out of the gray: the impact of provincial institutions on...

Electronic copy available at: http://ssrn.com/abstract=1669688

Out of the Gray: The Impact ofProvincial Institutions on Business

Formalization in Vietnam

Edmund Malesky and Markus Taussig

Scholars have long argued that institutional context significantly influencesbusiness strategy and economic performance. Research on the relationship be-tween institutions and business strategy, however, has overwhelmingly focusedon the decisions of larger, established corporations, mostly neglecting thestrategic thinking of smaller, more entrepreneurial ventures. This article seeksto correct this bias by focusing the analysis directly on the critical decision ofsmall-scale entrepreneurs to move from the informal and largely unregulatedsector into operation as formal companies. Using a unique dataset and rankingof provincial governance institutions from Vietnam, the authors show that im-provements in institutions make firms more likely to choose the formal sectorfrom the start and, for those who do not, to spend less time in the informal sec-tor. The study also finds that property rights have a more salient impact on for-malization than other types of institutions.

KEYWORDS: formalization, property rights, Vietnam, land title, governance, in-stitutions, provincial competitiveness index, PCI, institutions, economic growth

Scholars have long argued that institutional context significantlyinfluences business strategy, as governance influences the

propensity of firms to invest and their ability to grow (Porter 1980;DiMaggio and Powell 1991; North 1991; Weingast 1993; Knack andKeefer 1995; Grief 2006). Research on the interaction between institu-tions and business strategy, however, has overwhelmingly focused onthe decisions of larger, established corporations rather than smaller,more entrepreneurial ventures. This oversight is significant because,given the higher mortality they face, early-stage firms are much moreaverse to uncertainty than their more mature counterparts (Aldrich andFiol 1994; Santos and Eisenhardt 2006). As a result, there is a signifi-

Journal of East Asian Studies 9 (2009), 249–290

249

04_JEAS9.2_Malesky.qxd 5/14/09 11:12 AM Page 249

Electronic copy available at: http://ssrn.com/abstract=1669688

cant need for further research to understand how institutions influencenew firms and what role they play in driving aggregate economicgrowth and employment creation.

In this article, we consider the impact of the quality of governanceon an especially fundamental strategic decision faced by every nascententrepreneur: the decision to stay in the informal sector or to submit toformal government regulation. In practice, formalization means regis-tering as a company and thereby acknowledging the size and scope ofbusiness operations to tax officials and regulators. More conceptually,we analyze how the quality of local institutions influences the decisionto transition from avoiding government attention to accepting it.

Better understanding the microlevel dynamic of this entrepreneurialdecision is of great importance to our understanding of both politics andeconomic growth. When entrepreneurs choose to stay in the informaleconomy, governments lose tax revenues, the public faces danger fromineffective health and environmental regulation, and a vicious circle en-sues that undermines rule of law. This vicious circle involves poor gov-ernance, which leads to greater informality, which leads in turn to evenworse governance as policymakers lose out on information and other re-sources needed to properly regulate competition and protect the public.Indeed, the problem may even affect a country’s economic growth, as in-formal entrepreneurs self-select into lower-growth trajectories. SimeonDjankov (2008), for example, argues that the movement of firms fromthe informal economy into the formal economy is the most importantbenefit from improving governance and institutions.

The so-called underground economy is substantial even in ad-vanced economies like Singapore and South Korea (Loayza, Oviedo,and Serven 2005) but is even more germane in developing countries,where the informal economy often totals as much as half the value ofofficial GDP (Loayza, Oviedo, and Serven 2005; Schneider 2005). TheWorld Bank suggests that more than 30 percent of total output in de-veloping countries lies outside the realm of government regulation andthat this share may actually be growing (Kenyon 2007). Cross-nationalresearch designs have consistently found that weak institutions andgovernance are associated with lower shares of business activity oc-curring within the formal, government-regulated economy (Frye andShleifer 1997; Johnson, Shleifer, and Kaufmann 1997; La Porta et al.1999; Johnson et al. 2000; Friedman et al. 2000; Ihrig and Moe 2004;Antunes and Cavalcanti 2007; Djankov et al. 2008). These findings areconsistent with theoretical work suggesting that the probability of

250 Out of the Gray

04_JEAS9.2_Malesky.qxd 5/14/09 11:12 AM 50

“going formal” increases with prospects for future profitability (Ben-nett and Estrin 2007).

While the cross-national work analyzing the relationship betweeninstitutions and formalization is illustrative, its conception of both in-stitutions and causality is expansive. In general, there has been a fail-ure to disentangle an extremely broad spectrum of highly correlated in-stitutional factors that affect the environment facing business. This lackof precision in identifying the mechanics of the relationship betweeninstitutions, business strategies, and growth has naturally led to signif-icant criticism from political scientists and global management schol-ars (Kurtz and Schrank 2007; Ghemawat 2007).

This article tests the hypothesis that higher-quality governance in-creases the probability that individual entrepreneurs will transitionfrom the informal economy to the formal economy (an action hence-forth referred to as “formalization”). We test this hypothesis on aunique firm-level dataset covering all sixty-four provinces in nominallycommunist, but economically liberalizing and rapidly growing, Viet-nam. The subnational design allows us to hold constant cultural andhistorical factors that plague cross-national analyses (King et al. 2003),while focusing our attention on the quality of specific institutions,which varies considerably among subnational units in Vietnam.

As noted earlier, research on formalization around the world hasbeen bedeviled by the fact that it is usually not possible to access in-formation on completely informal firms. This is true in Vietnam aswell. But it is possible to use survey data to place firms along a spec-trum of formality. Near the informal end of this spectrum, most smallbusinesses are registered at the local level as household businesses butface substantially less (and more subjective) regulation than those firmsregistered as companies. Toward the more formal end of the spectrum,even fully registered companies are likely to engage in deception andnegotiation in their dealings with tax and other regulatory authorities.Because of this complex reality, we operationalize formalization asmovement by an entrepreneur from the relatively informal householdbusiness sector to the relatively formal company sector. Even thoughour quantitative analysis measures a dichotomous choice, our concep-tualization of formalization recognizes a more complex process.

Our results show that entrepreneurs in Vietnamese provinces withhigher-quality governance are significantly more likely to submit tomore extensive government regulation from the start of their operations.Further confirming our key hypothesis, we find that entrepreneurs who

251Edmund Malesky and Markus Taussig


initially select into the informal sector spend relatively less time therewhen local institutions are better.

We then address the serious issues of endogeneity endemic to re-search on institutions and business decisionmaking by adopting a two-stage, instrumental variable approach. Our strategy takes advantage ofVietnam’s war legacy to identify an exogenous determinant of the qual-ity of governance. The findings of these models strengthen our argu-ment that the causal direction runs from institutions to formalization.

Finally, after demonstrating the strength of our aggregate measureof governance, we disaggregate the composite index into its subcompo-nents to see what types of institutions are the primary drivers of businessformalization. This question, of course, is of critical importance to pol-icymakers in developing countries. Scholars have highlighted propertyrights (North 1991; North and Weingast 1989; Grief 2006), contract en-forcement institutions (Coase 1937; Laeven and Woodruff 2007), andregulatory institutions (Djankov et al. 2002). Some effort has been madeto disaggregate these effects (Acemoglu and Johnson 2005), but theanalysis has relied on cross-national regression, where the dependentvariable was GDP and not firm-level decisions. We find preliminary mi-crolevel evidence that the key types of institutions driving formalizationare property rights protection and confidence in local legal institutions.

Our study offers noteworthy contributions to three distinct litera-tures. First, we provide empirical evidence in support of the optimisticperspective, made famous by Hernando De Soto (1989), that informalenterprises represent a latent and underappreciated source of potentialeconomic growth in developing countries. Our findings complementprevious work (Di Tella, Galiani, and Schargrodsky 2007; Galiani andSchargrodsky 2007; Field 2007) by demonstrating that policy initia-tives aimed at clarifying property rights are a crucial, relatively cheap,and expeditious way to unlock this potential.

The second contribution of this article is to the rapidly growing en-trepreneurship literature. Our study represents a relatively rare large-Nempirical test of how particular environmental factors influence a clearlydefined strategic decision faced by all entrepreneurial ventures. Giventhat formality is far more conducive to firm growth, the study’s findingsdemonstrate a way in which external conditions shape the strategic ori-entation of entrepreneurs (Ardagna and Lusardi 2008). In particular, theyindicate that better institutions increase the probability of subsistence-oriented entrepreneurs turning into growth-oriented entrepreneurs.

Finally, we address the extensive literature on institutions andgrowth. Particularly germane is the recent debate in the pages of Jour-

252 Out of the Gray


nal of Politics on the relationship between cross-national measures ofgovernance and economic growth. Marcus Kurtz and Andrew Schrank(2007) argue that both the theoretical foundations and empirical analy-sis of the relationship are weak, while Daniel Kaufmann, Aart Kraay,and Massimo Mastruzzi (2007), architects of the most commonly citedmeasure of governance, defend their index and analyses of the impactof governance on growth.

Our research design allows us to address many of the criticismsleveled against the governance-growth research program. First, as dis-cussed earlier, we bypass the large problems of unobserved hetero-geneity in cross-national designs by focusing on units of analysis wheremost confounding historical and cultural factors can be held constant.Second, our decision-level analysis tests the causal logic directly andnot through macrolevel proxies for entrepreneurial decisions such asGDP growth or even aggregate investment. This allows a clearer diag-nosis of the mechanism by which institutions foster better economicoutcomes. Third, our research design is not vulnerable to the charge of“halo effects” that dogs much of the relevant work in which firm-levelsurveys are used—the charge that survey respondents may be morelikely to assign higher governance scores when located in areas char-acterized by recent economic growth (Seligson 2006). Kurtz andSchrank (2007) demonstrate the power of controlling for recent growthby showing that the significance of governance variables disappears incross-national regressions. We avoid the halo effect because our de-pendent variable is the individual entrepreneur’s one-time decision toformalize operations, not an aggregate measure of previous formaliza-tion that could color respondents’ views.

The article is organized as follows. In the first section, we discussthe specific case of Vietnam. Second, we introduce our data source andkey variables. In the next two sections, we lay out our results anddemonstrate our identification strategy for assessing the causal argu-ment. In the final section, we disaggregate governance institutions anddemonstrate which particular components of our aggregate governancemeasure drive formalization decisions.

The Case of Vietnam

Private business has a long history in Vietnam, despite the extendedrule of the country’s Communist party.1 Upon coming to power in1945, Ho Chi Minh’s government made a concerted effort to promote



local private business vis-à-vis dominant French business interests. In-deed, even in the headiest days of central planning, when all forms ofprivate business were formally illegal, the majority of Vietnameseworked in the informal private sector (Abrami and Henaff 2004).Throughout, entrepreneurs remained crucial to the functioning of theeconomy and to social welfare, filling in whenever and wherever theclumsy government system failed.

The pragmatic practice of certain local authorities’ turning a blindeye to illegal, but socially beneficial, entrepreneurial activity during thecentral planning period has been referred to as “fence breaking”(Fforde and de Vylder 1996). Such behavior characterized both Com-munist North Vietnam and subsequently unified, prereform Vietnam.Illegally operating entrepreneurs, however, were always at the mercy ofpolitical campaigns, serving as easy targets of blame in periods of in-flation and unemployment. This history and the continued lack of trans-parency and consistency of government policy and policy implementa-tion still have an impact even today on entrepreneurs’ attitudes towardmore open approaches to doing business. This is reflected, for example,in the continued value placed on political connections as business re-sources—even in the face of evidence that connected entrepreneurs donot actually perform any better than their less connected counterparts(Malesky and Taussig 2008).

These findings on fence breaking are not meant to minimize thesignificance of the launch in 1986 of doi moi (renovation) reforms forprivate business. One of the first economic reform moves was the le-galization of private sector activity and soon thereafter the establish-ment of a registration process for household enterprises.2 This reformhappened fully two years before the establishment of a legal frameworkfor private companies.3 Two decades later, the household sector re-mains the leading employer in not only agriculture, but also industryand services. Household enterprises played the particularly importantrole of absorbing a large share of the many workers cut loose from thestate sector and military in the early years of the reform period and con-tinue to serve as the country’s de facto primary welfare safety net.

While there are many data problems in identifying precisely thenumber of household businesses, it is clear that they have grown rap-idly. In a recent comprehensive survey of household activities, Viet-nam’s General Statistics Office (GSO) identified 3.7 million householdbusinesses. This compares to about 1 million in 1992 and around 2.5million in 2002. Most operations employed only one or two workers,with approximately 80 percent of these workers not receiving a regular

254 Out of the Gray


salary (e.g., family members) (VCCI 2008). While sources differ,household businesses appear to account for approximately 15 percentof GDP (Vietbao.vn 2007). This is about two times the share of regis-tered sole proprietorships—the smallest, and thus most comparable,form of private company (Vijverberg 2006).

A domestic company sector also emerged in the early doi moi erafollowing introduction of the country’s first enterprise laws in 1990 and1991, but growth stalled in the late 1990s at about 7–8 percent of GDP.Following a couple of years of slowed growth that coincided with thebroader East Asian economic crisis, Vietnam introduced an updated en-terprise law in January 2000 that dramatically reduced entry barriers(including time costs and requisite licenses) to setting up a company.4

The new enterprise law (henceforth “the Enterprise Law”) man-dated a simplified registration process that essentially eliminated dif-ferences in entry costs across provinces. Table 1 shows dramatic evi-dence of the reduction in registration times. In 1998, the median firmwaited nearly a month to fulfill registration procedures. By 2000, wait-ing periods had been halved to fifteen days. This leveling of the play-ing field was achieved through aggressive implementation of the law,which included public castigation of particular agencies and officialsseen to not be on board (Nguyen et al. 2004).

By the end of 2006, after seven consecutive years of increased an-nual registration totals, there were over 120,000 registered private com-panies operating in Vietnam—nearly six times the number in operationbefore the Enterprise Law came into effect in January 2000. A 2005 re-port by the Vietnam Chamber of Commerce and Industry suggestedthat a significant share of the increased registrations were, in fact, theformalization of firms formerly operating as household businesses(Taussig and Pham 2004). Table 1 further illustrates that companies thatregistered in 2000 or later were significantly less likely to have previ-ously operated as household enterprises, and those that had been house-hold enterprises spent significantly less time in the informal sector afterthe law. This is particularly true for firms in business services andhigher-technology industries.

As noted earlier, our concept of formalization does not involve amove from completely informal to completely formal, as it would bemisleading to refer to household businesses in Vietnam as fully infor-mal. Indeed, if the term informal is meant to mean government author-ities do not know about the existence of a particular business, Vietnamis likely to have very few informal businesses at all. It is true that, inthe earlier-mentioned comprehensive GSO survey, only about a third of



256

Table

1Im

pac

tof

the

2000

Ente

rprise

Law

onBus

ines

sFo

rmal

izat

ion

inVie

tnam

Off

icia

lDat

aPr

ovin

cial

Com

petit

iven

ess

Inde

xD

ata

2007

New

lyTo

talA

ctiv

ePC

I20

07Fi

rms

Est

ablis

hmen

tsR

egis

tere

dPr

ivat

eR

espo

nden

tsW

aitin

gov

erE

stab

lishm

ents

Beg

inni

ngG

apB

etw

een

Firm

s(M

inis

try

Firm

s(G

SOE

stab

lishe

dM

edia

nM

edia

nT

hree

Mon

ths

toB

egin

ning

Ope

ratio

nsas

Est

ablis

hmen

tof

Plan

ning

Ent

erpr

ise

Dur

ing

Day

sfo

rN

umbe

rof

Be

Fully

Leg

alO

pera

tions

asIn

form

alan

dR

egis

trat

ion

Yea

ran

dIn

vest

men

t)C

ensu

s)Y

ear

(#)

Reg

istr

atio

nL

icen

ses

Com

pani

es(%

)In

form

al(%

)(%

Self

-Dec

lare

d)(M

onth

s)

Pre-

Ent

erpr

ise

Law

1991

6160

253

13.8

93.8

68.8

43.2

019

923,

500

6430

510

.980

.069

.332

.11

1993

4,23

951

898

304

6.4

72.3

63.4

15.1

119

942,

843

924

100

304

13.8

67.7

60.6

14.6

819

954,

222

14,7

6291

303

14.0

72.9

66.7

5.60

1996

2,76

41,

486

100

305

11.5

70.6

66.1

9.46

1997

2,76

425

,027

124

305

10.3

73.3

65.8

6.54

1998

3,63

025

,965

103

305

11.7

64.1

58.3

6.24

1999

3,60

125

,440

153

254

7.0

70.3

59.4

5.40

Post

-Ent

erpr

ise

Law

2000

8,78

535

,004

457

153

8.0

63.5

59.6

3.09

2001

19,6

6244

,314

554

153

7.0

59.5

57.2

1.01

2002

17,9

9255

,236

619

153

8.1

60.7

57.9

0.83

2003

25,9

8264

,526

792

153

7.1

60.6

58.6

0.98

2004

48,3

3684

,003

1,06

815

36.

961

.859

.80.

8120

0539

,918

105,

169

1,09

315

37.

659

.757

.00.

3720

0639

,984

123,

392

803

152

5.8

62.4

59.6

0.90


household businesses had both registered with district authorities andfulfilled all obligations to pay an annual license fee and taxes. GivenVietnam’s administrative system and the fact that these firms were in-deed captured by the GSO survey, however, it is highly unlikely thateven the subset that have not registered are operating entirely under theradar of officials at even the subdistrict (i.e., ward) level. As a result,local officials can be seen as complicit in allowing small-scale enter-prises to opt out of registering with responsible higher-level authorities.

The basic argument behind this conception of formalization is that,in Vietnam, companies face a relatively more rigorous and transparentregulatory system than do household businesses. Officially, regulatoryresponsibilities for household businesses (ho kinh doanh ca the) andcompanies (doanh nghiep) are divided across two distinct administra-tive levels. Generally speaking, household businesses are overseen bydistrict-level authorities, while companies are the responsibility ofprovincial-level authorities. An exception is Vietnam’s largest cities,Ho Chi Minh City and Hanoi, where oversight of smaller companies isalso delegated to the district level.

As such, businesses that are of similar size and that undertake sim-ilar activities but are not of the same legal form are covered by differ-ent regulations, for example, with regard to environmental standards,fire prevention, and health insurance for workers (Vietbao.vn 2006). Atthe same time, similar-sized household businesses can be treated verydifferently depending on their location and the attitude of local author-ities. In general, the regulatory framework that covers the activities ofhousehold businesses is less comprehensive, less transparent, and lessstrictly implemented across localities than the one that covers compa-nies. With specific regard to differences across localities, for example,there is extensive anecdotal evidence that a 2004 decree requiring allbusinesses with more than ten workers or more than one business loca-tion to register as companies, not household businesses, has not beenenforced in certain parts of the country (Taussig and Pham 2004).

The Role of Provincial Government in Vietnam

There are two important elements to note about the formalization deci-sion. First, registration at the provincial level entails an entirely differ-ent legal regime than district-level oversight. The relative difference inthe quality of district versus provincial governance is not of concern inthis decision: officials are enforcing different rules and their activitiesare not directly comparable. Second, district officials are subordinate to



provincial leaders. Their appointments, budgets, and career prospectsare all determined by provincial-level benefactors. Although there issome variance in the level of authority that has been delegated to dis-trict officials by provinces, the quality of district-level governance isgenerally a reflection of provincial decisionmakers. As a result, the ap-propriate level of analysis of institutions is the provincial level ratherthan the district-province dyad level.

In this article, we test the hypothesis that the quality of provincialgovernance plays an important role in the formalization decision. It istrue that there are clear business benefits to formalization in Vietnam.Firms are better able to expand their operations through easier accessto export licenses, customs certificates, and opportunities to bid on lu-crative government contracts. Companies also have the legal right toopen up branches and operate beyond the bounds of their own district.In the case of more sophisticated corporate governance structures, suchas shareholding companies, formalization also provides the additionalprotection of limited liability and allows firms to take on additional eq-uity investors.

Formalization, however, also entails costs that must be weighedagainst these benefits. First, firms must absorb the direct costs of for-malization in terms of fees for licenses and waiting periods to legallystart operations. Second, formalization entails better documentation ofthe true extent of firms’ sales and profitability. More rigorous account-ing procedures can entail some learning costs and even necessitate em-ployment of a certified accountant.

These direct costs, while important, can be dwarfed by the indirectcosts of transparency. Provincial bureaucrats are now aware of the sizeof a firm’s operations. In poorly governed Vietnamese provinces, char-acterized by corruption and inadequate property rights protections, thisknowledge can be easily exploited by officials. Entrepreneurs with es-tablished arrangements with district-level officials may fear that open-ing themselves up to regulation by provincial government will only ex-acerbate the frequency and size of bribe payments. Thus, the quality ofgovernance matters a great deal for the cost-benefit calculus of an en-trepreneur considering formalization. The more restrained the “grab-bing hand” of provincial bureaucrats is, the more incentive a firm hasto formalize. In contrast, when corruption is high and property rightsare poorly protected, entrepreneurs may feel that they are better off re-maining in the gray economy.

Previous analyses have demonstrated wide variation in economicgovernance among Vietnamese provinces and also have shown that this

258 Out of the Gray


variance is critical in explaining key indicators of economic perfor-mance, such as growth in the number of private companies, firm in-vestment (Tran, Grafton, and Kompas 2008), foreign direct investment(Vu, Le, and Vo 2007; Riehl 2008; Malesky 2008), and economicgrowth (Malesky 2007). All of these analyses rely on an annual rank-ing of economic governance in Vietnam’s sixty-four provinces by theVietnam Chamber of Commerce and Industry and Vietnam Competi-tiveness Initiative called the Provincial Competitiveness Index (PCI).At its most basic level, the PCI is the collective voice of about 7,000domestic private firms.5 Private entrepreneurs’ opinions regarding eco-nomic governance in their provinces are collected in a twenty-page sur-vey. Responses are paired with relevant, published economic data toaddress perception biases and are combined to create provincial-levelscores.

The final outcome is a composite index ranking Vietnam’s sixty-fourprovinces according to the following ten dimensions of governance:

1. Entry costs: A measure of the time it takes a firm to register, thetime to receive all the necessary licenses needed to start a business, thenumber of licenses required to operate a business, and the perceived de-gree of difficulty to obtain all licenses/permits.

2. Land access and security of tenure: A measure combining twodimensions of the land problems confronting entrepreneurs—how easyit is to access land and the security of tenure once land is acquired.

3. Transparency and access to information: A measure of whetherfirms have access to the proper planning and legal documents neces-sary to run their businesses, whether those documents are equitablyavailable, whether new policies and laws are communicated to firmsand predictably implemented, and the business utility of the provincialWeb page.

4. Time costs of regulatory compliance: A measure of how muchtime firms waste on bureaucratic compliance after registration, as wellas how often and for how long firms must shut their operations downfor inspections by local regulatory agencies.

5. Informal charges: A measure of how much firms pay in infor-mal charges, how much of an obstacle those extra fees pose for theirbusiness operations, whether payment of those extra fees results in ex-pected results or “services,” and whether provincial officials use com-pliance with local regulations to extract rents.

6. Competition environment and state-owned enterprise (SOE)bias: A measure focusing on the perceived bias of provincial govern-



ments toward SOEs, equitized firms, and other provincial champions interms of incentives, policy, and access to capital.

7. Proactivity of provincial leadership: A measure of the creativ-ity and cleverness of provinces in implementing central policy, design-ing their own initiatives for private sector development, and workingwithin sometimes unclear national regulatory frameworks to assist andinterpret in favor of local private firms.

8. Business development services (BDS): A measure of provincialservices for private sector trade promotion, provision of regulatory in-formation to firms, business partner matchmaking, provision of indus-trial zones or industrial clusters, and technological services for firms.

9. Labor and training: A measure of the efforts by provincial au-thorities to promote vocational training and skills development forlocal industries and to assist in the placement of local labor withprovincial businesses.

10. Confidence in legal institutions: A measure of the private sec-tor’s confidence in provincial legal institutions; whether firms regardprovincial legal institutions as an effective vehicle for dispute resolutionor as an avenue for lodging appeals against corrupt official behavior. 6

More detail on the specific survey questions and indicators used inthe index construction is available in the annual reports.7 Table 2 showsthat subindexes in the index are all positively correlated; however, onlyfour of the subindex dyads have bivariate correlations greater than 60.Table 3 takes the analysis of underlying correlations a bit further,through the use of factor analysis, to assess the latent variables that un-dergird the correlations between subindexes.

Three significant factors are revealed to be responsible for about 64percent of the cumulative variance between subindexes. The first fac-tor contains the three most strongly correlated subindexes and two oth-ers that are generally concerned with postregistration policies and reg-ulation in the provincial business environment. Transparency, labor,BDS, proactivity, and time costs all involve the local-level policy ini-tiatives or decisions to implement those policy choices. The second fac-tor is clearly related to a general conception of property rights, consist-ing of the ability to access and the security of business premises (landaccess), the faith firms have that provincial courts will enforce con-tracts (confidence in legal institutions), and firm perceptions of the cor-ruption of provincial officials (informal charges). These subindexeshave less to do with implementation of policies and more to do with theformal restraints placed on the grabbing hand of bureaucrats. Good

260 Out of the Gray


261

Table

2Biv

aria

teCo

rrel

atio

nBet

wee

nGov

erna

nce

Subin

dex

es

Var

iabl

es1

23

45

67

89

1011

1213

1G

apbe

twee

nre

gist

ratio

n1

and

esta

blis

hmen

t(m

onth

s)2

Prov

inci

alec

onom

ic0.

0083

1go

vern

ance

3D

ista

nce

from

the

–0.0

209

0.21

23*

1se

vent

eent

hpa

ralle

l(km

)Su

bind

exes

4E

ntry

cost

s(b

usin

ess

0.01

0.39

78*

–0.1

864*

1re

gist

ratio

nan

dlic

ensi

ng)

5L

and

acce

ssan

dse

curi

ty–0

.014

20.

3791

*0.

3716

*0.

0400

*1

ofte

nure

6T

rans

pare

ncy

ofbu

sine

ss0.

0047

0.77

26*

0.05

59*

0.18

22*

0.23

18*

1in

form

atio

n7

Tim

eco

sts

ofre

gula

tory

0.04

33*

0.47

28*

0.08

99*

0.12

63*

0.06

96*

0.33

79*

1co

mpl

ianc

e8

Info

rmal

char

ges

0.00

770.

4364

*0.

1270

*0.

3012

*0.

4026

*0.

2006

*0.

1583

*1

(cor

rupt

ion)

9B

ias

tow

ard

stat

e-ow

ned

0.00

230.

5566

*–0

.023

90.

4336

*0.

1156

*0.

3814

*0.

0349

*0.

2792

*1

sect

or10

Proa

ctiv

ityof

prov

inci

al0.

0061

0.85

07*

0.26

06*

0.18

48*

0.36

74*

0.61

92*

0.25

90*

0.38

79*

0.46

86*

1le

ader

ship

11B

usin

ess

deve

lopm

ent

0.00

190.

8095

*0.

1205

*0.

1801

*0.

0384

*0.

7075

*0.

3549

*0.

1606

*0.

3417

*0.

6625

*1

serv

ices

12L

abor

trai

ning

and

0.00

050.

7755

*0.

0951

*0.

2276

*0.

2397

*0.

6287

*0.

2735

*0.

1601

*0.

3368

*0.

5496

*0.

6989

*1

mat

chm

akin

g13

Con

fide

nce

inle

gal

–0.0

138

0.52

03*

0.38

00*

0.14

42*

0.25

83*

0.16

28*

0.12

02*

0.17

92*

0.24

23*

0.49

71*

0.26

14*

0.30

10*

1in

stitu

tions

*Si

gnif

ican

tatt

he.0

5le

vel.


scores on property rights represent state retreat from intervention in theaffairs of private firms, as opposed to the first factor, where good scoresrepresent policy interventions. The final factor, responsible for thesmallest portion in cumulative variance, comprises two subindexes—entry costs and bias to the state sector—which address entry barriers toprivate entrepreneurs. Entry costs capture the direct financial costs ofentry, whereas SOE bias represents the implicit barriers to private sec-tor entry posed by the economic strength of the existing state sector orthe ideological convictions of provincial bureaucrats.

Our core analysis utilizes the final 100-point PCI index. However,in the final section, we perform a separate analysis by subindex in orderto explore which types of institutions are predominantly responsible forformalization decisions.

Figure 1 demonstrates the strong negative correlation between totaleconomic governance (the unweighted PCI) and the total level of in-formality in the economy, which is operationalized by the percentageof total business investment in 2006 accounted for by the household

262 Out of the Gray

Table 3 Factor Analysis of Governance Subindexes

Factor 1 Factor 2 Factor 3Variable Policies Property Rights Entry Costs Uniqueness

Land access and security 0.0684 0.8838 –0.0593 0.2107of tenure

Informal charges 0.0612 0.6881 0.3724 0.3841(corruption)

Confidence in legal 0.3276 0.4241 0.2274 0.6611institutions

Business development 0.9064 –0.0553 0.1074 0.1639services

Transparency of 0.835 0.0982 0.0916 0.2847business information

Labor training and 0.8054 0.095 0.1282 0.3259matchmaking

Proactivity of provincial 0.7404 0.4251 0.2119 0.2261leadership

Time costs of regulatory 0.5174 0.0385 –0.0988 0.7211compliance

Entry costs (business 0.0623 0.006 0.8587 0.2587registration and licensing)

Bias toward state- 0.2906 0.1525 0.7511 0.3282owned sector

Eigenvalue 3.83 1.50 1.10Cumulative variance 38.34% 53.37% 64.36%

explained

Note: Factor analysis performed with varimax rotation.


sector. The bivariate correlation is robust to a multiple regression withstructural control variables for infrastructure and provincial wealth, aswell as alternative definitions of informality.8

Data

To test which institutions, if any, facilitate the formalization process,we use data from the 2007 version of the Vietnam PCI Survey, a com-prehensive governance survey of 6,700 firms distributed across Viet-nam’s sixty-four provinces. The survey team randomly sampled from alist of private companies registered with each province’s Tax Authority.Stratification was based on firm size, age, and broad sector (agriculture,services, construction, and industry) in order to accurately reflect pop-ulations in each province. The result is the most comprehensive andmethodologically rigorous tool for assessment of the environment forprivate sector development in Vietnam to date. The subnational designof the study allows us significant variance on provincial endowments,


Figure 1 Relationship Between Formalization and Aggregate Economic Governance

Source: General Statistics Office (2006). Non-Farm Individual Business EstablishmentsCurrent and Solutions. Hanoi: Statistical Publishing House. Vietnam Provincial Competitive-ness Index 2007 (www.pcivietnam.org).

95% Confidence Interval

Fitted Values Based on 2007 Unweighted PCI

Actual Values

40 50 60 70 80Unweighted Provincial Competitiveness Index 2007

Hou

seho

ldIn

vest

men

t/Reg

iste

red

Priv

ate

Inve

stm

ent

020

4060

8010

0


even after holding constant the tricky differences in culture and historythat create problems for cross-national studies of the same issues.

One limitation of the dataset is that it consists only of those busi-nesses that eventually decided to formally register as companies. Itdoes not include household companies that never registered. As a re-sult, we have only found evidence of the influence of institutions on theformalization decision for a subset of this broader population at risk offormalization. We do show that our argument is robust to inclusion ofthe broader set of firms by demonstrating that institutions are correlatedwith aggregate formalization in each province. This test is admittedlyimperfect but does provide some indication of the robustness of ourfindings, in the absence of a random sample of household enterprises.A second issue is that our institutional measures generally postdate theformalization decisions we study. This problem is relatively minor, ascorrelation of scores is extremely high across the years that the measurehas been calculated, showing that institutions are sticky and do notchange rapidly over time. Furthermore, our instrumental variables helpsort out the causal process directly.

We proceed with data analysis in three main stages. First, we con-sider the role of the general local institutional environment in shapingan entrepreneur’s decision of whether a new business should be set upas a household business or as a private company formally registeredwith provincial authorities. Second, we look deeper at how this envi-ronment influences the subset of firms that do begin in the relatively in-formal household sector and the length of time until they transition intothe formal company sector. Finally, we consider which particular insti-tutions have the greatest impact on the timing of formalization by thislatter group.

Dependent Variables

The dependent variable of formal entry in the first stage of analysis isa simple dichotomous variable that equals “1” if the business either (1)answered no to a question about whether it had operated as a householdbusiness previous to registering as a company; or (2) did not report be-ginning operations in a year prior to the year it reported formally reg-istering as a company. For robustness, we also test a more conservativedummy based only on the first group.

For the second stage, we create a count variable for all firms thatscored a “1” on the broader of the first test dependent variables equalto the difference in months between year of establishment and the year

264 Out of the Gray


of registration. In the vast majority of cases, this count variable equals“0,” meaning that the difference between establishment and registrationhappened during the same year and so was less than a year. Neverthe-less, a substantial number of firms did have significant wait periods,leading to an average wait of 13.4 months and a maximum of fifteenyears.

Explanatory Variables

The first and second stages of analysis focus on the influence of thebroader institutional environment on formalization. We test this by uti-lizing each province’s unweighted 2007 PCI score as our explanatoryvariable. Each of the ten PCI subindexes consists of both “hard” and“soft” measures. By hard measures, we mean statistics that reflect aparticular outcome of the institutional environment and are availablefrom published sources such as the GSO or Ministry of Finance. Softmeasures, in turn, are those based on entrepreneurs’ own subjective re-sponses to survey questions about their institutional environment.

Measures of institutional development generally fall into one ofthese two categories. Hard measures are the more objective of the two,in that they are less subject to the biases of individual respondentsacross different localities and different industries. But, being more pre-cisely defined, hard measures are also more likely to be incompletemeasures of the reality of the institutional environment. In the case ofa hard measure of particular legislation, for example, it is usually notpossible to take into account the full spectrum of degrees of imple-mentation. Similarly, in the case of a hard measure that measures thefrequency of a particular legal action or outcome, issues of quality ofaction or outcome are neglected. These weaknesses in hard measurescan be addressed through the use of soft measures—that is, qualitativesurveys of actors in the field—that gauge the business community’stake on the actual implementation and overall effect of governmentpolicies. Because both hard and soft data have strengths and weak-nesses, the most appropriate methodology should clearly distinguishbetween the two but consider both (Woodruff 2006).

After establishing the general importance of the institutional envi-ronment to the formalization decision through use of the aggregateindex, we conclude with initial consideration of which specific institu-tions are most relevant. We do this by replicating our earlier regressionon time spent in the informal sector, successively replacing the aggre-gated PCI with each of its individual subindexes.



Control Variables

Our analysis controls for a series of firm-, entrepreneur-, and location-level characteristics that could be thought to influence the formaliza-tion decision.

At the level of the firm, we include a control for the timing of es-tablishment, which in some regressions we represent with years sinceestablishment and in other cases represent as years since establishmentpaired with its squared term to account for the possibility that institu-tional change has been more rapid in recent years.9 We further controlfor a firm’s status as an exporter (or not), the zoning of the land onwhich it operates (residential or not), equity capital at time of estab-lishment, broad economic sector, and ISIC code. Broad economic sec-tor refers to dummies for classification as primarily agriculture, indus-try, services, or construction, while ISIC codes are considered at boththe one- and two-digit levels.

The entrepreneur characteristics we consider relate to past personalexperience working for the state. Considering the highly political natureof regulation in Vietnam, such connections have been shown to provideadvantages to entrepreneurs (Nguyen et al. 2004; O’Conner 2000; Web-ster and Taussig 1999). These controls include a dummy for whether anentrepreneur is a former government bureaucrat or SOE manager andanother for whether the entrepreneur is a former SOE employee.

In addition to our key explanatory variables, we include a numberof other location-based control variables. To control for infrastructureand socioeconomic conditions, we control for telephones per capita,percentage of population that represents secondary school graduates,share of the population composed of people not of the majority Kinhethnicity, and GDP per capita; we also include a dummy for provincialstatus cities. To consider the effect of Vietnam’s unique history, we in-troduce a dummy for location in what was once commonly known asSouth Vietnam and regional-fixed effects.10

Results

The decision tree of formalization for entrepreneurs in Vietnam hasthree nodes. First, business owners must decide whether to start opera-tions as a formal or household enterprise. Entrepreneurs who begin inthe household sector must next decide whether or not to move from theinformal sector. If they opt in favor of such a movement, how long will

266 Out of the Gray


they wait until formalization? In this section, we test direct firm-levelevidence on the initial start-up decision and the length of time spent inthe household sector.11

Analysis of Initial Start-up Decision

In Table 4, using a probit model, we test the likelihood that an individ-ual entrepreneur will choose to begin initial operations in the formal sec-tor. Model 1 contains the key causal variable of economic governance,measured by the PCI 2007 along with firm-level controls. Model 2 thenadds information on the personal characteristics of entrepreneurs. Mod-els 3 and 4 add in location-specific controls. Model 5 addresses unob-served heterogeneity with sector-fixed effects, based on whether thefirm operated primarily in agriculture, services, construction, or indus-try. Models 6 and 7 substitute more refined industry-fixed effects usingthe ISIC 1- and 2-digit coding systems. Model 8 adds regional-fixed ef-fects, while Model 9 replaces entry timing control variables with estab-lishment year–fixed effects. For robustness, we also include Model 10,which replaces the main dependent variable with a new variable basedon the narrower definition of each firm’s self-assessment of whether itbegan as a household enterprise.

Our key explanatory variable for provincial economic governancehas a significantly positive effect on selection into the formal sector ineach of the seven models. In the fully specified Model 5, a 10-point in-crease in economic governance increases the probability of formalentry by 2.5 percent. This is consistent with our general expectationthat better institutions will encourage greater participation in the formalsector.

Our firm-level control for equity capital at establishment showsthat formality is more likely when entrepreneurs start up larger firms.Start-up capital has a strong positive impact, below the 0.01 level ofstatistical significance, in each model. In contrast, we find 12.5 timeslower likelihood that firms operating on land zoned for residential pur-poses will choose to start up in the formal sector. This is consistent withthe idea that businesses that start out of the home are likely to begin ina more cautious, exploratory fashion than those that are set up on landspecifically designated for business. To our surprise, we find no signif-icant relationship in any of the models between a firm’s status as an ex-porter and the odds that it entered as a formal enterprise.

Personal connections to the state significantly increase the oddsthat an entrepreneur will select into the formal sector. This is true



268

Table

4Im

pac

tof

Prov

incial

Econ

omic

Gov

erna

nce

onDec

isio

nto

Ente

rth

eM

arke

tFo

rmal

ly

Dep

ende

ntva

riab

le:

Prob

itA

naly

sis

Ent

ryas

afo

rmal

firm

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

Prov

inci

al0.

0037

4***

0.00

252*

0.00

203

0.00

251*

0.00

255*

*0.

0025

9**

0.00

226*

0.00

179*

0.00

262*

*0.

0032

1**

econ

omic

(0.0

0125

)(0

.001

42)

(0.0

0133

)(0

.001

30)

(0.0

0128

)(0

.001

29)

(0.0

0129

)(0

.000

918)

(0.0

0126

)(0

.001

61)

gove

rnan

ceY

ears

sinc

efi

rm0.

0026

40.

0125

0.01

170.

0119

0.00

939

0.01

53*

0.01

61*

0.01

53*

esta

blis

hmen

t(0

.005

99)

(0.0

0781

)(0

.007

73)

(0.0

0775

)(0

.007

82)

(0.0

0854

)(0

.009

06)

(0.0

0864

)Y

ears

sinc

e–0

.001

43**

*–0

.002

10**

*–0

.002

08**

*–0

.002

09**

*–0

.001

97**

*–0

.002

23**

*–0

.002

34**

*–0

.002

21**

*es

tabl

ishm

ent

(0.0

0040

4)(0

.000

501)

(0.0

0050

3)(0

.000

503)

(0.0

0049

9)(0

.000

545)

(0.0

0058

6)(0

.000

557)

squa

red

Ent

erpr

ise

law

–0.0

782*

–0.0

892*

–0.0

929*

–0.0

932*

–0.0

971*

*–0

.091

6*–0

.109

**–0

.094

3**

dum

my

(0.0

417)

(0.0

487)

(0.0

491)

(0.0

492)

(0.0

483)

(0.0

480)

(0.0

497)

(0.0

476)

Ow

ner

isa

form

er0.

558*

**0.

580*

**0.

580*

**0.

580*

**0.

578*

**0.

581*

**0.

582*

**0.

582*

**0.

578*

**0.

602*

**go

vern

men

t(0

.016

2)(0

.015

9)(0

.016

0)(0

.015

9)(0

.016

3)(0

.016

3)(0

.016

8)(0

.016

2)(0

.016

2)(0

.012

7)of

fici

alor

SOE

man

ager

Ow

ner

isa

form

er0.

616*

**0.

627*

**0.

627*

**0.

628*

**0.

625*

**0.

631*

**0.

632*

**0.

633*

**0.

627*

**0.

642*

**SO

Eem

ploy

ee(0

.012

4)(0

.013

7)(0

.013

6)(0

.013

5)(0

.014

1)(0

.013

6)(0

.014

7)(0

.013

9)(0

.013

3)(0

.011

4)E

xpor

ting

firm

0.01

990.

0125

0.01

140.

0308

0.03

410.

0395

0.03

380.

0341

0.00

179

(0.0

349)

(0.0

312)

(0.0

310)

(0.0

308)

(0.0

319)

(0.0

316)

(0.0

313)

(0.0

331)

(0.0

345)

Equ

ityca

pita

l0.

0268

***

0.02

72**

*0.

0279

***

0.02

39**

*0.

0304

***

0.02

45**

*0.

0288

***

0.03

06**

*0.

0374

***

attim

eof

(0.0

0749

)(0

.007

60)

(0.0

0758

)(0

.007

92)

(0.0

0740

)(0

.007

99)

(0.0

0718

)(0

.007

51)

(0.0

0855

)es

tabl

ishm

ent

Firm

oper

ates

–0.1

20**

*–0

.116

***

–0.1

17**

*–0

.125

***

–0.1

11**

*–0

.115

***

–0.1

10**

*–0

.115

***

–0.1

36**

*on

resi

dent

ial

(0.0

154)

(0.0

163)

(0.0

163)

(0.0

162)

(0.0

169)

(0.0

163)

(0.0

168)

(0.0

171)

(0.0

207)

purp

oses

land

Tele

phon

espe

r0.

284*

**0.

384*

**0.

406*

**0.

369*

**0.

366*

**0.

320*

**0.

357*

**0.

608*

**ca

pita

1995

(0.0

702)

(0.0

951)

(0.0

950)

(0.0

876)

(0.0

840)

(0.0

472)

(0.0

865)

(0.0

629)

Seco

ndar

ysc

hool

-0.0

0008

500.

0002

800.

0004

140.

0001

13–0

.000

139

–0.0

0010

0–0

.000

292

–0.0

0279

grad

uate

s20

06(0

.001

68)

(0.0

0184

)(0

.001

82)

(0.0

0184

)(0

.001

97)

(0.0

0174

)(0

.001

41)

(0.0

0297

)

cont

inue

s


269

Table

4Co

ntin

ued

Dep

ende

ntva

riab

le:

Prob

itA

naly

sis

Ent

ryas

afo

rmal

firm

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

Perc

enta

geof

0.00

376

0.00

378

0.00

121

0.00

322

–0.0

0253

–0.0

0093

70.

0008

82–0

.005

94m

inor

ities

in(0

.007

06)

(0.0

0717

)(0

.007

27)

(0.0

0719

)(0

.007

87)

(0.0

108)

(0.0

0696

)(0

.009

36)

prov

ince

Prov

ince

loca

ted

sout

h0.

0063

10.

0245

0.03

020.

0205

0.02

82–0

.005

34of

17th

para

llel

(0.0

281)

(0.0

288)

(0.0

292)

(0.0

282)

(0.0

295)

(0.0

355)

Dis

tanc

efr

omH

anoi

–0.0

0000

858

–0.0

0004

78–0

.000

0606

–0.0

0005

71–0

.000

0637

–0.0

0008

33an

dH

oC

hiM

inh

(0.0

0003

48)

(0.0

0003

80)

(0.0

0003

88)

(0.0

0003

74)

(0.0

0003

95)

(0.0

0005

25)

City

(km

)G

DP

per

capi

ta(l

n)–0

.040

2**

–0.0

381*

*–0

.045

3***

–0.0

395*

*–0

.059

6***

–0.0

314*

–0.0

726*

**(0

.017

5)(0

.017

7)(0

.016

5)(0

.018

4)(0

.015

3)(0

.016

1)(0

.020

3)Se

ctor

-fix

edef

fect

sno

nono

noye

sno

nono

nono

ISIC

-fix

edef

fect

sno

nono

nono

yes

noye

sye

sye

sIS

IC2

digi

t–fi

xed

nono

nono

nono

yes

nono

noef

fect

sE

stab

lishm

enty

ear–

nono

nono

nono

nono

yes

yes

fixe

def

fect

sR

egio

nal-

fixe

def

fect

sno

nono

nono

nono

yes

nono

Obs

erva

tions

6162

5103

5103

5103

5103

5062

5021

5062

5008

4416

Prov

inci

alcl

uste

rs64

6464

6464

6464

6464

64Ps

eudo

R-s

quar

ed0.

246

0.27

80.

280

0.28

00.

286

0.28

90.

302

0.29

20.

286

0.32

5L

oglik

elih

ood

–305

9–2

433

–242

6–2

424

–240

4–2

376

–230

9–2

366

–236

7–2

025

Not

es:

Coe

ffic

ient

sre

pres

entm

argi

nalp

roba

bilit

ies

with

robu

stst

anda

rder

rors

,clu

ster

edat

prov

inci

alle

vel(

inpa

rent

hese

s).*

**p

<0.

01,*

*p

<0.

05,*

p<

0.1

Dep

ende

ntva

riab

le:F

orm

ality

isco

ded

as0

iffi

rmco

mm

ence

dbu

sine

ssop

erat

ions

befo

rere

gist

ratio

nan

d1

othe

rwis

e.M

odel

1co

ntai

nsth

eke

yca

usal

vari

able

ofec

onom

icgo

vern

ance

,m

easu

red

byth

ePC

I20

07,

and

cont

rols

for

the

timin

gof

esta

blis

hmen

tan

dow

ner

back

grou

nd;

Mod

el2

adds

infi

rm-l

evel

cont

rols

;M

odel

3ad

dsin

loca

tion-

spec

ific

cont

rols

,not

incl

udin

gG

DP

per

capi

ta;

Mod

el4

adds

inG

DP

per

capi

ta;

Mod

el5

addr

esse

sun

obse

rved

hete

roge

ne-

ityw

ithbr

oad

sect

or-f

ixed

effe

cts;

Mod

els

6an

d7

subs

titut

em

ore

refi

ned

indu

stry

-fix

edef

fect

sus

ing

the

ISIC

1-an

d2-

digi

tcod

ing

syst

ems;

Mod

el8

uses

the

ISIC

1in

dust

ryco

n-tr

olan

dad

dsre

gion

al-f

ixed

effe

cts;

and

Mod

el9

repl

aces

entr

ytim

ing

cont

rolv

aria

bles

with

esta

blis

hmen

tyea

r–fi

xed

effe

cts.

Fina

lly,M

odel

10re

runs

Mod

el9,

repl

acin

gth

ede

-pe

nden

tvar

iabl

ew

itha

firm

’sse

lf-a

sses

smen

tof

whe

ther

itbe

gan

asa

hous

ehol

den

terp

rise

.


whether the connection is experience as a government bureaucrat orSOE manager or, on a lower level, as a former SOE employee. This isconsistent with a number of possible explanations. First and foremost,there is evidence that such connections are useful in accessing scarcebusiness resources needed for growth in Vietnam’s still imperfect busi-ness environment (Malesky and Taussig 2008). Second, in some indus-tries, state employment was, in the recent past, still the best place to ac-cumulate relevant experience for starting a growth-oriented business.And finally, for those entrepreneurs who left state employment of theirown will, only a growth-oriented business opportunity is likely to havetempted them away from the stability of state employment. Results forthese entrepreneur-level variables are again strongly significant acrossall models and substantively important. SOE managers and employeesare 58 percent and 62 percent more likely to begin operations formally.

Only two of the location-based control variables prove to have anyinfluence on our probit analyses. In Models 2 through 7, higher num-bers of telephones per capita in a firm’s province consistently lead to ahigher likelihood of that firm’s choosing formal status upon entry. Incontrast, across the same set of models, higher logged per capita GDPmakes the choice of formality less likely. This result is somewhat sur-prising, given the reverse causality expectation that higher levels of for-mality would mean higher levels of growth-oriented businesses andtherefore higher levels of wealth. Higher levels of wealth, however,also likely mean greater demand for local services and, holding institu-tions constant, may make it less worthwhile for authorities to closelyregulate smaller businesses. We find no significance at all for the effectof secondary education levels, share of minorities, or location in thehistorically more market-oriented southern half of the country.

Analysis of Wait Between Establishment and Formalization

Our second test considers the relationship between provincial institu-tions and the length of time a firm spent as a household enterprise be-fore making the formalization decision. In Table 5, we analyze the sub-sample of firms that did make the original decision to begin operationsin the informal sector. In sum, it appears that the impact of governanceinstitutions on formalization is statistically significant, substantivelylarge, and robust across a range of different specifications.

270 Out of the Gray


271

Table

5Ef

fect

ofEc

onom

icGov

erna

nce

onth

eGap

Bet

wee

nFi

rmEs

tablis

hmen

tan

dReg

istr

atio

n

Dep

ende

ntva

riab

le:

Gap

inm

onth

sbe

twee

nre

gist

rati

onO

LS

Neg

ativ

eB

inom

ialR

egre

ssio

n

and

esta

blis

hmen

t(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)

Prov

inci

al–0

.175

**–0

.184

*–0

.022

1**

–0.0

214*

*–0

.018

8*–0

.034

9***

–0.0

461*

**–0

.041

4***

–0.0

526*

**–0

.045

6***

econ

omic

(0.0

811)

(0.0

949)

(0.0

0873

)(0

.010

7)(0

.010

6)(0

.012

7)(0

.010

1)(0

.011

5)(0

.013

6)(0

.015

4)go

vern

ance

Yea

rssi

nce

firm

1.10

61.

658

0.24

4***

0.31

8***

0.30

4***

0.30

0***

0.33

0***

0.30

5***

esta

blis

hmen

t(0

.977

)(1

.088

)(0

.050

8)(0

.034

4)(0

.036

0)(0

.036

6)(0

.038

2)(0

.036

2)Y

ears

sinc

e0.

192*

**0.

181*

**–0

.002

26**

–0.0

0343

***

–0.0

0319

***

–0.0

0315

***

–0.0

0351

***

–0.0

0322

***

esta

blis

hmen

t(0

.030

5)(0

.032

9)(0

.000

900)

(0.0

0053

4)(0

.000

569)

(0.0

0058

6)(0

.000

617)

(0.0

0058

6)sq

uare

dE

nter

pris

ela

w12

.02*

**13

.93*

**–0

.883

***

–0.5

47**

–0.5

82**

–0.6

06**

–0.6

45**

*–0

.721

***

dum

my

(3.5

74)

(3.8

09)

(0.2

86)

(0.2

34)

(0.2

42)

(0.2

37)

(0.2

49)

(0.2

47)

Equ

ityca

pita

lat

2.76

2***

0.05

010.

0298

0.04

340.

176*

*0.

111

0.15

4**

0.14

6*tim

eof

(0.6

45)

(0.0

573)

(0.0

634)

(0.0

646)

(0.0

779)

(0.0

704)

(0.0

760)

(0.0

785)

esta

blis

hmen

tE

xpor

ting

firm

5.28

6***

-0.0

657

-0.0

0411

-0.0

313

0.14

50.

199

0.42

60.

626*

(1.6

85)

(0.2

08)

(0.2

18)

(0.2

22)

(0.2

40)

(0.2

61)

(0.3

09)

(0.3

72)

Firm

oper

ates

on–3

.455

***

–0.1

16–0

.107

-0.0

909

0.00

175

–0.0

501

0.05

52–0

.008

00ho

useh

old

land

(1.0

48)

(0.1

67)

(0.1

69)

(0.1

78)

(0.1

75)

(0.1

76)

(0.1

79)

(0.1

89)

Ow

ner

isa

form

er9.

828*

**0.

891*

**0.

878*

**1.

037*

**0.

868*

**0.

917*

**0.

887*

**go

vern

men

t(3

.347

)(0

.198

)(0

.207

)(0

.229

)(0

.207

)(0

.234

)(0

.241

)of

fici

alor

SOE

man

ager

Ow

ner

isa

form

er7.

777*

*1.

175*

**1.

246*

**1.

078*

**1.

154*

**1.

434*

**1.

559*

**SO

Eem

ploy

ee(3

.243

)(0

.394

)(0

.403

)(0

.345

)(0

.365

)(0

.470

)(0

.554

)Te

leph

ones

per

–5.2

660.

0768

–1.7

11**

–1.5

53*

–2.1

35*

–1.5

01ca

pita

1995

(9.2

44)

(0.6

06)

(0.7

70)

(0.8

02)

(1.2

37)

(1.0

64)

Seco

ndar

ysc

hool

–0.1

05–0

.002

770.

0088

3–0

.002

00–0

.000

574

0.00

686

grad

uate

s20

06(0

.124

)(0

.013

2)(0

.013

4)(0

.015

4)(0

.017

7)(0

.020

1)co

ntin

ues


272

Table

5Co

ntin

ued

Dep

ende

ntva

riab

le:

Gap

inm

onth

sbe

twee

nre

gist

rati

onO

LS

Neg

ativ

eB

inom

ialR

egre

ssio

n

and

esta

blis

hmen

t(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)

Perc

enta

geof

–0.6

11–0

.171

**–0

.084

6–0

.079

9–0

.148

0.07

61m

inor

ities

in(0

.478

)(0

.077

3)(0

.069

5)(0

.077

5)(0

.091

3)(0

.148

)pr

ovin

cePr

ovin

celo

cate

dso

uth

–1.7

530.

0225

0.06

460.

0031

40.

0884

of17

thpa

ralle

l(1

.515

)(0

.188

)(0

.208

)(0

.249

)(0

.297

)N

atio

nal-

leve

lcity

2.48

1–0

.210

–0.0

947

0.02

49–0

.023

6(3

.006

)(0

.179

)(0

.191

)(0

.337

)(0

.324

)G

DP

per

capi

ta(l

n)–0

.376

0.40

1***

0.38

6***

0.44

2***

0.32

2**

(1.0

63)

(0.0

829)

(0.1

02)

(0.1

03)

(0.1

27)

Dis

tanc

efr

omth

e17

thpa

ralle

lC

onst

ant

–9.9

91–3

.018

1.50

0**

0.59

40.

453

1.95

7-2

.935

-0.8

346.

550*

**5.

082*

*(6

.675

)(1

5.35

)(0

.724

)(0

.742

)(0

.743

)(1

.417

)(1

.810

)(2

.059

)(1

.950

)(2

.027

)Se

ctor

-fix

edef

fect

sno

nono

nono

noye

sno

nono

ISIC

-fix

edef

fect

sno

yes

nono

nono

noye

sye

sye

sE

stab

lishm

enty

ear–

nono

nono

nono

nono

yes

yes

fixe

def

fect

sR

egio

nal-

fixe

def

fect

sno

nono

nono

nono

nono

yes

Obs

erva

tions

3726

3078

3726

3097

3097

3097

3097

3078

3078

3078

Prov

inci

alcl

uste

rs64

6464

6464

6464

6464

64L

oglik

elih

ood

–170

30–1

4112

–464

9–4

017

–401

7–4

017

–401

7–3

970

–243

3–3

970

Roo

tmea

nsq

uare

d23

.39

23.8

1er

ror

R-s

quar

ed0.

805

0.81

7

Not

es:

Rob

usts

tand

ard

erro

rs,c

lust

ered

atpr

ovin

cial

leve

l(in

pare

nthe

ses)

.***

p<

0.01

,**

p<

0.05

,*p

<0.

1D

epen

dent

vari

able

isde

fine

das

the

gap

inm

onth

sbe

twee

nes

tabl

ishm

enta

ndre

gist

ratio

n.M

odel

1is

anO

LS

Mod

elw

ithth

eke

yca

usal

vari

able

ofec

onom

icgo

vern

ance

,mea

sure

dby

the

PCI

2007

,and

cont

rols

for

the

timin

gof

entr

y;M

odel

2ad

dsow

ner-

,fir

m-,

and

prov

inci

al-l

evel

cont

rols

,as

wel

las

indu

stry

-fix

edef

fect

s;M

odel

s3–

10ar

ene

gativ

ebi

nom

ialr

egre

ssio

ns.M

odel

3in

clud

esth

eke

yca

usal

vari

able

and

cont

rols

for

timin

gof

entr

y;M

odel

4ad

dsin

firm

-lev

elco

ntro

ls;M

odel

5ad

dsco

ntro

lsfo

row

ner

back

grou

nd;M

odel

6ad

dspr

ovin

cial

-lev

elco

ntro

ls;M

odel

7ad

dsco

ntro

lsfo

rre

gion

,urb

anst

atus

,and

aver

age

wea

lth,a

sw

ella

sbr

oad

sect

orfi

xed

effe

cts;

Mod

el8

uses

ISIC

indu

stry

-fix

edef

fect

s;M

odel

9re

plac

esen

try

timin

gva

riab

les

with

esta

blis

hmen

tyea

r–fi

xed

effe

cts;

Mod

el10

adds

regi

onal

fixe

def

fect

s.


Model 1, an OLS model including only economic governance asthe key explanatory variable and controls for the general improvementin the national regulatory environment over time, provides initial sup-port for our hypothesis that higher-quality institutions will reduce thetime a firm spends in the informal sector before formalization. Thisfinding survives the addition of provincial-level controls and industry-fixed effects in Model 2.

OLS assumes continuous, normally distributed dependent vari-ables, not count variables as is the case in this analysis. As such, wefind more robust support for our hypothesis with Models 3 through 10,which apply a negative binomial (nbreg) methodology that accounts forthe unique properties of count variables with substantial numbers ofzero values.12 Results do not change substantially in response to addingsector-fixed effects in Model 7, age and industry–fixed effects in Model 8, and regional-fixed effects in Model 9. In sum, a 10-point im-provement on the 100-point unweighted economic governance indexyields a reduction of approximately half a month spent in the informalsector.

This point estimate is highly contingent on whether the firm choseto formalize before or after the Enterprise Law. Those registered beforethe Enterprise Law spent on average an extra fourteen months in the in-formal sector. Figure 2 divides the predicted effect of good governancebetween firms registered before and after the regulatory changes intro-duced by the Enterprise Law. Provinces with good governance are con-servatively defined as those with index scores above the median of58.5. Using this delineation, we find that both before and after the En-terprise Law, firms in well-governed provinces spent significantly lesstime in the informal sector. Before the Enterprise Law, the gap betweenwell-governed and poorly governed locations was an astonishingtwenty-six months. After the law, the gap decreased substantially to 0.6months (about five days) but remained statistically significant at the0.05 level.

The most consistent and strongly significant result among the con-trol variables is that entrepreneurs with backgrounds in state employ-ment spend substantially more time in the informal sector. This is in-teresting, in that it is the opposite relationship with informality from theone found at the start of operations. This contrast implies that there aretwo distinct types of connected entrepreneurs. The first group is madeup of those who use their connections primarily as means for accessingscarce business resources. These entrepreneurs are most likely to beginoperations as formal companies. The second group consists of those



who use their connections primarily to minimize government regula-tion of their operations. These would be the entrepreneurs who choosethe informal sector and stay there for a relatively longer period of time.These same entrepreneurs may also be more risk averse, taking longerto leave their state positions, preferring to maintain both the stability oftheir government jobs and to, at first, just supplement this income withminimal rents from their informal businesses. At this point, our data donot allow us to distinguish between these two distinct groups, but theseresults call out for future work.

274 Out of the Gray

Figure 2 Predicted Gap Between Registration and Establishment (by level of governance and age of firm)

Note: Low and High are defined relative to the median score (58.49) on the 2007 VietnamProvincial Competitiveness Index. The Enterprise Law was formally promulgated in 2000.Firms established in 2000 or later are considered after the law. Results are the predicted mar-ginal effects from Table 5, Model 8 using Clarify (King, Tomz, and Wittenberg 2000); 95 per-cent confidence intervals are drawn around each point estimate.


The only province-level control that is consistently significant inthe negative binomial models is GDP per capita, which always has asubstantively large and statistically significant coefficient. This is con-sistent with our probit results, indicating that, keeping the quality oflocal institutions constant, markets with larger demand and likelygreater numbers of larger firms make greater amounts of informalitymore likely. The negative binomial models also provide limited evi-dence that better physical infrastructure, proxied for by more tele-phones, means less time in the informal sector, as does a greater shareof minorities in the province. Other province-level controls are also in-significant with the negative binomial specification.

Identification Strategy for Isolating the Effect of Institutions on Formalization

We address the issues of endogeneity by implementing an instrumentalvariable approach. The objective is to bolster our causal argument byguarding against two possibilities. First, there is reason to suspect re-verse causality. Provinces with higher levels of prior formalization willhave a larger tax base to draw on for improving government capacity.As a result, the causal arrow may run from formalization to institutionalquality (Kurtz and Shrank 2007). If this is true, formalization wouldlead to good governance or at least the perception of good governancefrom PCI respondents. While reverse causality is unlikely, as even for-mal firms represent a relatively small share of current provincial bud-gets, the cross-sectional nature of our panel does not allow us to im-mediately dismiss the possibility.

Second, while our within-country design diminishes this problemto a large extent, we must still guard against unobserved heterogeneity.Underlying cultural and historical factors at the province level mayplay a role in encouraging both formalization and more rules-basedgovernance on the part of local officials. In Vietnam, the North’s andSouth’s very different histories with capitalist business activity is a par-ticular concern. It has been argued that politicians in the South mayhave had greater experience with entrepreneurial activity and thereforeare more likely to construct institutions conducive to its development(Turley and Womack 1998; Nguyen et al. 2004). In this case, bothgreater formalization and good governance would be influenced by his-torical experience. The models we have cited may be misattributing


04_JEAS9.2_Malesky.qxd 5/14/09 11:13 AM 5

causation. A simple southern dummy would be too blunt to address theproblem if the bivariate correlation between the continuous measuresof formality and governance was sufficiently strong.

To ensure that our results survive these two forms of endogeneitybias, we use a two-staged instrumental variables approach to ensure thatexogenously determined economic governance is correlated with the de-cision of individual entrepreneurs to formalize their operations. To dothis, we first need an instrument that is strongly correlated with economicgovernance but is not correlated with the error term in the second-stageequation predicting formalization. Doing so will allow us to solve a mod-ified form of the traditional regression equation, where we can be surethat the causal chain runs through governance to formalization.

As our instrument, we use the distance for each province between itscapital and the infamous seventeenth parallel. The seventeenth parallelwas quite arbitrarily chosen at the Geneva Accords in 1954 as the borderbetween the two new countries that would be known as North Vietnamand South Vietnam. In choosing this as our instrumental variable, webuild on a logic introduced by Edward Miguel and Gerard Roland (2006)and also followed by Matthew Kocher, Thomas Pepinsky, and StathisKalyvas (2008). Both papers argue evidence that damage from war grewas one approached the former border that once divided Vietnam, show-ing specifically that distance to the seventeenth parallel is a strongly sig-nificant predictor of bombing intensity during the war. Miguel andRoland, in particular, use the seventeenth parallel as an instrument for theintensity of bombing in an effort to show that bombing had minimallong-term impact on future poverty across Vietnam.

We do not disagree with the Miguel and Roland empirics. Theydemonstrate persuasively that the seventeenth parallel is associatedwith bombing intensity and that exogenous bombing is not associatedwith spatial variance in poverty alleviation. Nevertheless, we do dis-agree slightly with the causal implications of their argument. The lim-ited variance in poverty has less to do with similar growth and indus-trial production rates across bombed and unbombed regions than itdoes with an explicit effort by the Vietnamese government to transferfunds to war-damaged and poor regions (World Bank 2002). Migueland Roland (2006, 21) concede this point at the end of the paper whenthey analyze the correlation between distance from the seventeenth par-allel and state investment between 1975 and 1986.

The high level of transfers continues even to this day. Figure 3shows the correlation between distance and average transfers per100,000 citizens between 2000 and 2006. Panel A shows that the dis-

276 Out of the Gray


tance is negatively correlated with transfers, a relationship that is sig-nificant at the 0.01 level. But this relationship includes the poor, un-derdeveloped northwestern region of Vietnam (marked with circles),which under the formula-based transfer system also receives a dispro-portionately high share of programmatic block grants. Dropping thenorthwest region further enhances the already strong correlation. De-pendence on investment from SOEs shows a similar pattern (see Figure4). Provinces close to the seventeenth parallel had higher-level SOEsoperating well into the reform era (Panel A) and have been muchslower about privatizing or liquidating these operations (Panel B).Malesky (2009) demonstrates that due to the unique Vietnam financialsystem, which provides incentives for favoring a dominant sector,provinces with large state sectors tend to have governance institutionsthat are biased against private actors.

In sum, damage during the war led provincial governments to bemore dependent on budget transfers and state resources from the cen-


Figure 3 Relationship Between Distance from the 17th Parallel and Fiscal Transfers

Note: Circles in Panel A represent northwestern provinces, while triangles are fitted val-ues for all other provinces in the country. Shaded region represents a 95 percent confidenceinterval. Bivariate correlations depicted are –.33 (significant at 0.01 level) for Panel A and .61(significant at 0.01 level) for Panel B. Data on transfers is from the 2000–2006 annual budg-ets. See www.mof.gov.vn.

Ave

rage

Tran

sfer

spe

r10

0,00

0C

itize

ns20

00–2

006

(ln)

Ave

rage

Tran

sfer

spe

r10

0,00

0C

itize

ns20

00–2

006

(ln)Entire Country Without Northern Mountains

Distance from the 17th Parallel (km)


tral government. This, in turn, led them to be more susceptible to thechanging whims of factionalized political leadership throughout Viet-nam’s ongoing experiment with market reforms, and also less likely toengage in their own initiatives to build up local market institutions.Miguel and Roland’s finding that damage from bombing does not helppredict present-day poverty levels is evidence of the serious, and in-deed quite effective, efforts by Vietnam’s central government to rebuildthese parts of the country and to acknowledge their sacrifice. This logicthat regions close to the seventeenth parallel were less likely to developeffective local market institutions is reinforced by the Kocher et al.(2008) discovery that more heavily bombed regions tended to bestronger bastions for the National Front for the Liberation of SouthVietnam (Viet Cong), which later translated into greater ideologicalsupport for the Communist government in Hanoi.

As Table 2 shows, preliminary evidence for use of distance fromthe seventeenth parallel is compelling. The bivariate correlation be-tween distance and the PCI is 0.21 and significant at the 0.05 level,

278 Out of the Gray

Figure 4 Relationship Between Distance from the 17th Parallel and Dependence on State Sector

Note: Triangles represent fitted values for all provinces in the country. Shaded region de-notes a 95 percent confidence interval. Bivariate correlations depicted are –0.20 (significantat 0.1 level) for Panel A and 0.45 (significant at 0.05 level) for Panel B. Data on enterprisesare from the General Statistical Office Enterprise Census (www.gso.gov.vn).

Change in State Sector 2000–2007

Num

ber

ofS

tate

-Ow

ned

Ent

erpr

ises

in20

00(ln

)

Distance from the 17th Parallel (km)

State Sector Size at Maximum

Per

cent

age

Cha

nge

inN

umbe

rof

Sta

teF

irms

Sin

ce20

00


while the correlation between distance and time spent in the informalsector is –0.02 and not significantly different from zero. This is tenta-tive confirmation that our instrument meets the exclusion criterion ofIV-2SLS.

Table 6 contains a package of models associated with the IV-2SLSprocedure. Columns 2 and 3 provide the critical first and second stages.As is immediately obvious, the instrument is significantly correlatedwith economic governance; the farther a province is from the seven-teenth parallel, the better its PCI scores, all else equal. The second stageconfirms our overall theory that good institutions are associated withlower time spent in the informal sector. Diagnostic tests, presented atthe bottom of Table 6, confirm that our instrument is strong and appro-priately identified.

Alternative pathways between the instrument and formalization areaddressed by control variables. The percentage of secondary schoolgraduates accounts for the fact that economic elites had more resourcesavailable to escape from war-damaged provinces, and this group wasmore likely to start new businesses than laborers who were trapped inwar-torn regions. Distance from Hanoi and Ho Chi Minh City accountsfor the fact that distance from the seventeenth parallel is also correlatedwith distance from these major metropolises and the greater opportuni-ties their large markets provide for business success. The second-stageimpact of governance can be considered to be the net effect on formal-ization after addressing these other channels.

Disaggregating Institutions That Influence Formalization

Finally, we investigate how specific types of institutions may affect for-malization by replicating our analysis of the gap between establishmentand formalization seventeen times, each iteration replacing our generalinstitutions explanatory variable with the three factors generated inTable 7 (Panel 1), the ten PCI subindexes (Panel 2), and four individ-ual indicators of economic governance that are hard measures, free offirm perception bias and halo effects (Panel 3). We display the coeffi-cients on the key causal variables and standard errors in Table 7. Con-trol variables remain the same as the previous specification in Table 5.

The results are consistent across the panel of Table 7: The drivinginstitutional factor in the determination of formalization is propertyrights. Beginning at the broadest level, the only factor associated with



280

Table 6 Instrumental Variables Analysis of Formalization

Dependent variable: Gap in months between registration and establishment Baseline 1st Stage 2nd Stage

Provincial economic governance –1.126*(0.661)

Years since firm establishment 0.0818 0.0595 1.719(0.0717) (0.0644) (1.081)

Years since establishment squared –0.00148 –0.000906 0.180***(0.00124) (0.00106) (0.0326)

Enterprise Law dummy 0.00680 0.0681 13.84***(0.550) (0.507) (3.716)

Equity capital at time of establishment 0.0730 0.0826 2.833***(0.119) (0.108) (0.614)

Exporting firm 0.297 0.308 5.492***(0.445) (0.420) (1.740)

Firm operates on household land 0.0900 0.107 –3.386***(0.159) (0.152) (1.043)

Owner is a former government official or –0.328 –0.285 9.599***SOE manager (0.305) (0.298) (3.253)

Owner is a former SOE employee 0.0367 0.347 7.919**(0.387) (0.327) (3.125)

Telephones per capita 1995 –7.975 –2.904 –12.60(9.711) (10.28) (12.07)

Secondary school graduates 2006 0.106 0.251 –0.0239(0.184) (0.189) (0.250)

Percentage of minorities in province –1.921** –2.019*** –2.490(0.748) (0.655) (1.638)

Province located south of 17th parallel 3.868* 4.062** 1.641(2.015) (2.010) (3.940)

National-level city 0.956 –0.212 3.324(3.141) (3.679) (4.393)

GDP per capita (ln) 0.962 0.173 0.598(0.941) (0.925) (1.678)

Distance from the 17th Parallel 0.00509*(0.00271)

Constant 42.47** 32.10* 38.63(18.07) (18.53) (36.51)

Sector-fixed effects no no noISIC-fixed effects yes yes yesEstablishment year–fixed effects no no noRegional-fixed effects no no noObservations 3298 3298 3078Provincial clusters 63 63 63Log likelihood –10129 –10022 –14116Root mean squared error 5.219 5.053 24.21R-squared 0.332 0.374 0.809Cragg-Donald F-statistic 209.9***Anderson canonical correlation LR statistic 204.7***Endogeneity test 5.695**

Notes: Robust standard errors, clustered at provincial level (in parentheses). *** p < 0.01, ** p < 0.05,* p < 0.1

Dependent variable is defined as the gap in months between establishment and registration.This presents a series of three models for the IV-2SLS procedure. The first two columns show a base-

line first stage without the instrument, followed by the actual first stage where economic governance is re-gressed on distance from the 17th parallel. The final model shows the complete second-stage results.


281

Table 7 Effect of Subindexes and Indicators of Economic Governance on Formalization

Factors Coefficient

1 Policies –0.687(0.65)

2 Property rights –1.389(0.77)

3 Entry barriers –0.312(0.51)

Subindex Coefficient

1 Entry costs (business registration and licensing) –0.750(0.80)

2 Land access and security of tenure –2.258**(1.08)

3 Transparency of business information –0.764(0.49)

4 Time costs of regulatory compliance –0.395(0.68)

5 Informal charges (corruption) –1.059(1.32)

6 Bias toward state-owned sector –0.477(0.84)

7 Proactivity of provincial leadership –0.468(0.34)

8 Business development services –0.0814(0.45)

9 Labor training and matchmaking –0.632(0.46)

10 Confidence in legal institutions –1.030*(0.54)

Hard indicators of institutions Coefficient

1 Median days to register business –0.0268(0.11)

2 Percentage of titled land in province –0.0611**(0.028)

3 Web page transparency –0.0716(0.15)

4 Percentage of legal cases filed by private firms 8.881(10.5)

Notes: Robust standard errors, clustered at province level (in parentheses); *** p < 0.01,** p < 0.05, * p < 0.1.

Each row replicates Model 2 of Table 5, but with Total Economic Governance replacedby the subindexes that comprise it.

To save space only the coefficients and standard errors on key causal variables are reported.


changes in formalization is the one we labeled property rights, becauseof its strong correlations with land access and security, confidence inlegal institutions, and informal charges. A one standard deviationchange in this factor would yield a 1.4-month decline in the gap be-tween establishment and registration. Confirming this finding, thestrongest results for the individual subindexes are land access and con-fidence in legal institutions. The substantive effect of improving the se-curity of business premises is particularly large. A one-point change inthe subindex is associated with a 2.3-month decline in the waiting pe-riod for formalization. Informal charges is the only subindex stronglyassociated with property rights that is not independently significant.This is not to say that informal charges do not matter for formalization,but their effect is likely mitigated by more secure property rights andstronger appeals processes through legal institutions.

The insignificance of entry costs is less surprising, as the Enter-prise Law decreased the variance in entry costs across provinces sig-nificantly. Only a few days separate waiting periods in the highest andlowest provinces, and it is unlikely that this marginal difference hasmuch of an impact on entrepreneurs’ decisions.

Because these indexes are generated by firm surveys, there is po-tential that perception biases, particularly halo effects, may be respon-sible for the association. As a robustness test, we rerun the analysis onlywith hard indicators that are not derived from the survey. These include(1) a measure of the median days for firm registration recorded by theVietnamese Ministry of Planning and Investment as a proxy for entrycosts; (2) a measure of the total amount of land in each province thathas a formal land use rights certificate (LURC) from the Ministry ofNatural Resources and the Environment to gauge property rights; (3) atwenty-point evaluation of the business information available onprovincial websites to proxy for transparency; and (4) the percentage ofclaims filed in provincial courts by private firms, gauging private sec-tor confidence in legal institutions.

The results of this analysis are displayed in the third panel. Onceagain, only the proxy for the security of property rights is statisticallysignificant. Firms in provinces that have made advances in grantingLURCs are significantly less likely to remain in the informal sector. Inmany ways, this result distills the essence of the property rights. As aresult of the 1993 Land Law, provinces were obligated to grant long-term leases on land in the province to current holders. The leases calledLURCs were for periods as long as ninety years and could be ex-changed, sublet, and used as collateral on bank loans (Do and Iyer

282 Out of the Gray


2008). Holders of these certificates have much more secure protectionof their business activities (Haare 2008) and, as a result, titled land issignificantly more valuable than land without a LURC (Kim 2004).

The wide variation in the allocation of LURCs that has existedthroughout the reform era continues to this day (Ravallion and van deWalle 2006). Currently, the percentage of titled land ranges from 19.5percent in the northwestern province of Tuyen Quang to 98.8 percentin the Mekong Delta’s Vinh Long province. The data demonstrate thatthe simple process of allocating these certificates encourages formal-ization of business activities, confirming one of De Soto’s (2000) mostfamous hypotheses and the Sebastian Galiani and Ernesto Schargrod-sky (2007) discovery that land titles incentivize entrepreneurial behav-ior through the long-term security they provide.

Conclusion

The decision of whether or not to select into the formal economic sec-tor is a crucial strategic decision faced by all entrepreneurs. This deci-sion can be made at any point in the life of a firm. To a significant ex-tent, it is a choice between ambitions—that is, whether to be agrowth-oriented entrepreneur or a more risk-averse subsistence-orientedentrepreneur. This decision is particularly complex under the more dif-ficult conditions typical of developing countries.

In this article, we capitalize on variation in the tendency to formal-ize among entrepreneurs across provinces in Vietnam to investigate theinfluence of local institutions on formalization. We find strong evi-dence that better institutions make entrepreneurs more likely to choosethe more growth-oriented path of formality. This evidence providesempirical support for public policy agendas that predict that cleaner,more predictable government institutions lead to greater entrepreneur-ial dynamism and economic growth.

Furthermore, it appears as though higher-quality property rightsare particularly relevant to enterprise formalization. While there isclearly a need for further investigation on this important point, land ti-tling, in particular, may provide a cheap and effective policy tool forencouraging household businesses to formalize. By simply grantingbetter property rights for existing plots, local officials can spur on for-malization and achieve quick benefits in additional tax revenues andgrowth-oriented entrepreneurship. The strength of the land title mea-sure raises interesting avenues for further research.



There are two channels by which land titles could facilitate formal-ization. The first is that LURCs allow entrepreneurs to feel secure thattheir business premises will be protected for the foreseeable future. Thatknowledge lengthens the business time horizons of operations, allowingentrepreneurs to consider riskier and more expensive operations. In-deed, Galiani and Schargrodsky (2007) show that households with landtitles in Argentina were more likely to take on expensive residential im-provements, such as new roofs and walkways. Galiani and Schargrodskydid not explore business decisions, but the logic of titles incentivizinginvestment is analogous. There is some possibility that the greater open-ness that transparency entails may engender more bribe requests, but atleast entrepreneurs need not fear having the business premises yankedfrom under a successful operation through expropriation.

Second, land titles can promote greater access to credit, becausethey can be offered as collateral (De Soto 2000). Increased credit could,in turn, raise the prospects of business expansion and consequently theperceived benefits of formalization. The Argentina experiment foundlittle evidence of increased credit, but their unit of analysis was thefamily residence, so they primarily explored mortgage and grocerystore credit. They did not consider more entrepreneurial ventures,which would be primarily interested in bank loans. The value of LURCcollateral may be more attractive for bankers when the property is in abusiness district and not a residential area.

Future research will seek to sort out which of these two microlog-ics is more important for the LURC correlation with formalization.Also, more work is needed on the interactions of the institutions ex-plored in this article. Is it possible that institutions may serve as com-plements or substitutes? For instance, are changes in entry barriers(such as the Enterprise Law) more effective in regimes characterizedby high property rights? Do policies, such as Business DevelopmentServices and transparency of regulatory documentation, substitute forpoor property rights? This article has been only a first cut, but the PCIdata can be an effective instrument for answering these questions aswell.

Edmund Malesky is assistant professor of political science at the GraduateSchool of International Relations and Pacific Studies at the University of California–San Diego. He has published in leading political science and eco-nomic journals and has been awarded the Harvard Academy Fellowship andGabriel Almond Award for best dissertation in comparative politics. Maleskyserves as the lead researcher for the Vietnam Provincial Competitiveness Index

284 Out of the Gray


and Cambodian Business Environment Scorecard. He has consulted for theAsia Foundation, USAID, and the World Bank Group. Markus Taussig is adoctoral candidate with the Strategy Unit at Harvard Business School focusedon issues of entrepreneurial strategy and finance in emerging markets. Taussiglived in Vietnam over ten years, working with various international develop-ment organizations on implementation of enterprise surveys and auditing ofSOEs. He also helped launch an international rubber trading company andmanaged a tourism startup for three years.

Notes

Earlier versions of this article were presented at the annual meeting of theAmerican Political Science Association (APSA) in Boston (August 2008) andthe Harvard Developmental Economics Seminar (October 2008). The articlehas benefited from the valuable advice of Melani Cammett, Tom Kenyon,James Anderson, Shawn Cole, Toan Do, Ngoc Anh Tran, Nguyen Van Thang,Dau Anh Tuan, Lakshmi Iyer, Josh Lerner, Kim Ninh, Steve Parker, Jim Winkler, Chris Woodruff, two anonymous reviewers, and Stephan Haggard,editor of Journal of East Asian Studies. Research assistance was provided byLily Phan, Quinn Dang, Tran Thi Thuy Trang, and Nguyen Manh Tuan. Spe-cial thanks go to the Asia Foundation, Developmental Alternatives Incorpo-rated, the Vietnamese Chamber of Commerce and Industry, and the UnitedStates Agency for International Development, which were responsible for thesurvey, and to the Vietnam Competitiveness Initiative, which supported this re-search and generously made the data available for our analysis.

1. For an excellent overview of the prereform history of private enterprisein Vietnam, see Dang (1999). This article serves as the primary source of in-formation for the next two paragraphs.

2. Decree 27/ND and Decree 29 in 1988 set the legal framework forhousehold businesses.

3. The first Enterprise Law was introduced in December 1990, setting upthe framework for sole proprietorships, followed by 1991 legislation allowingregistration of limited liability and joint-stock companies as well.

4. The 1999 Enterprise Law also introduced a fourth type of company: part-nerships. Partnerships are the least common company form, tending to be moreprevalent for services companies such as law firms and consulting companies.

5. The total number of PCI respondents has grown over time. We exploitthe 2007 dataset for this analysis, which has 6,700 firms, but the most recentiteration had 7,820. More importantly, at least fifty respondents are recordedfor each province, allowing for valid statistical analysis. When incorrect ad-dresses and bankruptcies are taken into account, the response rate for the PCIis about 30 percent. This is quite high by international business survey stan-dards, but nonresponse bias remains a possibility, as seven out of ten entrepre-neurs chose not to respond. Further analysis reveals that the response bias is



negative. More frustrated firms are more likely to respond. Fortunately, the re-sponse bias is systematic across provinces. The variance in response rates isonly 5 percent, thus the negative impact on governance score applies generallyacross the sixty-four units. See the annual PCI reports for more detail(www.pcivietnam.org).

6. Each of the subindexes is scaled 1–10, so the final unweighted indexscore can be calculated by simply summing up the components. A weightedPCI is available, but it is not useful for this empirical analysis, because indi-vidual weights are based on a subindex’s contribution to private sector growthand performance.

7. Data and full reports are available at www.pcivietnam.org.8. See Online Appendix 3 (http://irps.ucsd.edu/faculty/faculty-directory/

edmund-malesky.htm) for more details.9. The rationale for this is the substantial number of tangible changes in

recent years, including another Enterprise Law and a Competition Law in2005, and the steadily increasing pressure for further regulatory reform—notonly due to obligations laid out in the bilateral trade agreement with the UnitedStates and in the accession agreement to the World Trade Organization, butalso due to market pressures as services and product markets continue to glob-alize at a rapid pace.

10. This closely parallels the regional groupings of provinces determinedby Vietnam’s GSO, though we made some small changes to better reflect truesimilarities in character and resources. It should be noted that GSO has changedmembership in these groupings on a fairly regular basis. In some cases,provinces have been moved from a less successful grouping of provinces to amore successful grouping because of an improvement in the province’s ownperformance.

11. While we do not have direct evidence on firms that never formalized,as noted earlier, analysis of aggregate levels of formalization in Figure 1 andOnline Appendix 3 indicates that this nondecision is also highly dependent onthe quality of provincial governance.

12. In this case, negative binomial is preferred to poisson regression be-cause of high dispersion in the data.

References

Abrami, Regina M., and Nolwen Henaff. 2004. “The City and the Countryside:Economy, State and Socialist Legacies in the Vietnamese Labor Market.”In Reaching for the Dream: Challenges of Sustainable Development inVietnam, ed. Melanie Beresford and Angie Tran. Honolulu: University ofHawaii Press.

Acemoglu, Daron, and Simon Johnson. 2005. “Unbundling Institutions.” Jour-nal of Political Economy 113, 5: 949–995.

286 Out of the Gray


Aldrich, Howard, and C. Marlene Fiol. 1994. “Fools Rush In? The InstitutionalContext of Industry Creation.” Academy of Management Review 19, 4:645–670.

Antunes, Antonio, and Tiago Cavalcanti. 2007. “Start-Up Costs, Limited En-forcement, and the Hidden Economy.” European Economic Review 51:203–224.

Ardagna, Silvia, and Annamaria Lusardi. 2008. “Explaining International Dif-ferences in Entrepreneurship: The Role of Individual Characteristics andRegulatory Constraints.” NBER Working Paper Series 14012. Cambridge,MA: National Bureau of Economic Research.

Bennett, John, and Saul Estrin. 2007. “Entrepreneurial Entry in DevelopingEconomies: Modeling Interactions Between the Formal and the InformalSector.” International Policy Center Working Paper Series 44. Available athttp://hdl.handle.net/2027.42/55758 (accessed February 20, 2009).

Coase, Ronald. 1937. “The Nature of the Firm.” Economica 4, 16: 386–405.Dang Phong. 1999. “The Private Sector in Vietnamese Industry from 1945 to

the Present.” In Vietnam’s Undersized Engine: A Survey of 95 Larger Pri-vate Manufacturers, ed. Leila Webster and Markus Taussig. Mekong Proj-ect Development Facility (MPDF) Private Sector Development Discus-sion Brief 8. Hanoi: MPDF.

De Soto, Hernando. 1989. The Other Path: The Invisible Revolution in theThird World. London: Tauris.

———. 2000. The Mystery of Capital: Why Capitalism Triumphs in the Westand Fails Everywhere Else. New York: Basic Books.

DiMaggio, Paul, and Walter Powell. 1991. The New Institutionalism in Orga-nizational Analysis. Chicago: University of Chicago Press.

Di Tella, Rafael, Sebastian F. Galiani, and Ernesto S. Schargrodsky. 2007.“The Formation of Beliefs: Evidence from the Allocation of Land Titlesto Squatters.” Quarterly Journal of Economics 122, 1: 209–241.

Djankov, Simeon. 2008. “A Response to Is Doing Business Damaging Busi-ness?” Journal of Comparative Economics. Available at www.doingbusiness.org/documents/Response_to_Arrunada_JCE.pdf? (accessed Feb-ruary 5, 2008).

Djankov, Simeon, Rafael La Porta, Florencio Lopez-De-Silanes, and AndreiShleifer. 2002. “The Regulation of Entry,” Quarterly Journal of Econom-ics 117, 1: 1–37.

Djankov, Simeon, Yingyi Qian, Gerard Roland, and Ekaterina Zhuravskaya.2008. “Entrepreneurship in Development: First Results from China andRussia.” Manuscript.

Do, Quy-Toan, and Lakshmi Iyer. 2008. “Land Titling and Rural Transition inVietnam.” Economic Development and Cultural Change 56: 531–579.

Fforde, Adam, and Stefan de Vylder. 1996. From Plan to Market: The Eco-nomic Transition in Vietnam. Boulder: Westview Press.

Field, Erica. 2007. “Entitled to Work: Urban Tenure Security and Labor Sup-ply in Peru.” Quarterly Journal of Economics 122, 4: 1561–1602.



Friedman, Eric, Simon Johnson, Daniel Kaufmann, and Pablo Zoido-Lobaton.2000. “Dodging the Grabbing Hand: The Determinants of Unofficial Ac-tivity in 69 Countries.” Journal of Public Economics 76, 3: 459–493.

Frye, Timothy, and Andrei Shleifer. 1997. “The Invisible Hand and the Grab-bing Hand.” American Economic Review 87, 2: 354–358.

Galiani, Sebastian, and Ernesto Schargrodsky. 2007. “Property Rights for thePoor: Effects of Land Titling.” Business School Working Papers, Univer-sidad Torcuato Di Tella, Buenos Aires.

Ghemawat, Pankaj. 2007. Redefining Global Strategy: Crossing Borders in aWorld Where Differences Still Matter. Boston: Harvard Business SchoolPublishing Corporation.

Grief, Avner. 2006. Institutions and the Path to the Modern Economy: Lessonsfrom Medieval Trade. Cambridge: Cambridge University Press.

Haare, Denise. 2008. “The Origins and Influence of Land Property Rights inVietnam.” Development Policy Review 26, 3: 339–363.

Ihrig, Jane, and Karine Moe. 2004. “Lurking in the Shadows: TheInformal Sector and Government Policy.” Journal of Development Economics

73: 541–557.Johnson, Simon, Daniel Kaufmann, John McMillan, and Chris Woodruff. 2000.

“Why Do Firms Hide?” Journal of Public Economics 76, 3: 495–520.Johnson, Simon, Andrei Shleifer, and Daniel Kaufmann. 1997. “The Unofficial

Economy in Transition.” Brookings Papers on Economic Activity 28, 2.Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2007. “Growth and

Governance: A Reply.” Journal of Politics 69, 2: 555–562.Kenyon, Thomas. 2007. “A Framework for Thinking About Enterprise For-

malization Policies in Developing Countries.” World Bank Policy Re-search Working Paper No. 4235. Washington, DC: World Bank. Availableat http://ssrn.com/abstract=988845 (accessed February 20, 2009).

Kim, Annette. 2004. “A Market Without the ‘Right’ Property Rights.” Eco-nomics of Transition 7, 3: 275–305.

King, Gary, Christopher J. L. Murray, Joshua A. Salomon, and Ajay Tandon.2003. “Enhancing the Validity and Cross-Cultural Comparability of Mea-surement in Survey Research.” American Political Science Review 97, 4:567–584.

King, Tomz, and Wittenberg. 2000. “Making the Most of Statistical Analyses:Improving Interpretation and Presentation.” American Journal of PoliticalScience 44, 2: 341–355.

Knack, Stephen, and Phillip Keefer. 1995. “Institutions and Economic Perfor-mance: Cross-Country Testing Using Alternative Institutional Measures.”Economics and Politics 7, 3: 207–228.

Kocher, Matthew, Thomas B. Pepinsky, and Stathis N. Kalyvas. 2008. “Intothe Arms of Rebels? Aerial Bombardment, Indiscriminate Violence, andTerritorial Control in the Vietnam War.” Manuscript. Yale University.

Kurtz, Marcus J., and Andrew Schrank. 2007. “Growth and Governance: Mod-els, Measures, and Mechanisms.” Journal of Politics 69, 2: 538–554.

288 Out of the Gray


Laeven, Luc, and Christopher Woodruff. 2007. “The Quality of the Legal Sys-tem, Firm Ownership, and Firm Size.” Review of Economics and Statis-tics 89, 4: 601–614.

La Porta, Rafael, F. López-de-Silanes, A. Shleifer, and R. Vishny. 1999. “TheQuality of Government.” Journal of Law, Economics and Organization15, 1: 222–279.

Loayza, Norman, Ana Maria Oviedo, and Luis Serven. 2005. “The Impact ofRegulation on Growth and Informality Cross-Country Evidence.” WorldBank Policy Research Working Paper No. 3623. Washington, DC: WorldBank.

Malesky, Edmund. 2007. “The Vietnam Provincial Competitiveness Index:Measuring Economic Governance for Private Sector Development: 2007Final Report.” Vietnam Competitiveness Initiative Policy Paper No. 12.USAID’s Vietnam Competitiveness Inititative, Hanoi.

———. 2008. “Provincial Governance and Foreign Direct Investment in Viet-nam.” In Twenty Years of Foreign Investment in Vietnam. Ho Chi MinhCity: Saigon Economic Times.

———. 2009. “Gerrymandering Vietnamese Style: The Political Geographyof Economic Transition in an Authoritarian State.” Journal of Politics 71,1: 1–30.

Malesky, Edmund, and Markus Taussig. 2008. “Where Is Credit Due? LegalInstitutions, Connections, and the Efficiency of Bank Lending in Viet-nam.” Journal of Law, Economics, and Organization. http://jleo.oxfordjournals.org/cgi/reprint/ewn011v1?maxtoshow=&HITS=2&hits=2&RESULTFORMAT=&andorexacttitle=and&titleabstract=vietnam&andorexacttitleabs=and&andorexactfulltext=and&searchid=1&resourcetype=HWCIT. Accessed March 10, 2009.

Miguel, Edward, and Gerard Roland. 2006. “The Long Run Impact of Bomb-ing Vietnam.” NBER Working Paper No. W11954. Cambridge: NationalBureau of Economic Research.

Nguyen Dinh Cung, Pham Anh Tuan, Bui Van, and David Dapice. 2004. “WhyDon’t Northern Provinces Grow Faster?” Hanoi: Central Institute for Eco-nomic Management and United Nations Development Program.

North, Douglass. 1991. “Institutions.” Journal of Economic Perspectives 5, 1:97–112.

North, Douglass, and Barry Weingast. 1989. “Constitutions and Commitment:The Evolution of Institutional Governing Public Choice in Seventeenth-Century England.” Journal of Economic History 49, 4: 803–832.

O’Conner, Dennis. 2000. “Financial Sector Reform in China and Vietnam: AComparative Perspective.” Comparative Economic Studies 42, 4: 45–66.

Porter, Michael. 1980. Competitive Strategy: Techniques for Analyzing Indus-tries and Competitors. New York: Simon & Schuster.

Ravallion, Martin, and Dominique van de Walle. 2006. Does Rising LandlessnessSignal Success or Failure for Vietnam’s Agrarian Transition? World BankPolicy Research Working Paper No. 3871. Washington, DC: World Bank.



Riehl, Evan. 2008. “Regional Regulations and Foreign Investment in Viet-nam.” Manuscript, Washington University, St. Louis.

Santos, Felipe, and Kathleen Eisenhardt. 2006. “Constructing Markets and Orga-nizing Boundaries: Entrepreneurial Action in Nascent Fields.” Manuscript.

Schneider, Friedrich. 2005. “Shadow Economies Around the World: What DoWe Really Know?” European Journal of Political Economy 21, 3:598–642.

Seligson, Mitchell. 2006. “The Measurement and Impact of Corruption Vic-timization: Survey Evidence from Latin America.” World Development34, 2: 381–404.

Taussig, Markus, and Pham Thi Thu Hang. 2004. “Private Enterprise Formal-ity and the Role of Local Government: Making Markets Work Better forthe Poor.” Asian Development Bank Discussion Paper No. 2. Hanoi:Asian Development Bank.

Tran, Thi Bich, R. Quentin Grafton, and Tom Kompas. 2008. “InstitutionsMatter: The Case of Vietnam.” International and Development Econom-ics Working Paper No. 08-01. Australian National University, Canberra.

Turley, William, and Brantly Womack. 1998. “Asia’s Socialism’s Open Doors:Guangzhou and Ho Chi Minh City.” The China Journal 40: 95–119.

VCCI (Vietnam Chamber of Commerce and Industry). 2008. Vietnam Business2008. Hanoi: VCCI.

Vietbao.vn. 2006. “Ho kinh doanh ca the chua muon chuyen thanh DN.”March 22.

———. 2007. “Ho kinh doanh ca the gop 13% GDP.” October 14.Vijverberg, Wim. 2006. “Non-Farm Household Enterprises in Vietnam,”

World Bank Research Working Paper. Washington, DC: World Bank.Vu Thanh Tu Anh, Le Viet Thai, and Vo Tat Thang. 2007. “Provincial Extrale-

gal Investment Incentives in the Context of Decentralisation in Viet Nam:Mutually Beneficial or a Race to the Bottom?” United Nations Develop-ment Program Vietnam Working Paper Series. Hanoi: UNDP.

Webster, Leila, and Markus Taussig, eds. 1999. Vietnam’s Undersized Engine:A Survey of 95 Larger Private Manufacturers. MPDF Private Sector De-velopment Discussion Brief 8. Hanoi: MPDF.

Weingast, Barry. 1993. “Constitutions as Governance Structures: The PoliticalFoundations of Secure Markets.” Journal of Institutional and TheoreticalEconomics 146, 1: 286–311.

Woodruff, Christopher. 2006. “Measuring Institutions.” In International Hand-book on the Economics of Corruption, ed. Susan Rose-Ackerman. Chel-tenham, UK: Edward Elgar Publishing Ltd.

World Bank. 2002. Vietnam Development Report 2002: Implementing Reformsfor Faster Growth and Poverty Reduction. Hanoi: World Bank.

290 Out of the Gray


out of the gray: the impact of provincial institutions on...

Documents