosscountries. The total number of procedures ranges from 2 in Can-ada to 21 in the Dominican Republic and averages 10.48 for thewhole sample.Very few entry regulations cover tax and laborissues. The worldwide average number oflabor and tax proce-dures are 1.94 and 2.02, respectively. Procedures involving envi-ronmental issues and safety and health matters are even rarer(0.14 and 0.34 procedures on average, respectively).Instead,much of what governments do to regulate entry falls into the


category of screening procedures. The worldwide average numberof such procedures facing a new entrant is 6.04.

The number of procedures is highly correlated with both thetime and cost variables (see Table VI). The correlation of the (log)number of procedures with (log) time is 0.83 and with (log) cost is0.64.Translated into economic terms, this means that entrepre-neurs pay a steep price in terms of fees and delays in countriesthat make intense use of ex ante screening.For example,com-pleting 19 procedures demands 149 business days and 111.5percent of GDP per capita in Mozambique. In Italy the completionof 16 procedures takes up 62 business days and 20 percent of GDPper capita. The Dominican Republic is in a class of its own:completing its 21 procedures requires 80 business days and feesof at least 4.63 times per capita GDP.These gures are admit-�tedly extreme within the sample,yet meeting the ofcial� entryrequirements in the average sample country requires roughly 47days and fees of 47 percent of GDP per capita.

When we aggregate time and out-of-pocket costs into anaggregate cost measure,the results for some countries becomeeven more extreme. The world average full cost measure rises to66 percent of per capita GDP, but varies from 1.7 percent of percapita GDP for New Zealand to 4.95 times per capita GDP in theDominican Republic.

Panel B of Table III reports averages of the total number ofprocedures and its components, time and cost by quartiles of percapita GDP in 1999. Two patterns emerge. First, the cost-to-per-capita-GDP ratio decreases uniformly with GDP per capita. Theaverage cost-to-per-capita-GDP ratio for countries in the topquartile of per capita GDP (“rich countries”)is 10 percent andrises to 108 percent in countries in the bottom quartile of percapita GDP. This pattern merely reects the fact that the income�elasticity of fees (in log levels) is about 0.2. Second, countries inthe top quartile of per capita GDP require fewer procedures andtheir entrepreneurs face shorter delays in starting a legal busi-ness than those in the remaining countries.7 The total number ofprocedures in an average rich country is 6.8 which is signicantly�lower than the rest-of-sample average of11.8 (t-statistics are

7. One objection to this nding is that entrepreneurs in rich countries might�face more postentry regulations than they do in poor countries. We have data onone aspect of postentry regulation, namely the regulation of labor markets (seeDjankov et al. [2001a]). The numbers of entry and of labor market regulations arepositively correlated across countries, contrary to this objection.


TABL

E II

THE

VARI

ABLE

STh

is ta

ble

desc

ribes

the

varia

bles

col

lect

ed fo

r the

85

coun

tries

inclu

ded

in o

ur st

udy.T

he rs

t col

umn

give

s the

nam

e of

�th

eva

riabl

e. T

he se

cond

col

umn

desc

ribes

the

varia

ble

and

prov

ides

the

sour

ces f

rom

whi

ch it

was

col

lect

ed.

Varia

ble

Desc

riptio

n

Num

ber o

fpr

oced

ures

The

num

ber o

f diff

eren

t pro

cedu

res t

hat a

sta

rt-up

has

to c

ompl

y wi

th in

ord

er to

obt

ain

a le

gal s

tatu

s, i.

e., t

o st

art o

pera

ting

as a

lega

l ent

ity. S

ourc

e:Au

thor

s’own

cal

cula

tions

.Sa

fety

& H

ealth

The

num

ber o

f diff

eren

t saf

ety

and

heal

th p

roce

dure

s tha

t a s

tart-

up h

as to

com

ply

with

to st

art o

pera

ting

as a

lega

l ent

ity. S

ourc

e: A

utho

rs’

own

calcu

latio

ns.

Envi

ronm

ent

The

num

ber o

f diff

eren

t env

ironm

enta

l pro

cedu

res t

hat a

sta

rt-up

has

to c

ompl

y wi

th to

star

t ope

ratin

g as

a le

gal e

ntity

. Sou

rce:

Auth

ors’o

wnca

lcula

tions

.Ta

xes

The

num

ber o

f diff

eren

t tax

pro

cedu

res

that

a st

art-u

p ha

s to

com

ply

with

to s

tart

oper

atin

g as

a le

gal e

ntity

.So

urce

: Aut

hors

’own

calcu

latio

ns.

Labo

rTh

e nu

mbe

r of d

iffer

ent l

abor

pro

cedu

res

that

a st

art-u

p ha

s to

com

ply

with

to st

art o

pera

ting

as a

lega

l ent

ity. S

ourc

e: A

utho

rs’

own

calcu

latio

ns.

Scre

enin

gTh

e nu

mbe

r of d

iffer

ent s

teps

that

a st

art-u

p ha

s to

com

ply

with

in o

rder

to o

btai

n a

regi

stra

tion

certi

cate

that

are

not

ass

ocia

ted

with

saf

ety

and

�he

alth

issu

es, t

he e

nviro

nmen

t, ta

xes,

or l

abor

. Sou

rce:

Aut

hors

’ow

n ca

lcula

tions

.Ti

me

The

time

it ta

kes t

o ob

tain

lega

l sta

tus

to o

pera

te a

rm,

�in

busin

ess d

ays.

A we

ek h

as v

e bu

sines

s day

s and

a m

onth

has

twen

ty-tw

o. S

ourc

e:�

Auth

ors’

own

calcu

latio

ns.

Cost

The

cost

of o

btai

ning

lega

l sta

tus t

o op

erat

e a

rm a

s a

shar

e of

per

cap

ita G

DP in

199

9. It

inclu

des

all i

dent

iabl

e of

cial e

xpen

ses (

fees

, cos

ts o

f�

��

proc

edur

es a

nd fo

rms,

phot

ocop

ies,

scal

stam

ps, l

egal

and

not

ary

char

ges,

�et

c.).

The

com

pany

is a

ssum

ed to

hav

e a

star

t-up

capi

tal o

f ten

tim

es p

erca

pita

GDP

in 1

999.

Sou

rce:A

utho

rs’o

wn ca

lcula

tions

.Co

st 1

time

The

cost

of o

btai

ning

lega

l sta

tus t

o op

erat

e a

rm a

s a

shar

e of

per

cap

ita G

DP in

199

9. It

inclu

des a

ll id

entia

ble

ofcia

l exp

ense

s (fe

es, c

osts

of

��

�pr

oced

ures

and

form

s, ph

otoc

opie

s, sc

al st

amps

, leg

al a

nd n

otar

y ch

arge

s,�

etc.

) as w

ell a

s the

mon

etize

d va

lue

of th

e en

trepr

eneu

r’s ti

me.

The

tim

e of

the

entre

pren

eur i

s val

ued

as th

e pr

oduc

t of t

ime

and

per c

apita

GDP

in 1

999

expr

esse

d in

per

bus

ines

s day

term

s. Th

e co

mpa

ny is

ass

umed

to h

ave

ast

art-u

p ca

pita

l of t

en ti

mes

the

GDP

per c

apita

leve

l in

1999

. Sou

rce:

Aut

hors

’ow

n ca

lcula

tions

.GD

P/PO

P 199

9Gr

oss

dom

estic

pro

duct

per

cap

ita in

cur

rent

U. S

. dol

lars

in 1

999.

Sou

rce:

Wor

ld B

ank

[200

1].

Qual

ity st

anda

rds

Num

ber o

f ISO

900

0 ce

rtica

tions

per

thou

sand

inha

bita

nts i

ssue

d by

the

Inte

rnat

iona

l Org

aniza

tion

for S

tand

ardi

zatio

n as

of 1

999

to e

ach

coun

try in

�th

e sa

mpl

e. “I

SO s

tand

ards

repr

esen

t an

inte

rnat

iona

l con

sens

us o

n th

e st

ate

of th

e ar

t in

the

tech

nolo

gy c

once

rned

. . .

. ISO

900

0 is

prim

arily

conc

erne

d wi

th q

ualit

y m

anag

emen

t. . .

.IS

O de

velo

ps v

olun

tary

tech

nica

l sta

ndar

ds th

at c

ontri

bute

to m

akin

g th

e de

velo

pmen

t,m

anuf

actu

ring

and

supp

ly o

f pro

duct

s an

d se

rvice

s mor

e ef

cient

,�

safe

r and

cle

aner

. . .

. ISO

stan

dard

s also

serv

e to

saf

egua

rd c

onsu

mer

s. .

. . W

hen

an o

rgan

izatio

n ha

s am

anag

emen

t sys

tem

cer

tied

to a

n IS

O 90

00. .

. , t

his m

eans

that

the

p�

roce

ss in

u�enc

ing

qual

ity (I

SO 9

000)

. . .

con

form

s to

the

rele

vant

stan

dard

’sre

quire

men

ts.”

Sour

ce: I

nter

natio

nal O

rgan

izatio

n fo

r Sta

ndar

diza

tion

(www

.iso.

ch).

Wat

er p

ollu

tion

Emiss

ions

of o

rgan

ic wa

ter p

ollu

tant

s (k

ilogr

ams p

er d

ay p

er w

orke

r) fo

r 199

8. M

easu

red

in te

rms

of b

ioch

emica

l oxy

gen

dem

and,

which

refe

rs to

the

amou

nt o

f oxy

gen

that

bac

teria

in w

ater

will

cons

ume

in b

reak

ing

down

was

te. E

miss

ions

per

wor

ker a

re to

tal e

miss

ions

div

ided

by

the

num

ber o

fin

dust

rial w

orke

rs. S

ourc

e:Wor

ld B

ank

[200

1].


Deat

hs fr

omac

ciden

tal

poiso

ning

Log

of th

e nu

mbe

r of d

eath

s cau

sed

by a

ccid

enta

l poi

soni

ngs (

inclu

ding

by

drug

s, m

edica

tions

, bio

-pro

duct

s, so

lid a

nd li

quid

sub

stan

ces,

gase

s and

vapo

rs) p

er m

illio

n in

habi

tant

s. Av

erag

e of

the

year

s 198

1 th

roug

h 19

94 (t

he m

ost r

ecen

t ava

ilabl

e gu

re).

�Sour

ce: T

he n

umbe

r of a

ccid

enta

l dea

ths

from

poi

soni

ng is

take

n fro

m W

orld

Hea

lth O

rgan

izatio

n [1

998]

.Po

pula

tion

gure

s ar

e ta

ken

from

Wor

ld B

ank

[200

1].

�De

aths

from

inte

stin

alin

fect

ions

Log

of th

e nu

mbe

r of d

eath

s cau

sed

by in

test

inal

infe

ctio

ns (i

nclu

ding

dig

estiv

e di

sord

ers)

per

milli

on in

habi

tant

s. Av

erag

e of

the

year

s 198

1 th

roug

h19

94 (t

he m

ost r

ecen

t ava

ilabl

e gu

re).

�So

urce

: The

num

ber o

f dea

ths f

rom

inte

stin

al in

fect

ions

is ta

ken

from

Wor

ld H

ealth

Org

aniza

tion

[199

8].

Popu

latio

n gu

res a

re ta

ken

from

Wor

ld B

ank

[200

1].

�Si

ze o

f the

uno

fcia

l�

econ

omy

Size

of t

he s

hado

w ec

onom

y as

a p

erce

ntag

e of

GDP

(var

ying

tim

e pe

riods

). So

urce

:Au

thor

s’own

com

puta

tions

bas

ed o

n av

erag

ing

over

all

estim

ates

repo

rted

in S

chne

ider

and

Ens

te [2

000]

for a

ny g

iven

cou

ntry

as

well

as S

anan

ikon

e [1

996]

for B

urki

na F

aso,

Chi

dzer

o [1

996]

for S

eneg

al, T

urnh

aman

d Sc

hwar

tz [1

990]

for I

ndon

esia

and

Pak

istan

, and

Kas

nako

glu

and

Yayl

a [2

000]

for T

urke

y.Em

ploy

men

t in

the

unof

cial e

cono

my

�Sh

are

of th

e la

bor f

orce

em

ploy

ed in

the

unof

cial e

cono

my

in th

e ca

pita

l city

of e

ach

coun

try a

s a p

erce

nt o

f the

ofc

ial l

abor

.�

�Fi

gure

s ar

e ba

sed

onsu

rvey

s and

, for

som

e co

untri

es,on e

cono

met

ric e

stim

ates

. Sou

rce:

Sch

neid

er [2

000]

and

the

Glob

al U

rban

Indi

cato

rs D

atab

ase

[200

0](w

ww.u

rban

obse

rvat

ory.

org/

indi

cato

rs/d

atab

ase)

.Pr

oduc

t mar

ket

com

petit

ion

Surv

ey m

easu

re o

f the

ext

ent t

o wh

ich re

spon

dent

s agr

ee w

ith th

e fo

llowi

ng s

tate

men

t:“C

ompe

titio

n in

the

loca

l mar

ket i

s int

ense

and

mar

ket s

hare

suc

tuat

e co

nsta

ntly

.” Sc

ale

from

1 (s

trong

ly d

isagr

ee) t

hrou

gh 7

(stro

ngly

agr

ee).

Sour

ce: I

MD

[200

1].

�Co

rrupt

ion

Corru

ptio

n pe

rcep

tion

inde

x fo

r 199

9.Corru

ptio

n is

dene

d br

oadl

y as

“th

e m

isuse

of p

ublic

pow

er fo

r priv

ate

bene

ts,

��e.g.

,brib

ing

of p

ublic

ofc

ials,

�ki

ckba

cks i

n pu

blic

proc

urem

ent,

or e

mbe

zzle

men

t of p

ublic

fund

s.” T

he in

dex

aver

ages

the

corru

ptio

n sc

ores

giv

en b

y th

e fo

llowi

ng so

urce

s: (1

)Fr

eedo

m H

ouse

Nat

ions

in T

rans

it (F

H); (

2) G

allu

p In

tern

atio

nal (

GI);

(3) t

he E

cono

mist

Inte

lligen

ce U

nit (

EIU)

; (4)

the

Inst

itute

for M

anag

emen

tDe

velo

pmen

t,Lau

sann

e (IM

D);(5

) the

Inte

rnat

iona

l Crim

e Vi

ctim

Sur

vey

(ICVS

);(6) t

he P

oliti

cal a

nd E

cono

mic

Risk

Con

sulta

ncy,Ho

ng K

ong

(PER

C); (

7) T

he W

all S

treet

Jour

nal,C

entra

l Eur

opea

n Ec

onom

ic Re

view

(CEE

R); (

8) th

e W

orld

Ban

k an

d Un

iver

sity

of B

asel

(WB/

UB),

(9) t

he W

orld

Econ

omic

Foru

m (W

EF).

Desc

endi

ng sc

ore

from

1 (m

ost c

orru

pt) t

o 10

(lea

st c

orru

pt).

Sour

ce: T

rans

pare

ncy

Inte

rnat

iona

l (ww

w.tra

nspa

renc

y.de

/).Ex

ecut

ive

de fa

cto

inde

pend

ence

Inde

x of

“ope

ratio

n (d

e fa

cto)

inde

pend

ence

of c

hief

exe

cutiv

e.” D

esce

ndin

g sc

ale

from

1 to

7 (1

5pu

re in

divi

dual

; 2 5

inte

rmed

iate

cat

egor

y; 3

5slig

ht to

mod

erat

e lim

itatio

ns; 4

5inte

rmed

iate

cat

egor

y; 5

5subs

tant

ial l

imita

tions

; 6 5i

nter

med

iate

cat

egor

y; 7

5exe

cutiv

e pa

rity

or su

bord

inat

ion)

.Av

erag

e of

the

year

s 19

45 th

roug

h 19

98. S

ourc

e: Ja

gger

s and

Mar

shal

l [20

00].

Cons

train

ts o

nex

ecut

ive

powe

rIn

dex

of c

onst

rain

ts o

n th

e ex

ecut

ive

powe

r bas

ed o

n th

e nu

mbe

r of e

ffect

ive

veto

poi

nts i

n a

coun

try. V

eto

poin

ts in

clude

(1) a

n eff

ectiv

e le

gisla

ture

(repr

esen

ts tw

o ve

to p

oint

s in

the

case

of b

icam

eral

syst

ems)

; (2)

an

inde

pend

ent j

udici

ary;

and

(3) a

stro

ng fe

dera

l sys

tem

. Ave

rage

of t

he y

ears

194

5th

roug

h 19

98. S

ourc

e: H

enisz

[200

1].

Effec

tiven

ess

ofle

gisla

ture

Inde

x of

the

effec

tiven

ess o

f the

legi

slatu

re.

Asce

ndin

g sc

ale

from

1 to

4 (1

5no le

gisla

ture

;2 5

larg

ely

ineff

ectiv

e; 3

5par

tly e

ffect

ive;

4 5e

ffect

ive)

.Av

erag

e of

the

year

s 19

45 th

roug

h 19

98. S

ourc

e: T

he C

ross

-Nat

iona

l Tim

e-Se

ries

Data

Arc

hive

(www

.dat

aban

ks.si

teho

stin

g.ne

t/www

/mai

n.ht

m).

Com

petit

ion

in th

ele

gisla

ture

’sno

min

atin

g pr

oces

sInde

x of

the

com

petit

iven

ess

of th

e no

min

atin

g pr

oces

s for

sea

ts in

the

legi

slatu

re. A

scen

ding

scal

e fro

m 1

to 4

(1 5

no le

gisla

ture

; 2 5

nonc

ompe

titiv

e;3

5pa

rtly

com

petit

ive;

4 5c

ompe

titiv

e). A

vera

ge o

f the

yea

rs 1

945

thro

ugh

1998

.So

urce

: The

Cro

ss-N

atio

nal T

ime-

Serie

s Dat

a Ar

chiv

e(w

ww.d

atab

anks

.site

host

ing.

net/w

ww/m

ain.

htm

).Au

tocr

acy

Indi

cate

s the

“gen

eral

clo

sedn

ess o

f pol

itica

l ins

titut

ions

.” S

cale

from

0 to

10

with

0 b

eing

low

in a

utoc

racy

and

10

bein

g hi

gh in

aut

ocra

cy. A

vera

ge o

f the

year

s 19

45 th

roug

h 19

98.So

urce

: Jag

gers

and

Mar

shal

l [20

00].

Polit

ical r

ight

sIn

dex

of p

oliti

cal r

ight

s. H

ighe

r rat

ings

indi

cate

cou

ntrie

s tha

t com

e clo

ser t

o th

e id

eals

sugg

este

d by

the

chec

klist

que

stio

ns o

f (1)

free

and

fair

elec

tions

; (2)

thos

e el

ecte

d ru

le; (

3) th

ere

are

com

petit

ive

parti

es o

r oth

er c

ompe

titiv

e po

litica

l gro

upin

gs; (

4) th

e op

posit

ion

has a

n im

porta

nt ro

le a

nd p

ower

; and

(5) t

heen

titie

s hav

e se

lf-de

term

inat

ion

or a

n ex

trem

ely

high

deg

ree

of a

uton

omy.

Ave

rage

of t

he y

ears

197

2 th

roug

h 19

98. S

ourc

e: F

reed

om H

ouse

[200

1].

Lega

l orig

inId

entie

s the

lega

l orig

in o

f eac

h Co

mpa

ny L

aw o

r Com

mer

cial C

ode

of e

ach

coun

try.

�Th

ere

are

ve p

ossib

le o

rigin

s:�

(1) E

nglis

h Co

mm

on L

aw;(2

)Fr

ench

Com

mer

cial C

ode;(3

) Ger

man

Com

mer

cial C

ode;(4

) Sca

ndin

avia

n Co

mm

ercia

l Cod

e; a

nd (5

) Soc

ialis

t/Com

mun

ist la

ws. S

ourc

e:La

Por

ta e

t al.

[199

8];R

eyno

lds a

nd F

lore

s [19

89];

CIA

Wor

ld F

actb

ook

[200

1].


TABL

E III

THE

DATA

Pane

lA re

ports

the

tota

lnu

mbe

r ofp

roce

dure

s and

thei

r bre

akup

in th

e fo

llowi

ng v

e ca

tego

ries:

�(1

)saf

ety

and

heal

th;(

2)en

viro

nmen

t;(3)

taxe

s;(4)

labo

r;an

d (5

) scr

eeni

ng.Th

e ta

ble

also

repo

rts th

e tim

e,dire

ct c

ost (

as a

frac

tion

of G

DP p

er c

apita

in19

99) a

ssoc

iate

d wi

th m

eetin

g go

vern

men

t req

uire

men

ts,

and

dire

ct c

ost p

lus t

he m

onet

ized

valu

e of

the

entre

pren

eur’s

tim

e (a

s afra

ctio

n of

GDP

per

cap

ita in

199

9) a

s wel

l as t

he le

vel o

f GDP

per

cap

ita in

dol

lars

in 1

999.

Cou

ntrie

s are

sorte

d in

asc

endi

ng o

rder

on th

e ba

sis o

f(1)

the

tota

lnum

ber o

fpro

cedu

res;(

2) ti

me;

and

(3)c

ost.

Pane

lB p

rese

nts m

eans

of t

he v

aria

bles

by

quar

tiles

of

GDP

per c

apita

in 1

999.

Pane

lC p

rese

nts t

-sta

tistic

s for

diff

eren

ces i

n m

eans

acr

oss q

uarti

les o

f per

cap

ita G

DP in

199

9.Tabl

e II

desc

ribes

the

varia

bles

in d

etai

l.

Num

ber o

fpr

oced

ures

Safe

ty &

Heal

thEn

viro

nmen

tTa

xes

Labo

rSc

reen

ing

Tim

eCo

stCo

st 1

time

GDP/

POP 1

999

Pane

l A:D

ata

Cana

da2

00

10

12

0.01

450.

0225

19,3

20Au

stra

lia2

00

10

12

0.02

250.

0305

20,0

50Ne

w Ze

alan

d3

00

10

23

0.00

530.

0173

13,7

80De

nmar

k3

00

10

23

0.10

000.

1120

32,0

30Ire

land

30

01

02

160.

1157

0.17

9719

,160

Unite

d St

ates

40

01

12

40.

0049

0.01

6930

,600

Norw

ay4

00

11

218

0.04

720.

1192

32,8

80Un

ited

King

dom

50

01

13

40.

0143

0.03

0322

,640

Hong

Kon

g5

00

01

415

0.03

330.

0933

23,5

20M

ongo

lia5

00

10

422

0.03

310.

1211

350

Finl

and

50

01

31

240.

0116

0.10

7623

,780

Israe

l5

00

21

232

0.21

320.

3412

15,8

60Zi

mba

bwe

50

02

12

470.

1289

0.31

6952

0Sw

eden

60

01

14

130.

0256

0.07

7625

,040

Jam

aica

60

02

13

240.

1879

0.28

392,

330

Zam

bia

60

02

13

290.

6049

0.72

0932

0Pa

nam

a7

00

11

515

0.30

740.

3674

3,07

0Sw

itzer

land

70

02

14

160.

1724

0.23

6438

,350

Sing

apor

e7

00

12

422

0.11

910.

2071

29,6

10La

tvia

70

02

14

230.

4234

0.51

542,

470

Mal

aysia

70

01

15

420.

2645

0.43

253,

400


Sri L

anka

80

01

16

230.

1972

0.28

9282

0Ne

ther

land

s8

01

20

531

0.18

410.

3081

24,3

20Be

lgiu

m8

00

12

533

0.09

980.

2318

24,5

10Ta

iwan

, Chi

na8

00

12

537

0.06

600.

2140

13,2

48Hu

ngar

y8

00

11

639

0.85

871.

0147

4,65

0Pa

kist

an8

00

21

550

0.34

960.

5496

470

Peru

80

02

24

830.

1986

0.53

062,

390

Sout

h Af

rica

90

02

25

260.

0844

0.18

843,

160

Kyrg

yz R

epub

lic9

00

11

732

0.25

320.

3812

300

Thai

land

90

03

24

350.

0639

0.20

391,

960

Nige

ria9

01

21

536

2.57

002.

7140

310

Aust

ria9

00

21

637

0.27

280.

4208

25,9

70Tu

nisia

90

00

27

410.

1722

0.33

622,

100

Slov

enia

90

00

18

470.

2103

0.39

839,

890

Leba

non

90

01

17

631.

5672

1.81

923,

700

Urug

uay

100

01

45

230.

4949

0.58

695,

900

Bulg

aria

100

02

08

270.

1441

0.25

211,

380

Chile

100

03

25

280.

1308

0.24

284,

740

Germ

any

100

01

27

420.

1569

0.32

4925

,350

Ghan

a10

01

14

445

0.21

750.

3975

390

Lithu

ania

102

02

15

460.

0546

0.23

862,

620

Czec

h Re

publ

ic10

00

12

765

0.08

220.

3422

5,06

0In

dia

100

03

34

770.

5776

0.88

5645

0Ja

pan

110

02

27

260.

1161

0.22

0132

,230

Ugan

da11

20

21

629

0.30

400.

4200

320

Egyp

t, Ar

ab R

ep.

110

02

18

510.

9659

1.16

991,

400

Keny

a11

00

23

654

0.50

700.

7230

360

Arm

enia

110

01

19

550.

1267

0.34

6749

0Po

land

112

03

15

580.

2546

0.48

663,

960

Spai

n11

00

42

582

0.17

300.

5010

14,0

00In

done

sia11

00

21

812

80.

5379

1.04

9958

0Cr

oatia

121

02

36

380.

4503

0.60

234,

580

Kaza

khst

an12

00

13

842

0.47

470.

6427

1,23

0Po

rtuga

l12

00

22

876

0.18

440.

4884

10,6

00Sl

ovak

Rep

ublic

120

02

37

890.

1452

0.50

123,

590

Chin

a12

00

52

592

0.14

170.

5097

780


TABL

E III

(CON

TINU

ED)

Num

ber o

fpr

oced

ures

Safe

ty &

Heal

thEn

viro

nmen

tTa

xes

Labo

rSc

reen

ing

Tim

eCo

stCo

st 1

time

GDP/

POP 1

999

Kore

a,Re

p.13

00

24

727

0.16

270.

2707

8,49

0Ta

nzan

ia13

10

52

529

3.35

203.

4680

240

Ukra

ine

130

02

38

300.

2569

0.37

6975

0Tu

rkey

130

02

29

440.

1932

0.36

922,

900

Mal

awi

135

21

14

520.

1886

0.39

6619

0M

oroc

co13

10

33

657

0.21

260.

4406

1,20

0Ge

orgi

a13

20

11

969

0.60

480.

8808

620

Burk

ina

Faso

140

03

29

333.

1883

3.32

0324

0Ph

ilippi

nes

140

05

18

460.

1897

0.37

371,

020

Arge

ntin

a14

00

45

548

0.10

190.

2939

7,60

0Jo

rdan

141

02

110

640.

5369

0.79

291,

500

Vene

zuel

a14

11

33

610

40.

1060

0.52

203,

670

Gree

ce15

00

42

936

0.58

600.

7300

11,7

70Fr

ance

150

03

111

530.

1430

0.35

5023

,480

Braz

il15

00

75

363

0.20

140.

4534

4,42

0M

exico

151

22

37

670.

5664

0.83

444,

400

Mal

i16

10

32

1059

240

Italy

160

05

38

620.

2002

0.44

8219

,710

Sene

gal

160

03

211

691.

2331

1.50

9151

0Ec

uado

r16

20

24

872

0.62

230.

9103

1,31

0Ro

man

ia16

12

13

997

0.15

310.

5411

1,52

0Vi

etna

m16

01

15

911

21.

3377

1.78

5737

0M

adag

asca

r17

00

73

715

20.

4263

1.03

4325

0Co

lom

bia

182

04

57

480.

1480

0.34

002,

250

Moz

ambi

que

194

01

311

149

1.11

461.

7106

230

Russ

ian

Fede

ratio

n20

00

25

1357

0.19

790.

4259

2,27

0Bo

livia

200

12

710

882.

6558

3.00

781,

010

Dom

inica

n Re

publ

ic21

00

23

1680

4.63

094.

9509

191

Sam

ple

aver

age

10.4

80.

340.

142.

041.

946.

0447

.40

0.47

080.

6598

8,22

6


Pane

l B:M

eans

by

Quar

tiles

of G

DP p

er c

apita

in 1

999

1stQu

artil

e6.

770.

000.

051.

591.

144.

0024

.50

0.10

0.20

24,3

722nd

Quar

tile

11.1

00.

240.

142.

142.

386.

1949

.29

0.33

0.53

5,84

73rd

Quar

tile

12.3

30.

520.

142.

192.

337.

1453

.10

0.41

0.62

1,56

84th

Quar

tile

11.9

00.

620.

242.

241.

956.

9063

.76

1.08

1.34

349

Pane

l C: T

est o

f mea

ns (t

-sta

tistic

s)

1stvs

. 2nd

Quar

tile

24.2

0a22

.07b

20.8

721

.35

23.6

4a2

3.34

a23

.71a

23.

03a

23.

97a

12.0

3a

1stvs

. 3rd

Quar

tile

24.5

8a23

.02a

20.8

721

.64b

22.8

2a2

4.07

a24

.21a

22.

54b

23.

19a

16.3

5a

1stvs

. 4th

Quar

tile

24.0

4a22

.08a

21.5

521

.61

22.4

3b2

3.18

a24

.09a

23.

53a

24.

06a

17.3

1a

2ndvs

.3rd

Quar

tile

21.1

721

.34

0.00

20.1

10.

102

1.51

20.5

42

0.52

20.

596.

14a

2ndvs

.4th

Quar

tile

20.7

221

.17

20.6

120

.21

1.10

20.

8921

.46

22.

54b

22.

73a

8.05

a

3rdvs

.4th

Quar

tile

0.33

20.2

720

.61

20.1

10.

820.

2621

.06

22.

17b

22.

27b

8.53

a

a. S

igni

cant

at 1

per

cent

;�

b. s

igni

cant

at 5

per

cent

; c. s

igni

cant

at 1

0 pe

rcen

t.�

�


reported in Panel C). Rich countries also have fewer safety andhealth, tax, and labor start-up procedures than the rest ofthesample.Similarly, meeting government requirements takes ap-proximately 24.5 business days in rich countries, statisticallysignicantly lower than the rest-of-sample mean of 55.4 days. In�contrast, countries in the other three quartiles of per capitaincome are not statistically different from each other in the num-ber of procedures and the time it takes to complete them.

To summarize, the regulation of entry varies enormouslyacross countries. It often takes the form of screening procedures.Rich countries (i.e., those in the top quartile of per capita GDP)regulate entry relatively less than do all the other countries. Inprinciple, these ndings are consistent with both the public choice�and public interest theories. Market failures might be more per-vasive in countries with incomes just below the rst quartile of�GDP per capita, generating a greater demand for benign regula-tion in these countries. Alternatively, income levels may proxy forcharacteristics ofpolitical systems that allow politicians or in-cumbent rms to capture the regulatory process for their own�benet. In the next two sections we relate these patterns in the�data to the theories of regulation.

IV. WHO GETS THE RENTS FROM REGULATION ?Theories of regulation differ in their predictions as to who

gets its benets. The public interest theory predicts that stricter�entry regulation is associated with higher measured consumerwelfare. In contrast, the public choice theory sees regulation as atool to create rents for bureaucrats or incumbent rms. Stricter�regulation should then be associated with higher corruption andless competition.

Measuring rents is inherently extremely difcult, especially�across countries. In this section, we present some measures thatwe have been able to nd that bear—albeit quite imperfectly— on�the relevant theories. To begin, consider some variables bearingon the public interest theory. These variables reect the activities�of all rms in the country, and not just the entrants. The rst is� �a measure of a country’s compliance with international qualitystandards. It is a natural variable to focus on if the goal ofregulation is to screen out entrants who might sell output ofinferior quality. Second, we consider the level of water pollution,which should fall if entry regulation aims to control externalities


and does so successfully.8 Third, we consider two measures ofhealth outcomes that publicly interested entry regulation wouldguard against:the number of deaths from accidentalpoisoningand from intestinal infections.9 In addition, we include two mea-sures of the size of the unofcial economy based on estimates of�unofcial� output and employment,respectively.Since rms op-�erating unofcially avoid nearly all regulations, a large size of the�unofcial� economy in countries with more regulations under-mines the prediction of the public interest theory that regulationeffectively protects consumers.10 Finally, we use a survey mea-sure of “productmarket competition.” Stiffer entry regulationshould be associated with greater competition in the public inter-est theory,and lacking competition in the public choice theory,especially in its regulatory capture version.

Table IV presents the results on these seven measures ofconsequences ofregulation using the number of procedures asdependent variables.For two reasons, we run each regressionwith and without the log of per capita GDP. First, the number ofprocedures is correlated with income per capita, and we want tomake sure that we are not picking up the general effects of goodgovernance associated with higher income. Second, we use GDPper capita as a rough proxy of the prevalence of market failures ina country. Including per capita income as a control is a crude wayto keep the need for socially desirable regulation constant, whichallows us to focus on the consequences (and later causes) ofregulation separately from the need.

The results in Table IV show that compliance with interna-tional quality standards declines as the number ofproceduresrises. Pollution levels do not fall with regulation levels. The twomeasures of accidental poisoning are not lower in countries withmore regulations (if anything, the opposite seems to be true evencontrolling for per capita income). More regulation is associatedwith a larger unofcial economy, and statistically signicantly so� �if we use the unofcial� employmentvariable. Competition incountries with more regulation is perceived to be less intense,

8. We have tried measures of air pollution and obtained similar results.9. Due to reporting practices in poor countries,the second variable might

better capture deaths from accidental poisoning in the poor countries, according tothe World Health Organization [1998].

10. There is a large literature detailing how regulation can drive rms into�the unofcial economy, where they can avoid some or all of these regulations. See,�for example,Johnson, Kaufmann, and Shleifer [1997] and Friedman,Johnson,Kaufmann, and Zoido-Lobaton [2000].


TABLE IVEVIDENCE ON REGULATION AND SOCIAL OUTCOMES

The table presents the results of OLS regressions using the followingseven dependent variables:(1) Quality standards as proxied by the numberof ISO 9000 certications;� (2) Water pollution; (3) Deaths from accidentalpoisoning; (4) Deaths from intestinal infection; (5) Size of the unofcial�economy as a fraction of GDP; (6) Employment in the unofcial� economy;and (7) Product market competition. The independent variables are the logof the number of procedures and the log of per capita GDP in dollars in1999. Table II describes all variables in detail. Robust standard errors areshown below the coefcients.�

Dependent variableNumber ofprocedures

LnGDP/POP 1999 Constant

R2

N

Quality standards (ISOCertications)�

2 0.2781a 0.7649a 0.3311(0.0496) (0.1268) 85

2 0.1595a 0.0771a 2 0.1140 0.5384(0.0443) (0.0131) (0.1484) 85

Water pollution 0.0127b 0.1557a 0.0247(0.0084) (0.0174) 76

2 0.0037 2 0.0131a 0.2984a 0.2310(0.0076) (0.0027) (0.0314) 76

Deaths from accidentalpoisoning 0.6588a 1.6357a 0.1179

(0.2057) (0.4381) 570.0637 2 0.4525a 6.8347a 0.4109

(0.1958) (0.0933) (1.0929) 57

Deaths from intestinal infection 2.3049a 2 2.2697a 0.3451(0.3081) (0.6778) 611.0501a 2 0.8717a 7.8494a 0.6259

(0.2971) (0.1012) (1.3048) 61

Size of the unofcial economy� d 14.7553a 2 3.7982 0.2482(2.5698) (5.2139) 736.4849b 2 6.1908a 67.1030a 0.5187

(2.5385) (1.0834) (13.7059) 73

Employment in the unofcial�economy

19.4438a 2 4.1103 0.3132(2.5756) (5.9160) 4613.8512a 2 4.4585a 41.5133b 0.44772 3.6056 (1.3918) (17.6836) 46

Product market competition 2 0.4012a 5.7571a 0.1405(0.1213) (0.2511) 54

2 0.1418 0.2108a 3.3579a 0.3087(0.1202) (0.0680) (0.7749) 54

a. Signicant at 1 percent; b. signicant at 5 percent; c. signicant at 10 percent.� � �d. The regression on the size ofthe unofcial� economy controls for the log ofGDP per capita plus

unofcial economy income (i.e., GDP per capita* (1� 1unofcial economy)) and not just by GDP per capita as�all other regressions on the table do.


although this result is only statistically signicant without the�income control. We have also run all regressions using cost andtime as independent variables, and obtained qualitatively similarresults. While the data are noisy, none of the results support thepredictions of the public interest theory.11

The negative results in Table IV should be interpreted withcaution. First, some of our measures of public goods,such asdeaths from accidental poisoning, are probably more relevant forpoor countries, and in particular are unlikely to be inuenced by�entry regulation for rich countries. Accordingly, it might be moreappropriate to perform the analysis separately for countries atdifferent income levels. To this end, we divide the sample at themedian per capita income and rerun the regressions in Table IVfor each subsample. The data do not support the proposition that,in the subsample of poorer countries, heavier regulation of entryis associated with better social outcomes or more competition.

Second, an even deeper concern with the results in Table IVis that, despite our control for per capita income, there is impor-tant unobserved heterogeneity among countries correlated withregulation, which accounts for the results. For example, supposethat some countries have particularly egregious market failures,but also especially poor alternative mechanisms for dealing withthem, such as the press and the courts. Regulation, for example,might be less infected by corruption than either the press or thejudiciary. A publicly interested regulator in such countries wouldchoose to use more regulatory procedures because the alternativemethods of dealing with market failure are even worse, but stillend up with inferior outcomes.

We cannot dismiss this concern with the results of Table IV,although our later ndings cast doubt on its validity. We run the�regressions in Table IV using information on the freedom of thepress from Djankov,McLiesh, Nenova, and Shleifer [2001],andnd that,� holding constant various measures offreedom ofthepress and per capita income, the number of procedures is still notassociated with superior social outcomes. We also run the regres-sions in Table IV using a number of measures of citizen access to

11. Using data for publicly traded rms,� we have found no evidence thatcountries with heavier entry regulation have more protable rms, as measured� �by the return on assets. These protability numbers, however, are very crude. We�also measured protability using the return on World-Bank-nanced projects� �from the World Bank Operations Evaluation Department. These data also yieldno evidence that more regulations are associated with greater returns.


justice and of efciency of� the judiciary from Djankov et al.[2001b].Again, we nd that,� holding constant these measuresand per capita income, the number of procedures is associated, ifanything, with inferior social outcomes.

A direct implication of the tollboothhypothesis is that cor-ruption levels and the intensity of entry regulation are positivelycorrelated. In fact,since in many countries in our sample politi-cians run businesses, the regulation of entry produces the doublebenet of corruption revenues and reduced competition for the�incumbent businesses already afliated with the politicians. Fig-�ure III presents the relationship between corruption and thenumber of procedures without controlling for per capita GDP.12

Panel A of Table V shows statistically that,consistent with thetollbooththeory, more regulation is associated with worse corrup-tion scores. The coefcients are statistically signicant (with and� �without controlling for income) and large in economic terms. Theestimated coefcients imply that, controlling for per capita GDP,�reducing the number ofprocedures by ten is associated with a

12. We have tried a number of measures of corruption,all yielding similarresults. We have made sure that our results do not depend on “red tape” being partof the measure of corruption.

F IGURE IIICorruption and Number of Procedures

The scatter plot shows the values of the corruption index against the (log)number of procedures for the 78 countries in our sample with nonmissing data oncorruption.


reduction in corruption of .8 of a standard deviation, roughly thedifference between France and Italy.The results using the costand the time of meeting the entry regulations as independent

TABLE VEVIDENCE ON THE TOLLBOOTH THEORY

The table presents the results of OLS regressions using corruption as thedependent variable. The independent variables are (1) the log of the numberof procedures;(2) the log of time; (3) the log of cost;and the log of per capitaGDP in dollars in 1999. Panel A presents results for the 78 observationswith available corruption data. Panel B reports results separately for thesubsample of countries with GDP per capita in 1999 above and below thesample median. Table II describes all variables in detail. Robust standarderrors are shown in parentheses below the coefcients.�

Panel A: Results for the whole sample

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

Number ofprocedures

2 3.1811a 2 1.8654a(0.2986) (0.2131)

Time 2 1.7566a 2 0.8854a(0.1488) (0.1377)

Cost 2 1.2129a 2 0.4978a(0.1206) (0.1285)

Ln GDP/POP 1999 0.9966a 0.9765a 0.9960a(0.0864) (0.1014) (0.1118)

Constant 11.8741a 1.1345 11.0694a 0.0677 2.7520a 2 4.0893a(0.7380) (0.9299) (0.5932) (1.1176) (0.2414) (0.7867)

R2 0.4656 0.8125 0.4387 0.7662 0.4256 0.7306N 78 78 78 78 78 78

Panel B: Results for countries above and below the world median GDP per capita

Countries abovemedian GDP/POP1999

Countries belowmedian GDP/POP1999

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

Number ofprocedures

2 1.8729a 2 0.7841b(0.2971) (0.3304)

Time 2 0.8135a 2 0.0923(0.1762) (0.2850)

Cost 2 0.5327a 2 0.3408a(0.1894) (0.1021)

Ln GDP/POP 1999 1.4811a 1.5871a 1.7621a 0.3993b 0.3680c 0.2117(0.2265) (0.2789) (0.2913) (0.1735) (0.1802) (0.1718)

Constant 2 3.6970 2 5.9027c 2 11.3736a 2.3246c 1.0098 1.3125(2.4628) (2.9942) (2.5773) (1.2849) (1.8813) (1.1136)

R2 0.7820 0.7155 0.6728 0.2362 0.1324 0.2830N 40 40 40 38 38 38

a. Signicant at 1 percent; b. signicant at 5 percent; c. signicant at 10 percent.� � �


variables are also statistically signicant, pointing further to the�robustness of this evidence in favor of the tollbooth theory.

One way to reconcile the ndings in Table V with the public�interest theory is to argue that regulation has unintended conse-quences.Thus, benign politicians in emerging markets imitatethe regulations of rich countries with best intentions in mind, butare stymied by corruption and other enforcement failures.Thistheory is not entirely consistent with our earlier nding that�poorer countries in fact have more entry regulations than richcountries do.A further implication of this theory is that regula-tions should have a bigger impact on corruption in poorer coun-tries. Panel B of Table V addresses this hypothesis by examiningseparately the relationship between entry regulations and cor-ruption in countries with above and below world median income.The results show that regulations actually have a stronger effecton corruption in the subsample of richer countries.

On the second version of the unintended consequences argu-ment, it may be impossible for a benevolent government to screenbad entrants without facilitating corruption [Banerjee 1997; Ace-moglu and Verdier 2000]. In countries whose markets are fraughtwith failures, it might be better to have corrupt regulators thannone at all. Corruption may be the price to pay for addressingmarket failures. We turn next to the evidence regarding thepolitical attributes of countries that regulate entry to disentanglethe competing theories of regulation.

V. WHO REGULATES ENTRY ?In this section we focus on the political attributes of countries

that regulate entry. These attributes are intimately related to thecompeting hypotheses aboutregulation. In the public interesttheory, regulation remedies market failures.The implication isthat countries whose political systems are characterized byhigher congruence between policy outcomes and socialprefer-ences should regulate entry more strictly. In the empirical analy-sis that follows, we identify such countries with more represen-tative and limited governments.

In the public choice theory, despotic regimes are more likelyto be captured by incumbents and to have regulatory systemsaimed at maximizing the bribes and prots of a few cronies rather�than address market failures [Olson 1991; De Long and Shleifer1993]. Such dictators need the political support of various inter-


est groups,and use distortionary policies to favor their friendsand to abuse their opponents. The dictator’s choice of distortion-ary policies is not mitigated by public pressure, since he faces noelections. When the public is less able to assert its preferences,then, we expect more distortionary policy choices. Specically, we�expect more representative and limited government to be associ-ated with lighter regulation of entry.

One might argue,in contrast, that dictators should pursueefcient economic policies,� including light regulation of entry, ifthey are politically secure and can “tax” the fruits of entry andgrowth. One response,discussed by Olson [1991]and De Longand Shleifer [1993],is that while a few dictators are politicallysecure and pursue enlightened policies,most are not. Insecuredictators extract what they can from the economy as fast as theycan both to prolong their tenure,and to enrich themselves andtheir supporters while still in power. Democracy might notlengthen the horizons of politicians, but it does limit theiropportunities.

We collect data on a variety of characteristics ofpoliticalsystems,partly because we want to be exible regarding the�meaning of “good government.” Where possible, we use variablesfrom different sources to check the robustness of our results. Ourpolitical variables fall into four broad groups.The rst includes�the de facto independence of the executive and an index of con-straints on the executive. The second group includes an index ofthe effectiveness of the legislature and a measure of competitionin the legislature’s nominating process. The third group includesa measure of autocracy and one of political rights.

An additional variable that we focus on,used in the earlierwork by La Porta et al. [1998,1999] is legal origin.We classifycountries based on the origin of their commercial laws into ve�broad groups:English, French, German, Scandinavian,and So-cialist. Legal origin has been viewed as a proxy for the govern-ment’s proclivity to intervene in the economy and the stance ofthe law toward the security of property rights in a country [LaPorta et al. 1999].

Correlations among the political variables are presented inTable VI. Political variables tend to be strongly correlated withinblocks. For example, the measure of constraints on the executivepower is highly correlated with de facto independence of theexecutive (0.9761)and with the effectiveness ofthe legislature(0.9078).Yet, we report results on all three variables as each


comes from a different source. Similarly, blocks of variables tendto be correlated with each other. In particular, democracy tends tobe positively associated with competitive and limited executiveand legislative branches.Legal origin, in contrast, is insigni-�cantly correlated with other political variables (the exception isSocialist legal origin which has obvious correlations with democ-racy and limited government).13 Income levels are positively as-

13. Consistent with this nding, La Porta et al. [2001] nd that common law� �legal origin is associated with English constitutional guarantees of freedom, such

TABLE VICORRELATION TABLE FOR POLITICAL ATTRIBUTES

The table reports correlations among measures of regulation and thevariables used in Table VII. All variables are dened in Table II.� Signicance�levels are Bonferroni-adjusted.

Exec defacto

independence

Constraintson executive

powerEffectivenesslegislature

Competitionnominating Autocracy

Politicalrights

FrenchLO

Exec de factoindependence 1.0000

Constraints onexec. power 0.9761a 1.0000

Effectivenesslegislature 0.9210a 0.9078a 1.0000

Competitionnominating 0.8243a 0.8069a 0.8484a 1.0000

Autocracy 2 0.9085a 2 0.8844a 2 0.8514a 2 0.7819a 1.0000Political rights 0.8440a 0.8448a 0.8485a 0.7191a 2 0.8564a 1.0000French legal

origin 2 0.1814 2 0.1814 2 0.1901 2 0.1985 2 0.0258 0.0565 1.0000Socialist legal

origin 2 0.3321 2 0.2927 2 0.3236 2 0.3240 0.5475a 2 0.4572a 2 0.4169aGerman legal

origin 0.2101 0.2008 0.2023 0.1281 2 0.1920 0.2444 2 0.2141Scandinavian

legal origin 0.3391 0.3274 0.3378 0.2522 2 0.2978 0.3109 2 0.1727English legal

origin 0.2259 0.1998 0.1462 0.2412 2 0.2324 0.0778 2 0.4874aLn

GDP/POP1999 0.6900a 0.6703a 0.7483a 0.6123a 2 0.6389a 0.7519a 2 0.0767bLn(Number of

procedures) 2 0.5518a 2 0.5234a 2 0.5848a 2 0.4435b 0.4662a 2 0.4412a 0.4863a

Ln(Time) 2 0.5420a 2 0.5204a 2 0.5635a 2 0.4360b 0.4770a 2 0.4921a 0.3976b

Ln(Cost) 2 0.5070a 2 0.4937a 2 0.5656a 2 0.4177b 0.4075b 2 0.4588a 0.3472Ln(Cost 1

time) 2 0.5700a 2 0.5478a 2 0.6267a 2 0.4745a 0.4713a 2 0.5085a 0.3870b

a. Signicant at 1 percent; b. signicant at 5 percent; c. signicant at 10 percent.� � �


sociated with democracy as well as with competitive and limitedexecutive and legislative branches, but not with the legal origin.The fact that countries with severe market failures have moreabusive governments by itself limits the normative usefulness ofthe Pigouvian model.

In Table VII we present the results of regressing the number

as the independence of the judiciary and the accountability of the government tothe law. These constitutional guarantees of freedom are strongly associated witheconomic freedoms, but less so with political freedoms.

TABLE VI(CONTINUED )

SocialistLO

GermanLO

ScandinavianLO

EnglishLO

LnGDP/

POP1999

Ln(Number ofprocedures)

Ln(Time)

Ln(Cost)

Ln(Cost 1time)

1.0000

2 0.1479 1.0000

2 0.1192 2 0.0612 1.0000

2 0.3365 2 0.1729 2 0.0139 1.0000

2 0.1995 0.3409 0.3133 2 0.0742 1.0000

0.1538b 0.0030b 2 0.3413b 2 0.5069a 2 0.4745a 1.00000.1869 2 0.0640 2 0.2914 2 0.4291b 2 0.5014a 0.8263a 1.00000.0319 2 0.0727 2 0.3007 2 0.2172 2 0.5953a 0.6354a 0.6147a 1.0000

0.0851 2 0.0933 2 0.2786 2 0.3094 2 0.6244a 0.7434a 0.7793a 0.9605 1.0000


TABL

E VI

IEV

IDEN

CE O

NRE

GULA

TION

AND

POLI

TICA

LAT

TRIB

UTES

The

tabl

e pr

esen

ts th

e re

sults

of r

unni

ng re

gres

sions

for t

he lo

g of

the

num

ber o

f pro

cedu

res a

s the

dep

ende

nt v

aria

ble.

We

run

seve

n re

gres

sions

usin

g va

rious

pol

itica

lindi

cato

rs d

escr

ibed

in T

able

II a

nd (l

og)

GDP

per c

apita

.Ro

bust

stan

dard

erro

rs a

resh

own

in p

aren

thes

es b

elow

the

coef

cient

s.�

Depe

nden

t var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)

Exec

utiv

e de

fact

o in

depe

nden

ce20

.124

9a

(0.0

322)

Cons

train

ts o

n ex

ecut

ive

powe

r20

.104

8a

(0.0

352)

Effec

tiven

ess

of le

gisla

ture

20.3

301a

(0.0

778)

Com

petit

ion

nom

inat

ing

20.

2763b

(0.0

999)

Auto

crac

y0.

0545b

(0.0

178)

Polit

ical r

ight

s20

.347

0(0

.218

5)Fr

ench

lega

l orig

in0.

7245a

(0.0

916)

Socia

list l

egal

orig

in0.

4904a

(0.1

071)

Germ

an le

gal o

rigin

0.72

76a

(0.1

363)

Scan

dina

vian

lega

l orig

in2

0.00

85(0

.173

3)Ln

GDP

/POP

1999

20.0

491

20.0

634c

20.0

087

20.

0902b

20.0

867a

20.0

939b

20.

1434a

(0.0

331)

(0.0

352)

(0.0

401)

(0.0

358)

(0.0

321)

(0.0

386)

(0.0

270)

Cons

tant

3.17

82a3.

2040a

2.87

09a3.

3540a

2.74

57a3.

1850

a2.

9492a

(0.2

334)

(0.2

408)

(0.2

586)

(0.2

641)

(0.2

888)

(0.2

599)

(0.1

955)

R20.

3178

0.28

720.

3424

0.24

750.

2640

0.23

500.

6256

N84

8473

7384

8485

a. S

igni

cant

at 1

per

cent

;�

b. s

igni

cant

at 5

per

cent

; c. s

igni

cant

at 1

0 pe

rcen

t.�

�


of procedures on a constant and each ofthe political variablestaken one at a time and the log of per capita income.In inter-preting these regressions, we take the broad political measures oflimited and representative government as being exogenous toentry regulation. It is possible, of course, that both the politicaland the regulatory variables are simultaneously determined bysome deeper historical factors. Even so, it is interesting to knowwhat the correlation is. Does the history that produces goodgovernment also produce many or few regulations of entry? Thecontrol for the level of development is crucial(and in fact ourresults without this control are signicantly stronger).� Marketfailures are likely to be both more pervasive and severe in poorcountries than in rich ones. Moreover, our measures of goodgovernment are uniformly higher in richer countries. Withoutincome controls, our political variables may just proxy for incomelevels. Imagine, for example, that the consumers in poor countriesare exposed to a larger risk from bad rms entering their markets�and selling goods of inferior quality. The Pigouvian plannerwould then need more tools to screen entrants in the poorercountries.

Holding per capita income constant, countries with morelimited and representative governments have statistically sig-nicantly fewer procedures for entry regulation using ve out of� �six measures of better government.14 These results show thatcountries with more limited governments, governments moreopen to competition, and greater political rights have lighterregulation of entry even holding per capita income constant.Figure IV plots the number of procedures against the autocracyscore and shows that regulation is increasing in autocracy. Reg-ulation is heavy in autocratic countries such as Vietnam andMozambique and light in democratic countries such as Australia,Canada,New Zealand, and the United States.

The log of per capita GDP tends to enter these regressionssignicantly.� The interpretation of this result is clouded bothbecause there are problems of multicollinearity with the politicalvariables and because the direction of causation is unclear. In thepublic choice theory, burdensome regulation reects transfers�

14. Results are signicant in all� six regressions when we use time ratherthan number of procedures as the dependent variable.In contrast, results areinsignicant in three regressions (competition in the legislature’s nominating�process, autocracy, and political rights) when using cost as the dependentvariable.


from entrepreneurs or consumers, which are likely to be distor-tionary and, hence, associated with lower levels of income. Coun-tries may be poor because regulation is hostile to new businessformation.

Holding per capita income constant, countries of French,German, and Socialist legal origin have more regulations thanEnglish legal origin countries,while countries of Scandinavianlegal origin have about the same. The result that civil law coun-tries (with the exception of those in Scandinavia) regulate entrymore heavily supports the view that the legal origin proxies forthe state’s proclivity to intervene in economic life [La Porta et al.1999]. However, note that in itself this evidence does not discrimi-nate among the alternative theories in the same way as theevidence on democracy does: French origin countries mightmerely be more prepared to deal with market failures than com-mon law countries.

These results are broadly consistent with the public choicetheory that sees regulation as a mechanism to create rents forpoliticians and the rms they support.� The public choice theorypredicts that such rent extraction should be moderated by bettergovernment to the extent that outcomes in such regimes come

F IGURE IVAutocracy and Number of Procedures

The scatter plot shows the values of the (log) number of procedures against theautocracy score (higher values for more autocratic systems) for the 84 countries inour sample with nonmissing data for the autocracy score.


closer to representing the preferences of the public.In contrast,these results are more difcult to reconcile with public interest�unless one identies it with political systems of countries such as�Bolivia, Mozambique, or Vietnam, where corruption is wide-spread, governments are unlimited, and property rights insecure.Of course, it is possible that autocratic countries would performeven worse in the absence of heavy regulation because marketfailures are larger and alternative mechanisms of social controlare inferior. Such a possibility strikes us as remote, especiallysince we hold the level of development constant.

VI. CONCLUSION

An analysis of the regulation of entry in 85 countries showsthat, even aside from the costs associated with corruption andbureaucratic delay, business entry is extremely expensive, espe-cially in the countries outside the top quartile of the incomedistribution. We nd that heavier regulation of entry is generally�associated with greater corruption and a larger unofcial� econ-omy,but not with better quality of private or public goods.Wealso nd that the countries with less limited, less democratic, and�more interventionist governments regulate entry more heavily,even controlling for the level of economic development.

This evidence is difcult� to reconcile with public interesttheories of regulation but supports the public choice approach,especially the tollbooth theory that emphasizes rent extraction bypoliticians [McChesney 1987; Shleifer and Vishny 1993]. Entry isregulated more heavily by less democratic governments, and suchregulation does not yield visible social benets.� The principalbeneciaries� appear to be the politicians and bureaucratsthemselves.

WORLD BANKDEPARTMENT OF ECONOMICS , HARVARD UNIVERSITYSCHOOL OF MANAGEMENT , YALE UNIVERSITYDEPARTMENT OF ECONOMICS , HARVARD UNIVERSITY

REFERENCESAcemoglu, Daron, and Thierry Verdier, “The Choice between Market Failures and

Corruption,” American Economic Review,XC (2000), 194 –211.Banerjee, Abhijit, “A Theory of Misgovernance,” Quarterly Journal of Economics,

CXII (1997), 1289–1332.


Central Intelligence Agency, CIA World Factbook (2001), published online. http://www.cia.gov/cia/publications/factbook/

Chidzero, Anne-Marie, “Senegal,” in The Informal Sector and Micronance Insti-�tutions in West Africa, Leila Webster and Peter Fidler, eds. (Washington, DC:The World Bank, 1996).

De Long J. Bradford, and Andrei Shleifer, “Princes and Merchants: European CityGrowth before the Industrial Revolution,” Journal of Law and Economics,XXXVI (1993), 671–702.

Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shlei-fer, “The Regulation of Labor,” Harvard University manuscript in prepara-tion, 2001a.

Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shlei-fer, “Courts: The Lex Mundi Project,” Harvard University, 2001b.

Djankov, Simeon, Caralee McLiesh, Tatiana Nenova, and Andrei Shleifer, “WhoOwns the Media?” NBER Working Paper No. 8288, 2001.

De Soto, Hernando, The Other Path (New York, NY: Harper and Row, 1990).Freedom House, Freedom of the World (New York, NY: Freedom House, 2001).Friedman, Eric, Simon Johnson, Daniel Kaufmann, and Pablo Zoido-Lobaton,

“Dodging the Grabbing Hand: The Determinants of Unofcial Activity in 69�Countries,” Journal of Public Economics, LXXVI (2000), 459– 494.

Henisz, Witold Jerzy, “The Institutional Environment for Economic Growth,”Economics and Politics, XII (2000), 1–31.

Institute for International Management Development,World CompetitivenessReport (Lausanne, Switzerland: IMD, 2001).

Jaggers, Keith, and Monty G. Marshall, “Polity IV Project,” Center for Inter-national Development and Conict Management,� University of Maryland,2000.

Johnson, Simon, Daniel Kaufmann, and Andrei Shleifer, “The Unofcial Econ-�omy in Transition,” Brookings Papers on Economic Activity, 2 (1997),159–239.

Kasnakoglu, Zehra, and Mu¨ nu¨ r Yayla, “Unrecorded Economy in Turkey: A Mone-tary Approach” (1999), in Informal Sector in Turkey,Volume I, Tuncer Bu-lutay, ed. (Ankara, Turkey: SIS, forthcoming).

La Porta, Rafael, Florencio Lopez-de-Silanes,Andrei Shleifer, and Robert W.Vishny, “Law and Finance,” Journal of Political Economy, CVI (1998),1113–1155.

La Porta Rafael, Florencio Lopez-de-Silanes,Andrei Shleifer, and Robert W.Vishny, “The Quality of Government,” Journal of Law,Economics,and Or-ganization, XV (1999), 222–279.

La Porta, Rafael, Florencio Lopez-de-Silanes, Cristian Pop-Eleches,and AndreiShleifer, “Guarantees of Freedom,” manuscript, Harvard University, 2001.

McChesney, Fred S., “Rent Extraction and Rent Creation in the Economic Theoryof Regulation,” Journal of Legal Studies, XVI (1987), 101–118.

Olson, Mancur, “Autocracy, Democracy, and Prosperity,” in Richard Zeckhauser,ed., Strategy of Choice (Cambridge, MA: MIT Press, 1991).

Peltzman, Sam, “Toward a More General Theory of Regulation,” Journal of Lawand Economics, XIX (1976), 211–240.

Pigou, Arthur C., The Economics of Welfare, 4th ed. (London: Macmillan and Co.,1938).

Reynolds,Thomas H., and Arturo A. Flores, Foreign Law: Current Sources ofCodes and Basic Legislation in Jurisdictions of the World (Littleton, CO: F. B.Rothman, 1989).

Sananikone, Ousa, “Burkina Faso,” in The Informal Sector and Micronance�Institutions in West Africa, Leila Webster and Peter Fidler, eds. (Washington,DC: The World Bank, 1996.)

Schneider, Friedrich, “The Value Added of Underground Activities: Size andMeasurement of Shadow Economies and Shadow Economy Labor Force Allover the World,” mimeo, 2000.

Schneider, Friedrich, and Dominik H. Enste, “Shadow Economies: Size, Causes,and Consequences,” Journal of Economic Literature, XXXVIII (2000), 77–114.

Shleifer, Andrei, and Robert W. Vishny, “Corruption,” Quarterly Journal of Eco-nomics, CVIII (1993), 599 – 617.


Shleifer, Andrei, and Robert W. Vishny, The Grabbing Hand: Government Pa-thologies and their Cures (Cambridge, MA: Harvard University Press, 1998).

SRI International, International Practices and Experiences in Business StartupProcedures (Arlington, VA: SRI, 1999).

Stigler, George J., “The Theory of Economic Regulation,” Bell Journal of Econom-ics and Management Science,II (1971), 3–21.

Tullock, Gordon, “The Welfare Cost of Tariffs, Monopoly,and Theft,” WesternEconomic Journal, V (1967), 224–232.

Turnham, David, Bernard Salome,and Antoine Schwartz,The Informal SectorRevisited (Paris, OECD,1990).

World Bank, “Administrative Barriers to Investment in Africa: The Red TapeAnalysis,” FIAS, Washington, DC, 1999.

——, World Development Indicators (Washington, DC: The World Bank, 2001).World Economic Forum, The Global Competitiveness Report 2001, Klaus Schwab

et al., eds. (New York, NY: Oxford University Press, 2001).World Health Organization, Causes of Death and Life,Birth Statistics (Geneva,

Switzerland: World Health Organization, 1998).


db methodology regulation of entry - doing business

Documents