Income and income diversification of farm households in chau thanh a district , a case study of tan phu thanh village

Participating in non-farm activities brings higher income to farm households.Among significant influencing factors, labour ratio has a strongest positive effect onfarm households' income

VIETNAM- NETHERLANDS PROJECT FOR M.A IN DEVELOPMENT ECONOMICS

INCOME AND INCOME DIVERSIFICATION OF FARM HOUSEHOLDS IN CHAU THANH A DISTRICT:

A CASE STUDY OF TAN PHU THANH VILLAGE

Thesis submitted in partial fulfillment of the requirement for the degree of

Master of Arts in Development Economics

By

LE TAN NGHIEM Supervisor: MSc NGUYEN HUU DUNG

Ho Chi Minh City, December 2003

Finally, I especially thank my family I would have never made this thesis without theirloving and support.

Diversification is a basic strategy that farm households in the Mekong River Delta rely

on This study highlights this issue by using the sample of 217 farm households in TanPhu Thanh Village, Chau Thanh A District, Can Tho Province The theory of farmhousehold model and the literature on livelihood diversification are reviewed to set outfactors affected on income diversification of farm households A log-linear model and alogit model are then employed to seek the determinants of farm households' income andincome diversification in this area

The study finds that diversification significantly increased the farm households' totalincome Participating in non-farm activities brings higher income to farm households.Among significant influencing factors, labour ratio has a strongest positive effect onfarm households' income Finally, although land size is one of the most important inputs

of farm production, its positive effect on the total income of farm households is est

small-Identifying the determinants of farm households' income diversification, the study ployed the logit model of farm households' participation in non-farm activities There-sults show that labour ratio is the most decisive factor inducing farm households to work

em-in non-farm activities It has the strongest positive effect on households' em-incomediversification Farm households' decision to engage in non-farm activities is negativelyaffected by landowning Low agricultural income per capita pushes farm households tostabilize their income by joining in non-farm work Finally, farm households recentlyfaced difficulties in agricultural production are more likely to take up non-farm activi-ties

TABLE OF CONTENTS

CERTIFICATION i

ACKNOWLEDGEMENT ii

ABSTRACT iii

TABLE OF CONTENTS iv

LIST OF TABLES vii

LIST OF FIGURES ix

CHAPTER 1 INTRODUCTION I 1.1 Statement ofthe problem 1

1.2 Objectives ofthe study 4

1.3 Research questions 5

1.4 Research methodology 5

1.5 Scope ofthe study 6

1.6 Organization ofthe study 6

CHAPTER 2 LITERATURE REVIEW 8

2.1 Relevant concepts 8

2.1.1 Farm household 8

2.1.2 Sources of farm households' income 9

2.1.3 Income diversification 10

2.2 Theoretical backgrounds of farm households 11

2.2.1 Theory of farm household behaviours 11

2.2.2 Income diversification (Ellis, 2000) 16

2.3 Reviews of relevant studies 23

IV

2.3.1 A Case study of rural Southern Mali (Abdulai and CroleRees, 2001) 23

2.3.2 A Case study ofNorthem Ethiopia (Woldehanna and Oskam, 2001) 24

2.3.3 A Case study ofEthiopia (Block and Webb, 2001) 25

CHAPTER 3 ANALYTICAL FRAMEWORK AND MODEL SPECIFICATION 29 3.1 Sources ofthe data 29

3.2 Sources of income, income diversification and income inequality 30

3 2.1 Classification of farm households' income 30

3 2.2 Measurement of income diversification 31

3.2.3 Measurement of income inequality 32

3.3 Determinants of farm households' income 33

3.4 Determinants of farm households' participation in non-farm activities 3 7 3 4.1 Supplementary tests of the model 40

3 4.2 Marginal effects 41

CHAPTER 4 GENERAL INFORMATION ON CHAU THANH A DISTRICT AND TAN PHU THANH VILLAGE 43

4.1 Chau Thanh A District 43

4.1.1 Geography 43

4.1.2 Overview of socioeconomic context 44

4.2 Tan Phu Thanh Village 49

CHAPTER 5 INCOME AND INCOME DIVERSIFICATION OF FARM HOUSEHOLDS IN TAN PHU THANH VILLAGE 51

5.1 Socioeconomic characteristics 51

5.2 Sources of farm households' income 52

5.3 Income diversification- Herfindahl index 54

5.3.1 Number ofhouseholds' income generating activities, households' characteristics and total income 55

5.3.2 Herfindahl index across quintiles of total income 56

5.4 fucome inequality of farm households 51

5.5 Factors influencing farm households' income 58

5.5.1 Land size, labour ratio and households's income 58

5.5.2 Access to credits, risks in farm production and households' income 59

5.5.3 Herfindahl index and households' income 60

5.5.4 Main features of pure farm households versus non-pure farm households 61

5.6 Factors influencing farm households' participation in non-farm activities 62

5.6.1 Labour ratio and non-farm activities 63

5.6.2 Land size and non-farm activities 64

5.6.3 Household size and non-farm activities 64

5.6.4 Farm income per capita and non-farm activities 65

5.7 Farm households' income and its significant determinants 67

5.8 Households' participation in non-farm activities and its significant determinants 70 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS 75

6.1 Conclusions 75

6.2 Policy recommendations 78

REFERENCES 80

APPENDIX 84

Vl

LIST OF TABLES

Table 2.1: Typical Variables used in Empirical Studies Explaining Household Income

Diversification 27

Table 3.1: Variables of the Farm Household Income Model 36

Table 3.2: Variables of Farm Households' Participation in Non-farm Activities .40

Table 4.1: Structure of Gross Domestic Product (GDP), 1995-2002 44

Table 4.3: Trend of Land per Agriculturist in Chau Thanh A, 2000-2002 .48

Table 5.1: Socioeconomic Characteristics of the Sample 52

Table 5.2: Variety of Sources ofHouseholds' Income 53

Table 5.3: Numbers of Income-generating Activities And Households Characteristics 55

Table 5.4: Herfindahl Index across Quintiles ofFarm Households' Total Income 56

Table 5.5: Distribution of Households' Income across Diciles 57

Table 5.6: Land Size and Labour Ratio across Total Income Quintiles 58

Table 5.7: Access to Credits and Risks in Production across Total Income Quintiles 59

Table 5.8: Total Income across the quintiles ofHerfindahl Indices 60

Table 5.9: Some Different Features between Pure and Non-Pure Farm Households 61

Table 5.10: Percentage of Households' Participation in Non-Agricultural Activities, by Quintiles of Household's Labour Ratio 63

Table 5.11: Percentage of Farm Households' Participation in Non-Agricultural

Activities, by Cohorts of Land Size 64Table 5.12: Percentage of Households' Participation in Non-agricultural Activities, by

Number of Household Members 65Table 5.13: Percentage of Households' Participation in Non-agricultural Activities, by

Quintiles of Farm Income Per Capita 66

Table 5.14: Determinants ofFarm Households' Income 68

Table 5.15: Logistic Regression Results for Determinants ofFarm Households'

Participation in Non-farm Activities 711Table 5.16: Estimated Marginal Effects at the Mean for Household's Participation in

Non-farm Activities 72

V111

LIST OF FIGURES

Figure 2.1: Classification of Activities 9

Figure 4.1: Structure of Employment in Chau Thanh A District, 2001-2002 .45

Figure 4.2: Map ofChau Thanh A District .456

INTRODUCTION

1.1 STATEMENT OF THE PROBLEM

Diversification emerges as a usual standard Households seem rarely to keep all theirwealth in the form of any single asset as well as to put all assets in just one activity Theyare more likely to obtain their income from diverse portfolios of activities (Barrett andReardon, 2001) Ellis (2000), in his book entitled "Rural Livelihoods and Diversity in Developing Countries," also argues diversification is necessary for farm households' income

stabilization He explains diversification is affected by several reasons such as riskreduction, realization of economies of scope, diminishing returns to factor used in any givenincome-generating activity, responses to crises, liquidity constraints, and live-lihoodstrategies (Ellis, 2000) Households endowed with insufficient productive agrar-ian capital toabsorb their full labour endowment also compel to seek out off-farm or

non-farm income sources (Barrett eta!. 2001) In addition, exploiting comparative vantages from specialization in special skills or better technologies saves extra labourtime for households to make diversity in activities and incomes (Barrett and Reardon,2001)

ad-In rural areas of developing countries, diversity in activities and income becomes evenmore necessary due to specific characteristics High transaction costs induce many resi-dents to resort to self-provision in several goods and services Increasing populationpressures often result in landholdings being too small to absorb all of household's la-boursupply Limited risk-bearing capacity and weak financial institutions also create

1

strong incentives to select a portfolio of activities in order to stabilize income flows so as tostabilize consumption and minimize the risk of entitlements failure etc Respond-ing to each

of these factors, households often come to choose a common solution: diver-sifiedemployment and income patterns (Barrett and Reardon, 2001) This is also the reason toexplain for the evidence that non- farm income share, on average, accounts for 42% in

Africa, 40% in Latin America, and 32% in Asia (Reardon et al 1998).

Diversification and its implications for development and policy are now drawing siderable attention of scholars, policy makers, and donors Up to now, there have been agreat deal of formal studies on it but most of them were implemented in many African andLatin American developing countries These researches points out that income di-versification associates positively with either wealth accumulation or income stabiliza-tion,

con-or at least a partial consumption smoothing (Barrett et al 2000, 2001 a, 2001 b;

Ellis, 2000; Reardon, 1997, 2000, and Reardon et al 1998) Some findings showed that

diversification of income and assets continues being important in Africa for at least the next

several decades (Barrett et al 2001 ).

Vietnam is a developing agrarian country that the rice monoculture often leads to a vari-ety

of conflicts arising in the rice economy Using high external inputs, such as chemi-calfertilizers, pesticides, etc destroys the environment severely and reduces the eco-nomicefficiency of rice production Rice farmers have to get smaller amounts of reve-nue because

of low quality of rice and fluctuating prices In addition, the deterioration of soil conditions

in turn stagnate the rice productivity and so the yield These conflicts continuously lead ruralfarmers to low income from paddy production and challenge them to escape out of thepoverty or the low living standard (Iwamoto, 2001)

In an attempt to tackle these problems, farmers have applied many production strategies totheir farms Double or triple rice cropping methods, and other combinations, such as

rice-fruit, rice-vegetable, rice-livestock, rice-shrimp and etc were introduced to sify the rice economy in the delta (Dang, 1998) They also diversify farming activitiesthrough the introduction of crops other than rice, or animals, or fishery, or convertingpaddy fields to orchards, or to produce fruits and vegetable in upland areas Doing so isunderstood as agricultural diversification in rice cultivating areas.

diver-Though farm households have tried their best to absorb full labours' ability, abundance

of labour still exists in rural Vietnam The unemployment rate averages about 2.6

per-cent (Le et al 1999: 196) In addition, the level of the lack of additional occupations to

work of an at-work labourer' is often high in poor and pure agricultural areas It is evendiscernible when including the effects of seasonality of rural works in farm households'idleness A farmer can spend a thousand of hours for cultivating and harvesting in everythree-month cropping season; but he really works only about three hours per day duringnon-peak periods of time In the agricultural sector, in other words, workless time ac-

counts for thirty-five percents in total working hours (Le et al 1999: 197) Besides,

ex-pectation of gaining immediate cash income for daily consumption expense and ment in production enlarges the demand of working As an emergency requirement, ru-rallabour force itself finds some extra-works next to their existing main jobs

invest-In the Mekong River Delta (MRD), some less attractive traditional production activities have

pushed rural labour to seek income from other sources Le et al (1999) empha-sized that

45.5 percent of working agricultural labour force in this area have been en-gaging insupplementary activities due partly to the insufficient income from the main source.Moreover, there also exist a number of options for farmers to change their work motivated

by economically preferable opportunities (Le et al 1999) These motives

en-1 It refers to people who are already engaging in some daily jobs but still desire additional work.

3

courage rural residents to enroll in diversifying their resources in even farming activitiesitself and/or non-farm jobs (i.e., trading, manufacturing, carpentry, construction, hiring out,etc.) In general, farm households create diverse income portfolios from diversified activitiesfor their desire of higher or stable income.

Chau Thanh A, one of nine districts of Can Tho province, is a rural district located in MRDwith about 85.67 percent of labour force engaging in agriculture (CSO, 2002) Prices ofagricultural outputs often fluctuate, causing variability in income, and house-holds' wealthand consumption To overcome chaos, farmers have to achieve cash di-rectly from non-farmactivities to compensate the lost and stabilize their income In ad-dition, if a person in ruralVietnam changes his job from agriculture to non-agriculture,

his annual average income can increase thirty-to-fifty percent (Le et a!. 1999) Such benefitsmight have attracted all of farmers to work off their farms; but why there is still a number ofrural households relying on farm only instead of activities portfolios in both sectors in ChauThanh A District A study to address the question is necessary and meaningful

1.2 OBJECTIVES OF THE STUDY

To answer the above problem, the thesis is designed to identify factors of income sification, particularly farm households' participation in non-farm activities In other words,this thesis will find out factors leading agricultural households to diversifying their incomeout of agricultural activities The study also considers the main determi-nants of farmhouseholds' income, in which the effect of income diversification on total income is mostlypaid attention to

diver-Furthermore, based on final findings drawn from the study, we would like to set outrecommendations for local authority in planning strategies on economic development inthe coming years.

1.3 RESEARCH QUESTIONS

In order to reach the research objectives, the study tries to address these questions:

1 What are the determinants of farm households' income in Chau Thanh A District in 2002?

2 What are the determinants of farm households' mcome diversification m Chau Thanh

to study the determinants of farm households' income and income diversification in thecase of Tan Phu Thanh Village, Chau Thanh A District, Can Tho Province

Both qualitative and quantitative analyses are used in the research Firstly, the tive statistics will be utilized to discuss the performance of income and income diversi-fication in the village Then, econometric analysis techniques, particularly an ordinaryleast square (OLS) and a logit regression are employed in testing the models of house-holds' income and households' participation in non-farm activities respectively

descrip-5

In term of data, this thesis will make use of an available primary data from a field sur-veyconducted in Tan Phu Thanh Village in 2002 Details of the dataset and analyticalframework will be presented in Chapter 3.

1.5 SCOPE OF THE STUDY

Income diversification is a process in which households participate in a diverse range of

farm and non-farm activities Since beside the factors of households' characteristics, it is alsoaffected by other factors, such as effects of some macro policies, the overall trend ofdiversification in the village, etc., panel data is usually better for analysts to work with.However, this study only focuses on the effects of farm households' characteristics onhouseholds' income diversification by using cross-sectional data In addition, farmhouseholds that completely left the farm for participating in non-farm activities only are notstudied

1.6 ORGANIZATION OF THE STUDY

Following the introduction, the thesis will proceed as follows Chapter 2 gives a briefliterature review In this chapter, the theory on farm household model and other docu-ments

on diversification are reviewed in order to form a background for the research problem Inaddition, some case studies used for specifying the final model are also pre-sented here

Chapter 3 outlines the research methodology How these models are specified and whatestimations are employed will be discussed particularly Data used for analysis will be alsointroduced in this chapter

Chapter 4 gives brief information on the situation of income diversification in Chau Thanh ADistrict and Tan Phu Thanh Village

Chapter 5 is set for analyzing farm households' income and determinants of income sults of OLS and Logistics regression and interpretations will be displayed in line withdata description and bivariate analysis.

Re-Chapter 6 summarizes and concludes the thesis Some recommendations will also beformed based on the results of the empirical analysis

7

CHAPTER2 LITERATURE REVIEW

Identifying the determinants of income diversification of farm households is analyzinghousehold behaviours on both production and consumption Neo-classical economics hasprovided the theory of household economic model to investigate households' behav-iour

in making decisions to maximize their utility Regarding to diversification, the the-ory ofhousehold model predicts income diversification as a function of on-farm returns tolabour time compared to off-farm earning opportunities Besides, literature on thelivelihood diversification also denotes some other factors affecting on farm households'decision on working in non-farm activities such as risks in farm production, credit mar-kets failures, etc Therefore, the theoretical framework of this study will be based on:

(1) the Chayanov farm household model (1920), (2) the Barnum-Squire farm householdmodel (1979), and (3) the theory on livelihood diversification edited by Ellis (2000).This chapter starts with some relevant concepts

2.1 RELEVANT CONCEPTS

2.1.1 Farm household

Ellis (1993) defines farm households are 'households which derive their livelihoods mainly from agriculture, utilize mainly family labour in farm production, and are char- acterized by partial engagement in input and output markets which are often imperfect

or incomplete' (Ellis, 1993: 13).

According to Norton (1993), farm household is a social unit for mobilizing agricultural labour, managing productive resources, and organizing consumption The farm house-hold produces output for their own consumption and for markets It also supplies em-

ployment off the farm

In this study, farm households are referred to households that either engage in farmingproduction only or both farming and non-farming activities In addition, farm house-holds,agricultural households and peasants are used interchangeably with the same meaning It isalso similar to non-farm and non-agricultural household

2.1.2 Sources of farm households' income

This study employs farm and non-farm income or agricultural and non-agricultural income to imply two different amounts of earnings resulted from farm and non-farm activities re-spectively Moreover, in this context, these concepts are modified with a little difference from the usual understanding Figure 2.1 will be useful to discriminate between them

Figure 2.1: Classification of Activities

Source: Author's modification from Table 2 in Barrett and Reardon (200 1: 39)

Non-farm activities consist of all the activities mentioned in the shaded columns Farm activities are the activities such as crop cultivation, livestock rearing, and aquaculture.

9

Therefore, non-farm income refers to total amounts of revenues gained from all the farm activities and farm income is respectively from all the farm activities This classifi-cation allows non-farm and off-farm to be used interchangeably in the study.

non-Non-labour income refers to the total amounts of remittance from relatives, interest fromsaving accounts, and transfers from the government

Therefore, farm household's income is the sum of farm income, non-farm income, andnon-labour income

2.1.3 Income diversification

Diversification used to refer to the situation in which households allocate their resources

in portfolios of activities Barrett and Reardon (2001) describe, 'diversification is widelyunderstood as a form of self-insurance in which people exchange some foregoneexpected earnings for reduced income variability achieved by selecting a portfolio ofassets and activities that have low or negative correlation of incomes' (Barrett andReardon, 2000: 6) Ellis (2000) also defines livelihood diversification as 'the process bywhich rural household constructs an increasingly diverse portfolio of activities and as-sets in order to survive and to improve their standard ofliving' (Ellis, 2000: 15)

We can see that the ultimate goal of livelihood diversification is to improve income.Therefore, livelihood diversification and income diversification may have the samemeaning In this research, income diversification means that farm households gain addi-tional amounts of income by participating in non-farm activities and farm householdshaving income diversification are called non-pure farm households, otherwise are pure farm households.

2.2 THEORETICAL BACKGROUNDS OF FARM HOUSEHOLDS

2.2.1 Theory of farm household behaviours

Most people in developing agrarian countries earn a part of their livelihood by working

in their farms and then consume at least a proportion of output from the production.Consequently, households make decisions on both farm production and consumption Atheory integrated both households' production and consumption is known as 'agricul-tural household model' or 'farm household model' (Bardhan and Udry, 1999)

2.2.1.1 The basic theory of farm household behaviours

The original theory of peasant household economics is the Chayanov peasant model(1920), in which labour market is ignored In this model, Chayanov denotes that ahousehold has a trade-off between two objectives: an income objective (working onfarm) and an income-adverse objective (avoiding the drudgery and irksomeness of farmwork) The main factor influencing this trade-off is the demographic characteristics ofthe household, especially the size and composition of the household It means that deci-sion making to work in farm and non-farm depends on the characteristics of the house-hold, which is expressed by the ratio of consumers to workers in household The Cha-yanov model also argues that the household gets maximum utility when the opportunitycost of labour time (or the marginal rate of substitution of leisure and working for in-come) equals to the marginal value product of labour When the standard of consump-tion has been met, the disutility of an additional work is higher relatively than the utility

of an additional amount of income and so, the household prefers working off-farm ratherthan on farm (Ellis, 1993: 109)

However, the limitation of the Chayanov model remains in the ignoring of labour ket This leads a household impossibly to hiring in or hiring out its labour Therefore,

mar-11

from the basic ideas of the Chayanov model, many farm household models, such asBarnum-Squire (1979), Low (1986) and Bardhan and Udry (1999), consider the house-holds'responses to time allocation among their activities under the existence of a com-petitivelabour market The central theory on farm household behaviours discussed in the next partfocuses on the Barnum-Squire theory developed in 1979.

2.2.1.2 Barnum-Squire farm household model (1979)

Basically, the root of the Barnum-Squire model differs from the Chayanov model be-cause itconsiders farm households as producers as well as consumers This model is based on thefollowing assumptions:

1 There is a labour market with a single wage for off-farm work so that a house-hold isable to hire in or hire out labour

2 Land available for each farm household is fixed, at least within the research pe-riod

3 Time for 'home' activities and leisure are the same consumption item, which ferred to as homework time (Tz).

re-4 An importance choice for the household is that between own consumption of output (C)

and sale of output in order to purchase market goods (M).

5 Uncertainty and behaviour towards risk are ignored

Barnum and Squire assume that a household in this model has the option of hiring in labour

at the market wage as well as hiring it out and maximize its utility through deal-ing withthree problems: (1) the production problem, (2) the consumption problem, and (3) theallocation of time problem

The production function depends on the distribution of the production inputs, i.e the

cultivated land (A), the total labour used for farming production (L) (both household andhired), and other variable inputs into production (V):

Y = f(A,L, V)

In the Barnum-Squire model, the output from household production is not only used rectly for horne consumption, but also can be traded in the market Therefore, the householdutility contains three items: (1) utility from homework (Tz), (2) utility from the consumption

di-of horne output (C), and (3) utility from the consumption di-of market goods (M) The utility

Tw may be positive or negative if labour is hired in or hired out

The problems of farming households in the Barnum-Squire model are more complex thanthat in the Chayanov model Households have to choose their activities among

13

three pairs of consumption trade-off and three resources in the production function.Therefore, the conditions for equilibrium in both production and consumption sides of themodel are:

(1) In the production side, the marginal product of labour (MVPr) equals to the wage rate(w), and the marginal product of other variable inputs (MVPv) equals to the average prices ofthese inputs (v)

(2) In the consumption side, the marginal rate of substitution between each pair of threeitems in utility function, i.e homework (Tz) for consumption of home output (C);

homework (Tz) for consumption of market goods (M); and consumption of home output

(C) for consumption of market goods (M), equals to the price ratios between them (w/p, wlm,p/m).

The equilibrium conditions in the Barnum-Squire model suggests that in order to maxi-mizethe utility, a household has to make decision on time allocation among its activi-ties (farmand off-farm work; hiring in or hiring out labour; and consumption of own outputs andconsumption of market goods) Household decision-making not only de-pends on thepreferences of the household (that are influenced by the size of household and itscomposition), but also are strongly affected by changes in the inputs and output prices, themarket good price, and the market wage rate Any increase in the output prices (or decrease

in the price ratio) influences on farm household decisions through

substitution effect and income effect First, substitution effect in production raises theoutput C and the labour needed for farm work, and thanks to reduce the homework time Tz.

Second, substitution effect also occurs in consumption side and has a negative effect

on household consumption of own output Third, the 'profit effect' 2 determines the direc-tionand strength of household responses to changes in input or output prices The in-crease inoutput price causes a rise in farm profits and hence in household full-income Through theprofit effect, a rise in price effectively reduces rather than increases in-come Fourth,substitution and profit effects also take place in the decision on labour

time allocation of households An increase in the output price leads to the change in us-inglabour from family to hiring it in farm, and increases the non-farm time of house-hold

To conclude, the Barnum-Squire model stresses the importance of labour market for theworking of the farm economy This model is useful in predicting responses of the farmhousehold to changes in domestic (household size and structure) and market variables (pricesand wage rates) The Chayanov model as well as the Barnum-Squire model ex-plain at least

a part why household tends to diversify their activities by taking both farm and off-farmwork The determinants of household decision-making towards income diversification arethe wage rates, the prices of inputs or outputs, and the characteristics of household The farmhousehold economic model is, though useful in setting up a starting point to explore themotives for livelihood diversification, it is still leave behind some reasons relevant tointertemporal dimensions of livelihood strategies and situa-tions of survival under stress Thefollowing part will therefore include them in a more integrated study of determinants oflivelihood diversification edited by Ellis, 2000

2This is also the income effect in normal demand theory The profit effect is named in order to guish from it in normal demand theory.

distin-15

2.2.2 Income diversification (Ellis, 2000)

Ellis (2000) started at the farm household models and classified the reasons causing dividuals and households to pursue diversification as a livelihood strategy into two mainconsiderations, necessity versus choice They are sometimes referred to as 'push' and'pull' factors (Barrett and Reardon, 2000: 4) 'Necessity prefers to involuntary and dis-tress reasons for diversifying and by contrast, choice prefers to voluntary and proactivereasons for diversifying' (Ellis, 2000: 55) Such categorizations are not mutually exclu-sive as determinants of diversification, they constitute distinct but overlapping forces andprocess leading to diversification

in-Seasonality and risk and coping with shocks in production belong to push factors fordiversification, while labour market, credit market failures, and migration are of pullfactors How each ofthem affects diversification in tum will be made clear; and the fol-lowing is now coming up with the first factor of push group

2.2.2.1 Seasonality in crop production

All rural households confront seasonality as an inherent feature of their livelihoods(Ellis, 2000: 59) Seasonality, in economic view, means that returns to productive assets(i.e labour, land) vary across time, across individuals within a household, or acrosshouseholds within a community in both farm and non-farm sectors Households react tothe variation in returns by exhibiting diverse activities and income even if they havecomplete Ricardian specialization (Barrett and Reardon, 2001: 4) For example, if amember has a special talent for metal working, pottery, or some other skill-based trade,the household may let him specialize using his skill and other member of the household

to engage in less uniquely skilled occupations (e.g., farming)

The effects of seasonality on employment and income are more observable in farmhouseholds because of their dependence on cycles of agricultural production activities In thegrowing and harvesting periods, households' farming employment has to work extensively,

so the demand for labour in both farm and non-farm is high, leading to a shortage of labour.This shortage gives a good chance for wage earners to get higher income than before.Because the periods of peak demand for labour usually occur in a short time, farmhouseholds will allocate some of labour endowment to self-employ but still keep theirproduction unchanged and supply the rest for wage employment on the market to increasehouseholds' total income (Ellis, 2000) A household in the semi-arid tropics may allocate allits labour to on-farm production agriculture during the wet sea-son when the returns to on-farm labour are high and reallocate all its labour to non-farm activities during the dry season(Barrett and Reardon, 2001: 4)

The problem of seasonality may be more severe for small landholding or landlesshouseholds They try their best to earn as more income as possible during the harvestingperiods, but total income gained still does not cover all expense in the year For the sake ofsmoothing consumption and saving, their employment has to seek work to compen-sateincome Consequently, farm households were diversified their activities in both farm andnon-farm (Ellis, 2000)

At the household level, these cases will generally appear as diversified activities andincomes Therefore, it is no doubt that seasonality is a meaningful motivation to explain theincome diversification in farm households

2.2.2.2 Risk and coping with risk in livelihood diversification

Many neoclassical economists have developed theories on household economics, but

some of them, i.e., Barnum-Squire, Low, assume that ther i~GI ?Ifi@lfa:fid~A~~M&v In

TRUONG CH KINd Tt: 1 P.h- !V'I

reality, however, risk is relevant and so, Ellis (2000) takes risk into account in his study onlivelihood diversification as a determinant of diversified patterns of income.

Risk and coping with risk are considered as fundamental motive for livelihood fication Risk reducing strategies refer to making a forward plan for spreading risk over adiverse set of activities based on self-evaluation of the level of risk attached to each incomesource In contrast, coping is considered as a means of rescuing households out of

diversi-unpredicted crises To distinguish them, the former is named ex ante income

man-agement and the later, ex post consumption manman-agement (Ellis, 2000: 61).

Risk strategies are used to trade-off between a higher income accompanied to a highprobability of failure with lower income in a low chance of failure (Ellis, 2000: 60) Withinagriculture, if a peasant cultivates only one type of plant (e.g., paddy), when the price andyields of this product fluctuate, their income may be threaten more than a household withmore than two types of plants (e.g., paddy, beans, etc,) People therefore stabilize their totalrevenue by diversifying types of activities in agriculture In addition, non-farm activities areusually considered as less risky than agricultural ones and hence, participating in non-farmwork may help farm households stabilize their income Al-though households cannot knowexactly of whether or not spreading risk will affect the total income, they are ready to trade-off a lower income for a greater income security or a lower level of risk of a set of activities

In other words, farm households are

consid-ered to be risk averse, and for this reason, income diversification for risk reduction, ex

ante risk management, is an achievement of income portfolio (Ellis, 2000: 60)

Similarly, coping strategies, which usually appear after shocks, also compose portfolios ofactivities Vulnerability is used to explain it more discretely because of high level ofexposure to shocks and stress and proneness to food insecurity In the agricultural sec-tor,external hazards due to risk factor such as climates, markets or sudden disaster, etc,

sometimes destroy the results of farming production and hence, reduce revenue of farm-ers.For example, when crops fail or livestocks die, households must reallocate labour to otherpursuits, whether on or off-farm wage labour, informal off-farm earnings (hunt-ing) Thevariation in income makes producers unable to cover their expenses in whole year.Agricultural households now have to seek as much as possible from any farm or non-farmjobs to earn income for survival or wealth accumulation In other word, they are solving the

problem of ex post consumption management with crises (Ellis, 2000:

61-65)

In brief, either cases of ex ante risk control or ex post coping with shock also lead

households to engaging in a portfolio of income; so, in perspective of diversification risk is

an explainable determinant of income diversification

2.2.2.3 Labour markets

Like other studies on rural labour markets in developing countries, Ellis (2000) alsoconcentrates on the working of labour markets in agriculture itself However, he is morefavor in conditions in which the labour market works such as working site, workingcondition, costs of getting a job, costs of hiring labour and government regulations on usinglabour These factors influence the supply of and demand for labour; and hence, affect onincome diversification

Ellis (2000) argues that economic motivation in the household model, which refers to labourallocation into non-farm when the agricultural marginal return to labour time for anyindividual lowers than the non-farm wage rate or the marginal non-farm return to him/her, ismodified by the 'social exclusion' of individuals and households from par-ticular incomestreams He also emphasizes that the role of labour markets is more im-portant in developingcountries, especially where high population and shortage of land

19

often result in a large number of small landholding or landless Because households have

to work off the farm for income stabilization, they always consider many things on thelabour market

Households in these areas usually digest the opportunity cost of a non-farm job fully before making decision on whether or not to take it Costs of finding a job in terms

care-of time and money will also be considered If the cost is acceptable, they accept the job;otherwise, they may deny Job hunters also take into account the time and the costs oftravelling site The distance from horne to workplace may have negative effect on start-ing a new work (Ellis, 2000: 69-70)

On the other hand, employers who usually have larger landholding areas also, think ofamount of money paid to labour before hiring in labour If the costs of supervising la-bour are high, they will make use of self-employment or purchase modem technology forproduction instead of hiring outside labour The last factor that both employers andemployees must obey is regulations on the labour market If regulations prohibit hiring inand out labour, farming labours have fewer non-agricultural chances of generatingincome (Ellis, 2000: 69-70)

In short, regulations on the existence of rural labour market have nontrivial impacts onhouseholds' decision on participating in non-farm, thus on variety of income sources orincome diversification as well

2.2.2.4 Credit market failures

In some rural areas, farm households may face problems of missing credit markets Ellis(2000) identifies some reasons for credit market failures Costs of setting up bankingoperations are high, so is the cost to collect information on potential borrowers House-holds may not have enough collateral for loans In addition to high costs, banks suffer

the default risk Besides, private money lending based mainly on personalized tion between lenders and borrowers though appears in some regions, it still exposuressome constraints for people to access (Ellis, 2000: 74).

transac-Missing credit markets can either induce or reduce rural diversification (Ellis, 2000;Reardon, 1997) Imperfect or missing credit markets limit diversification into activities

or assets with high barriers to entry Small landholding farmers are often left behind ger landowners because they cannot afford to purchase better equipment that is oftenexpensive (Barrett, 1997) On the other hand, when thin or missing credit markets exist,cash funds generated outside agriculture become an extremely useful means either topurchase necessary variable inputs for farming (such as fertilizer, seeds, equipment, andlabour) or to make capital improvements such as bunds, ridges, and irrigation to one's

lar-farm (Ellis, 2000; Reardon eta! 1994; Savadogo eta! 1998; Reardon eta! 1999).

Barrett and Reardon (2001) add evidence to apprehend the role of credit markets to thepeasants in rural Under the absence of credit or insurance markets, households smoothconsumption in spite of a strong desire to do so When financial markets, credit and in-surance markets in particular are complete, they allocate part of permanent income forconsumption, and the rest is saved in the bank to offset any transitory negative earningsfrom consuming too much or shocks

In summary, the presence of credit market may help explain whether or not peasantsparticipate in non-farm activities If credit markets fail, it is more essential for house-hold to work off the farm for cash earnings to reinvest on the field Income diversifica-tion patterns are consequently resulted from the failure of the market

21

2.2.2.5 Migration

According Ellis (2000), migration happens when one or more family members leave theirhouseholds for a certain period of time in an effort to make new and different con-tributions to welfare (Ellis, 2000:70) Since this phenomenon changes the number ofpeople who can work, households' structures of income sources are also varied accord-ingly For this reason, migration is not only an influential factor of income diversifica-tion but also a special type of diversification that has a close relationship with labourmarket factors in households' decision making for living (Ellis, 2000: 70)

In this context, Ellis (2000) divides migration into some different sub-categories: sea-sonalmigration, circular migration, permanent migration, and international migration Despitesuch classification, common causes of migration it in are individual choices, temporarycontracts, the differences in the wage rate between rural and urban areas, and risk anduncertainty in crop production No matter what the reason, the ultimate objec-tive ofmigration is to send money back home for households' daily expenses, reinvest-ment onfarming and wealth accumulation for future living (Ellis, 2000: 70-72; Larson and Mundlak,1997; and Hoddinott, 1994) Migration really helps households stabilize or increase income,but policy makers are not finding of it (Ellis, 2000: 73) In sum, mi-gration is one of thefactors determining of income diversification

The subsection has provided some distinct but overlapping forces and process leading toincome diversification Seasonality, risk, labour markets, credit markets failure inducefarm households to select a diverse income portfolio These are also regarded as pullfactors because they bring benefits to households in terms of consumption and laboursmoothing, risk spreading and multiplying income by reinvesting in other activities Thecoping strategies are reflected as an involuntary or push reason for income diversifica-tion, because it forces households to engage in many activities to stabilize income The

remaining determinant itself contains both pull and push purpose for income tion.

diversifica-2.3 REVIEWS OF RELEVANT STUDIES

This part will review the empirical studies on how the factors mentioned in the previoussection influence households' income diversification As mentioned in the introduction, therehave been many empirical studies on income diversification in different countries on overthe world However, this study chooses three of them because these studies are implemented

in developing countries, which is similar to Vietnam They are also very close to my studyand available

2.3.1 A Case study of rural Southern Mali (Abdulai and CroleRees, 2001)

To examine the determinants of income diversification amongst rural households inSouthern Mali, Abdulai and CroleRees start with the theory of household model used toanalyze the factors influencing income diversification A two-stage sampling technique wasapplied to select 120 households from 15 villages of rural Southern Mali The data coversinformation on farm and non-farm activities, as well as demographic and loca-tioncharacteristics in the two years 1994/95 and 1995/96 The authors use the condi-tional fixedeffects logit model to control for the correlations between individual-specific effects thevariables included in the specification when identifying the determi-nants of incomediversification

In the model, the dependent variable is the probability of a household participating in farm work The explanatory variables consist of household characteristics, house-holdendowments, local conditions, and some multiplicative interaction terms (distance andlandholding, land and education and household and land) This study has some re-markablefindings:

non-23

1 Age of household head has positive relationship with participation in non-farmactivities of farm households, meaning that the older household head prefers toengaging in non-farm work than farm activities.

2 Households' participation in non-farm activities is positively affected by the numberofboth male and female adults, which means that households with more adult peopleare more likely to diversify into non-farm activities

3 Landholding (proxy of household's wealth) has a significant positive coefficient,implying that households with larger landowning have more opportunities in non-farm work, and hence more diversified income

4 Households received more non-labour income are unlikely to derive income fromnon-farm activities; in other words, there is a negative effect between an amount ofnon-labour income and participation in non-farm work

5 Households in remote areas are less likely to participate in the non-farm sector thantheir counterparts closer to local markets (this variable proxies for the la-bour marketfactor)

2.3.2 A Case study of Northern Ethiopia (Woldehanna and Oskam, 2001)

The paper identifies the determinants of households' income diversification in the NorthernEthiopia, in which the authors concentrate on the factors determining the deci-sions of farmhouseholds to engage in off-farm wage employment They base their re-search on the theory

of household model Tobit model is employed The panel data used comes from a survey on

201 heads of farm households, which were chosen randomly from a stratified sample areafor collecting information over the years 1996 and 1997 The survey was conducted in thetwo districts, Enderta and Adigudom, of Tigray Re-gion (Northern Ethiopia)

In the model, the dependent variable is the total labour hours supply for off-farm wageemployment The explanatory variables include farm characteristics, family characteris-tics,locations, and endogenous and exogenous household income Some conclusions are resultedfrom the estimated model as follows:

1 Age of the head of household is negative relationship with labour hours supply, implying that the older the head, the less likely to engage in off-farm

2 A larger household size is likely to devote more labour hours in off-farm wage employment

3 Labour hours supply for off-farm wage employment is negatively affected by number of dependents in a household

4 Land cultivated has a negative effect on labour hours supply, meaning thathousehold with a larger landholding is unlikely to work in off-farm wage em-ployment

5 Smaller output of farm production motivates the farm household to supply more labour hours in off-farm activities

6 Non-labour income reduces the total time supplied to off-farm wage employment

of household members or a negative relationship between these two variables

2.3.3 A Case study of Ethiopia (Block and Webb, 2001)

Block and Webb (200 1) study the factors determining income diversification in pia Based on panel data from two surveys (1989 and 1994) on 300 households, the au-thors study the associations among income diversification, household perceptions oflivelihood risks, and changes in consumption outcomes across two points in time in

Ethio-25

post-famine Ethiopia Particularly, Least Absolute Deviation (LAD) regression was plied to estimate income diversification as a function of income per capita, demograph-ics, agroecology, assets, human capital and risk perception In this model, income diver-sification is measured by the crop share of total income Their findings are:

ap-1 There is a strong evidence that income diversification is positively associated with age of household head and dependency ratio

2 Higher income is positively associated with greater diversification away from crop production

3 However, there is only a limited support for the hypothesis that households lieve diversification to be a viable risk minimization strategy

be-In short, empirical studies have documented some determinants of income tion Table 2.1 summarizes the typical variables which are commonly used in empiricalmodels to capture the effects of farm households characteristics (i.e., age of householdhead, numbers of adults and dependents, cultivated land, farm income) as well as risks inproduction, non-labour income on farm households' income and income diversifica-tion.Although different variables are used in different empirical models, the contents thatsuch variables reflect fall within the factors leading farm households to income di-versification as discussed in Section 2.2 In addition, the choice of variables to be in-cluded in empirical studies varies and depends on the purpose of each study, data avail-ability, and the characteristics of regions studied Therefore, these variables are consid-ered as the guidance of selecting variables for the models, which are discussed in details

diversifica-in Section 3.3 and 3 4

Table 2.1: Typical Variables used in Empirical Studies Explaining

Household Income Diversification

Author, Time,PlaceAbdulai and

-Land cultivated area (farm size)

- Farm income-Non-labour income

Block and - 300 observations

-More risk if lack of off-farm come (dummy)

in-In conclusion, this chapter has presented the theoretical background and relevant

em-pirical studies on diversification It provides a background for models specification toestimate the determinants of farm household income and income diversification The theory

of farm household model emphasizes that households' income diversification

27

mainly depends on farm households' characteristics The theory on diversification alsodenotes that risks in agricultural production, credit markets failures, and some other rea-sons also push and pull farm households to participate in non-farm activities Further-more, the impacts of these factors on diversification are examined in some selected em-pirical studies Therefore, these factors will be considered to include in the modelsspecification in Section 3.3 and 3.4.

CHAPTER3 ANALYTICAL FRAMEWORK AND MODEL SPECIFICATION

In order to identify factors affected the income diversification of farm households in TanPhu Thanh Village, this chapter outlines the model of farm households' participa-tion innon-farm activities The model of farm households' income is also specified to examinethe main influential factors of farm households' income, the effect of diversifi-cation onfarm households' income and the difference in total income between the diver-sified andthe non-diversified farm households as well Due to these purposes, a classi-fication ofincome sources, the measurements of income diversification and income ine-quality arereviewed The chapter starts with an introduction of the available data used in the study

3.1 SOURCES OF THE DATA

The data used in the empirical analysis of this thesis was drawn from a field survey inthe Tan Phu Thanh Village, Chau Thanh A District, Can Tho Province, conducted in

2002 The survey serves the research project "Effects of integrated farming systems on sustainable development of households in Tan Phu Thanh Village, Chau Thanh A Dis-

trict, Can Tho Province" directed by Can Tho University, funded by VNRP3

• tion collected covers households' characteristics, income sources, farm and non-farm em-

Informa-3 VNRP is the abbreviation of Vietnam-Netherlands Research Programme, managed by a Vietnamese Steering Committee.

29

ployment, difficulties in agricultural production, and socioeconomic context Such veyed data set is employed in this study with the permission of the author fu addition,sec-ondary data from official statistic by Can Tho Statistical Office and from Report andMaster Plan on Socio-economic Development of Chau Thanh A is also included in.

sur-The field survey was carried out in all ten hamlets of the Tan Phu Thanh Village Foreach hamlet, thirty households were selected randomly The sample has 300 observa-tions However, since this thesis will analyze the income portfolios offarm households,

households that did not involve in any agricultural production activity will be left out ofthe sample Due to this reason, information of 217 farm households will be in the em-pirical study of the thesis

The available dataset has some strengths to this study on income diversification Thedata covers all hamlets of the village, so it is highly representative for the village Thesample size is large enough to analyze However, it still exposes some weakenesses.Some variables which have effects on income and income diversification theoretically,are out of the data (i.e., households' assets, education levels of households' members,etc.,) Finally, crossectional data dose not observe intertemporal effects of some factors

on the process of diversification of farm households

3.2 SOURCES OF INCOME, INCOME DIVERSIFICATION AND INCOME

INEQUALITY

3.2.1 Classification of farm households' income

Households in the village collected income from a various sources of activities Forsimplicity in analysis, this study categories total income into six sources according totheir respective activities