DSpace at VNU: Farmland loss and livelihood outcomes: a microeconometric analysis of household surveys in Vietnam

DSpace at VNU: Farmland loss and livelihood outcomes: a microeconometric analysis of household surveys in Vietnam tài li...

On: 09 October 2014, At: 14:23

Publisher: Routledge

Informa Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of the Asia Pacific Economy

Publication details, including instructions for authors andsubscription information:

http://www.tandfonline.com/loi/rjap20

Farmland loss and livelihood outcomes:

a microeconometric analysis of household surveys in Vietnam

Tran Quang Tuyena, Steven Limb, Michael P Cameronb & Vu VanHuongbc

a Faculty of Political Economy, University of Economics andBusiness, Vietnam National University, Hanoi, Vietnamb

Department of Economics, University of Waikato, Hamilton, NewZealand

c Department of Economics, Academy of Finance, Hanoi, VietnamPublished online: 23 Apr 2014

To cite this article: Tran Quang Tuyen, Steven Lim, Michael P Cameron & Vu Van Huong (2014)

Farmland loss and livelihood outcomes: a microeconometric analysis of household surveys in

Vietnam, Journal of the Asia Pacific Economy, 19:3, 423-444, DOI: 10.1080/13547860.2014.908539

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the

“Content”) contained in the publications on our platform However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content Any opinionsand views expressed in this publication are the opinions and views of the authors,

and are not the views of or endorsed by Taylor & Francis The accuracy of the Contentshould not be relied upon and should be independently verified with primary sources

of information Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content

This article may be used for research, teaching, and private study purposes Any

substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden Terms &

and-conditions

Farmland loss and livelihood outcomes: a microeconometric analysis

of household surveys in Vietnam

Tran Quang Tuyena*, Steven Limb, Michael P Cameronband Vu Van Huongb,ca

Faculty of Political Economy, University of Economics and Business, Vietnam NationalUniversity, Hanoi, Vietnam;bDepartment of Economics, University of Waikato, Hamilton,New Zealand;cDepartment of Economics, Academy of Finance, Hanoi, Vietnam

Although there has been much discussion in the literature about the impacts offarmland loss (due to urbanization) on household livelihoods, no econometricevidence of these effects has been provided thus far This paper, hence, is the first toquantify the effects of farmland loss on household livelihood outcomes in peri-urbanareas of Hanoi, Vietnam Our study found no econometric evidence for negativeeffects of farmland loss on either income or expenditure per adult equivalent Inaddition, the results show that farmland loss has an indirect positive impact onhousehold welfare, via its positive impact on the choice of nonfarm-based livelihoods.Keywords: farmland loss; land acquisition; informal wage work; formal wage work;livelihood outcomes

JEL Classifications: Q12; O15; C 26

Increasing urban population and rapid economic growth, particularly in urban areas oflarge cities, have resulted in a great demand for urban land Taking Hanoi as an example,according to its land use plan for 2000–2010, 11,000 hectares of land, mostly annual cropland in Hanoi rural, was taken for 1736 projects related to industrial and urban develop-ment, and it was estimated that this farmland conversion caused the loss of agriculturaljobs of 150,000 farmers (Nguyen2009a) Moreover, thousands of households have beenanxious about a new plan of massive farmland acquisition for the expansion of Hanoi toboth banks of the Red River by 2020 This plan will induce about 12,000 households torelocate and nearly 6700 farms to be removed (Hoang2009)

*Corresponding author Email:tuyentq@vnu.edu.vn

Ó 2014 Taylor & Francis

Vol 19, No 3, 423–444, http://dx.doi.org/10.1080/13547860.2014.908539

In the setting of accelerating conversion of farmland for urbanization and zation in the urban fringes of large cities, a number of studies in Vietnam have addressedthe question of how farmland loss has affected rural household livelihoods (Do2006; Le

industriali-2007; Nguyen, Vu, and Philippe2011; Nguyen, Nguyen, and Ho2013; Nguyen2009b)

In general, these studies indicate that while the loss of agricultural land causes the loss oftraditional agricultural livelihoods and threatens food security, it can also bring about awide range of new opportunities for households to diversify their livelihoods and sources

of well-being In addition, similar impacts of farmland loss have been found elsewhere.Examples include negative impacts in China (Chen2007) and India (Fazal2000) Never-theless, other studies show positive impacts of farmland loss on rural livelihoods in China(Parish, Zhe, and Li1995) and Bangladesh (Toufique and Turton2002)

More importantly, when investigating the impacts of farmland loss on household lihoods, all above studies used qualitative methods or descriptive statistics, possibly due

live-to the unavailability of data Using a data-set from a 2010 field survey involving 477households in Hanoi’s peri-urban areas, this study, therefore, contributes to the literature

by applying microeconometric methods to answer the key research question: how, and towhat extent, has farmland loss affected household livelihood outcomes in Vietnam? Ourstudy found no econometric evidence for negative effects of farmland loss on eitherincome or consumption expenditure per adult equivalent In addition, we found that farm-land loss has an indirect positive impact on household welfare, via its positive impact onthe choice of nonfarm-based livelihoods

The paper is structured as follows: the next section describes an analytical frameworkthat is adapted for the specific context of the current study Section 3 reports the back-ground of the case study Data collection and methods are discussed in Section 4 Resultsand discussions are presented in Section 5, followed by the conclusion and policy impli-cations in Section 6

A household’s livelihood outcomes in turn can affect its future livelihood capitals Forinstance, better-off households tend to invest more in education and will therefore have ahigher level of human capital in the future Accordingly, livelihood capitals themselvesare endogenously determined by outcome influences The sustainable livelihood frame-work provides a conceptual description of dynamic and interdependent elements thattogether affect household livelihoods over time Given data limitations, our empiricalstudy only investigates the static impact of households’ livelihood assets and strategy ontheir livelihood outcomes In fact, such static models have been often used for quantifyingfactors determining household livelihood outcomes (Jansen, Pender, Damon, Wiele-maker, and Schipper2006; Pender and Gebremedhin2007) Following this approach, ourstudy only examines the static determinants of livelihood outcomes with a particularinterest in the setting of farmland loss due to escalated urbanization in Hanoi’s peri-urbanareas.

Figure 1 Conceptual framework for analysis of Hanoi peri-urban household livelihoods Source:Adapted from DFID’s sustainable livelihood framework (DFID1999), IDS’s sustainable rural live-lihood framework (Scoones1998) and Babulo et al (2008)

3 Background of the case study

3.1 The study site

Our research was conducted in Hoai Duc, a peri-urban district of Hanoi Of the districts

of Hanoi, Hoai Duc, has the largest number of farmland-acquisition projects and hasbeen experiencing a massive conversion of farmland for nonfarm uses (Huu Hoa2011).Hoai Duc is located on the northwest side of Hanoi, 19 km from the Central BusinessDistrict (CBD) The district has an extremely favourable geographical position, sur-rounded by various important roads namely Thang Long highway (the country’s longestand most modern highway), National Way 32, and is in close proximity to industrialzones, new urban areas and Bao Son Paradise Park (the biggest entertainment and tour-ism complex in North Vietnam) Consequently, in the period 2006–2010, around 1560hectares of farmland were compulsorily acquired by the State for 85 projects (Ha Noimoi2010)

Hoai Duc was merged into Hanoi City on 1 August 2008 The district occupies 8247hectares of land, of which agricultural land accounts for 4272 hectares and 91% of thisarea is used by households and individuals (Hoai Duc District People’s Committee2010).There are 20 administrative units under the district, including 19 communes and onetown Hoai Duc has around 50,400 households with a population of 193,600 people Inthe whole district, employment in the agricultural sector dropped by around 23% over thepast decade Nevertheless, a significant proportion of employment has remained in agri-culture, accounting for around 40% of the total employment in 2009 The correspondingfigures for industrial and services sectors are 33% and 27 %, respectively (StatisticsDepartment of Hoai Duc District2010)

3.2 Compensation for land-losing households

As revealed by the household survey, each household on average received a total pensation of 98,412,000 VND The minimum and maximum amounts were 4,000,000VND and 326,000,000 VND, respectively This might be a considerable source of finan-cial capital with which some households could initiate a new livelihood strategy or investmore in their current strategy However, most households have used this source for con-sumption purposes rather than production purposes.2This trend is also evident in otherperi-urban districts of Hanoi as described by Do (2006) and Nguyen (2009b) Therefore,

com-in the case of our sample, compensation might have little impact on livelihood choice,but could have a significant effect on expenditure

Also, Ha Tay Province People’s Committee issued the Decision 1098/2007/QĐ-UBand Decision 371/2008/QĐ-UB, which states that a plot of commercial land (đất dịch vụ)will be granted to households that lose more than 30% of their agricultural land Eachhousehold receives an area ofđất dịch vụequivalent to 10% of the area of farmland that

is taken for each project (Hop Nhan2008).Đất dịch vụis located close to industrial zones

or residential land in urban areas (WB2009), thus it can be used as a business premise fornonfarm activities such as opening a shop or a workshop, or for renting to other users.Thanks to this compensation with ‘land for land’, households will have not only anextremely valuable asset but also a potential new source of livelihood, particularly forelderly land-losing farmers.3In the remainder of this paper, households whose farmlandwas lost partly or totally by the State’s compulsory land acquisition will be referred to as

‘land-losing households’

4 Data and methods

4.1 Data

Adapted from the General Statistical Office (GSO) (2006) and Doan (2011), a householdquestionnaire was designed to gather a set of quantitative data on livelihood assets(human, social, financial, physical and natural capitals), economic activities (time alloca-tion) and livelihood outcomes (income and expenditure) A disproportionate stratifiedsampling method was used with two steps as follows: first, 12 communes with farmlandloss (due to the land acquisition by the State) were partitioned into three groups based ontheir employment structure The first group included three agricultural communes; thesecond one was characterized by five communes with a combination of both agriculturaland nonagricultural production while the third one represented four nonagriculturalcommunes From each group, two communes were randomly selected Second, fromeach of these communes, 80 households, including 40 households with farmland loss and

40 households without farmland loss, were randomly selected, for a target sample size of

480.4The survey was carried out from April to June 2010 Four hundred seventy-sevenhouseholds were successfully interviewed, among which 237 households lost some or all

of their farmland Among them, 113 households lost their farmland in early 2009 and 124households had farmland loss in the first half of 2008

4.2 Methods

4.2.1 Clustering livelihood strategies

We grouped households into distinct livelihood categories using partition cluster analysis.Proportions of time allocated for different economic activities before farmland acquisi-tion were used as variables for clustering past livelihood strategies Similarly, proportions

of income by various sources were used as variables for clustering current livelihoodstrategies or livelihood strategies after farmland acquisition A two-stage procedure sug-gested in Punj and Stewart (1983) was applied for cluster analysis First, we performedthe hierarchical method using Euclidean distance and Ward’s method to identify the pos-sible number of clusters At this stage, the values of coefficients from the agglomerationschedule were used to seek the elbow criterion for defining the optimal number of clusters(Egloff et al.2003) (see more in Tuyen[2013]) Then, the cluster analysis was rerun withthe optimal number of clusters, which had been identified using k-mean partitionclustering

4.2.2 Model specification for determinants of livelihood strategy choice

Once the whole sample was clustered into various groups of livelihood strategies, weapplied econometric methods to quantify the impact of farmland loss on household activ-ity choice and household welfare Because the choice of livelihood strategies is a poly-chotomous choice variable, we used a multinomial logit model (MNLM) to quantify thedeterminants of households’ activity choice (Train 2003) Following Van den Berg(2010) and Jansen, Pender, Damon, Wielemaker and Schipper (2006), we assumed that ahousehold’s livelihood choice is determined by fixed and slowly changing factors, includ-ing the household’s natural capital, human capital, and location variables In addition,other factors, in this case farmland loss and past livelihood strategies were included asregressors in the model Other types of livelihood capitals such as social capital, financialcapital and physical capital may be jointly determined with, even determined by, the

livelihood choice (Jansen, Pender, Damon, and Schipper2006) Therefore, we minimizedthe potential endogeneity problem by excluding such types of livelihood assets fromthe model Natural capital consists of the owned farmsize per adult (100 m2per adult(those aged 15 and over)) (more owned farmsize per adult stimulates farming activities),the size of residential land (10 m2) (can be used as a premise for household business) andthe location of houses or residential land plots (a prime location can be used for opening

a shop or a workshop).5Human capital is represented by household size and dependencyratio (this ratio is calculated by the number of household members aged under 15 andover 59, divided by the total members aged 15–59) (both reflect labour endowment), ageand gender of the household head, the number of male working members (male adultswho employed in the past 12 months) (influences the engagement in wage work), averageage of working members (younger members are more likely to work as wage earners) andaverage years of formal schooling of working members (requirements for formal wagework) were also included as explanatory variables

In fact, a number of households did not change their livelihood choices after farmlandacquisition, which indicates that their current livelihood strategies had been determinedprior to the farmland acquisition In such cases, current outcomes may be influenced bypast decisions; current behaviours may be explained by inertia or habit persistence(Cameron and Trivedi2005) Therefore, we included past livelihood strategy variables asregressors in the model of household livelihood choice Commune dummies were alsoincluded to account for commune fixed effects, which capture differences in inter-com-mune fertility of farmland, development of infrastructure, cultural, historical and geo-graphic communal level factors that may affect household livelihood strategies

In the present study, the loss of farmland of households is an exogenous variable,resulting from the State’s compulsory land acquisition.6Since the farmland acquisitiontook place at two different times, land-losing households were clustered into two groups:(1) households with farmland loss in 2008 and (2) those with farmland loss in 2009 Therationale for this division is that the length of time since farmland acquisition may berelated to the probability of livelihood change Moreover, the level of farmland loss variesamong households Some lost little, some lost part of their land while others lost all theirland As a result, the land loss in both years, as measured by the proportion of farmlandacquired by the State in 2008 and 2009, was used as the variable of interest.7

One might argue that compensation should be included as an explanatory variable inthe model of livelihood choice and in that of livelihood outcomes This is because thecompensation might have been invested in lucrative livelihood strategies, which in turnmight have resulted in higher income and greater consumption expenditure However, asmentioned in Section 3.2, only a very small proportion of households used their compen-sation for nonfarm production Hence, in the case of our sample, the compensation mighthave had little impact on the choice of nonfarm-based livelihoods In addition, there is anextremely high correlation between the amount of compensation and the levels of landloss since those with more land loss received more compensation.8If both variables wereincluded in the models, this would pose a serious multicollinearity problem Therefore,the compensation was not included as an explanatory variable in the model of activitychoice and that of livelihood outcomes

4.2.3 Model specification for determinants of livelihood outcomes

We used income and consumption expenditure per adult equivalent as indicators ofhousehold livelihood outcomes because they are considered as better measures of

well-being than income and consumption expenditure per capita (Haughton and ton2011).9The total annual income is constituted by different income sources (agricul-ture, animal husbandry, nonfarm self-employment, wage work and other income),whereas household expenditure is composed of total living expenses (food and nonfood,health care, education, housing, transportation, entertainment and other items) Note thatboth income and expenditure were measured accounting for own consumption of prod-ucts produced by households This is because most farm households are producers as well

Haugh-as consumers in developing countries Therefore, the consumption of home-produceditems, commonly vegetables and rice grown or poultry raised on the farm, are properlyrecorded as both income and consumption (Deaton1997)

Figure 1indicates that households’ livelihood outcomes are dependent on their holds’ livelihood strategy and assets As compared to the explanatory variables in theMNLM, we added some more asset-related explanatory variables that potentially affectlivelihood outcomes In the context of a simple conceptual framework, social capital can

house-be treated as one type of available assets of households, which can generate income ormake consumption possible (Grootaet et al.2004) Many studies have used group mem-berships as a proxy for social capital and evaluate their relationship with household well-being such as income or expenditure (Haddad and Maluccio 2003) Therefore, weincluded social capital in the form of the number of group memberships as an exogenouscapital like other capitals that can affect household income and expenditure We alsoincluded the value of productive assets per working member or ‘capital–labour ratio’ as aproxy for physical capital in the outcome models.10 Households with higher ‘capital–labour ratio’ were expected to obtain higher well-being Finally, we included dummy var-iables for financial capital in the form of access to formal and informal loan Householdsthat received formal or informal loans could use this resource for generating income ormaking consumption possible

Since three dummy variables of current livelihood choice (informal wage work, mal wage work and nonfarm self-employment, with farm work as base group) in the out-come equations were suspected to be endogenous, ordinary least square (OLS) estimation

for-of these models would be biased and inconsistent if these explanatory variables were related with the error term in the livelihood outcome models (Cameron and Trivedi

cor-2005) To control for this endogeneity, we employed the instrumental variable method(IV) estimator

First, following Pender and Gebremedhin (2007), we selected the livelihood strategychoice that households pursued prior to farmland acquisition as a potentially instrumentalvariable for the current livelihood strategy variables Second, we included the location of

a house (or a residential land plot) and the average age of working members as additionalinstruments As previously mentioned, households owning a house or a residential landplot in a prime location are more likely to open a shop as their livelihood strategy whilehouseholds with younger working members have greater opportunities to engage in wagework However, using the past livelihood strategy variables as an instrument may fail tomeet the assumption of instrument exogeneity because the lags from one to two yearsafter farmland acquisition may be less distant lags that will increase any correlationbetween these instruments and the error term of the livelihood outcomes equations Inaddition, the other instruments are likely to violate this assumption because these instru-ments may directly affect household livelihood outcomes For instance, households thatare endowed with a conveniently located house may gain greater income from lucrativehousehold businesses Similarly, households with younger workers may get higherincome from their highly paid jobs The above discussions imply that several necessary

IV tests must be conducted to determine whether both requirements of instruments vance and exogeneity) are satisfied or at least to ensure that a set of invalid and weakinstruments that generates imprecise estimates and misleading conclusions can beavoided.

(rele-In order to form an econometric foundation for instrumental variables, a series ofspecification tests was applied to the models We used the formal weak instrument testproposed by Stock and Yogo (2005) using the value of test statistic that is the F-statisticform of the Cragg–Donald Wald F-statistic (cited in Cameron and Trivedi2009) In bothexpenditure and income models, the values of Cragg–Donald Wald F-statistic are 28.615,which greatly exceeds the reported critical value of 9.53, so we can say that our instru-ments are not weak and satisfy the relevance requirement On the other hand, the validityrequirement of instruments was checked using a test of overidentifying restriction withboth two stage least squares (2SLS) and limited information maximum likelihood(LIML) estimates and the results came out similar The Hansen J-statistics were not statis-tically significant in both income and expenditure models and thus confirmed the validity

of the instrumental variables Combined, the above specification tests indicated that theselected instruments are in fact good instruments

Since the livelihood choice variables in both expenditure and income models werepotentially endogenous, an endogeneity test of these variables was conducted In bothmodels, the results showed that the null hypothesis of exogenous regressors was rejected

at the conventional level (5%), confirming that livelihood choice variables are nous This result, therefore, indicated that the IV model is preferred to the OLS model

endoge-5 Results and discussion

5.1 Description of household livelihood strategies

Table 1presents the four types of labour income-based strategies (strategies A–D) thathouseholds pursued before and after farmland acquisition that were classified via clusteranalysis Cluster analysis also identified 21 households that pursued a nonlabour income-based strategy (strategy E) after the farmland acquisition, as compared to 10 householdsfollowed this strategy before the farmland acquisition As shown inTable 1, the number

of households that followed a farm work-based strategy approximately halved rently, the number of households that pursued nonfarm-based livelihood strategies (A–C)

Concur-Table 1 Households’ past and current livelihood strategies

Changes in livelihood strategies of households

Whole sample Land-losing households Nonland-losing households

Livelihood strategy Past Current Past Current Past Current

Nonfarm self-employment 73 128 27 62 46 67

Note: Ten households that depended largely or totally on nonlabour income were excluded from cluster analysis

of the past livelihood strategy because they had very little or no time allocation to labour activities.

considerably increased A comparative look at two groups of households reveals thatthere is a more profound transition from the farm work-based strategy to the nonfarmwork-based strategies among land-losing households than that among nonland-losinghouseholds This suggests that the loss of farmland may have a considerable effect on thechoice of household livelihood strategy.

Table 2describes how much different income sources contributed to total householdincome for all households as well as for each livelihood group The results indicate thatfor the whole sample, farming activities remained the largest contribution to total house-hold income, accounting for around 28% of total income It is followed first by nonfarmself-employment (about 26%), and then by informal wage work (around 23%) Incomefrom formal wage work accounted for approximately 17% of total income and nonlabourincome constituted of around six% of total income

The main features of household livelihood strategies according to their livelihoodassets are presented inTable 3 Households pursuing livelihood A mainly derived incomefrom manual labour jobs The common kinds of such jobs were carpenters, painters, con-struction workers and other kinds of casual jobs Such jobs typically offered low andunstable income, without formal labour contracts Those who undertook these jobs hadbelow-average education and were younger than those in livelihood D The average farm-land per adult in this livelihood group was quite small compared to that in all other liveli-hood groups Moreover, households that followed this livelihood strategy also hold asmaller value of productive assets than those in other livelihoods Finally, the income andexpenditure per adult equivalent in this livelihood group were much lower than those innonfarm-based livelihood groups

Livelihood B consisted of households that on average derived around 75% of theirincome from formal wage work Formal wage earners were often employees who work inenterprises and factories, state offices or other organizations Such jobs often offered highand stable income, with formal labour contracts Working household members in thislivelihood group had a much higher than average education level and were younger thanthose in all other livelihood groups Households in this livelihood group also owned thesecond largest farmland per adult but income from farm work accounted for only around12% of total income Households adopting this strategy received the highest income, andhad the highest expenditure per adult equivalent

Regarding households in livelihood C, although about 40% of the household samplereported engaging in nonfarm household businesses, 29% of them depended on theseactivities as their main livelihood Such businesses included small-scale trade or produc-tion units, using family labour with an average size of 1.7 jobs Households’ businesspremises were mainly located at their homes or residential land plots, where they had aprime location for opening shop, a workshop or a small restaurant Working householdmembers in this livelihood group were somewhat older than those in group A and B, andattained the second highest level of education Finally, those in this group had the secondhighest income and expenditure per adult equivalent, just after those in livelihood B.Interestingly, while 83% of surveyed households maintained farm work, only about21% among them pursued this work as the main livelihood strategy Many householdscontinued rice cultivation as a source of food supply while others produced vegetablesand fruits to supply Hanoi’s urban markets The common types of crop plants consisted

of cabbages, tomatoes, water morning glory and various kinds of beans, oranges, fruits and guavas, etc Animal husbandry was mainly undertaken by pig or poultry breed-ing small-sized farms or cow-grazing households These activities, however, havesignificantly declined due to the spread of cattle diseases in recent years Households