land titling and rural transition in vietnam
TRANSCRIPT
Land Titling and Rural Transition in VietnamAuthor(s): Quy‐Toan Do /Lakshmi IyerSource: Economic Development and Cultural Change, Vol. 56, No. 3 (April 2008), pp. 531-579Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/533549 .
Accessed: 10/06/2014 11:17
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access toEconomic Development and Cultural Change.
http://www.jstor.org
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
� 2008 by The University of Chicago. All rights reserved. 0013-0079/2008/5603-0006$10.00
Land Titling and Rural Transition in Vietnam
quy-toan doWorld Bank
lakshmi iyerHarvard University
I. IntroductionThere is a certain amount of consensus among economists that better propertyrights institutions lead to improved economic outcomes (see North andThomas [1973]; Knack and Keefer [1995]; or Acemoglu, Johnson, and Rob-inson [2001] for analyses of general property rights institutions). Scholars suchas De Soto (2000) have argued that the major barrier to prosperity in devel-oping countries is the inability to convert property into usable assets, becauseof a lack of clear-cut legally recognized rights. However, the empirical evidenceon the importance of issuing formal titles to land is inconclusive, both on theoverall effect of having property titles and on which dimensions of land rightsare crucial. Alston, Libecap, and Schneider (1996) find that investments inland as well as land values are positively associated with the possession offormal titles in Brazil; an analysis of land titling in Indonesia reached similarconclusions (SMERU Research Team 2002). Galiani and Schargrodsky (2006)also find urban land titles leading to increased investment in housing in BuenosAires. Besley (1995) finds that more secure land rights lead to greater in-vestment, but investment on the land may also have been undertaken with a
We thank two anonymous referees, Daron Acemoglu, Abhijit Banerjee, Esther Duflo, ChristopherGibbs, Jonathan Haughton, Vu Quoc Huy, Hanan Jacoby, Simon Johnson, and Martin Ravallionfor helpful comments and suggestions. We also thank seminar participants at the Williams DavidsonInstitute–Center for Economic Policy Research 2004 International Conference on the economicsof transition, Harvard Business School, University of Michigan, Suffolk University, the World Bank,and Yale University. Special thanks go to Nguyen Phong and the General Statistical Office, Hanoi,for allowing us to access the Vietnam Living Standard Measurement Study Surveys; to Tran ThiMinh Ha, Do Duc Doi, and the Ministry of Natural Resources and Environment, Hanoi, forproviding us with land registration and land administration manpower data; and to the MITSchultz Fund for financial support. The findings, interpretations, and conclusions expressed in thispaper are entirely ours. They do not necessarily reflect the view of the World Bank, its executivedirectors, or the countries they represent. Contact the corresponding author, Quy-Toan Do, [email protected] [email protected].
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
532 economic development and cultural change
view to strengthening land rights. On the other hand, Braselle, Gaspart, andPlatteau (2002) review a number of studies in African countries that showvery little impact of land titling on investment. Looking at labor marketactivities, Field (2003) finds that urban land titling is associated with anincrease in formal labor market participation in Peru. Several studies on landtitling have examined the impact on credit markets, and the results are mixed.Feder et al. (1986) in Thailand and SMERU Research Team (2002) in Indonesiafind that possession of legal titles leads to an increase in credit access for thepoor, while Boucher et al. (2002) in Nicaragua and Honduras, and Field andTorero (2004) in Peru, show that access to credit remains low even after landreforms have been implemented. Galiani and Schargrodsky (2006) also findno impact of formal property rights on access to credit markets in BuenosAires.
This study investigates the impact of a specific legal change to land rightsin Vietnam. Land rights are an important issue in Vietnam, where agricultureaccounts for nearly a quarter of gross domestic product (GDP) and two-thirdsof the workforce is engaged in agricultural activities. Agricultural land inVietnam was decollectivized in 1988, and land-use rights were granted tohouseholds. We focus on the subsequent land law of 1993, which gave house-holds the right to inherit, transfer, exchange, lease, and mortgage their land-use rights. This was implemented by issuing land titles (or Land-Use Certif-icates as they are known in Vietnam) to all households. This law initiated anextensive land-titling program in Vietnam: by the year 2000, nearly 11 millionland titles had been issued to rural households, making this one of the largestrural titling programs in the developing world, not only in scale but also inspeed of implementation. To view this program in comparative perspective,8.7 million land titles have been distributed in Thailand since the early 1980sand 1.87 million in Indonesia between 1996 and 2000 (SMERU ResearchTeam 2002); the largest urban titling program, aimed at squatters in Peru,distributed 1.2 million titles (Field 2003). The economic consequences arethus of interest not only to Vietnam but also to other developing countriescontemplating land-titling programs.
Our study contributes to the literature on land titling in several distinctiveways. First, we study a large nationwide titling program rather than onerestricted to certain areas or certain classes of people. This is in contrast toalmost all the studies cited above. This means that our results include anypotential general equilibrium effects of such titling programs. Second, ourwork focuses on the impact of granting new rights to land, including therights to mortgage and trade, rather than a formalization of existing rights(as is the case in most programs aimed at squatters). Further, by looking at
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 533
different outcomes, we are able to make inferences regarding which of thesenew rights was most effective. Third, our focus on the 1993 reform enablesus to distinguish the effects of land titles from the incentive effects of de-collectivization. Our work is complementary with other studies of agrariantransition in Vietnam, such as Benjamin and Brandt (2004) and Ravallionand van de Walle (2003). However, the use of measures of land reform progressacross provinces enables us to distinguish the incremental impact of the 1993land law from the overall effect of economic growth during this period.1
We use household-level data from two waves of the Vietnam Living Stan-dards Survey, which took place in 1992–93 and 1997–98, respectively. Sinceindividual-level information on possession of land titles is not available inthese surveys, we collected province-level information on the issuance of land-use certificates. We verify that there are no systematic differences in prereformcharacteristics between provinces that issued titles faster and those that wereslower in issuing titles. We then compare the change in outcomes for house-holds between the two time periods, across provinces at different stages of theland-titling process. We find that households in provinces that have madegreater progress in land titling increase the proportion of cultivated areadevoted to multiyear crops. While this effect is statistically significant, it isnot large in magnitude: a one standard deviation increase in land titling resultsin a 0.09 standard deviation increase in the proportional area devoted to long-term crops. We also find that households in provinces with more land titlesdevote more time to nonfarm activities. Providing titles to all householdswould result in an increase of 11–12 weeks of nonfarm work by the household.
Our results provide little support for the view that such increased investmentis financed by credit since we find no significant increase in household bor-rowing. Analysis of land market transactions is hampered by severe under-reporting; while we do find that the overall volume of transactions increasesafter the reform, the increase is not significantly different in provinces thatare at an advanced stage of the titling process. Overall, our results suggestthat the main driving force underlying these changes of Vietnam’s rural econ-omy is the increased security of tenure associated with the 1993 land law.
The paper is structured as follows: Section II describes the evolution of landrights in Vietnam, Section III presents a simple theoretical framework toanalyze the effects of land titling, and Section IV describes our data andempirical strategy. Section V documents the variation in the implementation
1 Deininger and Jin (2003) also analyze changes to property rights in Vietnam, with a focus onperceived security of tenure proxied by the possession of long-term use land. Ngo (2005) examinesthe combined impact of this proxy measure with the measure of land titling adopted in this study.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
534 economic development and cultural change
of the land reform across provinces. Our main results on crop and labor choiceare discussed in Section VI, and an investigation of the potential mechanismsat work is undertaken in Section VII. Section VIII concludes.
II. Land Rights in VietnamThe history of Vietnam in the second half of the twentieth century is punc-tuated by three key dates: 1954 marked the independence of the country fromthe French and its division into two parts, North and South; in 1975, theso-called Vietnam war ended with the reunification of North and South Viet-nam; and 1986 saw the implementation of sweeping economic reforms (the“Doi Moi” policy) and a move toward a market-oriented economy, whichcontinues to the present day. The material in this section is largely based onWiegersma (1988), Pingali and Xuan (1992), and Boothroyd and Pham (2000).
A. The Institutional Framework until 1988Before the Geneva Accord of 1954, Vietnam was under French colonial rule.During the colonial period, most farmland in Vietnam was owned either byFrench plantation owners or by large Vietnamese landlords: 52% of the landwas owned by only 3% of the indigenous population, and more than 60% offarmers across the country were landless in the mid-1940s.
After independence a major land reform was carried out in the North. Asa reward for their war efforts, land and ownership rights were distributed tofarmers, and a rapid increase in agricultural output and productivity followed.However, the policy was reversed, and land began to be collectivized in thelate 1950s, as communist ideology gained strength. As a result, 86% of allpeasant households and 68% of total farmland were brought into cooperativesby 1960. Despite significant declines in output, the collectivization processcontinued so that 90% of all peasant households in the North were workingin cooperatives by the mid-1960s. An illuminating stylized fact illustrates theimpact of such an incentive system: while individual rural households wereallocated 5% of private farmland, they derived 60%–70% of their earningsfrom this small plot (Boothroyd and Pham 2000, chap. 2).
Land institutions in the South during the same period were driven bypolitical conflicts. At times when the government sought support from thelocal elites, pro-landowner policies were adopted. When the war against NorthVietnam began, the government tried to gain popular support by adoptingthe Land-to-the-Tiller law in 1970. Tillers of the soil were to enjoy all thebenefits of their work, and this would be accomplished by providing ownershiprights to cultivators and putting a retention limit on landlords as low as 20hectares. However, the law found opposition from landlords, and the lack of
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 535
independence of the bureaucracy made enforcement uneven throughout SouthVietnam.
In 1975, when the war ended and the country was reunified, land collec-tivization started in the South but was implemented with little success: aslate as 1986, only 5.9% of farmers in the Mekong Delta and 20% in theSoutheastern region were part of cooperatives, while this figure amounted to85% in the Central Lowlands region (Pingali and Xuan 1992). Under thecollective system, all households were paid a share of output according to theirrecorded labor hours on the communal land. In 1981, the first changes weremade to these arrangements: workers were now allowed to keep all of thesurplus they produced over a contracted output. However, this policy was latermodified, and quotas were constantly renegotiated, resulting in a decline ofpublic confidence. Agricultural yields were extremely low in this period, and,even as late as 1985, Vietnam was a net importer of rice.
Faced with a worsening economic crisis, the government announced in 1986the program of Doi Moi, or Renovation, and began a gradual movementtoward a market economy. As part of a major structural adjustment program,production and consumption subsidies were eliminated from the state budget,government spending was reduced to 6% of GDP, the government workforcewas reduced by 15%, 500,000 soldiers were demobilized, interest rates onloans to state-owned firms were raised, and central bank credit was no longerused to finance the budget deficit. The economy started opening up to trade,and the central bank undertook a massive devaluation of the currency to theprevailing black market rate. Inflation rates were brought down from 400%in 1986–87 to 10% in 1993. Financial markets were partially deregulated,foreign banks were allowed to operate in Vietnam, and a stock exchange wasopened in 2000. In the agricultural sector, Resolution 10 of 1988 grantedland-use rights to individual households, while the land law of 1993 madethese rights pledgeable and tradable. These two changes are described in detailin following sections.
These reforms have had a dramatic impact on the economy. During theperiod 1990–2005, the Vietnamese economy experienced annual growth ratesof 7.5%. Crop production grew at a rate of 5.5% over the period 1990–2004and accounted for 29% of total non-oil exports in 2003. By 2002, Vietnamwas the third largest exporter of rice and the second largest exporter of coffee,pepper, and cashews. Exports and imports amounted to 75% of GDP by1995.2 The benefits from growth have been fairly widespread: poverty ratesare estimated to have declined from 75% in 1984 to 55% in 1993 (Dollar
2 World Development Indicators Online database, http://go.worldbank.org/3JU2HA60D0.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
536 economic development and cultural change
and Litvack 1998), 37% in 1998, and 29% in 2002. In January 2007, Vietnamjoined the World Trade Organization (WTO) as its 150th member.
B. The 1988 Land Law
The 1988 land law was aimed at liberalizing the agricultural sector in Vietnam.Resolution 10 of this law consisted of transferring control and cash-flow rightsfrom the farming cooperative to the individual household. Land was allocatedto households with 10–15 years of secure tenure, output markets were pri-vatized, and investment decisions were decentralized and left to households.Private property was virtually instituted. However, as land-use rights weregiven to families without the possibility of trading such rights, a proper landmarket did not develop despite some informal transactions.
Land allocation to individual households was conducted by the communeauthorities and encountered some difficulties across the country. In the Northand in some regions of the South, land was distributed on a fairness basis,taking into account soil and sociodemographic characteristics of the region.Comparing the realized land allocation process with a simulated market-basedoutcome, Ravallion and van de Walle (2003) conclude that the realized processgenerated lower inequality and made the poorest better off. However, theprocess sometimes relied on arbitrary considerations, leading to favoritism anddisputes. For instance, Hayami (1994, 13) reports that “a farmer . . . in HaiHung Province complained that he received too small a land allocation becausehis eldest son was in military service and his other children were so youngthat they received only one-third of an adult’s allocation at a time. Thus, heexpects that his unfavorable allocation will be corrected at the end of the ten-year tenure period.” The situation in the South was complicated by the factthat Resolution 10 stipulated that farmers should be assigned the land theyowned prior to 1975, and this generated disagreement between farmers andformer landlords, although a 1989 ordinance gave rights to farmers. Theallocation was thus not immune to controversy, and disputes were still beingsettled in July 2001, as land allocation was being brought to completion inrural areas.
To many Vietnamese, Resolution 10 is perceived as the major land reformundertaken since 1975, and some scholars attributed Vietnam’s agriculturaloutput growth to such liberalization (see, e.g., Pingali and Xuan 1992). Thereis no denying that newly assigned property rights have unleashed farmers’incentives to invest and put an effort into their land, but much remained tobe done to achieve further economic efficiency. The 1993 land law was anadditional step toward this end.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 537
C. The 1993 Land Law and the Issuance of Land-Use CertificatesThe main focus of this study is the 1993 land law. The spirit of the law isin continuation of the reforms undertaken by the government since 1988.Despite the allocation of land and its corresponding use rights, no transactionscould be made officially. The 1993 land law made up for this deficiency. Itgranted five rights to the household: the right to transfer, exchange, inherit,rent, and mortgage.3 The law also extended the lease term to 20 years forannual crop land and 50 years for perennial crop land (Land Law of Vietnam1993, article 20). The implementation of the land law consisted of providingland-use certificates (LUCs) for the purpose of exercising these rights: forinstance, if a landowner wanted to use his land as collateral for a loan, theLUC would be handed over to the lender. As the actual procedure for issuingthe land titles has some relevance for our empirical strategy, it is worth goinginto in some detail.
The issuance of LUCs is done as follows: individuals first have to apply fora land-use certificate (alternatively known as Land-Tenure Certificate or RedBook) through the commune-level People’s Committee. The District Bureauof Land Administration then does the groundwork, which includes makinga list of all land users, training the staff, purchasing materials, and checkingand updating the documents related to land such as cadastral maps, landsurvey records, and so forth. In the meantime, a land registration committeeis set up, which includes members from the District Bureau of Land Admin-istration as well as officials from the commune-level, district-level, and some-times province-level People’s Committees. This process takes about 4–5 weeks.Application forms for land registration are then given out to all the land usersin the commune, who are asked to list all the plots of land owned or allocatedto them. This form has finally to be signed, not only by the land user himselfbut also by all neighboring households, in order to certify the absence ofdisputes regarding claims on the land.
The land registration committee scrutinizes all these forms and then decideswhether a given land user is eligible or not. Land users are classified as ineligibleif (i) they obtain the land through an illegal land transfer, that is, withoutregistering the transaction, without paying transfer taxes, or without a legalcontract; (ii) they inherit the land from parents without a formal inheritanceletter, so that old documents are still in the parents’ names; (iii) they haveno legal documents to prove their claim to the land; (iv) they are illegally
3 There were further modifications to the land law in 1998 and in July 2001. The 1998 revisionsgranted further rights by making it possible to sublease land and also allowed Vietnamese entre-preneurs to use such rights as their contribution to a joint venture with a foreign company. The2001 additions simplified procedures in urban areas.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
538 economic development and cultural change
occupying unallocated land; (v) they have not paid all their land taxes in thepast; or (vi) there are disputes regarding their ownership or the boundaries ofthe land they claim. Within 10 days of sending these application forms, apublic meeting is held in which information regarding eligibility is madepublic. At this time, the land administration also tries to resolve these disputes.The list of land users who are eligible for receiving the LUC is then sent tothe district-level People’s Committee. Unresolved disputes are referred to aspecial working group within the General Department of Land Administration(GDLA, now part of the Ministry of Natural Resources and Environment,MONRE). After approval at the district level, work begins on making theactual LUC for the land user. This stage is estimated to take about 1,500–2,000man days per commune in urban areas, and this figure is likely to be similarin rural areas. In 1998, there were (on average) 160 communes per province,and 1,300 households per commune.
Like some previous land reforms, the 1993 land law was unevenly imple-mented throughout the country. Because province-level differences in the speedof implementation of this reform are the key to our empirical strategy, weinvestigate the sources of such heterogeneity. According to Vo (1997), districtBureaus of Land Administration have on average five members, and mostcommunes have only one land officer, which makes registration a lengthyprocess. Haque and Montesi (1996, 7) also report the major reasons for thisslow progress to be “a lack of adequate finances, a lack of trained cadres, alack of interest and enthusiasm on the part of officials, a lack of proper directionand supervision and disputes among the cadres,” which is consistent with theinformation given to us by the GDLA. The GDLA also pointed out thatanother major factor that slowed down the process was the number of disputesthat needed to be resolved prior to LUC issuance. An additional reason fordelay may also be due to the fees related to registration and the backlog oftaxes that some households may be required to pay to become eligible. How-ever, fees are not very high, below VND 20,000 (less than US$1.50) in mostareas.
Some aspects of this large land-titling program gave rise to concerns re-garding the long-term sustainability of this reform. The guidelines for im-plementation of the five rights was formulated and approved by the GDLAin late 1998. In the meantime, substantial volumes of transactions had notbeen reported to the commune authorities, raising the concern that commune-level land registries would be outdated in a few years’ time (Asian DevelopmentBank 1998). The certificates were issued at the household level rather thanthe plot level, which caused problems in updating the land-use certificate in
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 539
the event of a land transaction. Following the passage of the New Land Lawin November 2003, the government began issuing certificates at the plot level.
III. What Is the Potential Impact of the Land Law?What is likely to be the impact of the land law and, more specifically, theissuance of LUCs? The longer lease term and the right to inherit decrease thelikelihood that an individual and her offspring will be expropriated by thestate, hence increasing the security of tenure on land and providing an incentiveto undertake long-term investments. If such investment (e.g., planting pe-rennial crops) is labor saving, there may also be a shift to nonagriculturalactivities. The rights to transfer, exchange, and lease LUCs create a formalmarket for land, which may achieve a better allocation of land than a cen-tralized/informal system. The right to mortgage LUCs may allow farmers toundertake investments that have high up-front costs, such as planting multi-year crops; however, access to credit markets might reduce the need to diversifyeconomic activities as a means of consumption or income smoothing. Theincremental impact of this right will depend on the relative inability of thecommune to put in place implicit or explicit contracts aimed at providingincentives to reimburse loans.4 We formalize these intuitions in a formal modelbelow.
A. The Basic SettingConsider a small open economy populated with farmers , eachi � {1, ... , N}having one plot of land. Each individual is endowed with one unit of laborthat can be split between the three available technologies: nonfarm activity,rice cultivation, and perennial crop cultivation. The production functions aresimply specified as follows: for any individual , and plot of sizei � {1, ... , N}m,
nonfarm (l) p f(l) p l,
j 1�jrice (l, m) p g(l, m) p Gl m ,
g 1�gperennial (l, m) p h(l, m) p Hl m ,
where l is the amount of labor invested by individual i. We assume thatagricultural production exhibits constant returns to scale. Let represent{n, r, p}the amounts of time devoted to nonfarm activities, rice cultivation, and pe-rennial crop cultivation, respectively. The key difference is that nonfarm ac-tivities and rice cultivation are assumed to yield returns immediately, while
4 Anecdotal evidence collected during fieldwork suggests that in some places, before 1993, com-mune authorities would commit not to reallocate land to households that defaulted on a loan.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
540 economic development and cultural change
returns from perennial crops are realized only in the future. Let r denote theshare of land area devoted to perennial crop cultivation; the remaining 1 �
is used to grow rice. Aggregate land area is normalized to one. Farmersr
have access to a credit market, where they can borrow in order to buy landor pay for any opportunity costs of undertaking long-term investments.
Economic activities are undertaken over two periods as follows:
1. : Financial contracts between farmers and lenders are signed.T p 0Land exchange arrangements and investments are undertaken. Returnson nonfarm activity and rice cultivation are realized. Intermediate con-sumption takes place.
2. : Farmers choose to repay loans or default, in which case theirT p 1land is seized by the lenders. Output on perennial crops is realized.Final consumption takes place.
We assume that expropriation is characterized by a constant probability ofexpropriation a (reflecting the general insecurity of property rights and thepossibility of land being taken over by the state for reallocation) and a prob-ability of expropriation that is contingent on default b, where default is definedby the nonreimbursement of a loan contracted at time . In summary,T p 0agents have a probability a of losing their land even if they do not defaulton their loans and a probability if they do. The land expropriateda � b
unconditionally is redistributed to agents in a lump-sum manner, while theland expropriated under default is given to creditors in case of default. Thetiming of the economy also implies that expropriation matters only if perennialcultivation is undertaken. We model the land law as a change in the probabilityof land expropriation as follows.5T p 1
Land law. The land law is the transition of from topre pre(a, b) (a , b )( ), such that (improved security of tenure) andpost post pre post prea , b a ≤ a b ≤
(increased ability to pledge land as collateral). We assume that the landpostb
law is unanticipated by agents.Land markets. A land allocation is a partition of the overall surface, that
is, each individual i is allocated a plot of area , so that . TheN
m � m p 1i iip1
vector denotes the land allocation schedule at the end of time(m )i ie{1,...,N}
, while the initial distribution of land is exogenous and is denotedT p 0. In the pre-1993 economy, m is not a choice variable, that is,0(m )i ie{1,...,N}
. In the post-1993 economy, m is the outcome of a land market equi-0m p m
librium, where h denotes the price of a marginal piece of land. We set theland price in the pre-1993 economy.h p 0
5 Parameters and equilibrium values of the pre-1993 economy will henceforth have a superscript“pre” while post-1993 variables will be noted with a superscript “post.”
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 541
Credit markets. Agents have access to an intertemporal saving technologywith the risk-free rate normalized to one. Furthermore, there exists an atomisticmarket, in which risk-neutral lenders can access funds at the risk-free rate andlend these funds back to farmers. We assume free entry in the credit supplysector, so that lenders break even in equilibrium. A financial contract forindividual i consists of the following: (i) an amount B borrowed at time
and (ii) an interest rate t so that is due at the beginning of timeT p 0 tB. We denote if agent i defaults on her loan at .T p 1 d p 1 T p 1
Individual preferences. Farmers’ utility functions are defined over con-sumption in both time periods as follows:
U[C , C ] p ln C � ln C .0 1 0 1
Farmers make consumption decisions ; savings decisions S; borrowing{C }T Tp0,1
and reimbursement decisions ; and land, labor, and capital investment{B, d}decisions ; in order to maximize their utility. In the post-1993{r, n, r, p, k}economy, they also make land market decisions.
We can decompose this optimization problem into two stages. In the firststage, farmers have amount of land and one unit of labor, and take them0
vector as given. The optimization problem results in a reduced-form(a, b, m)utility function as follows:
V(a, b, m ) p max U[C , C ]i 0 1C ,C ,S,B ≥ 0,d p {0,1}0 1
r,n,r,p � [0,1]n � r � p ≤ 1
subject to
h(m � m ) � C � S ≤ B � f(n) � g[r, (1 � r)m]0 0
(1 � d)tB � C ≤ S � (1 � a � bd)h( p, rm)1{[1 � (1 � d)t]B ≤ bdh( p, rm).
In the second stage, farmers make land market decisions to maximize utility.In the pre-1993 economy, the land distribution is not a choice variable, sothat
pre pre pre pre pre pre 0V (a , b , m ) p V(a , b , m ).
In the post-1993 economy, farmers take land prices h as given in order tomaximize their reduced-form utilities, so that
post post post post post postV (a , b , m ) p max V(a , b , m).m
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
542 economic development and cultural change
The land market clearing condition determines land price h, given that.
N post� m p 1iip1
B. Equilibrium OutcomesWe solve for a subgame-perfect equilibrium of the economy described above.Choosing to invest in perennial crops will depend on the ability to borrowin order to cover land purchases and the opportunity cost of intermediateconsumption. Since perennial crops face a risk of expropriation a, farmers willallocate land across crops in order to maximize profits. Finally, depending oninvestment choices, the allocation of land across farmers will be either irrelevant(due to constant returns to scale) or driven by how stringent credit constraintsare. We summarize the main results below, deferring details of the proofs tothe appendix. For ease of exposition, we restrict attention to two extremecases: access to credit is either unrestricted or absent .(b p 1) (b p 0)
Result 1 (crop choice). If expropriation risk a is high enough, then farmersdevote the entire land area to rice cultivation. If a is not too high and accessto credit is unrestricted, then the entire land area is devoted to cultivatingperennial crops. However, if access to credit is absent, then the need forintermediate consumption requires farmers to diversify across the two typesof crops. Formally, the conditions are as follows:
1 11/1�j 1/1�g ∗(jG) ≥ [g(1 � a)H] ⇒ r p 0
j g
1 11/1�j 1/1�g ∗(jG) ! [g(1 � a)H] and b p 1 ⇒ r p 1 (1)
j g
∗ 01 1 [1�h(m �m )](1�g)1/1�j 1/1�g ∗(jG) ! [g(1 � a)H] and b p 0 ⇒ r p � g.
1/1�j ∗j g (1�j/j)(jG) m
The interesting case is when perennial crops are a more productive technologyin the absence of expropriation:
1 11/1�j 1/1�g(jG) ! (gH) . (2)
j g
In this case, if was high enough such thatpre 1/1�ja (1/j)(jG) ≥ (1/g)[g(1 �, then an increase in the security of tenure, that is, a decrease inpre 1/1�ga )H]
a, leads to an increase in the proportional area dedicated to perennials.Result 2 (labor choice). In the absence of credit constraints, aggregate off-
farm employment increases if and only if
1/1�j post 1/1�g(jG) ≥ [(1 � a )gH] . (3)
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 543
A necessary condition for this is that , that is, perennials need to bej 1 g
comparatively less labor intensive than rice cultivation. In the presence ofcredit constraints, is a sufficient condition to have an increase in ag-j 1 g
gregate off-farm employment.Result 3 (consumption). When households have access to credit, they max-
imize aggregate output, and consumption is perfectly smoothed across periods.Thus, when ,postb p 1
post pre post preC 1 C and C 1 C .0 0 1 1
In the presence of credit constraints and in the absence of any land sales orpurchases, that is, , agricultural output at necessarily dropspre postm p m T p 0as farmers decide to partially move away from rice, so that we have
post preC ! C .0 0
Result 4 (land market activity). The first-order condition to determine h,the market price of land, is given by
∗df dg d C0∗ ∗ ∗ ∗ ∗[n (m)] � [r (m), (1 � r (m))m] � h[ p (m), r (m)m] p h. (4)∗dm dm dm C1
So there is activity in the land market only if (i) the initial land allocationwas not efficient and there is access to credit or (ii) households change theirdecisions with regard to perennial crops and off-farm activities and chooseoptimal land allocations accordingly.
IV. Data and Empirical StrategyA. DataOur major source of data is the two rounds of the Vietnam Living StandardsSurvey, conducted by the General Statistical Office of the government ofVietnam and funded by the United Nations Development Program (UNDP)and the Swedish International Development Agency (SIDA). The first roundof the survey was conducted in 1992–93 (henceforth VLSS-93) and the secondround was conducted in 1997–98 (henceforth VLSS-98). We take the formeras our prereform baseline data and the latter as our postreform outcomes.Multistage stratified sampling techniques were used to select 4,800 householdsin the first round. In 1998, the sample size was increased to 6,000, of which4,285 households had been interviewed in the first round as well.6 The surveyswere structured on the lines of the World Bank LSMS (Living Standards
6 The attrition rate is around 7% in rural areas; it is higher for smaller, urban, more educated,and richer households. The overall attrition rate is 11%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
544 economic development and cultural change
TABLE 1CHARACTERISTICS OF SURVEY HOUSEHOLDS SAMPLE: RURAL HOUSEHOLDS
1998 Survey
1993 SurveyPanel
Household HeadsReplacement
Household Heads
Number of households 3,840 3,375 894Age of household head 44.85 47.47 45.63
(14.79) (13.81) (14.07)Household size 4.97 4.79 4.84
(2.12) (1.90) (1.85)Sex of household head
(1 p male) .77 .77 .84(.42) (.42) (.37)
Ethnicity (1 p kinh) .86 .84 .82(.35) (.36) (.38)
Literate household head(1 p literate) .88 .88 .86
(.33) (.33) (.34)Years of education of
household head 5.96 6.48 6.03(4.07) (3.90) (3.85)
Real household expendi-ture (‘000 dong) 5,541.00 10,189.95 11,487.02
(3,856.02) (6,426.11) (7,278.54)Farming as main
occupation .83 .78 .76(.38) (.42) (.43)
Source. Vietnam Living Standards Measurement Study Surveys 1993 and 1998.Note. All means are weighted by sampling weights.
Measurement Study) format and are considered high-quality data. They containdetailed information on household size and composition, educational attain-ment, health, employment, fertility, migration, household expenditures, ag-ricultural activities, nonfarm economic activities, and borrowing and lendingactivities. Table 1 presents some basic characteristics of the rural householdsin the surveys, broken down into whether they were reinterviewed (panel)households or new households. We note the large increase in real householdexpenditure between 1993 and 1998, consistent with the high growth ratesenjoyed by the Vietnamese economy in this period.
Province-level data on progress of Land-Use Certificate issuance (numberof households and communes with LUC), as well as the number of landdepartment officials in each province, come from the records of GDLA inHanoi. We have data on province-level population, agricultural yields, andurbanization from the annual statistical yearbooks published by the GeneralStatistical Office. We also use the 1994 Agricultural and Rural Census con-ducted by the Ministry of Agriculture and Rural Development to constructmeasures of infrastructure facilities in rural areas as well as the extent of statefarms in agriculture.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 545
B. Empirical StrategyIdeally we would like to compare economic activity choices across two house-holds that differ only in the quality of land rights possessed by them. In oursetting, we take the possession of a land-use certificate as an indicator of havinggood land rights. However, we do not have household-level data on landregistration, since the VLSS-98 does not ask this question.7 We therefore usethe province-level proportion of households with LUC as a measure of theprobability that a given household would have an LUC. We will thus be usingthe differences in the level of LUC issuance across provinces to identify theimpact of the land law. We will also use an alternative measure (speed ofregistration) that captures how quickly the province issues LUCs to at least50% of the households, which takes into account the whole process of LUCissuance over time.
We estimate the impact of the land reform using a differences-in-differencesstrategy, by comparing the difference between 1993 and 1998 (before andafter reform) for the high-issuance provinces as compared to the low-issuanceprovinces. The regression equation we use is
y p a � a T � a R � a (T # R ) � X b � e , (5)ijt 0 1 t 2 j 3 t j ijt ijt
where represents the outcome of household i of province j at time t (1993yijt
or 1998), represents the time dummy (equal to zero for 1993 and one forTt
1998), is the measure of land reform implementation in province j (pro-Rj
portion of households with LUC in 1998 or number of years since LUC issuancecrossed 50% threshold [speed measure]), and are other household char-Xijt
acteristics. The two main outcomes we will consider are the proportion ofcultivated area devoted to long-term crops and the fraction of household labordevoted to nonfarm activities. The coefficient represents the change betweena1
1993 and 1998 for a province that had zero LUC issuance, while representsa2
the difference between high-issuance and low-issuance provinces in 1993 (pre-existing difference). Our coefficient of interest is , which tells us how mucha3
the high-issuance provinces have increased investment, compared to the low-issuance provinces over the period 1993–98. We will be controlling for house-hold characteristics like age, gender, and education of the household head;
7 The community questionnaire in VLSS-98 does ask the commune head the number of years forwhich households have the right to use annual and perennial crop land. The answers to this questionare sometimes wrong (e.g., some communes say they have the right to use annual crop land for50 years) and differ across annual and perennial categories (i.e., many communes that report 20years for annual crop land report less than 50 years for perennial) and are only weakly correlatedwith our province-level measures of LUC issuance. We therefore feel that this probably reflectssome perceptions of the commune head rather than actual possession of LUCs.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
546 economic development and cultural change
total household size; ethnicity; and total area cultivated. All our regressionsare for households in rural areas only, since our land reform figures are for therural sector. Since our main explanatory variable, the LUC issuance measure,is measured at the province level, we will also cluster all our standard errorsat the province level (see Bertrand, Duflo, and Mullainathan 2004).
Our identification assumption is that the household-specific error termis uncorrelated with the province-level LUC issuance measure . We shoulde Rijt j
note that our estimate of will be biased upward if there are province ora3
household characteristics that are correlated both with the issuance of LUCsand with our outcome variables or if households plant long-term crops in aneffort to obtain a longer lease term for their land.
The next section examines in greater detail whether LUC issuance is cor-related with province level characteristics and whether investments in long-term crops occur prior to land titling. We also control for all time-invarianthousehold and province characteristics by running specification (5) using onlythe panel households and including household fixed effects as follows:
y p a � a T � a (T # R ) � e ,ijt i 1 t 3 t j ijt
where is a fixed effect for household i. This reduces our sample size sinceai
many households were not interviewed in both years.
V. Implementation of the 1993 Land LawTable 2 documents the progress of LUC issuance in Vietnam. About 24% ofhouseholds had been issued land-use certificates at the end of 1994; by theend of 2000, this proportion had increased to 90%, consistent with the targetof issuing certificates to more than 11 million rural households by the end of2001. Table 2 also shows considerable variation across provinces in the speedof implementing this process. For instance, An Giang province issued LUCsto 91% of households in 1994, at which time Lai Chau and Lang Son hadmade negligible progress. Similarly, the proportion of households with LUCsin 1998 varied from 12% to 100% across different provinces. Another wayof seeing this is the variation in the number of provinces that attained certainlevels of LUC issuance over the years (table 2, panel B). For instance, 21provinces crossed the 25% threshold in 1994; by 1998, 60 out of 61 provinceshad attained this threshold. Sixteen provinces had issued LUCs to 50% ormore of households by 1995, and 48 had attained this threshold by 1998. Asof 2000, five provinces had yet to attain the 75% level of LUC issuance.
The identification strategy in this study relies on the observation that theland law was not implemented homogeneously throughout the country. Ourstrategy is likely to give biased results if the province-level registration levels
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 547
TABLE 2MEASURES OF LAND RIGHTS
A. Proportion of Households Registered
Mean North South Min Max
1994 .237 .246 .227 .001 .9141995 .367 .345 .391 .040 .9601996 .435 .404 .469 .074 .9521997 .632 .597 .667 .111 1.0001998 .713 .687 .740 .119 1.0002000 .902 .893 .911 .533 1.000
B. No. of Provinces withRegistration More Than
25% 50% 75%
1994 21 8 41995 33 16 61996 44 25 141997 57 43 281998 60 48 382000 61 61 56
C. Correlation among Registration Measuresand Manpower Measures
% households registered 1998 1.000% households registered 1996 .644 1.000% households registered 1994 .376 .595 1.000Years since reg 1 25% .492 .691 .697 1.000Years since reg 1 50% .828 .799 .648 .619 1.000Officials/commune 1994 .102 �.033 �.050 �.016 .072 1.000Officials/commune 1998 .113 .250 �.087 .136 .111 .062 1.000Officials/1,000 households 1994 .081 �.118 .077 �.016 .070 .465 �.070 1.000Officials/1,000 households 1998 �.007 �.051 .110 .076 .015 �.018 .304 .702
Source. Figures computed by authors from data provided by General Department of Land Administra-tion (GDLA), Hanoi.Note. Proportion of households registered in panel A refers to the number of households that possessa Land Use Certificate as a fraction of the total number of households. For 1996, 1998, and 2000, GDLAprovided the total number of households. For the remaining years, total number of households isestimated by authors based on total number of agricultural households. The entries in panel B are thenumber of provinces in which the proportion of households registered exceeds the specified threshold(25%, 50%, 75%). “Years since reg 1 25%” is 2001 minus the year in which LUC issuance in the provinceexceeded 25%. “Officials/commune” is the number of land department officials in the province dividedby the total number of communes in the province. “Officials/1,000 households” is the number of landdepartment officials in the province divided by the total number of agricultural households.
are correlated with other province-level characteristics that also affect ourdependent variables. We therefore try to see whether there are any observablesystematic differences between high-LUC-issuance provinces and low-issuanceones. Table 3 reports regressions in which the dependent variable is a measureof the LUC issuance in 1998 and the explanatory variables are province char-acteristics. Our two major measures of the prevalence of land rights due tothe reform are the proportion of households with LUC in 19988 and the speed8 We choose 1998 because our household survey data are from this year; see Sec. IV.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
548
TAB
LE3
WH
AT
DE
TER
MIN
ES
IMP
LEM
EN
TATI
ON
OF
LAN
DR
EFO
RM
?(S
AM
PLE
:P
RO
VIN
CE
S)
%H
ous
eho
lds
Reg
iste
red
1998
Spee
do
fR
egis
trat
ion
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(7)
(8)
Land
offi
cial
s/co
mm
une
(199
4).0
63.0
69.0
55.0
55.8
03.6
18.5
23(.0
72)
(.065
)(.0
52)
(.053
)(.7
20)
(.517
)(.5
24)
Po
pul
atio
nd
ensi
ty19
93�
.012
�.0
12�
.097
�.0
89(.0
15)
(.015
)(.1
15)
(.114
)To
tala
rea
of
pro
vinc
e(’0
00sq
km)
�.1
28�
.099
�.1
35.3
33(.1
23)
(.132
)(.9
20)
(1.0
46)
Pro
po
rtio
nur
ban
1993
.343
.230
�.1
47�
.854
(.310
)(.3
18)
(2.5
87)
(2.5
37)
Pro
po
rtio
nun
der
per
enni
alcr
op
s19
93�
.137
�.1
70�
2.13
3�
2.73
2(.1
91)
(.210
)(3
.039
)(3
.168
)P
rop
ort
ion
und
erp
eren
nial
cro
ps
1994
.005
(.154
)D
umm
yfo
rN
ort
h�
.030
�.0
12.0
33.1
13.1
15�
.723
�.6
65.0
92.5
27(.0
71)
(.071
)(.0
74)
(.120
)(.1
30)
(.646
)(.6
61)
(.589
)(.9
97)
Pro
po
rtio
no
fco
mm
unes
havi
ngm
arke
t.1
421.
539
(.207
)(1
.650
)P
rop
ort
ion
of
com
mun
esha
ving
hig
hway
�.1
84�
2.86
1(.1
59)
(2.0
51)
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
549
Pro
po
rtio
no
fco
mm
unes
havi
ngcl
inic
.400
.320
(.368
)(2
.844
)M
ean
age
of
hous
eho
ldhe
ad(1
993)
.022
**.0
22**
.088
(.009
)(.0
10)
(.086
)M
ean
hous
eho
ldsi
ze(1
993)
.049
.049
.914
(.060
)(.0
61)
(.561
)M
ean
year
so
fed
ucat
ion
(199
3).0
19.0
19.1
57(.0
26)
(.027
)(.2
62)
Log
per
cap
ita
exp
end
itur
e(1
993)
.204
.204
1.67
2(.1
66)
(.166
)(1
.398
)O
bse
rvat
ions
5957
5754
5457
5555
54R
2.0
6.0
7.1
1.1
8.1
8.0
5.0
6.0
9.1
4
No
te.
Ro
bus
tst
and
ard
erro
rsar
ein
par
enth
eses
.“%
hous
eho
lds
reg
iste
red
”is
the
pro
po
rtio
no
fru
ralh
ous
eho
lds
inth
ep
rovi
nce
who
have
aLa
ndU
seC
erti
ficat
e.“S
pee
do
fre
gis
trat
ion”
isco
mp
uted
as20
01m
inus
the
year
whe
n50
%o
fho
useh
old
sin
the
pro
vinc
eo
bta
ined
LUC
s.R
egre
ssio
nsex
clud
eH
ano
iand
Ho
Chi
Min
hC
ity,
whi
char
eo
utlie
rsin
term
so
fp
op
ulat
ion
den
sity
and
urb
aniz
atio
nra
tes.
Reg
ress
ions
(6)
and
(7)
excl
ude
old
Song
Be
pro
vinc
e,w
hich
isan
out
lier.
**Si
gni
fican
tat
5%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
550 economic development and cultural change
of registration measure that we compute as 2001 minus the year in whichLUC issuance reached 50% of households. While the first measure capturesthe status of issuance at a point of time, the second measure is based on thewhole process of land titling over all the years. These two measures are highlycorrelated (correlation p 0.83).
As mentioned in Section II.C, land officials in Vietnam cited two mainsources of delay in land titling: lack of manpower and the time taken to resolvedisputes. To see whether this is indeed the case, we obtained (from GDLA)data on land department manpower at the province level in different years.We find that most provinces had less than two land officials per commune inthese years. As expected, the number of land officials per commune (or per1,000 agricultural households) is somewhat positively correlated with measuresof the speed of LUC issuance (table 2, panel C). However, these correlationsare not very large, and none of them are statistically significant. We wereunable to obtain any quantitative information on the number of disputes ineach province.
Table 3 shows that land registration is not strongly correlated with theland department manpower levels or any other province characteristics suchas population density, urbanization, proportion of communes having a market,mean level of education, or per capita household expenditure. The results arequalitatively similar when we use land officials per 1,000 agricultural house-holds instead of land officials per commune, when we include paddy yieldsas an additional explanatory variable, and when we use proportion of householdswith LUC in years other than 1998 as a measure of the progress of land reform(results available upon request). We also tried specifications including weathervariables (rainfall and sunshine hours recorded in 1993) as well as a dummyfor whether the province was already reporting coffee production in 1996:these variables do not have much explanatory power for our dependent variable;in particular, the coefficient on the coffee dummy is negative and insignificant.
We also considered the possibility that households might be planting long-term crops in advance of the LUC issuance in an effort to get a longer leaseterm for their land. If this is a major reason for differences in LUC issuance,we would expect to find a significant relationship between early investmentsin long-term crops and the extent of LUC issuance in 1998. To check for this,we regressed the LUC issuance on the proportion of land under perennialcrops estimated from the 1994 Agricultural and Rural Census. By 1994, thelaw had been announced, and several provinces had already made progress inimplementing it. We find that this is not a significant predictor of LUCissuance in 1998 (table 3, col. 5). It could also be that the provinces differon dimensions such as “good governance” or “progressiveness,” which might
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 551
reflect itself in both faster LUC issuance and greater incentives toward long-term crops. As a partial check for this, we included as an additional regressorthe 2006 Provincial Competitiveness Index computed by the Vietnam Cham-ber of Commerce and Industry and the Vietnam Competitiveness Initiative(VNCI) and funded by the U.S. Agency for International Development. Again,this index is not a significant predictor of LUC issuance.9 While our data andreform predate the index, this is a reasonable proxy for province-level gov-ernance quality if such a characteristic changes slowly over time.
We note further that there is no significant difference in the implementationof the reform in the North and the South, despite the North’s longer historyof collectivization: in 1994, provinces in the North had on average 24% ofhouseholds registered, while provinces in the South had a registration levelof 23%. The corresponding figures for 1998 were 74% and 69%. The mapin figure 1 also demonstrates no specific geographical pattern in the progressof land reform across provinces. While it is still possible that there are someunobservable differences between high-issuance and low-issuance provinces, itis reassuring for our identification strategy that LUC issuance is not system-atically related to any of the several different observable province characteristics.Since it is extremely rare to find a “natural experiment” that assigns landrights in a truly random fashion (Galiani and Schargrodsky [2006] is anexception), this is probably the best that can be done to ensure comparabilitybetween treatment and control groups. The fact that our results are robust torestricting our sample to only the panel households also confirms that ourresults are not driven by some preexisting province or household time-invariantcharacteristics.
VI. Land Rights and Investment Decisions
We first look at the impact of the 1993 land law on crop and labor decisionsof rural households.
A. Crop Choice
As discussed earlier, the additional land rights conferred by the 1993 landlaw might induce households to undertake more long-term investments ontheir land. One way of measuring this is by looking at the allocation of landbetween annual crops and multiyear industrial or fruit crops, which typicallyyield returns only after a few years. The major multiyear crops grown by
9 Results available from the authors on request. The results are unchanged when we use only thenonland components of the index.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
552 economic development and cultural change
Figure 1. Geographical distribution of registration levels (1994–98)
Vietnamese farmers are coffee, tea, rubber, black pepper, and cashews; fruitcrops include citrus fruits, pineapples, bananas, and mangoes.
The difference-in-differences strategy is outlined in table 4, panel A. Prov-inces have been divided into two categories, depending on whether the pro-portion of households with land-use certificates in 1998 was above or belowthe sample median (80%). The numbers reported in panel A are the proportionsof cultivated land devoted to perennial industrial and fruit crops in each year,averaged across all rural households, in the two categories of provinces. Thus,the differences in the bottom row are the differences across time for each ofthe two categories of provinces, while the differences reported on the rightcolumn show for each year the difference in outcomes between low and highregistration provinces. Finally, the bottom right cell computes the difference-in-differences outcome: a household in a highly registered province in 1998on average increases its share of cultivated land area devoted to perennial
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 553
industrial and fruit crops by 5.6 percentage points more than a household ina low-registration province.
Panel B of table 4 presents a continuous version of panel A, followingspecification (5). Consistent with the results of panel A, panel B shows thatthe land reform led to a statistically significant increase in the proportion oftotal cultivated area devoted to multiyear crops: a household in a province inwhich everybody had an LUC would increase this proportion by 7.5 percentagepoints over the period 1993–98, compared to a household in a province inwhich nobody was issued an LUC (panel B, col. 2). This means that if theproportion of households registered goes up by one standard deviation, theproportional area devoted to long-term crops will increase by 0.09 standarddeviations. This increase comes at the expense of annual crops, which show adecrease of 6.5 percentage points in their share of total cultivated area (re-gressions not shown).10 We control for household characteristics like age,education, gender, household size, and ethnicity, as well as region fixed effects,while obtaining these estimates. The coefficients are also robust to otherchanges in the base specification, such as adding the household controls in-teracted with the time dummy, and to adding household income in 1993 orprovince-level mean per capita income as additional regressors (regressionsavailable upon request).
We perform a variety of robustness checks for our results. Table 4, column3, reports the results when we restrict our sample to panel households only,who are interviewed in both 1993 and 1998, and use household fixed effectsto control for any time-invariant household characteristics. The results fromthis are similar to the ones for the full sample, although the coefficient issmaller in magnitude.11 This is not surprising, given that the attenuation biasof measurement error gets exacerbated in a panel regression.
We get very similar results when we use our speed of LUC issuance measureinstead of the level of issuance in 1998 as the main explanatory variable: theresults in table 4, column 4, indicate that if a province were to reach the50% issuance level 1 year earlier, households in that province would on averageincrease the relative area devoted to long-term crops by 1.3 percentage points(0.06 standard deviations). Further, our results are also robust to alternativemeasures of crop choice: we find positive and statistically significant resultswhen we use a dummy for whether the household cultivates any long-term
10 We should note that in the household surveys, some respondents answer questions about long-term crops by mentioning the number of trees they have rather than the area devoted to them.Our results are robust to several ways of converting trees to areas.11 The results are identical when we control for time-varying effects of household characteristics.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
554
TAB
LE4
LAN
DR
IGH
TSA
ND
CR
OP
CH
OIC
E(S
AM
PLE
:R
UR
AL
HO
USE
HO
LDS)
A.
Diff
eren
ce-in
-Diff
eren
ceE
stim
ate
(Dis
cret
e)
%H
ous
eho
lds
wit
hLU
C
Bel
ow
Med
ian
( !80
%)
Ab
ove
Med
ian
(180
%)
Diff
eren
ce
1993
Surv
ey.1
017
.087
0�
.014
7(.0
347)
1998
Surv
ey.0
959
.137
2.0
413
(.038
2)D
iffer
ence
�.0
058
.050
2.0
560
(.014
9)(.0
132)
***
(.019
9)**
*
B.
Diff
eren
ce-in
-Diff
eren
ceE
stim
ate
(Co
ntin
uous
)
No
Co
ntro
ls(1
)R
egio
nFE
(2)
Pan
elH
ous
eho
lds
(3)
Alt
erna
tive
Land
Rig
hts
Mea
sure
(4)
Dis
cret
eM
easu
re(5
)%
Per
enni
al(6
)%
Frui
t(7
)
Land
rig
hts
#Ye
arp
1998
.080
***
.075
***
.053
*.0
13**
.148
*.0
41**
.034
**(.0
25)
(.025
)(.0
31)
(.005
)(.0
88)
(.021
)(.0
15)
Year
p19
98�
.033
�.0
33*
�.0
19�
.029
*.0
25�
.025
*�
.008
(.020
)(.0
17)
(.035
)(.0
17)
(.068
)(.0
13)
(.008
)La
ndri
ght
s.0
38�
.011
�.0
10*
�.0
87�
.003
�.0
08(.0
47)
(.027
)(.0
06)
(.094
)(.0
23)
(.010
)
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
555
Ag
eo
fho
useh
old
head
.002
***
.002
***
.004
***
.001
***
.001
***
(.000
)(.0
00)
(.001
)(.0
00)
(.000
)M
ale
hous
eho
ldhe
ad�
.020
**�
.021
**�
.010
�.0
17**
�.0
03(.0
10)
(.010
)(.0
15)
(.007
)(.0
05)
Year
so
fed
ucat
ion
of
head
.008
***
.008
***
.014
***
.005
***
.003
***
(.002
)(.0
02)
(.002
)(.0
01)
(.001
)H
ous
eho
ldsi
ze�
.002
�.0
03.0
11**
*�
.002
�.0
00(.0
02)
(.002
)(.0
04)
(.002
)(.0
01)
Maj
ori
tyet
hnic
gro
upd
umm
y.0
03.0
03�
.039
�.0
05.0
08(.0
30)
(.029
)(.0
58)
(.031
)(.0
08)
Tota
lare
acu
ltiv
ated
(#10
�5)
�.2
08**
�.2
14**
.127
�.0
56�
.151
***
(.090
)(.0
95)
(.104
)(.0
70)
(.046
)R
egio
nfix
edef
fect
sN
oYe
sN
oYe
sYe
sYe
sYe
sH
ous
eho
ldfix
edef
fect
sN
oN
oYe
sN
oN
oN
oN
oN
o.
of
ob
serv
atio
ns7,
469
7,46
96,
012
7,46
97,
469
7,46
97,
469
No
.o
fp
rovi
nces
5959
5959
5959
59R
2.0
1.2
2.8
1.2
2.0
7.2
2.0
6
No
te.
Dep
end
ent
vari
able
pP
rop
ort
ion
of
tota
lcu
ltiv
ated
area
dev
ote
dto
per
enni
alin
dus
tria
lcr
op
san
dfr
uit
cro
ps.
Stan
dar
der
rors
are
inp
aren
thes
es,
corr
ecte
dfo
rp
rovi
nce-
leve
lcl
uste
ring
.A
llre
gre
ssio
nsar
ew
eig
hted
by
sam
plin
gw
eig
hts.
“Lan
dri
ght
s”is
the
pro
vinc
e-le
vel
pro
po
rtio
no
fho
useh
old
sho
ldin
ga
Land
Use
Cer
tific
ate
(LU
C).
Reg
ress
ion
(5)u
ses
anal
tern
ativ
em
easu
reo
flan
dri
ght
s:th
esp
eed
ofr
egis
trat
ion
mea
sure
,whi
chis
2001
min
usth
eye
arin
whi
chLU
Cis
suan
cecr
oss
es50
%.“
Dis
cret
em
easu
re”
equa
lso
neif
the
hous
eho
ldre
po
rts
cult
ivat
ing
any
per
enni
alcr
op
san
d/o
rfr
uit
tree
san
dze
roo
ther
wis
e.%
per
enni
alis
the
pro
po
rtio
no
fto
talc
ulti
vate
dar
ead
evo
ted
top
eren
nial
ind
ustr
ialc
rop
s.%
frui
tis
the
pro
po
rtio
no
fto
talc
ulti
vate
dar
ead
evo
ted
tofr
uit
cro
ps.
*Si
gni
fican
tat
10%
.**
Sig
nific
ant
at5%
.**
*Si
gni
fican
tat
1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
556 economic development and cultural change
crops, as well as when we analyze multiyear crops and fruit crops separately.Finally, we should note that the coefficient on the land rights variable ( ina2
eq. [5]) is usually negative, consistent with the results reported in table 3.We thus feel that any bias in our estimates is likely to be downward ratherthan upward.
Another reason why our results are likely to be underestimates is the in-stitutional setting in Vietnam. While there is a fair amount of free choicegiven to households in choosing their crops, there are two main constraints.First, there can be externalities that might lead to restrictions on changingcrops: a typical example of this is the planting of a tree that might cast ashadow on the neighbor’s crops. Second, until recently, provincial level au-thorities planned aggregate crop allocations for food security purposes. Beyondthese minimum quotas, districts and then communes would not receive stricterdirectives in terms of crop choice. Both of these obviously limit the size ofthe impact we can hope to find.
Given that regulatory and other concerns might result in provincial gov-ernment influences on crop choice, we would like to verify that our resultsare not driven wholly by such influences. One way to do this is to directlycontrol for government involvement in agricultural activity. For this, we com-pute agricultural areas cultivated by state farms for each province, using the1994 Agricultural Census. We then use this variable as a control in ourregressions. As can be seen in appendix table A1, our main coefficients ofinterest are largely unchanged and, in fact, are slightly larger than in the basespecification.
B. Labor ChoiceThe 1993 land law is found to have had a significant impact on the numberof weeks worked in nonfarm activities per working member in the household(table 5). Our base estimates from panel B, column 3, indicate that a householdin a high LUC-issuance province increases its nonfarm activity by 2.7 weeksper working member between 1992–93 and 1997–98. The most commonnonfarm activities reported by households in the VLSS were sales in markets,food processing, woodworking, and work in the textiles and garment industry.In terms of standard deviations, our estimates mean that a one standard de-viation increase in the level of LUC issuance would increase nonfarm weeksworked per household member by 0.06 standard deviations. While the estimatemay not be too large in terms of weeks worked per household member, itcorresponds to an increase of between 11 and 12 weeks worked in the nonfarmsector for the household, as the average number of working members in thehousehold was 4.37 in 1992–93 and 4.46 in 1997–98.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 557
We perform robustness checks for these results in columns 4–6 in table 5,panel B. Results for panel households indicate an increase of 3.2 weeks workedper household member in the nonfarm sector (col. 4); this is quite close toour base estimate of 2.7 weeks. Our results are robust to using the speedmeasure of land rights: attaining the 50% level of LUC issuance one yearsooner results in an increase of 0.3 weeks worked per member in nonfarmactivities (col. 5). We also check the results using the number of hours workedper week in the previous 7 days as an alternative dependent variable. Theresults are fairly close to our base specification: assuming that households work8 hours a day and 52 weeks a year, the estimate of 0.365 hours per weekfrom column 6 translates to an increase of 2.37 weeks per year. Appendixtable A2 verifies that these results are not driven solely by state involvementin the economy, by controlling for the labor employed by state farms in 1994.
C. Results in Different SubsamplesTable 6 investigates these results further in different subsamples. The resultsseem to indicate that the land reform increased the proportional area devotedto long-term crops by a larger margin for households that owned more land(panel A, cols. 4 and 5). The impact is a little larger for households that livein communes where credit institutions are located nearby (cols. 6 and 7) andwhere there are alternative sources of credit available (cols. 8 and 9). Thisseems to indicate a role for credit institutions, though the estimates in columns6 and 8 are not statistically different from those in columns 7 and 9, respec-tively.12 We investigate the role of credit in greater detail in Section VII.B.The impact is also somewhat higher in provinces with more land inequalityin 1993 (cols. 10 and 11), indicating that the land distribution might playa role; we investigate this possibility further in Section VII.A. The results areremarkably similar across poor and rich households (cols. 2 and 3).
Similar analysis for the labor choice variable (table 6, panel B) indicatesthat the impact is very similar across poor and rich households but is greaterfor households having less land to begin with. The impact on labor choice ishigher for households living in communes in which credit institutions arefurther away and in which there are no alternative sources of credit: this isconsistent with the idea that nonfarm activities can be used as a consumption-smoothing device when credit markets do not function perfectly. The resultsare also somewhat higher in areas with higher land inequality. Again, as forcrop choice, none of these differences across subcategories is statistically sig-
12 The test of significance is based on combining these regressions and estimating triple interactioncoefficients. We chose to report separate regressions in the paper for expositional simplicity.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
558
TAB
LE5
LAN
DR
IGH
TSA
ND
NO
NFA
RM
AC
TIV
ITIE
S(S
AM
PLE
:R
UR
AL
HO
USE
HO
LDS)
A.
Diff
eren
ce-in
-Diff
eren
ceE
stim
ate
(Dis
cret
e)
%H
ous
eho
lds
wit
hLU
C
Bel
ow
Med
ian
( !80
%)
Ab
ove
Med
ian
(180
%)
Diff
eren
ce
1993
Surv
ey8.
016
7.22
7�
.789
(1.1
7)19
98Su
rvey
8.34
69.
043
.697
(1.2
1)D
iffer
ence
.330
1.81
61.
486
(.586
)(.3
28) *
**(.6
66)*
**
B.
Diff
eren
ce-in
-Diff
eren
ceE
stim
ate
(Co
ntin
uous
)
No
Co
ntro
ls(1
)
Ho
useh
old
Cha
ract
eris
tics
(2)
Reg
ion
FE(3
)
Pan
elH
ous
eho
lds
(4)
Alt
erna
tive
Land
Rig
hts
Mea
sure
(5)
Dep
end
ent
Vari
able
p
Ho
urs
Mea
sure
(6)
Land
rig
hts
#Ye
arp
1998
3.27
2***
2.80
8**
2.68
5**
3.46
2**
.299
*.3
65*
(1.1
94)
(1.1
58)
(1.1
46)
(1.3
46)
(.157
)(.1
96)
Year
p19
98�
1.32
2�
1.03
0�
.941
�.8
02�
.211
�.1
31(.9
44)
(.895
)(.8
89)
(2.2
15)
(.722
)(.1
29)
Land
rig
hts
�1.
547
�2.
911
�3.
306
�.4
02�
.368
(2.3
54)
(2.2
41)
(2.3
60)
(.343
)(.3
12)
Ag
eo
fho
useh
old
head
�.0
29**
�.0
31**
�.0
32**
�.0
05**
*(.0
12)
(.013
)(.0
13)
(.002
)
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
559
Mal
eho
useh
old
head
1.89
9***
1.91
9***
1.92
9***
.185
***
(.492
)(.4
83)
(.478
)(.0
62)
Year
so
fed
ucat
ion
of
head
.376
***
.380
***
.379
***
.042
***
(.066
)(.0
65)
(.063
)(.0
09)
Ho
useh
old
size
�.2
89**
*�
.293
***
�.2
93**
*�
.017
(.094
)(.0
89)
(.090
)(.0
13)
Maj
ori
tyet
hnic
gro
upd
umm
y3.
921 *
**3.
358*
*3.
303*
*.5
69**
*(.9
12)
(1.2
79)
(1.2
47)
(.171
)D
umm
yfo
rN
ort
h�
2.77
2 **
(1.0
52)
Reg
ion
fixed
effe
cts
No
No
Yes
No
Yes
Yes
Ho
useh
old
fixed
effe
cts
No
No
No
Yes
No
No
No
.o
fo
bse
rvat
ions
8,09
18,
091
8,09
16,
835
8,09
18,
094
No
.o
fp
rovi
nces
5959
5959
5959
R2
.00
.06
.07
.77
.07
.06
No
te.
Dep
end
ent
vari
able
pN
umb
ero
fw
eeks
wo
rked
inno
nfar
mac
tivi
ties
(per
wo
rkin
gm
emb
er)i
nth
ela
st12
mo
nths
.Sta
ndar
der
rors
are
inp
aren
thes
es,c
orr
ecte
dfo
rp
rovi
nce-
leve
lcl
uste
ring
.A
llre
gre
ssio
nsar
ew
eig
hted
by
sam
plin
gw
eig
hts.
“Lan
dri
ght
s”is
the
pro
vinc
e-le
vel
pro
po
rtio
no
fho
useh
old
sho
ldin
ga
Land
Use
Cer
tific
ate
(LU
C).
Reg
ress
ion
(5)u
ses
anal
tern
ativ
em
easu
reo
fla
ndri
ght
s:th
esp
eed
of
reg
istr
atio
nm
easu
re,w
hich
is20
01m
inus
the
year
inw
hich
LUC
issu
ance
cro
sses
50%
.“H
our
sm
easu
re”
isth
enu
mb
ero
fho
urs
wo
rked
inno
nfar
mac
tivi
ties
(per
wo
rkin
gm
emb
er)
inth
ep
revi
ous
7d
ays.
*Si
gni
fican
tat
10%
.**
Sig
nific
ant
at5%
.**
*Si
gni
fican
tat
1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
560
TAB
LE6
LAN
DR
IGH
TS,
CR
OP
CH
OIC
E,
AN
DLA
BO
RC
HO
ICE
IND
IFFE
RE
NT
SUB
CA
TEG
OR
IES
(SA
MP
LE:
PAN
EL
HO
USE
HO
LDS)
1993
Inco
me
Qui
ntile
1993
Land
Ho
ldin
gD
ista
nce
toC
red
itIn
stit
utio
nsA
lter
nati
veC
red
itP
rovi
ncia
lLan
dG
ini
Bas
eSp
ecifi
cati
on
(1)
q3
(2)
13
(3)
!7,
000
Squa
reM
iles
(4)
17,
000
Squa
reM
iles
(5)
kmq
5 (6)
km1
5 (7)
Yes
(8)
No
(9)
Low
(10)
Hig
h(1
1)
A.
Dep
end
ent
Vari
able
pP
rop
ort
ion
of
Tota
lCul
tiva
ted
Are
aD
evo
ted
toP
eren
nial
Ind
ustr
ialC
rop
san
dFr
uit
Cro
ps
Land
rig
hts
#Ye
arp
1998
.053
*.0
51*
.052
.047
*.0
60.0
60*
.046
.110
*.0
34.0
30.1
17(.0
31)
(.029
)(.0
49)
(.029
)(.0
42)
(.032
)(.0
55)
(.058
)(.0
35)
(.026
)(.0
74)
Year
p19
98�
.014
�.0
19�
.003
�.0
07�
.022
�.0
10�
.015
�.0
47�
.004
�.0
12�
.036
(.020
)(.0
17)
(.036
)(.0
16)
(.027
)(.0
20)
(.035
)(.0
44)
(.021
)(.0
19)
(.049
)H
ous
eho
ldfix
edef
fect
sYe
sYe
sYe
sYe
sYe
sYe
sYe
sYe
sYe
sYe
sYe
sN
o.
of
ob
serv
atio
ns6,
012
3,62
02,
392
3,03
02,
962
2,70
03,
252
1,38
44,
568
3,48
22,
530
R2
.80
.77
.83
.81
.80
.81
.79
.75
.82
.78
.80
B.
Dep
end
ent
Vari
able
pN
umb
ero
fW
eeks
Wo
rked
inN
onf
arm
Act
itiv
itie
sin
Last
12M
ont
hs,
per
Wo
rkin
gM
emb
er
Land
rig
hts
#Ye
arp
1998
3.24
6**
2.90
0*2.
886
3.92
1*2.
295
2.91
6*3.
584*
2.86
33.
422*
*3.
027*
*3.
653
(1.2
94)
(1.6
39)
(3.5
89)
(2.0
30)
(1.5
58)
(1.6
03)
(2.0
05)
(4.7
71)
(1.6
94)
(1.5
11)
(2.4
45)
Year
p19
98�
.824
�.3
91�
.937
�1.
328
�.0
45�
.405
�1.
195
�.7
95�
.838
�.9
83�
.714
(1.1
32)
(1.3
71)
(2.7
71)
(1.7
55)
(1.2
35)
(1.2
71)
(1.7
86)
(4.0
97)
(1.4
29)
(1.4
14)
(1.9
01)
Ho
useh
old
fixed
effe
cts
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
.o
fo
bse
rvat
ions
6,83
54,
545
2,29
03,
307
3,04
73,
026
3,68
61,
647
5,06
53,
687
3,14
8R
2.7
6.8
0.8
7.7
3.7
1.7
7.7
5.7
6.7
6.7
4.7
9
No
te.
Stan
dar
der
rors
are
inp
aren
thes
es,
corr
ecte
dfo
rp
rovi
nce-
leve
lcl
uste
ring
.A
llce
llsre
flect
resu
lts
of
reg
ress
ions
of
the
dep
end
ent
vari
able
on
land
rig
hts.
“Lan
dri
ght
s”is
the
pro
vinc
e-le
vel
pro
po
rtio
no
fho
useh
old
sho
ldin
ga
Land
Use
Cer
tific
ate
(LU
C).
1993
inco
me
qui
ntile
sar
eb
ased
on
hous
eho
ldex
pen
dit
ure.
Low
erq
uint
iles
imp
lyp
oo
rer
hous
eho
lds.
“Dis
tanc
eto
cred
itin
stit
utio
ns”
isth
eav
erag
ed
ista
nce
toal
lcre
dit
inst
itut
ions
inth
eco
mm
une,
mea
sure
dfr
om
the
Peo
ple
’sC
om
mit
tee
of
the
com
mun
e.“A
lter
nati
vecr
edit
”re
fers
toth
ep
rese
nce
of
acr
edit
inst
itut
ion
inth
eco
mm
une
that
isno
ta
go
vern
men
tb
ank.
All
reg
ress
ions
are
for
pan
elho
useh
old
so
nly.
*Si
gni
fican
tat
10%
.**
Sig
nific
ant
at5%
.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 561
nificant, so the results are only indicative and not conclusive. In the nextsection, we investigate the mechanisms behind our main results more closely.
VII. What Are the Mechanisms at Work?In the previous section, we found an increase in long-term investments, mea-sured by the percentage of agricultural land devoted to long-term perennialcrops, as well as an increase in labor devoted to nonagricultural activities. Inthis section, we try to deepen our understanding of the mechanisms underlyingthe observed transition, based on our results in Section III.
A. Land MarketsAnalysis of land market transactions is complicated by the possibility of sub-stantial underreporting by respondents. This is both because land transactionswere illegal before the land law of 1993 (but were nevertheless taking placeon an informal basis) and because there was a high tax (10%) imposed onland transactions until 1999.13 Nevertheless, there is an increase in reportedland market transactions between 1993 and 1998: the proportion of householdsthat report receiving land increases tenfold from 2.5% in 1993 to 25% in1998, and a similar tenfold increase is seen for households reporting sales ofland (from 1% to 10%). This could, however, simply reflect less underreportingafter the law was passed. The increase does not seem to be very different acrosshigh-issuance versus low-issuance provinces (see table 7). The participation ofhouseholds in the land rental market also increases between 1993 and 1998but, again, is not systematically related to the progress of land certificateissuance (results not shown).14 Our theoretical framework and the results ofthe next section suggest that this is most probably because of credit constraints,although it is possible that the initial allocation itself was the optimal one,as suggested by Ravallion and van de Walle (2003).
A province-level analysis of land distribution also suggests that there is notmuch difference between provinces with high and low LUC issuance (see table8). Province-level Gini coefficients of land ownership decline over time (panelA). The decrease is mainly for provinces in the South (from 0.58 to 0.50) thatstarted at much higher levels of inequality than the North, where the Giniremains constant at around 0.37. There is no difference in the change in Giniby province LUC issuance levels. Further, there is not much evidence that
13 Authors’ conversation with GDLA officials (July 2001).14 We should note that the data on renting are not fully comparable across the two surveys: the1993 survey asks separately about land rented in and land sharecropped, while the 1998 surveydoes not ask about sharecropping. In our analysis, we include sharecropping as part of land rentedin.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
562
TAB
LE7
LAN
DR
IGH
TSA
ND
LAN
DTR
AN
SAC
TIO
NS
Dis
tanc
eto
Cre
dit
Inst
itut
ions
Alt
erna
tive
Cre
dit
No
Co
ntro
ls(1
)R
egio
nFE
(2)
Pan
elH
ous
eho
lds
(3)
Alt
erna
tive
Land
Rig
hts
(4)
kmq
5 (5)
kmQ
5 (6)
Yes
(7)
No
(8)
A.
Dep
end
ent
Vari
able
p1
IfH
ous
eho
ldR
epo
rts
Acq
uiri
ngLa
ndin
Last
One
Year
Land
rig
hts
#Ye
ar19
98.0
03.0
19�
.023
.001
.096
4�
.171
0.3
970*
*�
.141
0(.0
96)
(.097
)(.1
42)
(.012
)(.1
416)
(.200
2)(.1
540)
(.156
1)La
ndri
ght
s.0
02.0
08.0
01(.0
12)
(.042
)(.0
06)
Year
p19
98.2
21**
*.2
09**
*.2
27**
.219
***
.146
8.3
371*
*�
.106
1.3
154*
*(.0
67)
(.070
)(.0
99)
(.053
)(.0
962)
(.145
9)(.0
985)
(.109
7)H
ous
eho
ldch
arac
teri
stic
sN
oYe
s…
Yes
……
……
Reg
ion
fixed
effe
cts
No
Yes
…Ye
s…
……
…H
ous
eho
ldfix
edef
fect
sN
oN
oYe
sN
oYe
sYe
sYe
sYe
sO
bse
rvat
ions
8,10
88,
108
6,84
88,
108
3,03
13,
694
1,64
95,
076
R2
.10
.15
.57
.15
.57
.57
.58
.57
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
563
B.
Dep
end
ent
Vari
able
p1
IfH
ous
eho
ldR
epo
rts
Selli
ngLa
ndin
Last
One
Year
Land
rig
hts
#Ye
arp
1998
�.0
86�
.079
�.0
93�
.011
�.0
775
�.1
482
.070
1�
.166
8*(.0
55)
(.057
)(.0
77)
(.008
)(.0
831)
(.122
9)(.0
732)
(.094
9)La
ndri
ght
s.0
11*
�.0
02.0
01(.0
06)
(.013
)(.0
02)
Year
p19
98.1
56**
*.1
49**
*.1
67**
*.1
36**
*.1
680*
*.2
053*
*.0
562
.218
5(.0
45)
(.047
)(.0
64)
(.038
)(.0
655)
(.102
9)(.0
578)
(.077
2)H
ous
eho
ldch
arac
teri
stic
sN
oYe
s…
Yes
Reg
ion
fixed
effe
cts
No
Yes
…Ye
s…
……
…H
ous
eho
ldfix
edef
fect
sN
oN
oYe
sN
oYe
sYe
sYe
sYe
sO
bse
rvat
ions
8,10
88,
108
6,84
88,
108
3,03
13,
694
1,64
95,
076
R2
.04
.05
.53
.05
.52
.54
.51
.54
No
te.
Stan
dar
der
rors
are
inp
aren
thes
es,
corr
ecte
dfo
rp
rovi
nce-
leve
lcl
uste
ring
.A
llre
gre
ssio
nsar
ew
eig
hted
by
sam
plin
gw
eig
hts.
“Lan
dri
ght
s”is
the
pro
vinc
e-le
vel
pro
po
rtio
no
fho
useh
old
sho
ldin
ga
Land
Use
Cer
tific
ate
(LU
C).
“Dis
tanc
eto
cred
itin
stit
utio
ns”
isth
eav
erag
ed
ista
nce
toal
lcre
dit
inst
itut
ions
inth
eco
mm
une,
mea
sure
dfr
om
the
Peo
ple
’sC
om
mit
tee
of
the
com
mun
e.“A
lter
nati
vecr
edit
”re
fers
toth
ep
rese
nce
of
acr
edit
inst
itut
ion
inth
eco
mm
une
that
isno
ta
go
vern
men
tban
k.“H
ous
eho
ldco
ntro
ls”
incl
ude
age,
gen
der
,an
ded
ucat
ion
of
hous
eho
ldhe
ad;
hous
eho
ldsi
ze;
ethn
icit
y;an
da
dum
my
for
No
rth.
Reg
ress
ions
(5)–
(8)
are
for
pan
elho
useh
old
so
nly.
*Si
gni
fican
tat
10%
.**
Sig
nific
ant
at5%
.**
*Si
gni
fican
tat
1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
564 economic development and cultural change
TABLE 8LAND RIGHTS AND LAND INEQUALITY (SAMPLE: PROVINCES)
% Households with LUC
Below Median(!80%)
Above Median(180%) Difference
A. Province Gini coefficient:1993 .4730 .4762 .0033
(.0467)1998 .4439 .4371 �.0069
(.0383)Difference �.0290 �.0391 �.0102
(.0425) (.0424) (.0603)B. Proportion of landless in province:
1993 .1191 .1159 �.0032(.0403)
1998 .0706 .0695 �.0011(.0277)
Difference �.0485 �.0464 .0021(.0350) (.0337) (.0486)
Note. Standard errors are in parentheses. Gini coefficients are computed by the authors from household-level data on landholdings.
formalizing land rights creates increasing landlessness, because people are nowable to sell land in times of need (table 8, panel B). The proportion of landlesshouseholds decreases from 11.4% to about 7% overall, but the trends aresimilar across high-issuance and low-issuance provinces.
B. Credit MarketsAs outlined earlier, the land reform can affect crop choice both by increasingthe incentive to invest in long-term crops and by making access to creditmarkets easier as farmers can now pledge land as collateral for a loan. Inparticular, having a pledgeable land-use certificate might be expected to in-crease a household’s access to credit from formal sources such as banks andcredit cooperatives. Table 9, however, shows that the issuance of LUCs doesnot lead to households being more likely to have an outstanding loan. Ap-proximately half of all households in our sample have outstanding loans (51%in 1993 and 54% in 1998): the probability of having a loan is 11 percentagepoints lower for households in high-issuance provinces (panel A, col. 2). Thisdifference is, however, statistically insignificant. Households in highly regis-tered provinces also do not show an increase in the proportion of loans fromformal sources (panel B), nor do they show any increase in the amount theyborrow (results available on request). We also see that households in communeswith lower access to credit institutions show a marginally significant declinein the probability of having an outstanding loan over this period (panel A,col. 6).
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 565
Overall, we do not find evidence that the 1993 land law resulted in increasedaccess to credit, which is somewhat surprising given that the right to mortgagewas one of the rights conferred by the law. However, in practice, it is veryhard for banks to seize land in cases of default, partly because regulations todeal with defaulting households had not been fully clarified at that time andalso because commune officials were not likely to support transfers of land topeople outside the village. Thus the usual practice in case of default, of theVietnam Bank of Agriculture and Rural Development, for instance, was totry and reschedule the loans or to stop lending to the defaulting householdin future.15
C. Within Households or Between Households?
Our model predicts that under the assumption of perennials being a labor-saving technology, the shift toward perennials should take place alongside adiversification of labor off the farm. An alternative mechanism implies spe-cialization across households: some households would specialize in agriculture,while others would devote time to off-farm employment. In order to findadditional evidence to discriminate these two potential channels, we testwhether households that undertake crop changes are also the same householdsthat devote more time off the farm. The results in table 10 indicate that thechanges in crop choice and labor choice documented in Section VI occur mainlywithin households, lending little support for a specialization channel. Weconduct this analysis as follows: we restrict attention to panel households andcompute the change in crop and labor choices between 1992–93 and 1997–98.Depending on whether these differences are positive (denoted 1) or nonpositive(denoted 0), households are divided into four categories (00, 01, 10, and 11,respectively).16 We then conducted a multinomial logit regression of thiscategorical variable on the measure of land rights, where the base category(denoted 00) was chosen to be the category of households that did not increaseeither their multiyear crop cultivation or their nonfarm weeks of work. Table9 reports the relative risk ratios that indicate how much more likely we areto observe a given category rather than the base category. In particular, anincrease of the level of LUC issuance from 0% to 100% makes it 3.169 timesmore likely for a given household to be of category 11 than of category 00.The transition toward the two other categories (namely, 01 and 10) is notsignificant. These results, together with those of Section VII.A, suggest that
15 Authors’ conversation with Vietnam Bank for Agriculture and Rural Development officials (July2001).16 The medians of such differences are close to zero.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
566
TAB
LE9
LAN
DR
IGH
TSA
ND
CR
ED
IT
Dis
tanc
eto
Cre
dit
Inst
itut
ions
Alt
erna
tive
Cre
dit
No
Co
ntro
ls(1
)
Ho
useh
old
Cha
ract
eris
tics
�R
egio
nFE
(2)
Pan
elH
ous
eho
lds
(3)
Alt
erna
tive
Land
Rig
hts
(4)
kmq
5 (5)
km1
5 (6)
Yes
(7)
No
(8)
A.
Dep
end
ent
Vari
able
p1
IfH
ous
eho
ldH
asO
utst
and
ing
Loan
s
Land
rig
hts
#Ye
arp
1998
�.1
11�
.107
�.0
87�
.013
.047
�.2
41*
.009
�.1
02(.0
71)
(.073
)(.0
93)
(.009
)(.0
92)
(.135
)(.1
57)
(.102
)La
ndri
ght
s.0
35�
.002
….0
18…
……
…(.0
70)
(.061
)…
(.009
) *…
……
…Ye
arp
1998
.110
.125
.091
.106
.003
.198
*�
.016
.114
(.054
) **
(.055
)**
(.071
)(.0
41)*
*(.0
67)
(.104
)(.1
38)
(.074
)H
ous
eho
ldch
arac
teri
stic
sN
oYe
sN
oYe
sN
oN
oN
oN
oH
ous
eho
ldfix
edef
fect
sN
oN
oYe
sN
oYe
sYe
sYe
sYe
sO
bse
rvat
ions
8,10
88,
108
6,68
68,
108
2,93
03,
688
1,61
85,
000
R2
.00
.06
.24
.05
.63
.61
.63
.61
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
567
B.
Dep
end
ent
Vari
able
pP
rop
ort
ion
of
Bo
rro
win
gfr
om
Form
alSo
urce
s
Land
rig
hts
#Ti
me
.061
.063
.075
.011
.071
.058
.358
�.0
23(.0
89)
(.090
)(.1
93)
(.015
)(.2
61)
(.257
)(.4
78)
(.155
)La
ndri
ght
s.2
25**
*.2
10**
*�
.017
�.0
26.2
45**
(.065
)(.0
66)
(.010
)(.3
68)
(.119
)Ye
arp
1998
�.0
54�
.124
**.1
81.2
10**
*.1
30.2
39(.0
73)
(.057
)(.1
38)
(.063
)(.1
91)
(.193
)H
ous
eho
ldco
ntro
lsN
oYe
s…
Yes
……
……
Reg
ion
fixed
effe
cts
No
Yes
…Ye
s…
……
…H
ous
eho
ldfix
edef
fect
sN
oN
oYe
sN
oYe
sYe
sYe
sYe
sO
bse
rvat
ions
4,31
54,
315
3,57
04,
315
1,57
91,
980
971
2,58
8R
2.0
9.1
1.7
5.1
1.7
4.7
6.7
2.7
6
No
te.
Stan
dar
der
rors
are
inp
aren
thes
es,
corr
ecte
dfo
rp
rovi
nce-
leve
lcl
uste
ring
.A
llre
gre
ssio
nsar
ew
eig
hted
by
sam
plin
gw
eig
hts.
“Lan
dri
ght
s”is
the
pro
vinc
e-le
vel
pro
po
rtio
no
fho
useh
old
sho
ldin
ga
Land
Use
Cer
tific
ate
(LU
C).
“Dis
tanc
eto
cred
itin
stit
utio
ns”
isth
eav
erag
ed
ista
nce
toal
lcre
dit
inst
itut
ions
inth
eco
mm
une,
mea
sure
dfr
om
the
Peo
ple
’sC
om
mit
tee
of
the
com
mun
e.“A
lter
nati
vecr
edit
”re
fers
toth
ep
rese
nce
of
acr
edit
inst
itut
ion
inth
eco
mm
une
that
isno
ta
go
vern
men
tban
k.“H
ous
eho
ldco
ntro
ls”
incl
ude
age,
gen
der
,an
ded
ucat
ion
of
hous
eho
ldhe
ad;
hous
eho
ldsi
ze;
ethn
icit
y;an
da
dum
my
for
No
rth.
Reg
ress
ions
(5)–
(8)
are
for
pan
elho
useh
old
so
nly.
*Si
gni
fican
tat
10%
.**
Sig
nific
ant
at5%
.**
*Si
gni
fican
tat
1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
568 economic development and cultural change
TABLE 10ARE CROP AND LABOR CHANGES WITHIN OR ACROSS HOUSEHOLDS? (SAMPLE: PANEL HOUSEHOLDS)
Category 01(Dcrop choice ,q 0Dlabor choice )1 0
Category 10(Dcrop choice ,Q 0Dlabor choice )q 0
Category 11(Dcrop choice ,1 0Dlabor choice )1 0
Land rights 1.1935* 1.749 3.169**(.6928) (.8240) (1.523)
Note. Table shows relative risk ratios obtained from multinomial logit regressions of the dependentvariable on the land rights measure. Dependent variable is categorical with four categories: category00 (Dcrop choice , Dlabor choice ), category 01 (Dcrop choice , Dlabor choice ), categoryq 0 q 0 q 0 1 010 (Dcrop choice , Dlabor choice ), and category 11 (Dcrop choice , Dlabor choice ), where1 0 q 0 1 0 1 0Dcrop choice p change in proportion of household cultivated area devoted to perennial industrialcrops and fruit crops, between 1993 and 1998. Dlabor choice p change in number of weeks worked(per household member) in nonfarm activities, between 1993 and 1998. The base category is category00 (Dcrop choice , Dlabor choice ). “Land rights” is the province-level proportion of householdsq 0 q 0holding a Land Use Certificate (LUC). Standard errors are in parentheses, corrected for province-levelclustering. All regressions are weighted by sampling weights.* Significant at 10%.** Significant at 5%.
the changes we observe in crop and labor choices occur more within householdsthan between households.
D. Income and ConsumptionAn improvement in the security of tenure is expected to translate into increasedlifetime income and expenditure. However, the relatively short time lag be-tween the implementation of the 1993 land law and the 1997–98 surveymight suggest that we do not capture steady-state outcomes but, rather,transitional outcomes. This is especially true because investments in crops suchas tea or coffee yield returns only after a minimum of 3 or 4 years. In theabsence of credit constraints, the transition toward cultivation of perennialcrops increases permanent income, which should translate to an increase incurrent consumption. Such an absence of credit constraints corresponds to therate of time preference and the rate of interest being equal for the household.17
However, when credit constraints are binding, consumption along the tran-sition path can decrease: if investment prospects are good enough, then agentswill accept a drop in consumption in the first period to enjoy higher incomesin the second period (see “Result 3” in Sec. III).
Table 11 shows results obtained from the estimation of (5), in which theleft-hand-side variables of interest are real household expenditure and a measureof farm income. We find no significant effect of the 1993 land law on eitherof these outcomes. This, together with the results of Section VII.B, indicatethat the land law has not been very effective in alleviating credit constraintsfor rural households. The absence of any effect on consumption can admittedly
17 We thank an anonymous referee for pointing this out.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 569
TABLE 11IMPACT OF LAND RIGHTS ON EXPENDITURE, INCOME, AND PRODUCTIVITY
(SAMPLE: RURAL HOUSEHOLDS)
Dependent Variable:Real Household
Expenditure
Dependent Variable:Value of AgriculturalOutput / Cultivated
Area
Units 1998 dongs (’000)1998 dongs (’000)/
square mile cultivatedMean (SD) in 1993 5,414.04 (3,480.40) .7985 (2.342)Mean (SD) in 1998 10,293.81 (5,885.25) 1.659 (6.728)Land rights #
Year p 1998 495.222 490.478 .691 .723(625.744) (599.934) (.470) (.470)
Year p 1998 4,486.901*** 4,511.802*** .323 .302(490.666) (473.702) (.227) (.233)
Land rights 37.810 336.781 .221 .338(323.907) (322.540) (.203) (.223)
Household controls Yes Yes Yes YesRegion fixed effects No Yes No YesObservations 7,501 7,501 7,466 7,466R2 .44 .46 .01 .02
Note. Standard errors are in parentheses, corrected for province-level clustering. All regressions areweighted by sampling weights. “Land rights” is the province-level proportion of households holding aLand Use Certificate (LUC). “Household controls” include age, gender, and education of household head;household size; ethnicity; and a dummy for North.*** Significant at 1%.
be caused by an attenuation bias due to measurement error. But beyond thisconcern, the results described in this paragraph are consistent with a transitioneconomy with credit constraints and imperfect consumption smoothing.
VIII. ConclusionVietnam’s 1993 land law made land rights secure, pledgeable, and tradableand was implemented by means of an extensive rural land-titling program.We show that this reform had a statistically significant impact on the decisionsof households to undertake long-term agricultural investments and at the sametime devote labor to nonfarm activities. However, these results were not verylarge in magnitude; in particular, issuing land titles to all households wouldresult in only a 0.3 standard deviation increase in the proportion of landdevoted to long-term crops. We find no significant impact on overall householdconsumption expenditure or agricultural income.
We tested several mechanisms that can account for this transition in therural economy. We found no evidence that land titles increased access to crediton the part of rural households; neither were they significant determinants ofland market activity. We also find no evidence that the 1993 land law resultedin major changes in the land distribution in Vietnam. We conclude that the
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
570 economic development and cultural change
results we observe stem primarily from increasing the security of tenure ofthe landholders, the formal title being interpreted as a guarantee against futureexpropriation by the state. Another conclusion of our study is that imple-menting land-titling programs without complementary changes in the bank-ing system and the rules regarding land transactions is not likely to result indramatic changes in the rural economy. This conclusion is subject to twocaveats related to the short period of our analysis: first, the increases in in-vestment we observe are likely to yield greater returns in the future, and wemay thus be underestimating the gains from this land reform. Second, wecannot rule out the possibility that the reform might lead to significant changesin the functioning of the land and credit markets over time, thereby allowinghouseholds to capture the full benefits of formal land titles.
AppendixEquilibrium OutcomesWe solve the model set out in Section III as follows.
Borrowing and consumption decisions. As lenders break even in equilibrium,agents internalize their default decisions so that and : default∗ ∗t p 1 d p 0does not occur on the equilibrium path. Savings and borrowings are thusperfect substitutes, and we adopt the convention that , with a negative∗S p 0value of denoting net savings. The incentive-compatibility constraint for∗Bthe borrower determines her pledgeable income or, formally, ∗B ≤
. Consumption decisions are such that time and∗ ∗ ∗bh( p , r m ) T p 0 T p 1budget constraints are binding:
∗ ∗ ∗ 0 ∗ ∗ ∗ ∗C p B � h(m � m ) � f(n ) � g(r , (1 � r )m ),0∗ ∗ ∗ ∗ ∗{ C p (1 � a)h( p , r m ) � B ,1
and the level of savings/borrowings allows individuals to smooth consumptionacross time periods:
∗ ∗ ∗ ∗ ∗B p min [bh( p , r m , W )], (A1)
in which equalizes and . Thus, credit markets have two functions:∗ ∗ ∗W C C0 1
smooth consumption over time when land purchases need to be paid up-frontor when intermediate consumption needs to be financed when householdsadopt perennial crops.18
Labor and crop choices. When credit access is unrestricted, labor allocations
18 We ignore any potential set-up cost that perennials will require. They are assimilated to op-portunity costs of intermediate consumption.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 571
are determined by the equalization of marginal productivity of labor with theprevailing wage rate, which is normalized to one. Thus, for given areas
and dedicated to perennials and rice cultivation, respectively,∗ ∗ ∗ ∗r m (1 � r )mthe first-order conditions imply:
∗ 1/1�g ∗ ∗p p [(1 � a)gH] r m. (A2)∗ 1/1�j ∗ ∗{ r p (jG) (1 � r )m
Since consumption can be perfectly smoothed over time, only total outputmatters. Constant returns to scale implies that only one crop is going to becultivated: and∗r � {0, 1}
1 1∗ 1/1�j 1/1�gp p 0 ⇔ (jG) ≥ [g(1 � a)H] . (A3)j g
Then, we can rewrite (A2) as
1 11/1�g ∗ 1/1�j 1/1�g[(1 � a)gH] m , 0 if (jG) ! [g(1 � a)H]( )∗ ∗( p , r ) p (A4)j g1/1�j ∗ ∗{ 0, (jG) (1 � r )m otherwise.( )
When credit constraints bind , optimal crop and labor∗(b p 0 p1 B p 0)choices will trade off the opportunity cost of intermediate consumption againsthigher final consumption levels. The first-order conditions are now
∗ 1/1�j ∗ ∗r p (jG) (1 � r )m , (A5)
while
∗� � C0∗ ∗ ∗ ∗ ∗ ∗g(r , (1 � r )m ) p h( p , r m ) (A6)∗�l �l C1
and
∗� � C0∗ ∗ ∗ ∗ ∗ ∗g(r , (1 � r )m ) p h( p , r m ) . (A7)∗�m �m C1
The left-hand sides of (A6) and (A7) capture the opportunity cost of movingaway from rice, while the right-hand side measures the gains due to increasedperennial crop cultivation. As consumption is not smoothed across periods,second-period consumption is valued relatively less than in the first-best case.If (A3) holds, then households dedicate land and labor to rice cultivation andsave in order to smooth consumption over time. Otherwise, access to creditallows households to borrow against future earnings, so that they dedicateland and labor to perennial cultivation.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
572 economic development and cultural change
Otherwise, when perennials are more productive but credit is not available,consumption cannot be financed by credit. As ∗T p 0 (�/�l)g(r , (1 �
, the ratio of consumption levels is equal to∗ ∗r )m ) p 1
∗� C0∗ ∗ ∗h( p , r m ) p 1,∗�l C1
so that, after rearranging, we obtain
g 1 � j∗ 1/1�j ∗ ∗p p (jG) r m , (A8)1 � g j
and
g 1 � j∗ 1/1�j ∗ ∗ ∗n p 1 � (jG) 1 � r � r m . (A9)[ ]1 � g j
The optimal crop allocation is then given by
∗ ∗ ∗ 0 ∗ ∗r p arg max ln [ f(n ) � h(m � m ) � g(r , (1 � r)m )]r
∗ ∗� ln [(1 � a)h( p , rm )].
The first-order conditions give the following optimal crop choice as an interiorsolution
∗ 0[1 � h(m � m )](1 � g)∗r p � g. (A10)1/1�j ∗(1 � j/j)(jG) m
Combining (A3) and (A10), we obtain result 1.Suppose that is such thatpre 1/1�j pre 1/1�ga (1/j)(jG) ≥ (1/g)[g(1 � a )H] p
. Given (2), a decrease in a eventually makes perennial crops morepre1 r p 0attractive than rice. Thus, households either entirely or partially shift towardperennials, so that the share of land devoted to perennial crops unambiguouslyincreases.
pre postr ! r .
Labor choice. On the labor market, , so that off-farmpre 1/1�j prer p (jG) m
activities are given by
pre 1/1�j pren p 1 � (jG) m .
When transition takes place with access to credit, then
post post 1/1�g postn p 1 � [(1 � a )gH] m ,
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 573
so that off-farm employment increases if and only if
1/1�j pre post 1/1�g post(jG) m ≥ [(1 � a )gH] m .
In the aggregate, land area is unchanged, so that, by linearity, aggregate off-farm employment increases if and only if
1/1�j post 1/1�g(jG) ≥ [(1 � a )gH] .
A necessary condition for this to hold is that , that is, perennials needj 1 g
to be comparatively less labor intensive than rice cultivation in order to savelabor when shifting away from rice.
In the presence of credit constraints, off-farm employment is given by
g 1 � jpost 1/1�j post post postn p 1 � (jG) 1 � r � r m[ ]1 � g j
so that an individual will increase off-farm employment if and only if
g 1 � jpost post post pre1 � r � r m ≤ m .[ ]1 � g j
In the aggregate, implies that ,post postj 1 g 1 � r � (g/1 � g)(1 � j/j)r ! 1so that
post pren ≥ n .� �Consumption choices. We have seen that transition creates the demand to
borrow in order to smooth consumption over time as the newly adoptedperennial crop delivers output only in period . The result is reinforcedT p 2by the need to borrow in order to cover set-up costs (not modeled here) orto purchase land. When households have access to credit, they maximizeaggregate output and consumption is perfectly smoothed across periods. Thus,when ,postb p 1
post pre post preC 1 C and C 1 C .0 0 1 1
The result might differ when credit constraints are binding. As agents tradeoff the marginal utility of consumption against the marginal utilityT p 0of consumption, they might decrease consumption in order toT p 1 T p 0enjoy more wealth at the second period. To see this, let us formally comparepre-land-law and post-land-law consumption levels. Prereform, assum-T p 0
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
574 economic development and cultural change
ing that , agriculture consists exclusively of rice, so thata p 0
1 � jpre 1/1�j preC p 1 � (jG) m .0 [ ]j
In the presence of credit constraints and in the absence of any land sales orpurchases, consumption is determined by (A5) and (A9).
g 1 � jpost 1/1�j post post postC p 1 � (jG) 1 � r � r m0 [ ]1 � g j
11/1�j post post� (jG) (1 � r )m[ ]
j
1 � j g1/1�j post post postp 1 � (jG) (1 � r ) � r m ,[ ]j 1 � g
so that if , thenpre postm p m
post preC 1 C .0 0
As farmers decide to partially move away from rice, agricultural output atnecessarily drops, which translates into a drop in consumption as creditT p 0
market access is restricted. If we relax the assumption that , thenpostb p 0households can borrow against agricultural output, so that consumptionT p 1does not necessarily drop and eventually increases along the transition path.
Land allocation. When land is entirely devoted to either rice or perennials,the agricultural production function exhibits constant returns to scale, so thatland distribution is irrelevant. On the other hand, in the presence of creditconstraints and crop diversification, the optimal allocation of land equalizesmarginal product of land across households. The price of the marginal pieceof land is such that the marginal farmer owning is indifferent between∗m
selling, purchasing, or keeping his piece of land. In the presence of creditconstraints, land sales and purchases trade off the increased agricultural outputagainst the marginal utility of consumption as land purchases need toT p 0be financed out of retained earnings. Demand and supply∗ 0 0(m 1 m ) (m 1
of land is then driven by∗m )
∗ ∗ ∗ 0 ∗ ∗m p arg max ln [ f(n ) � h(m � m ) � g(r , (1 � r )m)]m
∗ ∗� ln [(1 � a)h( p , r m)],
where h is such that , and equilibrium labor and crop investments∗� m p 1are given by (A5), (A8), (A9), and (A10) and are all functions of m. First-
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 575
order conditions are such that
∗df dg d C0∗ ∗ ∗ ∗ ∗[n (m)] � [r (m), (1 � r (m))m] � h[ p (m), r (m)m] p h. (A11)∗dm dm dm C1
The left-hand side of (A11) captures the marginal utility gain from increasingland area, while the cost is land price h.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
576
TABLE A1LAND RIGHTS AND CROP CHOICE (ROBUSTNESS CHECKS)
BaseSpecification
(1)
% State FarmArea(2)
Controlling for% State FarmAgricultural
Area(3)
ProvincialCompetitive-ness Index
(4)
Land rights # Year p 1998 .075*** .087*** .078*** .095***(.025) (.028) (.025) (.030)
Year p 1998 �.033* �.060** �.046** .139**(.017) (.023) (.019) (.065)
Land rights �.011 �.019 �.016 �.025(.027) (.033) (.027) (.037)
Age of household head .002*** .002*** .002*** .002***(.000) (.000) (.000) (.000)
Male household head �.020** �.019 �.017* �.020**(.010) (.011) (.010) (.010)
Years of education of head .008*** .008*** .008*** .008***(.002) (.002) (.002) (.002)
Household size �.002 �.002 �.002 �.003(.002) (.002) (.002) (.002)
Majority ethnic group dummy .003 �.000 .004 .000(.030) (.028) (.031) (.030)
Total area cultivated (# 10�5) �.208** �.265*** �.212** �.205**(.090) (.083) (.082) (.090)
% state farm area 1994 2.783(1.889)
% state farm area 1994 # Year p
1998 1.445(1.085)
% state farm agricultural area .745**(.366)
% state farm agricultural area #
Year p 1998 .274(.196)
PCI .003(.002)
PCI # Year p 1998 �.004**(.001)
Region fixed effects Yes Yes Yes YesHousehold fixed effects No No No NoNo. of observations 7,469 6,253 7,469 7,469No. of provinces 59 56 59 59R2 .22 .21 .25 .22
Note. Standard errors are in parentheses, corrected for province-level clustering. All regressions areweighted by sampling weights. “Land rights” is the province-level proportion of households holding aLand Use Certificate (LUC). All state farm variables are computed from the 1994 Agricultural Census. %State farm area is the proportion of total province area devoted to state farms. % State farm agriculturalarea is the ratio of state farm agricultural area to total province agricultural area. PCI is the ProvincialCompetitiveness Index 2006.* Significant at 10%.** Significant at 5%.*** Significant at 1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
577
TABLE A2LAND RIGHTS AND NONFARM ACTIVITIES (ROBUSTNESS CHECKS)
Controlling for
BaseSpecification
(1)
Labor Employedin State Farms
(2)
ProvincialCompetitiveness
Index(3)
Land rights # Year p 1998 2.685** 2.831** 2.880**(1.146) (1.142) (1.301)
Year p 1998 �.941 �3.455 .739(.889) (2.365) (2.398)
Land rights �3.3.6 �1.331 �3.099(2.360) (.916) (2.430)
Age of household head �.031** �.031** �.031**(.013) (.013) (.013)
Male household head 1.919*** 1.920*** 1.921***(.483) (.483) (.485)
Years of education of head .380*** .381*** .381***(.065) (.065) (.066)
Household size �.293*** �.296*** �.291***(.089) (.089) (.089)
Majority ethnic group dummy 3.358** 3.342** 3.474**(1.279) (1.286) (1.325)
Labor employed in state farms 1994 .010(.007)
Labor employed in state farms #
Year p 1998 1994 �.014(.009)
PCI �.032(.073)
PCI # Year p 1998 �.035(.052)
Region fixed effects Yes Yes YesHousehold fixed effects No No NoNo. of observations 8,091 8,091 8,091No. of provinces 59 59 59R2 .07 .07 .07
Note. Dependent variable p Number of weeks worked in nonfarm activities (per working member) inthe last 12 months. Standard errors are in parentheses, corrected for province-level clustering. All re-gressions are weighted by sampling weights. “Land rights” is the province-level proportion of householdsholding a Land Use Certificate (LUC). All state farm variables computed from the 1994 AgriculturalCensus. “Labor employed in state farms” is the ratio of the number of workers employed in state farmsto the number of agricultural households in province. PCI is the Provincial Competitiveness Index 2006.** Significant at 5%.*** Significant at 1%.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
578 economic development and cultural change
ReferencesAcemoglu, Daron, Simon Johnson, and James A. Robinson. 2001. “The Colonial
Origins of Comparative Development: An Empirical Investigation.” American Eco-nomic Review 91, no. 5:1369–1401.
Alston, Lee, Gary Libecap, and Robert Schneider. 1996. “The Determinants andImpact of Property Rights: Land Titles on the Brazilian Frontier.” Journal of Law,Economics and Organization 12, no. 1:25–61.
Asian Development Bank. 1998. “Vietnam Land Information System and Agricul-tural Taxation Study.” Working Paper TA 2225-VIE, Asian Development Bank,Manila.
Benjamin, Dwayne, and Loren Brandt. 2004. “Agriculture and Income Distributionin Rural Vietnam under Economic Reforms: A Tale of Two Regions.” In EconomicGrowth, Poverty and Welfare: Policy Lessons from Vietnam, ed. Paul Glewwe, NishaAgrawal, and David Dollar. Washington, DC: World Bank.
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan. 2004. “How MuchShould We Trust Differences-in-Differences Estimates?” Quarterly Journal of Eco-nomics 119, no. 1:249–76.
Besley, Timothy. 1995. “Property Rights and Investment Incentives: Theory andEvidence from Ghana.” Journal of Political Economy 103, no. 5:903–37.
Boothroyd, Peter, and Pham Xuan Nam. 2000. Socioeconomic Renovation in Vietnam:The Origin, Evolution and Impact of Doi Moi. Ottawa: International DevelopmentResearch Center.
Boucher, Steve, Brad Barham, Michael Carter, and Juan Sebastian Chamorro. 2002.“Market-Friendly Reforms and the Operation of Credit and Land Markets inCentral America.” Photocopy, Agricultural and Resource Economics Department,University of California, Davis.
Braselle, Anne-Sophie, Frederic Gaspart, and Jean-Philippe Platteau. 2002. “LandTenure Security and Investment Incentives: Puzzling Evidence from Burkina Faso.”Journal of Development Economics 67:373–418.
Deininger, Klaus, and Songqing Jin. 2003. “Land Sales and Rental Markets inTransition: Evidence from Rural Vietnam.” World Bank Policy Research WorkingPaper 3013, World Bank, Washington, DC.
De Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Triumphs in theWest and Fails Everywhere Else. New York: Basic.
Dollar, David, and Jennie Litvack. 1998. “Macroeconomic Reform and Poverty Re-duction in Vietnam.” In Household Welfare and Vietnam’s Transition, ed. DavidDollar, Paul Glewwe, and Jennie Litvack. Washington, DC: World Bank.
Feder, Gershon, Tongroj Onchan, Yongyuth Chalamwong, and C. Hangladoran.1986. Land Policies and Farm Productivity in Thailand. Baltimore: Johns HopkinsUniversity Press.
Field, Erica. 2003. “Entitled to Work: Urban Property Rights and the Labor Supplyin Peru.” Photocopy, Department of Economics, Harvard University.
Field, Erica, and Maximo Torero. 2004. “Do Property Titles Increase Credit Accessamong the Urban Poor? Evidence from Peru.” Photocopy, Department of Eco-nomics, Harvard University.
Galiani, Sebastian, and Ernesto Schargrodsky. 2006. “Property Rights for the Poor:
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions
Do and Iyer 579
Effects of Land Titling.” Photocopy, Business School, Universidad Torcuato DiTella.
Haque, T., and L. Montesi. 1996. “Tenurial Reforms and Agricultural Developmentin Vietnam.” Land Reform Bulletin, Food and Agricultural Organization, Rome.
Hayami, Yujiro. 1994. “Strategies for the Reform of Land Property Relations.” InAgricultural Policy Analysis for Transition to a Market-Oriented Economy in Viet Nam,ed. R. Barker, 9–59. FAO Economic and Social Development Paper 123. Rome:Food and Agriculture Organization.
Knack, Steven, and Philip Keefer. 1995. “Institutions and Economic Performance:Cross-Country Tests Using Alternative Measures.” Economics and Politics 7:207–27.
Ngo, Thi Minh-Phuong. 2005. “Establishing a Quasi-Private Property Rights Sys-tem: Tenure Security and Investment Incentives in Vietnam.” Photocopy, LondonSchool of Economics.
North, Douglass C., and Robert P. Thomas. 1973. The Rise of the Western World: ANew Economic History. Cambridge: Cambridge University Press.
Pingali, Prabhu L., and Vo-Tong Xuan. 1992. “Vietnam: Decollectivization and RiceProductivity Growth.” Economic Development and Cultural Change 40, no. 4:697–718.
Ravallion, Martin, and Dominique van de Walle. 2003. “Land Allocation in Viet-nam’s Agrarian Transition.” World Bank Policy Research Working Paper 2951,World Bank, Washington, DC.
SMERU Research Team. 2002. “An Impact Evaluation of Systematic Land Titlingunder the Land Administration Project (LAP).” Photocopy, SMERU ResearchInstitute, Jakarta.
Vo, Dang Hung. 1997. “Land Administration Reform in Vietnam.” Paper presentedat the Federation Internationale des Geometres (International Federation of Sur-veyors) Commission 7 Symposium, Madison, WI.
Wiegersma, Nancy. 1988. Vietnam: Peasant Land, Peasant Revolution: Patriarchy andCollectivity in the Rural Economy. New York: St. Martin’s.
This content downloaded from 77.100.35.84 on Tue, 10 Jun 2014 11:17:27 AMAll use subject to JSTOR Terms and Conditions