Research Proposal

Effects of Contract Farming on Productivity and Income of Farmers in Production of Tea in Phu Tho Province, Vietnam

Anh Tru Nguyen MSc (University of the Philippines Los Baños)

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in Management

Newcastle Business School Faculty of Business and Law The University of Newcastle Australia

February 2019

Statement of Originality

I hereby certify that the work embodied in the thesis is my own work, conducted under normal supervision. The thesis contains no material which has been accepted, or is being examined, for the award of any other degree or diploma in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made. I give consent to the final version of my thesis being made available worldwide when deposited in the University’s Digital Repository, subject to the provisions of the

Anh Tru Nguyen

Date: 19/02/2019

i Acknowledgement of Authorship

I hereby certify that the work embodied in this thesis contains a published paper work of which I am a joint author. I have included as part of the thesis a written statement, endorsed by my supervisor, attesting to my contribution to the joint publication work.

Anh Tru Nguyen

Date: 19/02/2019

ii Publications

This thesis comprises the following published article and conference papers.

I warrant that I have obtained, where necessary, permission from the copyright owners to use any third party copyright material reproduced in the thesis, or to use any of my own published work in which the copyright is held by another party.

Anh Tru Nguyen

Date: 19/02/2019

Published articles:

1. Anh Tru Nguyen, Janet Dzator, Andrew Nadolny (2015). Does contract farming improve productivity and income of farmers? A review of theory and evidence. Journal of Developing Areas, 49(6) (Special Issue), 531-538.

2. Anh Tru Nguyen, Janet Dzator and Andrew Nadolny (2018). Contract farming, agriculture productivity and poverty reduction: evidence from tea estates in Viet Nam.

Asia-Pacific Sustainable Development Journal, 25(1), 107-143.

iii Conference papers:

1. Anh Tru Nguyen, Janet Dzator and Andrew Nadolny (2015). Does contract farming improve productivity and income of farmers? A review of theory and evidence. Selected paper presented at the Australasian Conference on Business and Social Sciences,

Sydney, Australia, 13–14 April 2015.

2. Anh Tru Nguyen, Janet Dzator and Andrew Nadolny (2017). The impact of contract farming on tea productivity: a case study in Phu Tho province, Vietnam. Selected paper presented at the Third HDR Colloquium, Faculty of Business and Law, the University of Newcastle, Australia, 6–7 November 2017.

3. Anh Tru Nguyen, Janet Dzator and Andrew Nadolny (2018). Contract farming and farm household income in Vietnam: the case of tea production in Phu Tho province,

Vietnam. Selected paper presented at the 14th International Conference, Western

Economic Association International, Newcastle, Australia, 11–14 January 2018.

Statement of Contribution of Others

I, Dr Janet Dzator, attest that Higher Degree by Research candidate Anh Tru Nguyen made a significant and original contribution to the joint publications included as part of this thesis.

Dr. Janet Dzator

Date: 19/02/2019

I, Dr Andrew Nadolny, attest that Higher Degree by Research candidate Anh Tru

Nguyen made a significant and original contribution to the joint publications included as part of this thesis.

Dr. Andrew Nadolny

Date: 19/02/2019

v Acknowledgements

It is impossible to complete a thesis without the support of a number of important people. First and foremost, I would like to acknowledge and express in-depth thanks to my supervisors, Dr Janet Dzator and Dr Andrew Nadolny, from Newcastle Business

School at the University of Newcastle, who have always given me insightful supervision and effective comments on my PhD studies over a long period. Both supervisors have a vast and impressive knowledge of their disciplines, and the many valuable lessons I have learnt from them motivate, encourage and inspire me in my future career.

I am grateful for the support and resources provided by the Ministry of

Education and Training of Vietnam, Vietnam International Education Cooperation

Department, and the University of Newcastle, Australia. I also acknowledge the support and assistance of the Ministry of Agriculture and Rural Development, Faculty of

Accounting and Business Management, Vietnam National University of Agriculture that allowed me to undertake the PhD program at the University of Newcastle. In addition, I would like to send a special thanks to the lecturers and staff of Newcastle Business

School at the University of Newcastle who have always been approachable and helpful, in particular Professor Morris Altman, Professor Martin Watts, Associate Professor

Frank Agbola, Mrs Wendy Jones, and Mrs Kerri Foulds.

I appreciate the valuable collaboration of local officials, enterprises, and farmers in Phu Tho province, Vietnam in completing the surveys and thank the research assistants at the Vietnam National University of Agriculture for gathering primary data.

I would have been unable to complete the research without their contributions.

I also wish to acknowledge my mother, brother, sister, and my little family for their support in not only my PhD studies but also in my career. Thanks to my lovely wife, Le Huyen Trang, and little boy, Nguyen Anh Nguyen, who are always ready to overcome all challenges to encourage me not only in my PhD studies but also in life. I would like to acknowledge all my colleagues, friends and fellow students for their support in studies and life.

Above all, this thesis is dedicated to my beloved father, who did not live long enough to see me complete this stage of my life. His memory burns bright and I thank him for encouraging me in whatever I wanted to do.

vii

Contents

Statement of Originality i

Acknowledgement of Authorship ii

Publications iii

Statement of Contribution of Others v

Acknowledgements vi

Contents viii

List of Figures xiii

List of Tables xiv

List of Abbreviations xvi

Abstract xvii

Chapter 1: Introduction 1

1.0 Introduction 1

1.1 Overview of contract farming and tea production in Vietnam 1

1.2 Research questions and aims 7

1.3 Significance of research 8

1.4 Thesis organisation 9

Chapter 2: Literature Review 11

2.0 Introduction 11

2.1 Theoretical framework 11

2.1.1 Concepts of contract 11

2.1.2 Concepts and types of contract farming 12

2.1.3 History of contract farming 14

2.1.4 Contract theory 15

viii

2.1.5 Motivation for contract farming 23

2.1.6 Advantages and disadvantages of farmers in contracts 25

2.1.7 Model for theoretical framework of the study 32

2.2 Empirical review 35

2.2.1 Effects of contract farming on productivity 35

2.2.2 Effects of contract farming on income 39

2.3 Chapter conclusions 44

Chapter 3: Tea Production in a Global and Vietnamese Context 46

3.0 Introduction 46

3.1 Tea production, consumption and trade in a global context 46

3.1.1 Tea production 46

3.1.2 Tea consumption 49

3.1.3 Tea trade 50

3.2 Tea production, consumption and exports of Vietnam 52

3.2.1 Tea production 52

3.2.2 Tea consumption 55

3.2.3 Tea exports 55

3.3 Tea production in Phu Tho province 57

3.3.1 Planted area, yield and productivity of tea 58

3.3.2 Tea varieties 58

3.3.3 Tea processing 59

3.3.4 Tea price and exports 60

3.4 Chapter conclusions 62

Chapter 4: Research Methods 63

4.0 Introduction 63

4.1 Selection of research sites 63

4.2 Sample 65

4.3 Data collection 68

4.3.1 Secondary data 68

4.3.2 Primary data 69

4.4 Data analysis 71

4.4.1 Descriptive analysis 71

4.4.2 Quantitative analysis 71

4.5 Chapter conclusions 83

Chapter 5: Tea Productivity and Technical Efficiency of Tea Production in Phu Tho Province 85

5.0 Introduction 85

5.1 Overview of analytical results 85

5.2 Characteristics of tea households in Phu Tho province 86

5.3 Estimation of tea productivity and technical efficiency of tea production 89

5.3.1 Tea productivity and technical efficiency of contracted farmers 89

5.3.2 Tea productivity and technical efficiency of independent farmers 92

5.4 Discussion 96

5.4.1 Factors affecting tea productivity 96

5.4.2 Factor affecting technical efficiency of tea production 98

5.5 Chapter conclusions 101

Chapter 6: Effects of Contract Farming on Tea Productivity, Income and

Poverty of Farmers 103

6.0 Introduction 103

6.1 Overview of analytical results 103

6.2 Effects of contract farming on tea productivity and farmer income 104

6.2.1 Estimation of determinants affecting contract participation 104

6.2.2 Demonstration of the common support region 108

6.2.3 Selection of a matching algorithm 109

6.2.4 Assessment of the match quality 111

6.2.5 Estimation of average treatment effects on the treated for tea productivity and farmer income 113

6.2.6 Implementation of sensitivity analysis 114

6.3 Influences of determinants on poverty of tea farmers 114

6.4 Discussion 115

6.4.1 Effects of contract farming on tea productivity 116

6.4.2 Influences of contract farming on income of farmers 116

6.4.3 Effects of determinants on poverty of tea farmers 118

6.5 Chapter conclusions 118 Chapter 7: The Relationship between Economic Growth, Tea Exports and Poverty in Vietnam 120 7.0 Introduction 120 7.1 Overview of analytical results 120 7.2 Economic growth, tea exports and poverty in Vietnam: An overview 120

7.3 The relationship between economic growth, tea exports and poverty in Vietnam 121

7.3.1 Implementation of the unit root test 122

7.3.2 Determination of the lag length 123

7.3.3 Estimation of the VAR model 124 xi

7.3.4 Testing the Granger causality 125

7.3.5 Examination of eigenvalue stability 126

7.3.6 Performance of the Johansen co-integration test 127

7.4 Discussion 128

7.5 Chapter conclusions 130

Chapter 8: Conclusions and Policy Implications 132

8.1 Conclusions 132

8.2 Policy implications 134

8.2.1 The Government 134

8.2.2 Phu Tho province 138

8.2.3 Enterprises and tea farmers 139

8.3 Research limitations and direction for further studies 141

References 143

Appendices 167

Appendix A (Chapter 3) 167

Appendix B (Chapter 6) 172

Appendix C (Chapter 7) 175

xii

List of Figures

Figure 2.1: Three approaches of contract theory 16

Figure 2.2: Model for theoretical framework of the research 33

Figure 3.1: Annual per capita tea consumption in selected countries in 2016 50

Figure 3.2: Quantity of tea exported by Vietnam to major markets (2012–2014) 56

Figure 3.3: Value of tea exported by Vietnam to major markets (2012–2014) 57

Figure 3.4: Producer and retail prices of tea in Phu Tho province 60

Figure 3.5: Export volume and turnover of processed tea by Phu Tho province 61

Figure 4.1: Administrative map of Phu Tho province, Vietnam with research sites 64

Figure 4.2: Sampling procedure in Phu Tho province 65

Figure 4.3: Measurement of technical efficiency 72

Figure 7.1: Eigenvalue stability condition 127

xiii

List of Tables

Table 3.1: Tea area, production and yield of major producers 46

Table 3.2: Value of tea production, percentage of agricultural value added and agricultural employment in total labour force of main producers 48

Table 3.3: Producer price of tea in selected countries in 2015 and 2016

(US$/tonne) 49

Table 3.4: Tea exports and imports of the world 51

Table 3.5: Planted area of tea in Vietnam (thousand hectares) 52

Table 3.6: Harvested area of tea in Vietnam (thousand hectares) 52

Table 3.7: Tea production in Vietnam (thousand tonnes) 53

Table 3.8: Yield of fresh tea in Vietnam (100kg/hectare) 54

Table 3.9: Planted area and productivity of leaf tea in Phu Tho province 58

Table 4.1: Sample of households interviewed in Phu Tho province 67

Table 4.2: Number of contracted households interviewed in Phu Tho province 70

Table 4.3: Description of covariates in the stochastic frontier model 75

Table 4.4: Description of covariates in the treatment effect model for tea productivity 80

Table 4.5: Description of covariates in the treatment effect model for income of tea farmers 80

Table 4.6: Description of covariates in the logit model 82

Table 4.7: Description of covariates in the VAR model 83

Table 5.1: Characteristics of tea households in Phu Tho province 88

Table 5.2: Maximum likelihood estimates of the stochastic frontier model and inefficiency model for contracted farmers 89

xiv

Table 5.3: Technical efficiency of tea production of contracted farmers 92

Table 5.4: Maximum likelihood estimates of the stochastic frontier model and inefficiency model for independent farmers 93

Table 5.5: Technical efficiency of tea production of independent farmers 95

Table 6.1: Logistic model of factors determining contract participation for tea productivity 105

Table 6.2: Logistic model of factors determining contract participation for farmer income 107

Table 6.3: Distribution of estimated propensity scores 108

Table 6.4: Matching performance of matching methods for tea productivity 110

Table 6.5: Matching performance of matching methods for farmer income 110

Table 6.6: Propensity scores and covariate balancing for the NNM(5) 111

Table 6.7: Propensity scores and covariate balancing for the KM(0.25) 112

Table 6.8: Quality of matching methods before and after matching 112

Table 6.9: Estimation of average treatment effects on the treated for tea productivity and farmer income 113

Table 6.10: Number of poor and non-poor households for tea producers surveyed 114

Table 6.11: Determinants of poverty of tea producers in Phu Tho province 115

Table 7.1: Characteristics of economic growth, tea exports and poverty in Vietnam (1977–2016) 121

Table 7.2: Augmented Dickey-Fuller test for the unit root 123

Table 7.3: Selection of the lag length 124

Table 7.4: Results of the Granger causality Wald test 126

Table 7.5: Results of trace statistic in the Johansen co-integration test 128

List of Abbreviations

ATT Average Treatment Effects on the Treated

EUR European Euro

FAO Food and Agriculture Organization of the United Nations

GDP Gross Domestic Product

HA Hectare

KG Kilogram

KM Kernel Matching

MARD Ministry of Agriculture and Rural Development, Vietnam

NNM Nearest-neighbour Matching

NPK Nitrogen (N) Phosphorus (P) Potassium (K) (Mixed Fertiliser)

NTPHEPR National Target Program on Hunger Eradication and Poverty Reduction

OLS Ordinary Least Square

PSM Propensity Score Matching

SFM Stochastic Frontier Model

TE Technical Efficiency

VAR Vector Autoregressive

VBCSD Vietnam Business Council for Sustainable Development

VND Vietnam Dong

UNCTAD United Nations Conference on Trade and Development

US$ United States Dollar

xvi

Abstract

The research investigates the impacts of contract farming on tea productivity, income and poverty of farmers in Phu Tho province, Vietnam. Moreover, this study examines the causal relationship between economic growth, tea exports and poverty in Vietnam.

In Vietnam, there is only a limited number of research studies on contract farming of tea production. The findings in the limited literature about the effects of contract farming on crop productivity and farmers’ welfare are not only inconclusive but the studies were based on cross-sectional data, which only measure association and not causality.

The paucity of specific knowledge on the impacts of contract farming on

Vietnamese tea farmers especially in rural areas like Phu Tho province motivates the current study. Consequently, this study made three empirical contributions to the extant literature.

First, this study employs both cross-sectional and time series data to quantitatively explain the relationship between businesses and farmers in contract farming of tea production in Phu Tho province and in Vietnam. The study found that contracted farmers had superior technical efficiency that is statistically significant compare to non-contracted farmers in Phu Tho province. This finding confirms the positive effect of contracting on technical efficiency, which dominate the extant literature unlike in Ngoc et al. (2014) who found that there were no significant difference in technical efficiency between contracted and non-contracted farmers in Phu

Tho province. Second, the current research is unique and different from existing studies because it explores mechanisms of the influence of contract farming on crop productivity and income of farmers in Phu Tho province by using the propensity score matching. Ngoc et al. (2014) only employed the stochastic frontier production function

xvii

to estimate technical efficiency of tea production and Oanh et al. (2016) estimated only the revenue and cost of tea farmers in Phu Tho province. The third investigation used time series data and confirmed the importance of tea production to economic growth and poverty reduction in Vietnam. The time-series analysis using data over the last four decades (1977–2016) was the first of its kind in Vietnam to the knowledge of the author. The limited existing studies only assess contract farming of tea production at the provincial level (Saigenji and Zeller, 2009; Ngoc et al., 2014; and Oanh et al., 2016).

Along with macro policies recommended to the government and Phu Tho province, the research also addresses recommendations to firms and tea farmers in Phu

Tho province to facilitate the achievement of contract farming and reduce poverty. First, specification of the relationship in the contract. Second, construction of the contents of the contract through discussions between enterprises and tea growers. Third, signing the contract through intermediate organisations such as farmer associations and cooperatives. Fourth, specification of the consistency of the contents of a contract with the conditions and nature of the contracting parties. Fifth, adoption of a tea farming information system. Lastly, specification of communities and tea farmers that are consistent with contract farming of tea production.

xviii

CHAPTER 1: INTRODUCTION

1.0 Introduction

This chapter discusses the importance of contract farming on tea production in Vietnam.

The chapter also presents the aims, research questions, significance of the research and thesis organisation.

1.1 Overview of contract farming and tea production in Vietnam

Tea production makes a significant contribution to the national economies of many countries. It is considered an important crop to contribute to poverty reduction in developing countries, specifically as it can involve contract farming. In 2014, China had the highest value of tea production at US$10.1 billion, followed by India (US$3.2 billion), Kenya (US$1.2 billion), Sri Lanka (US$899 million) and Vietnam (US$315.7 million). In these countries, smallholders play a crucial role in tea production being more than 90 percent of producers in China, more than 80 percent in Vietnam, nearly 70 percent in Kenya and Sri Lanka, and more than 30 percent in India (FAO, 2016).

The harvested tea areas of producer countries have changed significantly in recent years. From 1990 to 2014, the harvested tea area of China increased by 1.4 million hectares, India increased by nearly 150,000 hectares and Kenya by more than

100,000 hectares. Vietnam increased its harvested area from 80,000 hectares in 2000 to

128,000 hectares in 2014. In the same period, the tea production value of China

increased by 1,290 percent, followed by Vietnam (545 percent), Kenya (199 percent),

India (120 percent), and Sri Lanka (90 percent) (FAO, 2016).

International trade in tea is complex because of requirements in blending, packaging and branding and differences among export countries. Currently, blending, packaging and branding are often implemented in producer or intermediate countries due to lower costs. For example, in 2013, Argentina exported 74,466 tonnes of tea to more than 40 countries. However, the United States of America was the largest export market of this country, accounting for more than 65 percent of tea exports (FAO, 2014).

In 2014, the global volume of exported tea was 1.73 million tonnes, with a value of

US$5.61 billion. In 2014, Sri Lanka, China, Kenya, India and Vietnam were the dominant producer exporting countries of tea, while the volume of tea imported by major importing countries such as the Russian Federation, the United States of America, the United Kingdom, Egypt and Pakistan was 640,000 tonnes (FAO, 2016).

In Vietnam, tea trees are mainly planted in three areas, the Northern midlands and mountainous area, North Central and Central coastal area, and Central Highlands, because of favourable soil and climatic conditions. Currently, tea is grown by small farms in poor areas and therefore tea production has a strong potential for reducing poverty because it requires few inputs and it is a labour-intensive production (Asian

Development Bank, 2004).

According to Vietnam Tea Association, in 2014, the harvested area of tea in

Vietnam was 113,000 hectares and the average yield was 8 tonnes per hectare. In 2014, the exported volume of Vietnamese tea was 130,000 tonnes with value of US$230 million. The volume of tea consumed in the domestic market accounted for 33,000 tonnes with a revenue of VND2,300 billion. Vietnam is ranked fifth by volume in

exporting tea countries in the world, behind China, India, Kenya and Sri Lanka

(VBCSD, 2015). However, the tea sector in Vietnam has numerous difficulties related to low productivity, high transaction costs and poor quality (Asian Development Bank,

2004). Other threats faced by the tea sector of Vietnam include small-scale production, weak coordination among stakeholders in production, shortages in raw materials and lack of finance to upgrade to international standards (Khoi et al., 2015; and VBCSD,

2015).

In recent years, contract farming has become a significant development in global agriculture (Trang, 2014). There are various definitions. Contract farming is known as an agreement between farmers and a buyer, which specifies production and marketing of agricultural products (Minot, 1986). Contract farming is an agreement between farmers and firms in producing and providing agricultural products with a certain price

(Eaton and Shepherd, 2001). Contract farming is a type of vertical integration which continues to evolve (Rehber, 2007).

In Vietnam, contract farming (agricultural contract) has been promoted in 2002 after the Prime Minister released the Decision 80/2002/QD-TTg to encourage the linkage between enterprises and farmers in terms of purchasing agricultural produce via contracts. Further, in 2013, the Decision 62/2013/QD-TTg of the Prime Minister has been implemented to facilitate coordination among producers, cooperatives and enterprises in production, processing and marketing of agricultural commodities based on large-scale production. Contract farming is considered to research in different agricultural products such as rice (Nhan et al., 2013), vegetable (Wang et al., 2014); dairy (Saenger et al., 2014); and catfish (Trifkovic, 2014).

In Vietnam, the proportion of tea products traded via contract farming between producers and enterprises has grown rapidly in recent years. For instance, in 2010, only

9 percent of tea production in Vietnam was traded between producers and enterprises via contracts (Khoi, 2014), but this proportion increased to more than 40 percent in

2014 (ActionAid, 2015). Therefore, it is important to research both the theoretical and empirical aspects of contract farming in tea production in Vietnam to provide evidence on the effects of contract farming on tea productivity and the income of farmers in tea production.

Contract farming has theoretical appeal as a production model because it is seen as providing an opportunity for farmers to access inputs, extension services and markets from their contractors. However, results from the existing literature about the impact of contract farming on productivity and income are mixed. While some studies found that contract farming achieves high output (Tripathi et al., 2005; Igweoscar, 2014; and

Mishra et al., 2018) or income (Miyata et al., 2009; Narayanan, 2014; Maertens and

Velde, 2017; and Ton et al., 2018), others found no increase in output or income

(Dhillon and Singh, 2006; Mwambi et al., 2016; and Dube and Mugwagwa, 2017). It has been suggested that possibly more educated farmers understand the terms of a contract better than less educated farmers and so are better able to use contract opportunities such as access to superior inputs to increase production. This research investigates the reasons for such differences in the context of Vietnam.

As in most developing countries, the Government of Vietnam has implemented the National Target Program on Hunger Eradication and Poverty Reduction

(NTPHEPR) in recent years. The NTPHEPR promotes economic growth, facilitates production forces and contributes to stability and sustainability of socio-economic

development. Through the NTPHEPR, Vietnam has achieved the Millennium

Development Goals in poverty reduction. The poverty rate in the country dropped from

58 percent in 1993 to below 4.5 percent by the end of 2015 and the national poverty rate declined by 2 percent annually. The livelihood of the poor and infrastructure provision in mountainous and ethnic minority areas has improved (Chuyen, 2016).

The NTPHEPR is implemented across the whole country. Located in the

Northern midlands and mountainous areas of Vietnam, Phu Tho province has 277 commune-level administrative units, with 218 mountainous communes, and 72 extremely poor communes. At the end of 2016, the growth rate of the agriculture, forestry and aquaculture sector of this province was 6.18 percent annually and this sector contributed 25.56 percent to the provincial economy. At the same time, 10.51 percent of households in Phu Tho province were poor, 8.03 percent were near poor, and average income per capita was US$1,454 (Phu Tho Statistics Office, 2017).

In this context, tea has been considered as a main crop in the program of developing the agricultural economics of Phu Tho province (Decision No 99/2008/QD-

TTg of the Prime Minister of Vietnam, 2008). The planted area for producing tea in this province is expected to reach 15,500 hectares in 2020 (the People’s Committee of Phu

Tho Province, 2012). By 2017, in Phu Tho province, the harvested area of tea was

15,533.9 hectares, average yield was 11.1 tonnes per hectare and production of tea was

172,742 tonnes (Phu Tho Statistics Office, 2018). In Vietnam, Phu Tho province is third in production and fourth in planted area of tea. The total volume of tea processed through enterprises, agencies and villages is 50,000 to 55,000 tonnes annually, with 70 percent black tea and 30 percent green tea (Department of Agriculture and Rural

Development in Phu Tho province, 2015). A major target of the NTPHEPR in Phu Tho

province is to promote tea production in Phu Tho province because climatic conditions are favourable. However, there has been no study on contract farming in tea production in Phu Tho province compared to other areas such as Son La province (Saigenji and

Zeller, 2009) and Thai Nguyen province (Nghia, 2009).

The Phu Tho province is selected for this research because, unlike the other tea producing areas of Thai Nguyen, Ha Giang, Tuyen Quang and Son La in Northern midlands and mountainous areas of Vietnam, tea production is the main source of income to farm households and contributes to the NTPHEPR goal of the province. In addition, almost all enterprises in this province procure tea materials from producers and then process the tea into black tea only for export. Since poverty rates and literacy rates vary across Vietnam, the impact of contract farming at different locations is an important empirical issue. This study contributes to knowledge of the impact of contract farming on tea productivity and income in Phu Tho province. Existing studies by Ngoc et al. (2014) and Oanh et al. (2016) assessed contract farming of tea production in Phu

Tho province, Vietnam. However, a research by Ngoc et al. (2014) only focused on estimating technical efficiency of tea production in Phu Tho and Thai Nguyen provinces, Vietnam. Further, Oanh et al. (2016) described characteristics of tea production of contracted and non-contracted farmers in Phu Tho province. However, both studies ignore to investigate impacts of contract farming of tea production on tea productivity and farmer’s income in Phu Tho province. In addition, none of these examines the role of tea production in poverty reduction of tea growers in Phu Tho province and the relationship between tea exports, economic growth and poverty in

Vietnam in recent years. The study, therefore, needs to be carried out to narrow down these gaps of existing research. Moreover, the study is also timely given the

Government’s NTPHEPR and the direction for development of trade in rural areas, for example, to reduce the poverty rate by 2 percent and increase the per capita income of poor households by 3.5 times in 2020 compared to 2010, and increase the proportion of agricultural products sold via contracts by 45–50 percent in 2020.

1.2 Research questions and aims

Due to the gaps in knowledge about certain theoretical aspects of contract farming and the paucity of specific knowledge on the impacts of contract farming on Vietnamese tea farmers, five research questions are generated as follows:

1. What factors affect tea productivity and the technical efficiency of tea

production in Phu Tho province?

2. What is the effect of contract farming on tea productivity in Phu Tho

province?

3. What is the influence of contract farming on farmers’ income in Phu Tho

province?

4. What are the determinants of poverty of tea farmers in Phu Tho province; and

5. What is the relationship between tea exports, economic growth and poverty in

Vietnam?

The major objectives of this study are to:

1. Investigate factors affecting tea productivity and the technical efficiency of

tea production in Phu Tho province;

2. Estimate the impacts of contract farming on tea productivity in Phu Tho

province;

3. Estimate the influence of contract farming on farmers’ income in Phu Tho

province;

4. Examine the determinants of poverty of tea farmers in Phu Tho province; and

5. Investigate the causal relationship between economic growth, tea exports, and

poverty in Vietnam.

1.3 Significance of research

This study is important for four reasons.

First, it augments existing theoretical knowledge about contract farming and applies streams of contract theory such as transaction cost economics, property rights theory and agency theory to explain the relationship between businesses and farmers in contract farming of tea production.

Second, it provides information on factors affecting tea production in Phu Tho province, specifically under contract regimes. Tea is known as a key export product which contributes to improving income, generating employment and reducing poverty for farm households, especially in the Northern midlands and mountainous area of

Vietnam, including Phu Tho province which is the location of the field work. However, despite the importance of tea production in the socio-economic development of the region, there is limited information on the role of tea production in poverty reduction of tea growers and the effect of contract farming on tea productivity and income of farmers

(Saigenji and Zeller, 2009; Ngoc et al., 2014; and Oanh et al., 2016).

Third, it contributes to knowledge for improved policy development and delivery in economic development, specifically the design and effectiveness of the

Vietnamese government’s NTPHEPR in Phu Tho province. By exploring in depth the impact of contract farming on tea productivity and incomes of farmers as well as assessing determinants of poverty of tea producers in this region, lessons can be learnt that could be applied to other regions.

Lastly, it re-examines the role of tea exports in economic growth and poverty reduction in Vietnam by investigating the relationship between tea exports, economic growth, and poverty for the last four decades (1977_2016) using a time-series data. This analysis assists the government in terms of drawing feasible policies to foster tea production, economic growth and reduce poverty in the country.

1.4 Thesis organisation

This thesis has eight chapters. Chapter 1 covers background, research questions, aims and significance of the research. Chapter 2 examines the literature including the theoretical framework and empirical studies. The theoretical framework section discusses concepts of contracts and contract farming, types of contracts, the history of contract farming, contract theory, motivation for contracting, and advantages and disadvantages to farmers of contracts. The empirical review analyses the effects of contract farming on productivity and income, the context of production, consumption and trade of tea in the world, and the tea sector in Vietnam. Chapter 3 firstly provides an overview of global production, consumption and trade of tea. It then sets the context for tea production in Vietnam. Chapter 4 discusses research methods, including selection of research sites, sample selection, data collection and data analysis. Chapter 5 assesses tea productivity and the technical efficiency of tea production in Phu Tho province. Chapter

6 investigates influences of contract farming on tea productivity, income and poverty of farmers. Chapter 7 examines the causal relationship between economic growth, tea exports and poverty in Vietnam. Conclusions and policy implications are summarised in

Chapter 8.

CHAPTER 2: LITERATURE REVIEW1

2.0 Introduction

This chapter presents the theoretical framework and empirical review of the study. The theoretical framework of this research focuses on discussing definitions of contracts and contract farming, the history of contract farming, contract theory, types of contracts, motivation for contracting, and advantages and challenges for farmers in contracts. The framework helps analysis of the results in subsequent chapters. The empirical review analyses influences of contract farming on productivity and income of farmers, in the context of Vietnam. The combined theoretical and empirical review identify the gaps in knowledge which lead to the research questions.

2.1 Theoretical framework

2.1.1 Concepts of contracts

Concepts of contracts have been defined by many scholars in the literature. Contracts are legally binding promises (MacNeil, 1974). Contracts can be defined as a formal legal commitment in which each party gives interests but these are not necessarily written and approval (Masten, 2000). A contract is an agreement in which two parties express their actions based on reciprocal commitments (Brousseau and Glachant, 2002).

1 Anh Tru Nguyen, Janet Dzator, Andrew Nadolny (2015). Does contract farming improve productivity and income of farmers? A review of theory and evidence. Journal of Developing Areas, 49(6) (Special Issue), 531-538.

A contract describes the legally binding rights and duties of buyers and sellers which can be formally written or informally understood but controlled by public law (Hart and

Moore, 1990; O’Looney, 1998; Tirole, 1999; and Bajari and Tadelis, 2001). Contracts can be specified as integrated mechanisms which allocate value, risk and decision rights among buyers and sellers (Paulson et al., 2010). In general, most scholars indicate that contracts are agreements which refer to the exchange of parties based on their rights and obligations under the legal framework.

2.1.2 Concepts and types of contract farming

There are numerous definitions of contract farming in the literature. Contract farming can be defined as a written or oral agreement between a firm and farmers which indicates production and/or marketing conditions (Roy, 1963). In contracts, a firm buys agricultural products from local producers with conditions on planting time, volume and prices signed prior between parties, and contracted enterprises often provide credit, inputs, farm machinery rentals and technical assistance to famers (Glover and Ghee,

1992). Contract farming is a system where processing or exporting units purchase the agricultural products of farmers with given terms in the contract (Baumann, 2000).

Contract farming is an agreement between contractors and producers in terms of producing and providing agricultural products at given prices and timing of delivery

(Eaton and Shepherd, 2001; and Minot, 2007). Contract farming reflects the relation between state or private enterprises and growers instead of transactions in spot markets

(Rehber, 2007). Contract farming is defined as a component of market liberalisation and agricultural transformation which combines operations of corporations and smallholders

(Simmons et al., 2005). Contract farming is able to have positive or negative influences on producers based on specific situations of the economy (Singh, 2000). Contract farming is a means of connecting smallholders in developing countries to export markets and modern economies (Kristen and Sartorius, 2002). Concepts of contract farming concentrate on specifying the relation between contractors and producers in producing and delivering agricultural products under an agreement at given conditions in prices, volume, quality and delivery time.

Contracts have been classified into three categories: market specification contracts, resource providing contracts, and production management contracts. Market specification contracts refer to terms and conditions of future agreements which determine quantity, timing and prices of products to be sold. This contract type is often signed during planting time, as producers can independently decide production processes and contractors only consider the degree of product satisfaction when they purchase the products. Resource providing contracts specify provision of inputs and technical assistance from buyers to producers. In this contract type, due to the resources provided by buyers, farmers are obtained some managerial assistances and supervisions of integrators. Prices are set based on open markets and the income of producers is guaranteed at a minimum level. Production management contracts regulate production methodologies, input standards, planting and harvesting specifications. In this contract type, integrators take a dominant decision in production because market and price risks are transferred from producers to contractors (Kohls and Uhl, 1985; Baumann, 2000; and Hernandez, 2011).

2.1.3 History of contract farming

Contract farming first appeared more than a century ago and involved agreements between corporations of capitalist countries and producers in colonies. For example, the

Japanese implemented sugar production in Taiwan in the last decades of the nineteenth century and US firms operated in the banana industry in Central America in the early part of the twentieth century (Watts, 1994). Since then, contract farming has expanded rapidly in vegetable production in the United States, the seed industry in Europe in the decades before the Second World War (Rehber, 2007), and pig production in the United

States (Hamilton, 2008).

Contract farming plays an important role in agricultural production. Contract farming accounts for about 15 percent of agricultural output in developed countries

(Rehber, 2007). Contract farming accounted for 39 percent of the total value of US agricultural production in 2001 (Young and Hobbs, 2002).

In Europe, contract farming also plays an important role in transitional countries.

For instance, contract farming accounts for 38 percent of the production of dairy, poultry and sugar in Germany (Prowse, 2012). In the Czech Republic, Slovakia and

Hungary, the proportion of corporate farms uses contracts varies from 60 percent to 85 percent. The rate of food companies employing contracts in Armenia, Georgia,

Moldova, Ukraine and Russia increased from 25 percent in 1997 to 75 percent in 2003

(Swinnen and Maertens, 2007).

The development of contract farming varies in different regions of the world.

Contract farming grew significantly in the 1950s including for bananas in Honduras, barley in Peru, and vegetables and grain in Mexico (Prowse, 2012). In Brazil, more than

70 percent of poultry and 30 percent of soya production are produced through contract farming (ActionAid, 2015). Similarly, contract farming has increased dramatically in

Southeast and South Asia. For example, in Indonesia, the operation of contract farming has increased significantly since 1956 when the federal government facilitated contract farming through the Federal Land Development Agency (Rehber, 2007). In Malaysia, contract farming has also expanded sharply based on state-promoted out-grower arrangements (Morrison et al., 2006). In Vietnam, more than 90 percent of cotton and fresh milk and 40 percent of rice and tea come from contract farming (ActionAid,

2015). In East Asia, contract farming has also expanded dramatically. For example, more than 18 billion hectares were planted under contract farming in 2001 in China, an increase of about 40 percent compared to 2000 (Guo et al., 2007).

The growth of contract farming is also significant in Sub-Saharan Africa in the late 1980s. In Mozambique, all cotton and tobacco are produced under contract farming and more than 12 percent of the rural population is implementing a contract (ActionAid,

2015). In Kenya, 50 percent of tea and sugar is produced under contracts and the number of contract growers for horticultural exports has increased rapidly (Prowse,

2012).

2.1.4 Contract theory

There are different approaches to the theory of contract in the literature. This research focuses on reviewing three approaches due to their association with the research questions: transaction cost economics, property rights theory and agency theory (Figure

2.1). Contract farming of tea production involves transactions between farmers and

firms, illustrated in transaction cost economics. Further, contracts can be incomplete because of limitations in information and knowledge (bounded rationality) of parties, as discussed in property rights theory. Transactions between enterprises and farmers in the contract may be seen as transactions between the principal and agents, analysed by agency theory.

Figure 2.1. Three approaches of contract theory

Contract theory

Transaction cost economics Property rights theory Agency theory

Uncertainty Asset specificity Frequency Incomplete Specific Moral Adverse contract investment hazard selection

Source: Author, 2015

Transaction cost economics

Transaction cost economics was developed by observations by Coase (1937) and then developed by other scholars (Williamson, 1975 and 1985; North, 1990 and 2004). This theory illustrates transactions among stakeholders in the economy to take advantage of specialisation and division of labour in an organisation. During transaction processes, rights to use goods or services are transferred among economic units and this generates costs (Vavra, 2009). Transaction costs are known as the costs of operating the economic system (Arrow, 1969), or costs related to information, negotiation, enforcement and monitoring of transactions (Matthews, 1986; Kherallah and Kirsten, 2001; and Vavra,

2009). Therefore, transaction cost economics refers to the existence of a firm by minimising its transaction costs.

There are three elements of a transaction: uncertainty, asset specificity and frequency (Coase, 1937; Williamson, 1979; and Gregory, 2011). Uncertainty refers to variations of unpredictable variables in a transaction (Gregory, 2011). For example, technology can change in the future and it affects transactions among parties

(Choudhury, 1997; and Song and Montoya-Weiss, 2001). In addition, a transaction can be influenced by the variability of existing demand for commodities or services

(Choudhury, 1997). On the other hand, uncertainty occurs because of either volatility or ambiguity. Volatility refers to changes of the environment and therefore parties in the contract have to adapt to these variations. To deal with this issue, renegotiation conditions should be considered in the contract. Ambiguity appears due to lack of clear information, uncertainty of causal-effect relationships between variables, and uncertainty of actions and their potential effects (Carson et al., 2006). Asset specificity refers to the transferability of assets in a given transaction (Williamson, 1985).

Williamson (1981) identified three types of asset specificity: location specificity, physical asset specificity, and human asset specificity. Location specificity means firms minimise costs when assets are located near each other. Physical asset specificity explains that assets have little value in their ownership, but have value in a transaction.

Human asset specificity expresses that human skills can be learned and then used in a transaction. Rent seeking can appear during the process of asset transferability. Rent seeking is known as the waste of individuals and groups in seeking wealth transfers

(Pasour, 1983). Rent seeking can be seen as a result of the combination of incomplete contracts and relationship specific investment. Due to the rigidity of fundamental

variables, decision governance has become a main source of inefficiency in transaction cost economics theory (Wu, 2006). Frequency refers to the number or volume of transactions. The degree of frequency is important because it affects justification of alternative governance structures. For example, a larger number or volume of transactions can lead to an increase in justification for alternative governance structures of a firm (Williamson, 1979 and 1985).

There are differences between transaction cost economics and neoclassical theories of the firm. Firstly, transaction cost economics proposes behavioural assumptions related to opportunism and bounded rationality. In neoclassical economics, people are viewed as self-interested. However, according to transaction cost economics, sometimes some individuals behave opportunistically because they want to take advantage in a transaction and that generates costs in exchanges. Opportunism is a key assumption in transaction cost economics since it is associated with the motivations of human behaviour (Williamson, 1985). In terms of bounded rationality, neoclassical theory assumes that individuals have perfect information. However, based on transaction cost economics, humans cannot access perfect information because of limitations in their cognitive abilities and imperfect information conditions and, as a result, this leads to right or wrong decisions of parties in a transaction (Williamson,

1985; and Martins et al., 2010). Secondly, transaction cost economics views the transaction as the unit of analysis, not the firm, industry or individual. Thirdly, according to transaction cost economics, the firm can be known as a governance structure, not a production function. Lastly, transaction cost economics views property rights as costly and problematic to enforce (Williamson, 1996; and Cook and Barry,

2004).

Property rights theory

Property rights can be defined as the rights of individuals to use resources which are supported by the force of etiquette, social custom and legislation. Property rights indicate norms of human behaviour because these criteria allow people to use legal resources (Alchian, 1965; and Furuboth and Pejovich, 1972). Property rights theory was considered in studies on strategic management implemented by Coase (1937, 1959 and

1960) and then developed by Grossman and Hart (1986), Hart and Moore (1990), Hart

(1995), Barzel (1997) and others. According to neoclassical economic theory, the market is an economic system where the price mechanism operates efficiently in terms of coordinating economic activities. An externality may be internalised between two parties if there are no transaction costs and property rights are consistently established.

The outcome depends on the initial allocation of property rights (Coase, 1960). The existence of a firm can be explained by the price mechanism because the price mechanism does not function efficiently (Kim and Mahoney, 2005). Three characteristics of universality, exclusivity and transferability were used in property rights theory to analyse the relation between economic efficiency and property rights

(Cook and Barry, 2004).

Incomplete contracts and relationship specific investments are two main components of property rights theory (Grossman and Hart, 1986; and Hart and Moore,

1990). Incomplete contracts occur because of bounded rationality and contracting costs

(Kherallah and Kirsten, 2001). Incomplete contracts can lead to opportunism and economic rents since it is impossible to design contingent contracts for relationship specific investment (Anderson and Hill, 1991; and Kherallah and Kirsten, 2001).

Property rights theory identifies that subsequent investments and effort have become the main source of inefficiency (Wu, 2006). Grossman and Hart (1986), Hart and Moore

(1990), Hart (1995) and Barzel (1997) focused their attention on the ownership issue in an economic organisation because they note that when contracts are incomplete, ownership becomes a source of power. As a consequence, residual control rights appear and this can lead to an increase in residual risk bearing. According to these researchers, property rights theory assumes efficient negotiation, and where rent seeking is relevant to contractible prior specific investment, property rights theory does not require contractible specific investment. Based on property rights theory, efficient negotiation is a reason that explains decisions of parties in sharing the surplus from their specific investment because the surplus share of each party determines their investment incentive and, in turn, asset ownership influences the surplus share of parties.

To sum up, property rights theory considers maximisation of the investment incentives of parties and the coordination decision determines prior investment and total surplus (Gibbons, 2005). Furthermore, this theory recognises that different types of property rights arise to deal with the economic issue of allocating scarce resources and these also affect economic behaviour and economic efficiencies (Coase, 1960; and

Pejovich, 1982 and 1995).

Agency theory

Agency theory illustrates the relation between a principal and an agent, in which a principal delegates work to an agent who does the work (Jensen and Meckling, 1976; and Cook and Barry, 2004). Agency theory has been researched and developed based on

the risk sharing literature because agency issues occur when there are differences in goals and the labour division of parties (Ross, 1973; and Jensen and Meckling, 1976).

The agency problem arises when the principal and agent have different objectives and the existence of asymmetric information creates difficulties for the principal in examining the agent’s behaviour. This issue occurs in a firm due to the relationship between managers and outside shareholders and, as a result of this, the principal cannot control the actual actions of agents. Another issue is risk sharing which arises when the principal and agent have different attitudes toward risks. Each party has different options and actions in response to risks and these can affect their relations in the contract (Eisenhardt, 1989; Cook and Barry, 2004; and Boland et al., 2008).

There are two major contributions of agency theory to the organisational behaviour. The initial contribution is the treatment of information. According to agency theory, information is similar to a commodity which can be purchased in the market.

Hence, information plays an important role in establishing formal information systems in an organisation. This implies that organisations need to build up efficient information systems to deal with the actions of agents. Another contribution of this theory is risk implications. Organisations face an uncertain future due to the effects of state regulation, demand, competition and technological innovation that influence the outcomes of organisations. Variation of outcomes can impact the degree of risk acceptance of parties and then this affects contracts between the principal and agent

(Eisenhardt, 1989).

Agency theory refers to asymmetric information between contracting parties due to moral hazard and adverse selection (Larsen, 2008). Asymmetric information demonstrates the situation when one party knows more information than another party.

Moral hazard exists when the principal is unable to observe behaviours performed by agents and, as a consequence, the principal has to rely on outcomes reported by agents.

For example, in contract farming of tea production in Vietnam, sometimes contractors

(principals) cannot control production practices such as land preparation, fertiliser and chemical applications and harvest techniques undertaken by tea growers (agents) because production zones are broad and the number of tea producers is large. Therefore, growers may produce tea based on their own practices and then sell a partial volume of products to open markets to gain higher prices. In contrast, adverse selection occurs when the agent knows more information than the principal and, as a result, the outcome can be inefficient because it is very difficult for the principal to write a contract which provides consistent incentives for agents. For example, in tea production in Phu Tho province in Vietnam, in some cases local growers (agents), due to their experience in tea production, know more about weather conditions and characteristics of soil and tea trees than contractors and these producers accept more risk in contract price and require higher incentives given by contractors (principals) in a contract. If contractors pay a higher price, this leads to an increase in their costs. In contrast, if contractors pay a lower price, tea growers are able to refuse the contract.

Agency theory recognises that the relation between a principal and agent should consider efficient information and risk-bearing costs. Unlike transaction cost economics, agency theory selects contracts between the principal and agent as the unit of analysis. Several assumptions have been proposed in this theory, including human, organisational and information assumptions. In terms of human assumptions, three elements are considered: self-interest, bounded rationality and risk aversion. Regarding organisational assumptions, this theory refers to conflicting objectives between parties,

efficiency as the key criteria, and asymmetric information between the principal and agent. Agency theory assumes that information is a product which can be purchased in the market. Agency and risk sharing problems are two main issues addressed in this theory and these appear because of the differences in goals and risk preferences between the principal and agent in contracts (Eisenhardt, 1989; and Cook and Barry, 2004).

2.1.5 Motivation for contract farming

Several reasons explain the participation of parties in contracts for farming.

First, contract farming can be defined as a type of vertical coordination. Parties who operate in the market have to consider integration with others due to the effects of externalities and imperfect or asymmetric information. These effects can generate higher exchange or transaction costs among parties. Hence, contract farming can assist stakeholders to reduce transaction costs.

Second, a reason for participating in contracts is uncertainty and risk reduction.

Uncertainty and risks can occur in prices of inputs and outputs, quantity and quality characteristics, and food safety. In a contract, input and output prices are negotiated by parties before implementing the contract, which helps decrease price uncertainties.

Moreover, based on a contract, a party produces and provides a volume of commodities at a certain quality, but this nearly never happens in the spot market. Another risk is food safety, a serious problem in society in recent years. This issue affects human life and environment and therefore requires more coordination among parties in terms of producing, processing and marketing agricultural products (Rehber, 2007).

Third, market structures can change because of the relationship between actors in a contract. Due to consumer requirements for food safety and nutrition, quantity and quality attributes of commodities should be more controlled and it is necessary to improve the linkages between producers, processors, retailers and others in food supply chains (Buccola and French, 1981; and Rehber, 2007).

Fourth, the improvement of production technologies is another motivation for contracting. Inconsistent provision of raw materials can lead to higher unit costs for a producer firm and, as a result, a firm can obtain lower revenues and inefficient production. In the contract farming framework, sustainable arrangements among parties allow the provision of stable raw materials and these help exploit machinery capacity more efficiently (Roy, 1963; and Harryman, 1994). Therefore, contract farming is a motivation for innovating technologies and providing more efficient production

(Rehber, 2007).

Lastly, contract farming is a vehicle for commercialisation and industrialisation in agriculture, especially in developing countries. In many developing countries, small farm households play an important role in agricultural production. However, they face many constraints such as capital shortage, lack of market information, price variations, backward production technologies, and weak negotiating position. Due to the terms agreed in a contract, contract farming is seen to be an affordable alternative which assists small farmers to deal with these challenges (Roy, 1963).

These motivations for contracting are also appropriate to the context of contract farming in Vietnam. For example, contract farming can be seen as a type of vertical integration among parties in producing, processing and marketing agricultural products such as tea. Without contract farming, both firms and growers have to face high

transaction costs to search for partners, procure materials and sell their products. In contrast, with contracts, the transaction costs of parties can be decreased because enterprises are able to establish a production zone which provides stable materials for processing and exporting, while producers benefit in terms of input provisions (such as seed, fertiliser, chemical and credit), technical assistance and output procurement.

Furthermore, contract farming is a means to reduce uncertainties and risks in agricultural production. In Vietnam, agricultural production faces many uncertainties and risks related to natural disasters such as typhoon, drought, flood and climate change, diseases, market shocks and price fluctuations. Hence, contract farming is necessary since this helps reduce the adverse impacts of uncertainties and risks. Due to competition pressures in the market, actors need to coordinate with each other in food value chains. Contract farming may assist to improve relations between stakeholders in producing, processing and marketing agricultural commodities by efficiently exploiting the strengths of each party in a contract. As a consequence of this, agricultural products can satisfy the requirements of consumers for quantity, shape and quality (nutrition, hygiene and food safety). Finally, because of globalisation trends, contract farming can assist small farmers to access markets and improve technologies and professionalism in agricultural production. Therefore, contract farming can increase industrialisation and commercialisation of agricultural production, especially in developing countries like

Vietnam.

2.1.6 Advantages and disadvantages of farmers in contracts

Advantages

Farmers often gain benefits when they participate in contract farming.

First, in a contract, inputs can be provided by firms at reasonable prices, adequate quantity and better quality. This assists producers to avoid uncertainties and reduce risks in input provisions when they purchase these in spot markets and, as a consequence, producers can reduce production costs, improve productivity and obtain a higher income. In this arrangement, contractors also gain benefits in terms of selling their products such as seeds, fertilisers, pesticides and so on. Furthermore, through providing production inputs, contractors expect to control and guarantee the quantity and quality of outputs which they will purchase in the future (Glover, 1994; Simmons,

2002; and Silva, 2005). In developing countries, most producers in agriculture are smallholders and therefore they do not have enough capital or face constraints in credit access. With a contract, producers can overcome this issue because contractors pre- invest by providing input production and farmers can reimburse after harvesting their products (Simmons, 2002).

Technical assistance is another advantage to producers in contract farming.

Farmers often adopt traditional technologies in agricultural production and this can lead to lower productivity and heterogeneous product quality. This may be appropriate for a self-sufficient economy or when selling agricultural products in open markets where buyers do not have strict requirements for product quality. In contrast, producers have to follow production procedures and criteria on quantity, quality and delivery time based on certain terms in a contract. In this case, farmers need technical assistance or technology transfer provided by contracting firms. In addition, assistance can support producers in reducing costs when searching for adequate production technologies

(Glover, 1994; and Simmons, 2002).

Due to the effects of technical assistance, the production management skills of farmers can be improved through a contract. Unlike in the past when most agricultural products were produced for domestic consumption, in a contract, some agricultural products are processed and then exported to international markets which require a huge volume and strict quality of products at the same time. Additionally, risks can be decreased when producers adopt advanced management skills in their production.

Hence, the production management skills of producers can be improved to satisfy these requirements (Glover, 1994; Simmons, 2002; and Silva, 2005).

Employment generation for households is a benefit of contract farming.

Especially in developing countries, there is no work after the harvest season for some labour in farm households and it is difficult for workers to find an off-farm job.

Therefore, some labour must migrate to cities to look for unskilled jobs with a low salary. By contrast, in a contract, some agricultural products such as coffee, cotton, tea and aquaculture products are more labour intensive which creates employment for labour in small farm households (Simmons, 2002) and reduces the negative impacts of the migration of labour from rural to urban areas of countries. Hybrid seed contracts increase the hiring of non-family labour and the use of female labour and provide off- farm work opportunities for neighbours of the contracting farmers in East Java,

Indonesia (Winters et al., 2005). In Punjab, India, contract farming creates the labor intensity by 640 hours per hectare for potato, 3,600–4,000 hours per hectare for tomato, and 740 hours per hectare for paddy rice. Therefore, employment in vegetable crops such as tomato is ﬁve times higher than that for paddy rice (Gill, 2001).

Lastly, market access is also an advantage to farmers when they participate in contract farming. In many developing countries, one of the most difficult issues for

agricultural producers is seeking markets for their products. Without contract farming, producers have to spend more of their time transporting products and selling in open markets with a fluctuating price. In some cases, farmers can gain higher income because prices in spot markets are higher than those in a contract, but their income is unstable because of variations of supply, demand and prices in the market. Contractors often guarantee to purchase outputs of producers and this not only assists farmers to have a stable and sustainable market but also reduces transaction costs. For example, in developing countries, domestic consumption of high-value crops is increasing significantly and exports of traditional commercial crops have also grown rapidly due to effects of contract farming (Swinnen et al., 2015).

In Vietnam, farmers gain benefits from contract farming. First, in a contract, producers often prefer to use inputs such as seed, fertilisers and chemicals provided by contractors because they think these inputs are better quality at a reasonable price compared to open markets. In practice, it is very difficult for local authorities and functional agencies such as police, market management and plant protection to control the quality of seed, fertilisers and pesticides sold in open markets. Second, producers expect to receive technical assistance from contractors since they want to develop and use new production technologies which can create products with better quality and higher yield. Additionally, new production technologies are able to assist farmers to reduce production costs and improve their production skills. Lastly, market access of farmers may be encouraged through participating in contract farming. For example, in the contract of rice production signed between agribusinesses and farmers in An Giang province, Vietnam, enterprises are in charge of supplying inputs such as seed, fertilizer, and pesticide to contracted farmers, supervising farming techniques, and purchasing

paddy rice from farmers with fixed price or market price mechanism. After processing and packaging, rice products will be exported to Japan (Nhan et al., 2013). In the contract of Kernel peanut production signed between agribusinesses and farmers in

Nghe An province, Vietnam, farmers can access inputs and equipment such as seed, fertilizer, plastic row cover and plough machines provided by contracting firms.

Transaction costs such as screening, monitoring and managing the contract are covered by contracting enterprises and contracting firms commit to purchase peanut of farmers based on the market price. As a result, the peanut yield of contracted households is higher than that of their counterparts by 20 percent (Tuan, 2012).

Disadvantages

Apart from the advantages mentioned above, farmers face challenges in contracts.

One issue is the dependence of producers on the specific terms of contract farming. Farmers must accept and meet terms relating to quantity, quality, prices and delivery time in a contract. Dependence on conditions specified by contractors can push farmers to lose flexibility in their production. It may be very difficult for farmers to adjust production along with uncertainties or variations of the market because of fixed terms in a contract. When participating in a contract, producers must invest more of their resources (land, labour, capital) to produce certain products based on the contract and, as a result, this may reduce their opportunities to produce other products.

Moreover, production technologies specified by contractors can be irrelevant to the production habits of farmers. Some farmers prefer traditional cultivation methods rather than modern ones. Therefore, contract farming can be an issue for farmers since it

creates dependence by producers on the terms given by contractors in a contract. For example, in the contract of seed production in India, contracting firms are protected from all and even unforeseen obligations by the contractual agreements, while the farmer must meet the contract obligations under all circumstances. There is no compensation to farmers even in the case of crop failure due to natural disasters (Singh,

2002).

Another challenge for farmers is the transparency in a contract. Terms regarding quantity, quality, technologies, prices and delivery time in a contract are often prepared and set up by contractors and they have an advantage when signing contracts with farmers. In some cases, producers have no or little clear understanding of terms and articles given in a contract because of limitations in their knowledge and the absence of explanations from contracting firms. Due to small-scale production of farm households, land concentration and labour exploitation have become motivations for contractors to use contracts. Ambiguity is also revealed in pricing methods. Some firms propose complex formulas for calculating prices and this leads to inadequate understanding by producers. Prices are pre-established in a contract and, in some cases, it is very difficult to adjust according to market prices at procurement time. For instance, in the contract of seed production in India, multinational companies use English, while local farmers are familiar with local language and consequently, they are unable to understand all terms mentioned in the agreement (Singh, 2002).

Debt is also an issue for farmers in contract farming. This problem occurs when producers receive pre-investment by contractors through providing seeds, fertilisers, pesticides, credit and so on. This pre-investment can be an advantage for farmers because they have opportunities to obtain adequate volume, great quality and reasonable

prices of these inputs and repay costs after harvesting. However, due to the negative effects of natural disasters and uncertainties of the market, producers may get into difficulties because of low productivity and be unable to repay their debts. Furthermore, some households may use credit inefficiently or for the wrong purposes, losing their repayment ability. Some households use credit capital to purchase facilities or for consumption purposes instead of investing in their production. Hence, debt is also a potential problem for farmers in contract farming. For example, in the contract of sugarcane production in Thailand, some farmers were worried about the debts because a household had an average debt of Baht101,250. The factory supports the farmers by loaning input production, such as providing seeds, fertilizers, and chemicals. The expense will be deducted from the products that the farmers sell to the factory

(Pouncgchompu et al., 2016).

Lastly, disruptions of socio-cultural structures in families and communities in rural areas can be a challenge for farmers. Clearly, contract farming requires more specialisation in agricultural production. However, this may become a constraint for farmers because of the effects of traditional production habits and simple labour divisions in farm households in rural areas. An agricultural labourer can perform several operations in their production, such as land preparation, fertiliser and pesticide applications, and harvesting, so it is very difficult to divide a certain job for a member in a family. In some rural areas, gender discrimination is a constraint, which reduces the possibility of women deciding to participate in contract farming. For example, in the contract of tea production in Kericho district, Kenya, about one third of all tea plots were partly or completely neglected largely because of conflicts between spouses. Thus,

the problem of low productivity in smallholder tea production is defined as the result of the prevailing gender relations in a local community (Bülow and Sorensen, 1993).

2.1.7 Model for theoretical framework of the study

The model for the theoretical framework of the study is presented in Figure 2.2. The purpose is to capture both the macro and micro determinants of contract farming as explained in the proceeding paragraphs.

Figure 2.2. Model for theoretical framework of the research

Geographical Business Socio-economic Legal conditions environment conditions regulations

Advantages for Contractual agreements Risks for famers

farmers - Production risks Firm Farmer - Market access - Forms - Market risks characteristics characteristics - Technical - Terms - Financial risks

assistance - Volume of products - Cultural and

Product - Input - Quality of products social risks Crop - Prices attributes provisions - Contract breach characteristics - Expected - Product delivery by firms higher income - Dispute resolution

Tea productivity Farmer income

Poverty of tea farmers in Phu Tho province

Tea exports Economic growth Poverty National level Source: Author, 2017 33

Contracts are established through negotiations between firms (contractors) and farmers (producers) in tea production. Negotiations on forms, terms, volume, quality, price, delivery and dispute resolution are specified in the contract. Macro determinants affecting contractual agreements such as location, socio-economics, legislation and business environment are analysed in the research.

Moreover, the study also discusses micro factors, such as firm characteristics, product attributes, farmer characteristics and crop characteristics, which affect the contract. For example, in Vietnam, most tea is harvested by hand and only 8 percent of tea areas are harvested by machine. Due to degradation of tea leaves 4–6 hours after harvesting, harvested material needs to be delivered quickly to processing plants.

Vietnamese farmers are not interested in improving harvesting techniques, and consequently, product quality decreases, while labour costs increase.

There are various types of businesses operating in producing and processing tea in Vietnam, including state-owned enterprises, joint ventures, foreign enterprises and private enterprises. In 2014, there were about 500 units operating in producing and processing tea with total capacity of more than 500,000 tonnes of dry tea a year.

However, only 30 percent of enterprises operate at maximum capacity in the season. A range of tea products are processed and sold to the market such as green tea, black tea and Oolong tea (VBCSD, 2015).

The research analyses the advantages and risks to farmers of contracts. The next chapters estimate factors affecting tea productivity and technical efficiency of tea production and assess the influence of contract farming on tea productivity, farmer income and determinants on poverty of tea farmers in Phu Tho province. Lastly, the

relationship between economic growth, tea exports, and poverty in Vietnam are discussed (Figure 2.2).

2.2 Empirical review

2.2.1 Effects of contract farming on productivity

There are mixed results on the effects of contract farming on productivity. Some previous studies found contract farming has a positive impact on agricultural productivity.

A research by Paul et al. (2004) investigated productivity, economics and efficiency in US agriculture by using a stochastic production frontier. Results showed that although contracts have no strong impacts, lower contract levels are less efficient than contract-intensive entities. Moreover, the value of marketing and production contracts for crop and livestock species and on manure nitrogen production are two dominant contributors to higher productivity, scale economies and technical efficiency.

Tripathi et al. (2005) assessed impacts of contract farming on potato yield in

Haryana, India. They found that the potato yield of contracted households is higher than that of their counterparts by 8.84 percent because non-contracted farmers have higher yield and price uncertainty than contracted farmers. Likewise, Swain (2013) estimated effects of contract farming on farm productivity and efficiency of hybrid paddy seed production in Andhra Pradesh, India by using the Heckman sample selection model. He found that contract farmers are efficient in producing contract crops, while independent

farm households tended to gain more efficiency in planting non-contract crops. Small farmers obtained a higher level of efficiency compared to large farmers.

Effects of contract farming on productivity are also carried out in African countries. A study by Igweoscar (2014) investigated effects of contract farming on cassava productivity in Southeastern Nigeria. Results showed that the cassava productivity of contracted households is higher than that of non-contracted households because education, farm size, and farming experience have been found higher and labour cost has been found lower for contracted farmers than their counterparts.

Likewise, Ajao and Oydele (2013) estimated economic efficiency of contract farming in the tobacco sector in Oyo state, Nigeria by using data envelopment analysis. Results demonstrated that contract farming can create opportunities for farmers to obtain maximum outputs by improving efficiency of certain resources and farmers’ education and household size are two influential factors contributing to efficiency of available resources. A research by Obare and Kariuki (2003) assessed effects of informal contract farming on French bean productivity in Central Kenya. They found that the productivity of French bean of contracted households is higher than that of their counterparts because of positive and significant effects of fertilizers, pesticides, and high-yielding variety seed use. Likewise, Kalimangasi et al. (2014) investigated the contribution of contracts of cocoa production to the livelihood of small farmers in Tanzania using

Cobb-Douglas production functions. They found that participation of farmers in contract farming can lead to an increase in production quantity. Mpeta et al. (2018) analysed effects of contract farming on technical efficiency and productivity of sunflower production in Kongwa district, Tanzania. They found that the average technical efficiency of sunflower production of contracted farmers is higher than that of

non-contracted farmers by 4.5–7.4 percent and consequently, productivity of contracted farmers is higher than that of their counterparts by 24 percent. These are result of the use of high-quality seeds and service provision from contract firms. Parirenyatwa and

Mago (2014) assessed the development of contract farming in Zimbabwe. Result showed that contract farming has a potential to boost productivity of the farming sector and empower the emerging black farmers in this country because contracted households can borrow seeds, fertilizers, and pesticides from agribusinesses.

Other studies also found a positive relationship between contract farming and agricultural productivity. Begum et al. (2012) assessed the farm efficiency of the poultry sector in Bangladesh by using data envelopment analysis. Results indicated that contract farming is a feasible alternative to foster efficiency at the farm level because of technological transfer from integrators to producers. In addition, an excess amount of inputs used on poultry farms in a rural area of Bangladesh can lead to inefficient production. Similarly, Mishra et al. (2018) estimated the impact of contract farming on the yield of lentil in Nepal. Results showed that the lentil yield of contracted farmers is higher than that of independent ones by 80 kilogram per hectare due to risk management strategies, land ownership and lower transportation costs.

Moreover, Saenger et al. (2014) assessed effects of contract farming on input use and output levels of dairy farmers in Vietnam. They found that the provision of third-party contract enforcement has a positive impact on input use and output levels because it reduces opportunistic behaviour and asymmetric information.

While the majority of the studies cited found a positive relationship between contract farming and productivity, there is other research that question this positive relationship. Miyata et al. (2009) examined the influence of contract farming on the

income of farmers in production of apples and green onions in China by using an instrumental variable method and found mixed results. Specifically, contracted apple producers gained a higher yield compared to non-contracted households due to technical assistance and specialised inputs provided by apple packers. In contrast, the yield of contracted green onion growers was lower than that of independent producers. Mishra et al. (2018) evaluated influences of contract farming on the yield of organic basmati rice in India. They found that although contract participation generates a lower yield of organic basmati rice, but it enhances prices received for contracted farmers. Ngoc et al.

(2014) investigated technical efficiencies of tea production in Thai Nguyen and Phu Tho provinces, Vietnam by using stochastic frontier production functions. They concluded there is no difference in statistical significance of technical coefficients in both contracted and non-contracted farmers.

In general, the majority of studies indicated that contract farming has a positive impact on productivity of agricultural products, except green onion in China (Miyata et al., 2009) and tea in Thai Nguyen and Phu Tho provinces, Vietnam (Ngoc et al., 2014).

The main reasons for the higher productivity of contracted farmers include opportunities to access better inputs such as seeds, fertilisers, pesticides, feed, technical assistance, and credit and preferences in purchasing products at reasonable prices.

In Vietnam, the effect of contract farming on productivity is still a controversial matter. For example, in the tea sector, one study claimed that contract farming can increase productivity of tea farmers (Saigenji and Zeller, 2009), while another study concluded there is no difference in productivity of contracted and independent tea farmers (Ngoc et al., 2014). Mixed results can be explained by differences in research duration and sites. For instance, Saigenji and Zeller (2009) examined influences of

contract farming on tea productivity in Moc Chau district, Son La province, Vietnam from June to October 2007, while Ngoc et al. (2014) accessed effects of contract farming on tea productivity in Thai Nguyen and Phu Tho provinces in 2013. Saigenji and Zeller (2009) found that the member of the Communist Party is a key factor determining contract participation. However, Ngoc et al. (2014) indicated that access to credit, market, and extension technology is a major motivation of farmers in participating in a contract in Thai Nguyen province, and provision of technology and information is the most important reason for farmers participating in a contract in Phu

Tho province. Thus, more studies are needed on the influences of contract farming on tea productivity in Vietnam.

2.2.2 Effects of contract farming on income

Some previous studies found a positive relationship between contract farming and farmers’ income. Ton et al. (2018) examined impacts of contract farming on income of smallholders in 13 developing countries. They found that smallholders can benefit from contract participation because 61 percent of contracted farmers have larger land size or more assets than their counterparts in the region, but the poorest farmers are rarely included. Similarly, Warning and Key (2002) examined social performance and distributional consequences of contract farming of peanut production in Senegal by using probit models. Results showed that the Arachide de Bouche program has a positive impact on income of contracted farmers because it links farmers to export markets. Simmons et al. (2005) analysed the emergence and benefits of contract farming of seed corn production in East Java, seed rice in Bali, and broiler chickens in

Lombok, Indonesia by using probit models. They found that contract farming can increase returns to capital and create better off participants in seed corn and broiler production. In contrast, contracts did not improve returns to capital of seed rice producers, but it assisted them to secure market access. Likewise, Dhillon and Singh

(2006) assessed challenges and opportunities of contract farming in the agricultural sector in Punjab, India. Their results demonstrated that contract farming may improve income of contracted households. However, the expansion of contracts should be considered carefully because of problems faced by farmers in contract farming such as exploitation of the labour force by contractors.

Other studies also found that contract farming has a positive impact on farmers’ income. Maertens and Velde (2017) investigated impacts of contract farming on rice production in Benin. Results indicated that income of contracted households is higher than that of non-contracted households due to positive effects of the rice area, intensification of rice production, increased commercialisation of rice, and higher farm- gate prices. Dedehouanou et al. (2013) analysed the effect of contract farming on welfare of mango and bean growers in the Niayes region of Senegal. They concluded that contract participation has a positive impact on farmer income. However, bean contracts contribute less to farmers’ happiness, while mango contracts have a strong positive effect on happiness of farmers. Similarly, Bolwig et al. (2009) assessed the effects of contract participation on income of coffee growers in tropical Africa by using ordinary least squares and full information maximum likelihood regressions. Results showed that contract participation leads to a 75 percent increase in net revenue of coffee producers, which is equivalent to 12.5 percent of average total household income.

There are several reasons identified in the literature to explain why contract farming can have a positive effect on farmer incomes. Girma and Gardebroek (2014) examined effects of contract farming on income of organic honey producers in southwestern Ethiopia. They found that household income of contracted beekeepers is

US$404–411 annually higher than that of their counterparts because they are able to improve quality of honey delivery and obtain higher prices. Similarly, Mulatu et al.,

(2017) assessed impacts of contract farming of vegetable production on household income in the East Shewa zone, Ethiopia. Results addressed that contracted farmers have a higher income than their counterparts because credit, the membership, access to market information and livestock ownership significantly increase the probability of farmers’ participation in contract farming. Bellemare (2012) estimated welfare impacts of contract farming on income of farm households in cotton, green beans, snow peas, leeks, barley, rice, tomatoes, maize and oats production in Madagascar. Results demonstrated that a one percent increase in households participating in contracts generates a 0.5 percent increase in household income.

Some studies also found that there is a positive relationship between contract participation and farmer’s income in Asia. Miyata et al. (2009) evaluated the influence of contract participation on farmer income in Shandong province, China. They found that contracted households earn more than their counterparts because of positive effects of household labour, education, farm size, share of land irrigated, and proximity to the village leader. Cahyadi and Waibel (2016) examined the relationship between contract farming and poverty of oil palm smallholders in Indonesia. They found that contract participation has a positive effect on income of smallholders because it can reduce the negative impact of price shocks, but contract smallholders also remain vulnerable to

poverty. Trifkovic (2014) examined contract farming in the catfish sector of Vietnam.

Results showed that contract participation has a positive impact on welfare of farmers after controlling for both observable and unobservable farmer characteristics, but the effect of contracts on farm employment is not significant. Nhan et al. (2013) assessed the effects of contract farming on income of rice producers in An Giang province,

Vietnam. Results indicated that contracted farmers had a higher net return compared to independent farmers due to adequate inputs, technical assistance, and product procurement provided by contractors. Sokchea and Culas (2015) investigated the effects of contract farming with farmer organisations on farmers’ income in Cambodia. Results showed that contract farming with the Reasmey Stung Sen Agricultural Development

Cooperative assists farmers to gain high returns through improving product quality and targeting market niches.

While the majority of the studies cited found a positive effect of contract farming on farmer income, other research argues that contract farming either has no effects or negatively impacts farmer income (Ze-ying et al., 2018; Narayanan, 2014;

Tongchure and Hoang, 2013; Wang et al., 2014; and Ragasa et al., 2018). Ze-ying et al.

(2018) evaluated influences of contract farming on welfare of broiler producers in

China. Results showed that contract farming may not generate a higher welfare for small broiler producers who focus on seeking alternative market opportunities rather than their comparative advantages. Narayanan (2014) investigated effects of contract farming of four agricultural commodities on farmer income in Southern India. He found that income of contracted farmers is lower than that of their counterparts for marigold production because these households heavily depend on hired labours. Tongchure and

Hoang (2013) assessed participation of small farmers in contract farming of cassava

production in Nakhon Ratchasima, Thailand. They did not find improvement of farmer income. However, results showed that contract farming may reduce production costs, increase market efficiency, provide lower interest rates, and create symmetric information for cassava growers. Similarly, Wang et al. (2014) examined the economic impact of direct marketing and contracts in safe vegetables in northern Vietnam. Results showed that contracts have less influence on the income of safe vegetable producers compared to direct sales. Lastly, Ragasa et al. (2018) investigated limitations of contract farming of maize production in Upper West Ghana. Their results showed that contract farming negatively affects profit of maize growers since an increase in maize yield is not enough to compensate for input expenditure and capital costs.

Contract farming in developed and developing countries has been researched by scholars (Minot and Sawyer, 2016; Otsuka et al., 2016; and Wang et al., 2014).

Although contract farming has played an important role in terms of boosting farmers’ income by introducing new crops and production methods, there is room for strengthening its effects on poverty reduction (Otsuka et al., 2016). Third-party certification is necessary to deal with one of the most common issues in contract farming is side-selling when farmers sell their products to other buyers to avoid repaying loans or to obtain a higher price. Further, another problem of contract farming is that it is only suitable for high-value commodities which sold to large-scale buyers for quality-sensitive markets. However, in most of developing countries, the rate of farmers involved in contract farming varies from 1 to 5 percent (Minot and Sawyer, 2016).

Studies on contract participation in most developing countries find that neither the number of laborers employed on the farm nor farm assets contribute effectively to

contracting, but that government policies are affordable drivers contributing to the achievement of contract farming (Wang et al., 2014).

In Vietnam, studies on effects of contract farming on income have been conducted on different products such as tea (Saigenji and Zeller, 2009), rice (Nhan et al., 2013) and vegetables (Wang et al., 2014). These studies concluded that contract farming is able to generate a higher income for producers, to different degrees.

Therefore, it is important for more research to provide evidence on the impacts of contract farming in tea production on farmer income in Vietnam.

2.3 Chapter conclusions

This chapter presented the theoretical framework of the study and empirical review of previous research. While the theoretical framework discussed definitions of contracts and contract farming, the history of contract farming, contract theory, types of contracts, motivation for contracting, and advantages and challenges of farmers in contracts, the empirical review summarised outcomes of previous studies on effects of contract farming on agricultural productivity and farmers’ income. There are mixed results on the impacts of contract participation on productivity and farmer income. For instance, some studies found that contract farming has a positive influence on productivity and farmer income, while others either did not find a positive relationship between contract participation and productivity and income or found contract farming negatively affects farmers’ income. We can see here after the review that it is worthwhile to ask if contract farming does really benefit poor farmers, given the literature of contract benefits is not

100 percent conclusive – even though the majority of studies suggest it is a good thing for farmers.

CHAPTER 3: TEA PRODUCTION IN A GLOBAL AND VIETNAMESE

CONTEXT

3.0 Introduction

This chapter firstly provides an overview of global production, consumption and trade of tea. It then sets the context for tea production in Vietnam. Relevant government policies and regulations are also discussed. This chapter is important because it specifies the position of the tea sector of Vietnam in the global market.

3.1 Tea production, consumption and trade in a global context

3.1.1 Tea production

Tea area, production and yield of major producers are shown in Table 3.1.

Table 3.1. Tea area, production and yield of major producers

Country Area (1,000 ha) Production Yield (kg/ha) (1,000 tonnes) 1990 2000 2014 1990 2000 2014 1990 2000 2014 Argentina 38 38 41 53 68 82 1382 1789 2008 Bangladesh 49 49 58 46 53 64 942 1085 1093 Burundi 9 8 10 4 7 9 469 890 900 China (Mainland) 1062 1089 2741 540 683 1950 509 627 711 India 417 504 567 721 850 1211 1730 1686 2137 Indonesia 135 154 123 161 163 145 1189 1056 1177 Iran 35 35 32 46 47 27 1314 1350 828 Japan 50 50 45 90 89 81 1798 1786 1806 Kenya 97 126 203 197 237 449 2031 1883 2210 Malawi 19 19 19 40 42 46 2079 2217 2413 Sri Lanka 222 189 204 234 309 340 1055 1635 1666 Thailand 10 6 22 5 6 15 526 908 698 Turkey 79 77 77 123 139 246 1554 1802 3195 Uganda 21 21 38 7 29 65 319 1394 1720 Vietnam 75 80 128 32 72 180 429 896 1406 Source: FAO, 2016 Note: ha means hectare 46

In 2014, China had the largest planted area of tea with 2.7 million hectares, following by India (567,000 hectares), Sri Lanka (204,000 hectares) and Kenya

(203,000 hectares). Burundi had the smallest area with 10,000 hectares. From 1990 to

2014, the tea planted area of China increased by about 1.4 million hectares, India by nearly 150,000 hectares, and Kenya by more than 100,000 hectares. Vietnam expanded its tea area by more than 1.7 times or 128,000 hectares from 1990 to 2014 (Table 3.1).

The expansion of tea planted area generated an increase in the production volume of countries. By 2014, tea production of Sri Lanka increased rapidly by more than 68 percent compared to this in 1990, followed by India (more than 59 percent),

Kenya (more than 43 percent), China (more than 27 percent), and Vietnam (more than

17 percent). By contrast, tea production of Iran decreased sharply by 24 percent from

1990 to 2014, while Indonesia and Japan declined about 6 percent in the same period

(Table 3.1).

Tea yield also varied by country. From 1990 to 2014, India had the strongest growth of tea yield by nearly 81 percent, followed by China (more than 71 percent),

Argentina (more than 68 percent), Turkey (more than 48 percent), and Vietnam (more than 30 percent). An increase of tea yield in most countries in the past three decades resulted from the introduction of new clones and adoption of good agricultural practices

(FAO, 2016).

The value of tea production, proportion of agricultural value added in gross domestic product (GDP) and agricultural employment in the total labour force of the main producers are shown in Table 3.2.

Table 3.2. Value of tea production, percentage of agricultural value added and agricultural employment in total labour force of main producers

Country Value of tea production Percentage of Percentage of (million US$) agricultural value agricultural added in GDP (%) employment in total labour force (%) 1990 2000 2014 1990 2000 2014 1990 2000 2014 Argentina 41 52 123 8.1 5.1 8.3 0.4 0.7 0.5 Bangladesh 93 95 169 32.8 23.8 16.1 na 62.1 na Burundi 8 13 24 55.9 48.1 39.3 na na na China (Mainland) 728 835 10118 26.7 14.7 9.2 53.4 46.3 na India 1454 1526 3203 29.0 23.0 17.8 na 59.9 na Indonesia 262 173 293 19.4 15.6 13.4 55.9 45.3 34.3 Iran 24 27 26 12.8 9.1 9.3 na na 17.9 Japan 1767 1996 1263 2.1 1.6 na 7.2 5.1 na Kenya 397 426 1187 29.5 32.4 30.3 na na na Malawi 80 76 121 45.0 39.5 33.3 na na na Sri Lanka 472 555 899 26.3 19.9 8.3 47.8 na 30.4 Thailand 7 8 78 12.5 8.5 10.5 63.3 48.5 na Turkey 192 126 1084 18.1 11.3 8.0 46.9 36.0 19.7 Uganda 14 53 173 56.6 29.4 27.2 na na na Vietnam 49 90 316 38.7 22.7 18.1 na 65.3 na Source: FAO, 2016 Note: na denotes not available

In 2014, the value of tea production by China was US$10.1 billion, followed by

India (US$3.2 billion), Kenya (US$1.2 billion), Turkey (US$1.08 billion), Sri Lanka

(US$899 million), Vietnam (US$316 million), Indonesia (US$293 million) and

Argentina (US$123 million). Both percentages of agricultural value added in GDP and agricultural employment in the total labour force tended to decrease in most countries between 1990 and 2014 (Table 3.2).

According to estimates by the Food and Agriculture Organization of the United

Nations (FAO), over the 10 years 2013 to 2023, world black tea production will grow slightly by 2.9 percent annually and reach 4.17 million tonnes in 2023. World green tea production will increase sharply at 8.2 percent annually because the production volume of tea in China is expected to reach 2.97 million tonnes in 2023 (Chang, 2015).

Table 3.3. Producer price of tea in selected countries in 2015 and 2016 (US$/tonne)

Country 2015 2016 Iran (Islamic Republic of) 511.5 602.7 Japan 12,119.6 13,907.2 Kenya 3,020.6 2,436.4 Nepal 446.5 449.9 Sri Lanka 441.2 478.2 Turkey 580.9 586.1 Vietnam 343.4 372.6 Source: Compiled from FAOSTAT, 2018a

As seen in table 3.3, the producer price of Japanese tea was higher than that of the rest countries for two years (2015_2016). By 2016, the producer price of tea in Japan rose sharply by nearly US$1,800 per tonne, while the producer price of tea in Vietnam increased by about US$29 per tonne and the tea price of Turkey had a slight increase.

Higher prices reflect advances in processing technologies and quality of Japanese tea in the international market. By contrast, at the same time, the producer price of tea tended to decline in other producer countries, particularly in Kenya by nearly US$585 per tonne and in Iran by US$91 per tonne (Table 3.3).

3.1.2 Tea consumption

Tea is one of the most popular and lowest cost beverages after water. There are many varieties of tea, but black tea is the most consumed all over the world. The demand for tea is growing because of its benefits for health such as weight loss functions and antioxidants (UNCTAD, 2016).

There are two directions for tea demand in the international market. In tea- dominant markets, demand for bagged black tea is stable, especially in emerging

countries. In non-tea markets, consumers drink tea since they want to avoid the negative effects of coffee or caffeine (UNCTAD, 2016).

From 2009 to 2013, global tea consumption rose by about 24 percent to reach

4.8 million tonnes in 2013. This development is caused by a sharp growth in per capita income in China, India and other emerging economies. In addition, an increase in the number of consumers entering the middle class led to a greater demand for tea consumption.

Figure 3.1. Annual per capita tea consumption in selected countries in 2016

8 6.96 7

6 4.83 5 4.28 4

pounds 3.05 3 2.68 2.63 2.23 2.2 2.13 1.98 2

0 Turkey Ireland United Russia Morocco New Egypt Poland Japan Saudi Kingdom Zealand Arabia Source: Compiled from STATISTA, 2018

Of the leading tea consumption countries of the world, Turkey has the tea highest consumption per capita of nearly 7 pounds a year, followed by Ireland (4.83 pounds), the United Kingdom (4.28 pounds), and Saudi Arabia stands at the end of the list by nearly 2 pounds (Figure 3.1).

3.1.3 Tea trade

Table 3.4. Tea exports and imports of the world

Item 2009 2010 2011 2012 2013 Export quantity (tonne) 1,822,227 2,022,762 1,983,292 1,805,977 2,051,373 Export value (thousand US$) 5,428,305 6,401,671 6,610,275 6,316,770 7,576,418 Import quantity (tonne) 1,597,878 1,719,606 1,899,444 1,934,971 1,893,353 Import value (thousand US$) 4,997,601 5,686,665 6,628,782 6,800,463 7,139,831 Source: Compiled from FAOSTAT, 2018b

The exported volume of tea was higher than the imported volume by 200–300,000 tonnes from 2009 to 2013, except in 2012. In 2013, both exported and imported values of tea increased by US$2.1 billion compared to values in 2009 (Table 3.4).

The value of world trade in tea reached US$5.61 billion in 2014 with 1.73 million tonnes of exported volume. In 2014, Sri Lanka was the leading tea exporter with value of tea exported of US$1,609 million, followed by Kenya (US$1,273 million),

China (US$1,272 million), India (US$656 million), Vietnam (US$229 million),

Indonesia (US$135 million), Argentina (US$115 million), Uganda and Malawi (US$86 million each) and Japan (US$75 million). The value of tea exported in other producer countries in 2014 was low, such as Turkey (US$20 million), Iran (US$17 million),

Burundi (US$14 million), Thailand (US$9 million), and Bangladesh (US$4 million)

(FAO, 2016).

Variations in tea prices lead to different revenues from exporting tea of producer countries. For example, Kenya was the largest tea producer with production of 414,000 tonnes in 2014, but Kenya’s value of tea exported is only ranked third because the average price of tea exported was US$2.36 per kg, while Sri Lanka exported tea at

US$4.90 per kg and China exported at US$4.22 per kg (FAO, 2016).

The volume of global tea imports was 640,000 tonnes in 2014, with key importers including the Russian Federation, the United States of America, the United

Kingdom, Egypt and Pakistan (FAO, 2016). 51

3.2 Tea production, consumption and exports of Vietnam

3.2.1 Tea production

The tea tree is mainly planted in three areas of Vietnam: the Northern midlands and mountainous area, North Central and Central coastal area, and Central Highlands

(VBCSD, 2015).

Table 3.5. Planted area of tea in Vietnam (thousand hectares)

Annual Country/Regions 2011 2012 2013 2014 growth rate (%) Red River Delta 4.7 4.8 4.6 4.7 0.03 Northern midlands and mountainous area 90.0 90.3 93.4 95.7 2.07 North Central and Central coastal area 8.3 8.9 8.8 8.8 1.97 Central Highlands 24.5 24.2 23.0 22.9 -2.20 Total Vietnam 127.5 128.2 129.8 132.1 1.19 Source: General Statistics Office of Viet Nam, 2015 and author’s calculation, 2016

The planted area of tea in the whole country and regions increased slightly over the period 2011_2014, except Central Highlands. In Vietnam, by 2014, a majority of tea was planted in Northern midlands and mountainous area (95,700 hectares) and Central

Highlands (22,900 hectares) because of favourable soil and adequate climate conditions in these two regions (Table 3.5).

Table 3.6. Harvested area of tea in Vietnam (thousand hectares)

Annual Country/Regions 2011 2012 2013 2014 growth rate (%) Red River Delta 4.5 4.6 4.4 4.6 0.34 Northern midlands and mountainous area 79.3 80.0 81.8 82.9 1.48 North Central and Central coastal area 6.8 6.9 7.1 7.1 1.66 Central Highlands 23.7 23.0 21.5 20.8 -4.21 Total Vietnam 114.3 114.5 114.8 115.4 0.32 Source: General Statistics Office of Viet Nam, 2015 and author’s calculation, 2016

Similar to the trend of planted area of tea, the harvested area of tea in the whole country and regions increased slightly in the period 2011 to 2014. For instance, the harvested area of tea in the whole country increased by only 0.32 percent a year, North

Central and Central coastal area by 1.66 percent, Northern midlands and mountainous area by 1.48 percent and the Red River Delta by 0.34 percent, while the harvested area of tea in Central Highlands fell over 4 percent a year. Due to the large planted area of tea in Northern midlands and mountainous area and Central Highlands, these regions also dominated in terms of harvested area of tea in 2014 with 82,900 hectares and

20,800 hectares, respectively. By contrast, in 2014 the harvested area of tea was small in

North Central and Central coast area (7,100 hectares) and the Red River Delta (4,600 hectares) (Table 3.6).

Table 3.7. Tea production in Vietnam (thousand tonnes)

Annual Country/Regions 2011 2012 2013 2014 growth rate (%) Red River Delta 30.3 30.3 30.0 31.9 1.72 Northern midlands and mountainous area 576.1 595.3 617.5 637.8 3.45 North Central and Central coastal area 64.0 68.3 70.6 70.8 3.43 Central Highlands 208.6 215.9 218.2 222.0 2.10 Total Vietnam 879.0 909.8 936.3 962.5 3.07 Source: General Statistics Office of Viet Nam, 2015 and author’s calculation, 2016

Tea production in the whole country and regions had a gradual increase in the period 2011 to 2014. For example, average growth of tea production in the whole country increased by more than 3 percent a year, while average growth in Northern midlands and mountainous area rose by 3.45 percent, following by North Central and

Central coast area (3.43 percent), Central Highlands (2.1 percent) and Red River Delta

(1.72 percent). In 2014, Northern midlands and mountainous area was the dominant region in terms of tea production with 637,800 tonnes, followed by Central Highlands

(222,000 tonnes), North Central and Central coast area (70,800 tonnes) and the Red

River Delta (31,900 tonnes) (Table 3.7).

Table 3.8. Yield of fresh tea in Vietnam (100kg/hectare)

Annual Country/Regions 2011 2012 2013 2014 growth rate (%) Red River Delta 66.7 65.9 68.2 69.3 1.30 Northern midlands and mountainous area 72.6 74.4 75.5 76.9 1.94 North Central and Central coastal area 94.7 99.0 99.4 99.7 1.74 Central Highlands 88.1 93.9 101.5 106.7 6.58 Total Vietnam 76.9 79.5 81.6 83.4 2.74 Source: General Statistics Office of Viet Nam, 2015 and author’s calculation, 2016

Tea yield had some variation in the period 2011 to 2014. For example, average growth of tea yield in the whole country increased by 2.74 percent a year, while Central

Highlands was the leading region in average growth of tea yield at 6.58 percent a year and average growth of tea yield in the other regions was below 2 percent a year. The average growth of tea yield in Phu Tho province was 4.85 percent in total over four years (2011–2014). In 2014, tea yield of the whole country was 83,400 kg/hectare, while Central Highlands had the highest yield at 106,700 kg/hectare, followed by North

Central and Central coast area (99,700 kg/hectare), Northern midlands and mountainous area (76,900 kg/hectare) and the Red River Delta (69,300 kg/hectare). Central

Highlands was the leading region in terms of tea yield because of advantages in soil, climate and advanced techniques (new varieties, fertiliser and chemical applications, irrigation, harvested methods) in producing tea in this region (Table 3.8).

Some volatility of tea yield in the whole country and regions can be explained by harvested area and production of tea. During the period 2011–2014, the harvested area of tea remained stable due to a limitation of land area and tea production increased

by a small proportion, which combined led to a small increase in tea yield. In terms of tea cultivation, tea yield can be improved by adopting new varieties, fertiliser and pesticide applications, and harvesting techniques rather than implementing intensification.

3.2.2 Tea consumption

Tea consumption in the domestic market of Vietnam of 90 million people is below

30,000 tonnes a year (or 0.3 kg per capita on average). This is much lower than in other countries like China (1 kg), Japan (2 kg), the Middle East (2 kg), Russia and the United

Kingdom (2.5 kg). By 2014, tea export price is lower than the price in the domestic market. For example, the current export price is US$1.6 per kg, while the domestic price is VND110,000–220,000 per kg (US$5–10 per kg) (VBCSD, 2015). Hence, domestic enterprises in the tea sector may not achieve high profitability if they only focus on the exported market, and ignore the domestic market.

3.2.3 Tea exports

In 2014, the exported volume of Vietnamese tea was 130,000 tonnes with a value of

US$230 million. Vietnam is ranked fifth of exporting tea countries in the world, behind

China, India, Kenya and Sri Lanka (VBCSD, 2015). The quantity of tea exported by

Vietnam in the period 2012–2014 is shown in Figure 3.2.

Figure 3.2. Quantity of tea exported by Vietnam to major markets (2012–2014)

4475 EU 6622 7070

9871 The USA 9909 8170

11450 Russia 11748 13896 2014 12897 2013 China 14011 14632 2012 23255 Taiwan 22477 22453

35039 Pakistan 22090 24045

0 5000 10000 15000 20000 25000 30000 35000 40000 tonnes Source: General Statistics Office of Viet Nam, 2015

Pakistan was the biggest importer of tea from Vietnam for three years (2012–

2014) and in 2014 Vietnam exported more than 35,000 tonnes of tea to Pakistan. The second biggest importer of Vietnamese tea was Taiwan which imported more than

23,000 tonnes of tea from Vietnam in 2014, following by China, Russia, the United

States of America, and EU. For two years (2013 and 2014), the volume of tea exported from Vietnam to Pakistan and Taiwan increased, but the volume of tea exported to

China, Russia, the United States of America and EU decreased. According to tea experts, the volume of exported Vietnamese tea declined in these markets due to strong competition from similar products produced by African countries (Figure 3.2).

The value of tea exported by Vietnam in the period 2012–2014 is shown in

Figure 3.3.

Figure 3.3. Value of tea exported by Vietnam to major markets (2012–2014)

6800 EU 9325 9246

10682 The USA 10871 8304

17381 Russia 17825 20013 2014

16003 2013 China 17853 2012 17877

29311 Taiwan 28626 27397

75187 Pakistan 42545 41948

0 10000 20000 30000 40000 50000 60000 70000 80000 thousand EUR Source: General Statistics Office of Viet Nam, 2015 Note: EUR means Euro (currency unit)

Similar to the volume of tea exported, Pakistan was the leading market for

Vietnamese tea with a value of more than EUR75 million in 2014, followed by Taiwan

(more than EUR29.3 million), Russia (more than EUR17.3 million), China (more than

EUR16 million), the United States of America (more than EUR10.6 million) and EU

(EUR6.8 million). The value of tea exported from Vietnam to Pakistan and Taiwan increased for three years (2012–2014), however other markets (China, Russia, the

United States of America and EU) had a declining trend, explained by a decrease in the volume of tea exported to these markets, where it is very difficult for Vietnamese tea to gain higher prices (Figure 3.3).

3.3 Tea production in Phu Tho province

3.3.1 Planted area, yield and productivity of tea

Phu Tho province is ranked third in tea productivity and fourth in planted area of tea in

Vietnam. Planted area and productivity of leaf tea in Phu Tho province are shown in

Table 3.9.

Table 3.9. Planted area and productivity of leaf tea in Phu Tho province

Annual growth Item Unit 2013 2014 2015 2016 rate (%) Total planted area ha 16,080.1 16,301.6 16,584.0 16,761.5 1.39 Harvested area ha 14,483.8 14,749.2 15,314.6 15,661.9 2.64 Productivity tonnes 136,195.2 152,219.5 154,753.3 162,388.0 6.03 Source: Phu Tho Statistics Office, 2017 and author’s calculation, 2017 Note: ha means hectare; na means not available

Over four years (2013–2016), total planted area of tea in Phu Tho province increased by 1.39 percent a year and harvested area increased 2.64 percent a year. In the same period, tea productivity rose by 39.74 percent in total (Table 3.9).

3.3.2 Tea varieties

The mix of tea varieties has changed in recent years. Several new varieties of tea such as LDP1, LDP2, Phuc Van Tien and Kim Tuyen have been planted and these have increased the area cultivated by new varieties to 71.5 percent in 2015. In this province, varieties which are relevant to processing green tea are Kim Tuyen, Phuc Van Tien and

Bat Tien; varieties for processing black tea include PH1 and PH11; and LDP1 and

LDP2 for used for both green and black tea. The planted area of safe tea was 3,600 hectares, with 1,640 hectares of planted area certified by the Rainforest Alliance

program of the European Union (Department of Agriculture and Rural Development in

Phu Tho province, 2015).

To improve food safety and restore the trust of consumers, the Ministry of

Agriculture and Rural Development of Vietnam implemented a program called “safe vegetable” in 1995. The program aims to educate farmers on the proper use of fertilizers and pesticides as well as of water from non-polluted sources. After that, in 2009, the program is extended to the tea sector and consequently, “safe tea” is encouraged to plant in tea production areas of this country (Bac et al., 2017).

Rainforest Alliance is a program of the European Union. The program aims to educate smallholder farmers and large estate tea growers on the dangers of land degradation and training them in sustainable farming and land management techniques by using methods that protect the health of farmers, their livelihoods, their land and the surrounding waterways (VBCSD, 2015).

Machinery is used in tea production. For example, in the province, 2,087 harvest machines harvest 56.3 percent of tea areas, and 1,416 pruning machines prune 80 percent of tea areas (Department of Agriculture and Rural Development in Phu Tho province, 2015).

3.3.3 Tea processing

There are 56 enterprises and agencies which process green tea, black tea and others with capacity of a tonne of leaf tea per day, with 1,261 mini-processing units and 12 villages producing and processing tea in the whole province. Some firms meet the national standard in food safety for tea processing such as Phu Ben Tea Company Limited, Phu

Da Tea Company, Hai Yen Tea Company Limited, and Bao Long Exported Tea

Company Limited. About 95 percent of leaf tea is dried by machines in the whole province. The total volume of processed tea in the whole province is 50–55,000 tonnes a year, with 70 percent black tea and 30 percent green tea (Department of Agriculture and

Rural Development in Phu Tho province, 2015).

3.3.4 Tea price and exports

Producer prices and retail prices of tea in Phu Tho province are shown in Figure 3.4.

Figure 3.4. Producer and retail prices of tea in Phu Tho province

120,000.00

107,234 100,000.00 99,787

87,401 86,735 80,000.00

60,000.00 Producer price of leaf tea

VND/kg Retail price of processed tea

40,000.00

20,000.00

5,055.00 4,983.00 4,562.00 4,769.00 0.00 2013 2014 2015 2016 Source: Compiled from Phu Tho Statistics Office, 2017 Note: VND means Vietnam Dong (Vietnamese currency unit), kg means kilogram

In 2016, the producer price of leaf tea in Phu Tho province was VND4,769/kg which declined by nearly VND300/kg compared to the price in 2013. By contrast, the retail price of processed tea in this province increased by about 41 percent from 2013 to

2016 and reached a peak of VND107,234/kg (US$4.71/kg) in 2016. The retail price of processed tea was approximately 23 times higher than the producer price of leaf tea in

2016. A big gap between the retail price and producer price means challenges for tea producers in capturing profit in the tea value chain since the majority of the value added to tea belonged to processors, traders and exporters (Figure 3.4).

The export of processed tea by Phu Tho province to the international market is presented in Figure 3.5.

Figure 3.5. Export volume and turnover of processed tea by Phu Tho province

25000 40000

2051635788 35000 20000 17528 30000 28667 16900 15097 25855 25000 15000 23552 20000 tonnes 10000 15000 US$ thousand

10000 5000 5000

0 0 2013 2014 2015 2016

Export volume Export turnover Source: Compiled from Phu Tho Statistics Office, 2017

Variation in exported volume and turnover of processed tea had a similar pattern for four years (2013–2016). Starting at 15,097 tonnes in 2013, the exported volume of tea rose significantly by more than 5,000 tonnes in 2014 and then dropped sharply in the next two years. Likewise, exported turnover of tea increased rapidly by more than US$7 million from 2013 to 2014 and then presented a rapid decline in the next two years

(Figure 3.5). Currently, tea products from Phu Tho province are exported to many countries all over the world such as India, China, Pakistan, Germany, the United States of America and the Netherlands.

3.4 Chapter conclusions

This chapter presented an overview of production, consumption and trade of tea in the world, Vietnam and Phu Tho province. In Vietnam, Phu Tho province ranks third in tea productivity and fourth in planted area of tea. There is huge difference between the producer price of leaf tea and the retail price of processed tea in this province.

Currently, tea products from Phu Tho province are exported to many countries all over the world such as India, China, Pakistan, Germany, the United States of America and the Netherlands. In order to assess the influence of contract farming of tea production on tea productivity and farmers’ income and to examine the role of tea production in poverty reduction, Phu Tho province is chosen as a representative case study.

CHAPTER 4: RESEARCH METHODS

4.0 Introduction

Having developed a theoretical framework to explain the motives for contracts in

Chapter 2, and then establishing a context for tea production in Chapter 3, this chapter discusses the development of the research tools and data used to answer the research questions. Research methodologies, including selection of research sites, sample, data collection, and data analysis are discussed. This chapter is important because it justifies the research tools and data used to answer the research questions in factors affecting the technical efficiency of tea production, the impact of contract farming on tea productivity and farmers’ income, determinants of poverty in Phu Tho province, and the relationship between economic growth, tea exports and poverty in Vietnam.

4.1 Selection of research sites

In order to investigate the impacts of contract farming on productivity and income of farmers in tea production, Phu Tho province, Vietnam was chosen as a research site for the following reasons: (1) tea is considered a main crop in the program of developing the agricultural economics of Phu Tho province because of its contributions to poverty reduction, employment generation and income improvement for farm households; (2) currently Phu Tho province is ranked third on production and fourth on planted area of tea in Vietnam; and (3) there are still issues in tea production in Phu Tho province,

including low yield, high chemical use, poor coordination between producers and enterprises, and lack of owned brands for tea products.

Figure 4.1. Administrative map of Phu Tho province, Vietnam with research sites

Research sites

The selection of districts in Phu Tho province as research sites depends on the following factors: (1) planted tea and harvested area of tea in these districts is in the top five of Phu Tho province; (2) existence of coordinated models between producers and enterprises in tea production in these districts; (3) possibility of receiving permission from provincial and district authorities to conduct the study; and (4) possibility of obtaining a list of contracted farmers from enterprises which operate in these districts.

In terms of harvested area of tea, the top five districts of Phu Tho province are: Tan Son district (2,670 hectares), followed by Thanh Son district (1,968.1 hectares), Thanh Ba district (1,867.1 hectares), Yen Lap district (1,661.7 hectares) and Ha Hoa district

(1,654.6 hectares) (Phu Tho Statistics Office, 2015). Some linkage models between growers and firms in tea production are operating in these districts including 64

coordinated models between Phu Ben Tea Company Limited and tea producers in

Thanh Ba district, Phu Da Tea Company and tea growers in Thanh Son district, Dai Loc

Company Limited and tea producers in Tan Son district, Ton Vinh Company Limited and tea growers in Thanh Son district, and Asian Company Limited and tea farmers in

Yen Lap district. The researcher contacted and received permission from provincial and district authorities to conduct this research. Lastly, enterprises agreed to provide a list of contracted farmers to the researcher. Therefore, based on above criteria, the five districts in Phu Tho province of Thanh Ba, Ha Hoa, Tan Son, Thanh Son and Yen Lap were chosen as research sites for this study (see research sites in Figure 4.1). Sample procedures are specified in the next section.

4.2 Sample

A sample for the household survey was chosen purposely from provincial to household levels, as shown in Figure 4.2.

Figure 4.2. Sampling procedure in Phu Tho province

Phu Tho province

5 districts

9 communes

18 villages

358 households

Source: Author, 2016

Due to constraints in time, human and financial resources, it was necessary to carry out sampling in this study. Purposely sampling was applied for this research because it can reduce the potential bias in selecting cases to be included in the sample and therefore this method provides us a highly representative sample chosen from the population (Sharma, 2017).

This research is conducted in Phu Tho province, where eleven districts plant tea.

However, farmers cannot be selected and interviewed in all districts because of the limitations mentioned above. Hence, a four-stage sample was designed to collect primary data. First, from the provincial level, five districts were purposely chosen which satisfy the criteria mentioned above. Second, nine communes were chosen purposely from five districts. Specifically, three communes (Thai Ninh, Dong Linh and Van Linh) were chosen in Thanh Ba district, one commune (Yen Ky) in Ha Hoa district, three communes (Tan Phu, Van Luong and Minh Dai) in Tan Son district, one commune

(Van Mieu) in Thanh Son district, and one commune (Ngoc Lap) in Yen Lap district.

The next step was to select purposely 18 villages from nine communes (two villages per commune). To select 18 villages for the survey, the researcher contacted and discussed the research with the chair of each commune and leaders of the villages to confirm there are at least two types of tea households, such as contracted and independent farmers, in a chosen village. Two criteria used to select nine communes and 18 villages include the survey is accepted by chairmen of communes and leaders of villages and there are both contracted and non-contracted farmers of tea production in chosen communes and villages. Finally, 358 households were chosen purposely for the survey from 18 villages

(Figure 4.2).

In addition, it is not known exactly how many tea households are in Phu Tho province or how many are engaged in contract farming because of limited data released by statistical offices. With an expectation to obtain a significance level at five percent in this research, the sample size for the household survey was determined as follows:

n = [Z P(1 P)]/d (Daniel and Cross, 2013). 2 2 Where: n is −the sample size; Z is Z-Score (with a 95% confidence interval, Z is equal to 1.96); P is expected prevalence or proportion (P = 0.5 is large enough for the sample); and d is precision (d = 5% can be accepted for this research).

Based on the above formula, the calculated sample size is 385 farm households

(n = 385). However, due to the isolation of some households in remote areas, only 358 households were interviewed. Therefore, information from 358 households was entered and analysed for this research.

Table 4.1. Sample of households interviewed in Phu Tho province

Type Thanh Ba Ha Hoa Tan Son Thanh Son Yen Lap Total district district district district district Contracted farmers 1 0 86 36 41 164 Independent farmers 87 5 97 5 0 194 Total 88 5 183 41 41 358 Source: Author’s calculation, 2016

Households interviewed were households who have ever or are cultivating tea in districts of Phu Tho province. Of the total of 358 households interviewed, there were more independent households (194) compared to contracted household (164) based on the selection procedure. Contracted farmers sign contracts with firms to produce tea, but cultivate tea on their own land. Normally, these contracts last from one to five years.

Independent farmers, who do not make contracts with firms, often grow tea on their own land and sell products to open markets. A list of independent households was

provided by commune authorities and leaders of villages. A list of contracted producers was provided after discussions with managers of enterprises (Table 4.1).

4.3 Data collection

4.3.1 Secondary data

In order to investigate the impacts of contract farming on productivity and income of farmers, a total of 34 scientific papers in academic journals were reviewed, with 15 papers on the effects of contract farming on productivity and 19 papers on influences of contract farming on income. To increase accuracy in assessing effects of contract farming on productivity and income of farmers, several agricultural products in different countries and Vietnam were researched. The effects of contract farming on productivity and income of farmers were analysed in the context of Vietnam.

Further, because the majority of Vietnamese tea production is exported to international markets and it is necessary to understand tea production, consumption and trade in the world and Vietnam, studies, reports and scientific papers published by the

Food and Agriculture Organization of the United Nations (FAO), the Government of

Vietnam, the Ministry of Agriculture and Rural Development (MARD) and Vietnam

General Statistics Office were reviewed. In order to identify directions for the development of contract farming and tea sector in Vietnam, laws, decisions, directives, circulars released by Vietnam National Assembly, the Government of Vietnam and the

MARD were explored. Finally, secondary information on nature, socio-economics and tea production and contract farming in Phu Tho province was collected from reports and

census databases of the Phu Tho People’s Committee, the Department of Agriculture and Rural Development and Statistics Office in Phu Tho province by Internet searching or direct discussions with officers.

4.3.2 Primary data

Primary data for this research was collected from a cross sectional survey in Phu Tho province, between March and June 2016. The purpose of the survey was to gather detailed information on the effects of contract farming on productivity and income of farm households in tea production in Phu Tho province. Interviewers included the researcher and six undergraduate students studying Business Administration and

Accounting disciplines at the Vietnam National University of Agriculture, Vietnam. The survey had 62 questions in four sections as follows: section 1 (Demographic information about the household: 10 questions); section 2 (Resources of the household:

9 questions); section 3 (Tea production, processing and marketing of the household: 26 questions); and section 4 (Contracts in production, processing and distribution of tea: 17 questions). Before carrying out interviews, the questionnaires were translated into

Vietnamese and all interviewers were trained by the researcher to ensure a full understanding of the survey. The survey was pre-tested by randomly choosing ten tea farmers, with five farmers from Dong Linh commune, Thanh Ba district and five farmers from Van Luong commune, Tan Son district, Phu Tho province. The purpose of pre-testing was to examine the relevance of the survey to real situations to obtain appropriate information from respondents and save time in interviews. The questionnaires were revised and completed before researchers conduct interviews.

The face-to-face method of survey was selected to minimise the non-response rate. Interviewers asked respondents the questions in the questionnaires and answers were recorded on answer sheets by interviewers. An interview took up to an hour and interview places were respondents’ homes, fields or village offices. Respondents expressed their willingness to participate in the interview by signing consent forms.

A list of independent farmers was provided by commune authorities and village leaders in Phu Tho province. To receive the list of contracted households, the researcher worked with managers of firms which are producing, processing and exporting tea products in Phu Tho province. The researcher contacted three enterprises to obtain the list of contracted households: Dai Loc Tea Company Limited in Tan Son district, Ton

Vinh Technology and Commerce Company Limited in Thanh Son district, and Asian

Tea Company Limited in Yen Lap district. Then, contracted households were chosen randomly from the list provided by businesses and the number of contracted households is presented in Table 4.2.

Table 4.2. Number of contracted households interviewed in Phu Tho province

Enterprises Number % Dai Loc Tea Company Limited 80 48.78 Ton Vinh Technology and Commerce Company 39 23.78 Asian Tea Company Limited 45 27.44 Total 164 100.00 Source: Author’s calculation, 2016

These companies are small and medium domestic private enterprises operating in the tea sector. Of the total of 164 contracted farmers chosen to interview (Table 4.2),

80 producers had contracts with Dai Loc Company, 39 producers had contracts with

Ton Vinh Company, and 45 producers had contracts with Asian Company.

The data collected was checked daily by the researcher and suspicious entries were verified before the research team left a community. The data collected was coded, entered onto a computer, checked, cleaned and then analysed.

4.4 Data analysis

4.4.1 Descriptive analysis

Statistical indicators such as average (Mean) and standard deviation (SD) were calculated to describe demographic and socio-economic features of tea households. The purpose of descriptive analysis is to present and compare demographic and socio- economics characteristics of tea households. Descriptive analysis was carried out by

Stata MP 14.2 software.

4.4.2 Quantitative analysis

Stochastic frontier model

In economics, productivity and efficiency terms are used interchangeably (Coelli et al.,

2005) and these are used interchangeably in this study. In this study, tea productivity is computed by the volume of leaf tea harvested for a year by each household.

Efficiency includes two components: technical efficiency and allocative efficiency. However, this research only focuses on technical efficiency to estimate a

maximum output of tea produced with a given input level which means examining the effects of contract farming on tea productivity without the influence of price factors.

Efficiency theory was proposed by Farrell (1957). He developed a production function with two inputs (X1 and X2) to produce output Y, in which the vertical and horizontal axes reflect the input-output ratios. SS’ represents the unit isoquant in production of the farm, while AA’ denotes the isocost of the farm (Figure 4.3).

Figure 4.3. Measurement of technical efficiency

X2/Y

Q’

0 A’ S’ X1/Y Source: Farrell, 1957

Technical efficiency (TE) can be measured as follows: TE = OQ/OP. The value of TE is between zero and one, and a farm will obtain full technical efficiency if TE is equal to one. QP reflects the technical inefficiency of the farm because, with a given quantity of inputs, to produce a unit of output, QP represents a proportional decrease of inputs without a reduction in output. Thus, a production function illustrates the maximum amount of output produced from a given quantity of inputs with a fixed technology (Figure 4.3).

In fact, technical efficiency of a farm tends to be associated with demographic characteristics (age, gender, education, etc.) and socio-economic characteristics (land

size, labour, asset value, income, etc.) of a farmer and farm (Hoppe et al., 1996 and

2001). Therefore, this research estimates technical efficiency of tea production and then assesses the impact of contract farming on tea productivity.

There are different methods used by previous studies to estimate technical efficiency in agricultural production. In order to estimate technical efficiency of tea production, methods applied include the translog model (Basnayake and Gunaratne,

2002) and the stochastic frontier model (Ajibefun et al., 2006; Baten et al., 2010; Phu and Nguyen, 2014; and Hong and Yabe, 2015). The stochastic frontier model has been used widely to estimate technical efficiency of other agricultural products such as maize, rice, groundnuts and pulses in Southern Malawi (Chirwa and Mwafongo, 1998), wheat in Punjab, Pakistan (Hassan and Ahmad, 2005), agricultural production in sixty provinces of Vietnam during 1990 to 2005 (Minh and Long, 2009), paddy in Sri Lanka

(Shantha et al., 2012), shrimp in Bangladesh (Begum et al., 2013), and maize in Central province, Zambia (Chiona et al., 2014).

In this research, the stochastic frontier model is chosen because of its advantages as follows: (1) it presents a disturbance term denoting statistical noise, measurement error and exogenous effects out of the control of production units; and (2) it allows statistical tests of hypotheses regarding the production structure and the degree of inefficiency (Hassan and Ahmad, 2005). First, stochastic frontier models are used to estimate factors affecting tea productivity. Second, efficient levels of tea production for both contracted and independent farmers are demonstrated.

In order to implement these stages, the framework developed by Bravo-Ureta et al. (2012) is applied by designing two stochastic frontier models, with the first one for contracted farmers and the second for independent farmers. However, an issue occurs

because of unobserved differences among contracted farmers and independent farmers.

Hence, the stochastic frontier model proposed by Greene (2010) is used to solve this problem due to its consistency in non-random sample selection.

The specification of this model can be represented as follows:

Yi = βc Χi + Ɛi for D = 1 (4.1) ′ i Yi = βn Χi + Ɛi for D = 0 (4.2) ′ i Di = 1[ Zi + Wi > 0] (4.3) ′ Ɛi = Vi α Ui (4.4)

Ui ~ N −(0, cu) for D = 1 (4.5) + i Ui ~ N (0, δcu) for D = 0 (4.6) + i Wi, Vi~ N [δ0, 0], (1, c, cv, c2v) for D = 1 (4.7) 2 i Wi, Vi~ N [0, 0], (1, λn, δnv,δ n2v) for D = 0 (4.8) 2 i Where: Yi representsλ theδ logarithmicδ output quantity of tea of producer i = 1, …,

N; Xi denotes a vector of logarithmic input quantity (planted area, family labour, capital, fertilisers, chemicals, etc.); Di is a dummy variable (1 for contracted farmers and

0 for independent farmers); Zi is a vector of covariates in the sample selection equation;

Ɛi denotes the error term of the stochastic frontier model; Vi represents noise

(managerial ability, etc.); Ui is inefficiency; Wi denotes the error term of the selection equation; α, βc and βn are parameters to be estimated.

Assumptions for this model are: Ui follows a half-normal distribution with

c n dispersion parameter u and u; Wi and Vi follow a bivariate normal distribution with

c2 n2 c n variances 1 and v andδ v; δλ is the correlation coefficient of contracted farmers; λ is the correlation coefficientδ δ of independent farmers; and non-zero values of λc and λn express self-selection.

Based on a two-stage estimation procedure proposed by Greene (2010), technical efficiency in tea production of contracted farmers and independent farmers

-Ui can be estimated [TEi = E(e )] (4.9).

Table 4.3. Description of covariates in the stochastic frontier model

Variable definitions Labels Unit Expected signs of coefficients 1. Stochastic model Dependent variable: Productivity (total Y kilograms (kg) volume of leaf tea harvested by a household in a year) Independent variables: Family labour X1 person +/- Land size for tea plantation X2 sao + Capital for tea production X3 VND1,000 + Quantity of mixed fertiliser (NPK) X4 kg/sao/year + Pesticide costs for tea X5 VND1,000/sao +/- Density: the average number of tea trees X6 tree/sao +/- grown in an area 2. Inefficiency model Dependent variable: the level of inefficiency Ui Independent variables: Age of household head X7 years +/- Gender (1=Male and 0=Female) D1 +/- Education level of household head X8 years - Household category* (1=Poor household D2 +/- and 0=Otherwise) Harvest times: number of harvest times in X9 times +/- a year Harvest methods (1=Machine and D3 +/- 0=Otherwise) Note: *poor households are specified based on criteria for ‘poor’ for the period 2016–2020 in Decision number 59/2015/QD-TTg of the Prime Minister of Vietnam. In Vietnam, poor households in rural areas are households that have average capita income less than or equal to VND700,000 per month (US$31/capita/month).

Propensity score matching

Regression analysis may be used to investigate the influence of contract farming on tea productivity and income of farmers. The specification of the model can be represented as follows (Heckman, 1979):

Y = X + A + (4.10)

i i i Where:β Y representsα Ɛ either tea productivity or income of households; Xi denotes explanatory variables; β and α are coefficients to be estimated; Ai is a dummy variable

(1 for contracted farmers and 0 for independent ones); and Ɛi is the error term.

However, participation of farmers in a contract is non-random since contract participation depends on either selection of contractors or the situation of farm households. Hence, the decision of farmers to participate in a contract is unobservable.

Correlation between explanatory variables (Ai) and the error term (Ɛi) violates one of key assumptions of ordinary least square (OLS) and this generates a bias estimation.

Different methods are proposed to deal with this issue. Some scholars have used the

Heckman two-step method, while others have used instrumental variable (IV) to fix selection bias (Meshesha, 2011; Bellmare, 2012; Key, 2013; and Kumar et al., 2016).

However, the Heckman two-step model depends heavily on the assumption of normal distribution in error terms. Instrumental variable is strict because it is very difficult to specify appropriate instruments in the estimation. In order to overcome these challenges, propensity score matching is used in this study to assess the impacts of contract farming on tea productivity and income of farmers.

Propensity score matching is the most common method which matches individuals in the comparison group to members in the treatment group with a set of observed characteristics in the form of a propensity score. The propensity is used to predict the probability of participation in an intervention (Baker, 2000). It is a non-

experimental method developed to estimate the average effect of social programs

(Rosenbaum and Rubin, 1983; and Heckman et al., 1998). In propensity score matching, average outcomes of participants and non-participants are compared under the condition of the propensity score value. Hence, the match is good if the treatment group is matched to the control group with a closer propensity score (Baker, 2000).

Propensity score matching is applied in various disciplines such as economics, international development, medicine and political science. In agricultural economics, propensity score matching has been used to investigate impacts of agricultural technology on poverty reduction (Becerril and Abdulai, 2010; Ali and Abdulai, 2010; and Tsadik et al., 2015), effects of training on pesticide use (Schreinemachers et al.,

2016; and Gautam et al., 2017), and influences of contract farming on crop productivity and income of farmers (Ragasa et al., 2018). Propensity score matching is useful because it uses observational data to estimate the treatment effect on an outcome and reduces the selection bias due to nonrandom treatment assignment (Dehejia and Wahba,

2002; and Garrido et al., 2014).

Propensity score matching is applied in this research because it can correct the potential selection bias which tends to increase due to systematic differences between the participants and non-participants (Ali et al., 2013). For example, contracted farmers can obtain a higher tea productivity compared to their counterparts because of technical assistance offered by enterprises in the contract of tea production. Moreover, contracted farmers may sell a bigger volume of leaf tea compared to independent farmers because businesses contract to purchase leaf tea after harvesting based on the terms of the contract.

Propensity score matching is carried out by six steps as follows. The first step is estimating determinants affecting contract participation by running a binary logistic model. The objective of this assignment is to estimate propensity scores of covariates in the sample of both contracted and independent farmers. The second step is demonstration of the distribution of households with respect to the estimated propensity scores of both contracted and independent farmers. The next step is selection of a matching algorithm. In this step, a matching algorithm is used to balance comparison groups by matching treatment group individuals with suitable controls. There are various matching algorithms such as the nearest-neighbour matching, caliper matching, radius matching, Kernel matching, and Mahalanobis metric matching (Ali et al., 2013).

The fourth step is assessment of the match quality. This step may be implemented by the psmatch2 routine proposed by Leuven and Sianesi (2003). Outputs of this step include the estimated treatment effect with its standard errors and confidence interval in addition to a number of diagnostics used to examine the match quality. In addition, covariate distributions before and after matching are compared. The matching is successful if the distribution of all covariates in treatment and control groups is balanced (Gemeci et al., 2012). The fifth step is estimation of average treatment effects on the treated (ATT). The objective of this step is to assess impacts of contract participation on tea productivity and income of farmers. The final step is implementation of sensitivity analysis. The purpose of this step is to examine the possibility of hidden bias from unobserved covariates.

To evaluate the impact of contract participation on tea productivity and farmer income, all observable characteristics need to be homogeneous between contracted farmers (the treatment group) and independent farmers (the control group). The

expected treatment effect of contract participation or average treatment effect on the treated (ATT) can be measured by the difference between the actual tea productivity and/or income and tea productivity and/or income if farmers did not participate in the contract. ATT may be specified as follows:

ATT = E(Y Y / P = 1) (4.11)

1i 0i i Where: ATT −denotes average treatment effect on the treated for tea productivity and/or income; Y1i represents tea productivity and/or income when the farmer participates in the contract; Y0i represents tea productivity and/or income when the farmer did not participate in the contract; and Pi denotes contract participation (1 for contract participation and 0 for otherwise).

The main objective of the research is to investigate effects of contract farming on tea productivity and farmer income. These are two distinguished goals. Further, tea productivity and farmer income are influenced by different factors. Thus, a different set of covariates used in the estimation of propensity score matching for tea productivity and farmer income (Tables 4.4 and 4.5).

Table 4.4. Description of covariates in the treatment effect model for tea productivity

Variable definitions Labels Unit Expected signs of coefficients 1. Treatment variable: contract participation (P=1 P for contract participation and P=0 for otherwise) 2. Covariates Age of household head X1 years +/- Gender (1=Male and 0=Female) D1 +/- Household category* (1=Poor household and D2 +/- 0=Otherwise) Family labour X2 person +/- Land size for tea plantation X3 sao + Tea age: the average years of tea trees grown by X4 years + household Density: number of tea trees grown in an area X5 tree/sao + Harvest times X6 times +/- Harvest methods (1=Machine and 0=Otherwise) D3 +/- 3. Outcome variable: tea productivity (total Y kg/year volume of leaf tea harvested by a household ) Note: *poor households are specified in Table 4.3

Table 4.5. Description of covariates in the treatment effect model for farmer income

Variable definitions Labels Unit Expected signs of coefficients 1. Treatment variable: contract P participation (P=1 for contract participation and P=0 for otherwise) 2. Covariates Age of household head X1 years +/- Gender (1=Male and 0=Female) D1 +/- Household category* (1=Poor D2 +/- household and 0=Otherwise) Family labour X2 person +/- Land size for tea plantation X3 sao + Tea yield X4 100kg/sao/year + Price of leaf tea X5 VND1,000/kg + 3. Outcome variables: Income of tea household Y1 VND1,000/year Income per capita Y2 VND1,000/year Income per labor Y3 VND1,000/year Note: *poor households are specified in Tables 4.3

Logit model

The logit model is widely employed to estimate the functional relationship between binary response variable and predictors. The logit is chosen for this study because it has a simple structural form and interpretable results. Specifically, the transformation of the logit model is directly interpreted by log-odds (Adekanmbi, 2017). In addition, the logit is better than the probit model because probability of observations increases when the sample size increases (Cakmakyapan and Goktas, 2013).

Thus, the logit model is used in this research to assess the impacts of determinants on poverty of tea producers. Poverty of tea producers is calculated by the monthly average income of the households. The model can be specified as follows:

Ln(Odds) = Logit(P) = Ln = (4.12) Pi 𝑛𝑛 1−Pi 𝑘𝑘=0 𝑘𝑘 𝑖𝑖𝑖𝑖 The transform form of the logit� can� be ∑specified𝛽𝛽 𝑋𝑋 as follows:

Ln = + + + = + Ɛ (4.13) Pi 1−Pi 0 1 1 𝑘𝑘 𝑖𝑖𝑖𝑖 𝑖𝑖 � � 𝛽𝛽 𝛽𝛽 𝑋𝑋 ⋯ 𝛽𝛽 𝑋𝑋 𝑋𝑋′ β

Where: The ratio (Pi/1-Pi) is the odd that Pi = 1 (P=1 for the poor household and

P=0 for otherwise); βs are unknown parameters to be estimated; X is a vector explanatory variables which include demographic and socio-economic characteristics of farm households; Ɛ denotes an independent normally distributed error term; and k =

1,…, n where n is number of observations.

Table 4.6. Description of covariates in the logit model

Variable definitions Labels Unit Expected signs of coefficients Dependent variable: poverty* (P=1 for P poor household and P=0 for otherwise) Covariates Age of household head X1 years +/- Gender (1=Male and 0=Female) D1 +/- Family labour X2 person +/- Land size for tea plantation X3 sao - Capital for tea production X4 VND1,000 - Tea productivity X5 kg/year - Price of leaf tea X6 VND1,000/kg - Contract participation D2 +/- Note: *poor households are specified based on criteria for ‘poor’ for the period 2016– 2020 in Decision number 59/2015/QD-TTg of the Prime Minister of Vietnam. In Vietnam, poor households in rural areas are households that have average capita income less than or equal to VND700,000 per month (US$31/capita/month).

Vector autoregressive (VAR) model

In this study, the VAR model is used to examine the causal relationship between tea exports, gross domestic product (GDP) and poverty in Vietnam for the last four decades

(1977–2016). This period is chosen for the study since it covers important events for the development of socio-economics of this country. Specifically, from 1977 to 1985,

Vietnam has implemented a planned central economy. A transitional process of this country has been started in 1986 when a planned central economy is eradicated and substituted by a market-oriented economy. After that, the integration of Vietnam into the international community has been enhanced by participating in the Association of

Southeast Asian Nations (ASEAN) in 1995 and the World Trade Organization (WTO) in 2006. A time-series dataset is gathered from the database released by the Food and

Agriculture Organization of the United Nations (FAO) and the World Bank. The VAR model is chosen because it explains the endogenous variables solely by their own 82

history, apart from deterministic regressors and therefore this method incorporates non- statistical a priori information (Pfaff, 2008). In addition, the VAR model is a popular method in economics and other sciences since it is a simple and flexible model for multivariate time-series data (Suharsono et al., 2017).

The specification of a VAR model can be defined as follows (Pfaff, 2008):

= + + + (4.14)

𝑡𝑡 1 𝑡𝑡−1 𝑝𝑝 𝑡𝑡−𝑝𝑝 𝑡𝑡 Where:𝑌𝑌 𝐴𝐴 𝑌𝑌Yt denotes⋯ a𝐴𝐴 set𝑌𝑌 of KƐ endogenous variables (GDP, export value of tea, proportion of population living below the poverty line, and poverty gap); Ai represents

(K x K) coefficient matrices for i = 1,…, p; and Ɛt is a K-dimensional process with E(Ɛt)

= 0.

Table 4.7. Description of covariates in the VAR model

Variable definitions Label Unit GDP Y1 million US$ Export value of tea Y2 thousand US$ The proportion of population living below the poverty line Y3 % Poverty gap: mean distance below the poverty line Y4 % Note: US$ means United States Dollar

In this study, the procedure of a VAR model includes six steps: (1) performing the unit root test; (2) determining lag length; (3) estimating the VAR model; (4) testing the Granger causality; (5) checking the stability of eigenvalues; and (6) implementing the Johansen test for co-integration. The VAR model is estimated by the Stata MP 14.2 software.

4.5 Chapter conclusions

This chapter discussed the development of the research tools used to collect and analyse the data to answer the research questions. Research methodologies, including selection 83

of research sites, sample, data collection, and data analysis were discussed. Stochastic frontier models are used to estimate factors affecting tea productivity and demonstrate efficient levels of tea production. The propensity score matching is employed to examine the effects of contract farming on tea productivity and farmer income. Further, the logit model is used to estimate influences on poverty of tea producers in Phu Tho province. Lastly, the VAR model is used to investigate the relationship between economic growth, tea exports and poverty in Vietnam over the last four decades (1977–

2016).

CHAPTER 5: TEA PRODUCTIVITY AND TECHNICAL EFFICIENCY OF

TEA PRODUCTION IN PHU THO PROVINCE2

5.0 Introduction

Having identified the knowledge gaps and then discussing the formation and implantation of the tools used to address the research questions, this chapter tackles the first research objective to investigate influences of factors on tea productivity and technical efficiency of tea production in Phu Tho province.

5.1 Overview of analytical results

This section provides a brief overview of the analytical results (see chapter 4 on methodology for details).

Yi = βc Χi + Ɛi for D = 1 (5.1) ′ i Yi = βn Χi + Ɛi for D = 0 (5.2) ′ i Di = 1[ Zi + Wi > 0] (5.3) ′ Ɛi = Vi α Ui (5.4)

Ui ~ N −(0, cu) for D = 1 (5.5) + i Ui ~ N (0, δcu) for D = 0 (5.6) + -Ui i TEi = E(e δ) (5.7)

2 Anh Tru Nguyen, Janet Dzator and Andrew Nadolny (2018). Contract farming, agriculture productivity and poverty reduction: evidence from tea estates in Viet Nam. Asia-Pacific Sustainable Development Journal, 25(1), 107-143. 85

Equations 5.1–5.6 (which are the same as equations 4.1–4.6 under methodology) are used to estimate factors affecting tea productivity and inefficiency levels. The stochastic frontier model and inefficiency model are simultaneously estimated to overcome an issue of unobserved differences among contracted and independent farmers. Variables in socio-economics and production of households such as family labour, land size, capital, quantity of mixed fertilizer, pesticide costs, and density, are chosen for the stochastic frontier model, while demographical and production variables such as age, gender, education, household category, harvest time, and harvest methods, are chosen to estimate inefficiency models.

Equation 5.7 (which is the same as equation 4.9 under methodology) is used to demonstrate technical efficiency of tea production in this chapter. Technical efficiency is analysed separately for contracted and independent farmers to compare technical efficiency of tea production of the two farmer groups that are of interest in the current study. Results in factors affecting tea productivity and technical efficiency of tea production of contracted farmers are presented in Tables 5.2 and 5.3. Similarly, Table

5.4 and 5.5 address outcomes on factors affecting tea productivity and technical efficiency of tea production of independent households.

5.2 Characteristics of tea households in Phu Tho province

Characteristics of tea households are summarised in Table 5.1. In terms of demographic characteristics, the average age of household heads is 48.7 years and the average age of contracted and independent farmers is similar. The majority of tea households is managed by the men because in Vietnam, especially in the rural area, important tasks

are often decided and carried out by the men. Most of independent households is dominated by King people, while contracted households are dominated by other minority ethnic groups such as Tay, Nung, Dao, and so on. The average number of school years of household heads is nearly eight years which implies that most household heads have only graduated from secondary school (Table 5.1).

For socio-economic characteristics, the average number of members in a family is 5 and labourers is 2.7, and contracted farmers have more family members and labourers than independent farmers. Total land size and land size for tea of contracted households are higher than those of independent ones by more than 8 sao and 6 sao, respectively. The average asset value of a tea household is VND53 million (US$2,311).

Capital for tea production of contracted farmers is nearly VND3 million higher than independent farmers. Similarly, tea tree density of contracted households is higher that of independent households by nearly 100 trees per sao. However, tea production experience and tea tree age of independent farmers are higher than those of contracted households. Both household types have on average more than five harvest times a year

(Table 5.1).

T-test results show that there are different characteristics between contracted and independent households, except age, hired land, and asset value. This implies that farmers chosen for the survey are not different in age. The majority of farmers plants tea in their own land and consequently, there is no different between two groups in hired land. Asset value of two groups is homogenous and this suggests that there is no different in asset ownership between two groups of farmers who live in the same community (Table 5.1).

Table 5.1. Characteristics of tea households in Phu Tho province

Items Unit Contracted households Independent households Overall T-test (n = 164) (n = 194) (n = 358) (Difference)

Mean SD Mean SD Mean SD 1. Demographic characteristics Age years 48.44 8.34 48.92 10.79 48.70 9.74 0.48 Gender (1=Male and 0=Female) 0.95 0.79 0.86 -0.16*** Ethnicity (1=Kinh and 0=Otherwise) 0.26 0.78 0.54 0.52*** Education school years 7.62 2.24 8.31 2.29 7.99 2.29 0.69*** Household categorya (1=Poor household and 0.04 0.12 0.14 0.18*** 0=Otherwise) 2. Socio-economic characteristics Household members person 4.53 1.08 4.12 1.15 5.01 1.17 0.90*** Family labourers person 3.01 0.97 2.55 1.05 2.76 1.04 -0.45*** Total land size sao 27.49 17.61 20.49 19.29 23.69 18.84 -6.99*** Land size for tea plantation sao 22.60 17.00 16.49 13.24 19.29 15.36 -6.10*** Hired land (1=Yes and 0=Otherwise) 0.01 0.01 0.01 -0.00 Asset value VND1,000 53202.9 33889 52252.11 40523.86 52687.67 37581.71 -950.78 Capital VND1,000 13682.93 11194.33 11039.69 8036.40 12250.56 9688.65 -2643.23*** Credit (1=Yes and 0=Otherwise) 0.17 0.32 0.25 0.15*** Experience years 15.26 7.17 19.94 9.08 17.79 8.57 4.68*** Tea tree ageb years 17.90 9.19 22.79 11.40 20.55 10.71 4.88*** Density trees/sao 603.10 91.01 513.91 119.99 554.77 116.39 -89.19*** Harvest times times/year 5.47 0.65 5.31 0.84 5.39 0.76 -0.15** Source: Author’s calculation, 2018 Note: SD means standard deviation apoor households are specified based on criteria for ‘poor’ for the period 2016–2020 in Decision number 59/2015/QD-TTg of the Prime Minister of Vietnam. In Vietnam, poor households in rural areas are households that have average capita income less than or equal to VND700,000 per month (US$31/capita/month). btea tree age is calculated by the average years of tea trees grown by a household. *** and ** mean statistical significance at 1% and 5%, respectively.

5.3 Estimation of tea productivity and technical efficiency of tea production

5.3.1 Tea productivity and technical efficiency of contracted farmers

Tea productivity

Maximum likelihood estimates of the stochastic frontier model and inefficiency model for tea production of contracted farmers are presented in Table 5.2.

Table 5.2. Maximum likelihood estimates of the stochastic frontier model and inefficiency model for contracted farmers

Variables Coefficient Standard Error Z P>[z]

Stochastic model LnLabour 0.063 0.104 0.60 0.545 LnLand size for tea 0.809*** 0.051 15.78 0.000 LnCapital 0.007 0.027 0.26 0.798 LnNPK 0.024 0.046 0.52 0.605 LnPesticide 0.062* 0.037 1.66 0.097 LnDensity -0.699*** 0.241 -2.90 0.004 Constant 11.065*** 1.429 7.74 0.000 Inefficiency model Age 0.028* 0.016 1.74 0.083 Gender 1.756*** 0.671 2.62 0.009 Education -0.195*** 0.065 -2.99 0.003 Household category 0.357 0.606 0.59 0.556 Harvest times -0.110 0.171 -0.64 0.521 Harvest methods -0.081 0.240 -0.34 0.735 Constant -1.537 1.262 -1.22 0.223 Noise variation ( v) 0.150 Inefficiency variation ( u) 0.830 2 2 2 Total error varianceδ ( v u ) 0.712 Signal-to-noise ratio ( δ u/ v) 5.499 Observations δ = δ + δ 164 LogLikelihood λ = δ δ -101.35 Prob > chi2 0.00 LR test of u = 0 chibar2(01) = 33.93 Prob >= chibar2 = 0.00 Source: Author’s calculation, 2018 Note: *** andδ * mean statistical significance at 1% and 10%, respectively

The values of loglikelihood of -112.14 and prob>chi2 of 0.00 support the fitness of the half-normal model. Noise variation and inefficiency variation are equal to 0.15 and 0.83, respectively. The value of lambda is greater than 1 ( = 5.499) and this suggests that the majority of variation in tea productivity is outweighedλ by inefficiency effects (Table 5.2).

To test technical inefficiency in the model, hypotheses are designed as follows:

Null hypothesis (Ho): There is no technical inefficiency in the model ( u = 0)

δ Alternative hypothesis (Ha): There is technical inefficiency in the model ( u > 0)

The null hypothesis is rejected if the p-value is either less than or equalδ to the predetermined value (0.05) and, in contrast, the null hypothesis is accepted if the p- value is greater than the predetermined value (0.05). In this case, the p-value is less than the predetermined value (0.00 < 0.05) and therefore it is concluded that technical inefficiency exists in the model (Table 5.2).

In the stochastic model, land size, pesticide cost and tree density are statistically significant, while the other variables (labour, capital and NPK) are not significant. Land size and pesticide cost have a positive relationship with tea productivity which implies that an increase in land size and pesticide cost increases tea productivity. Land is a key input for tea production and therefore tea productivity may be enhanced if farm households have more land area. According to the estimation, expenditure for tea production of farmers in Phu Tho province varies from VND700,000 to VND1 million per sao a year (US$31–44/sao/year). Of the total cost for tea production, labour cost is

47.2 percent, input cost (fertilisers, pesticides and materials) is 52.4 percent, and 0.4 percent is for other costs. Thus, about half the total cost is costs for fertilisers, pesticides and materials. In recent years, high use of pesticides and chemicals has become an issue

in the tea sector in Vietnam as it is not difficult for tea producers to purchase pesticides and chemicals in open markets due to convenience, cheap prices, and lack of control from market surveillance agencies. In contrast, tree density negatively affects tea productivity and this suggests that if farmers grow more tea trees in an area, then tea productivity tends to decline. Currently, the average number of tea trees planted by contracted farmers is 603 trees per sao. According to the recommendation of experts, the most suitable tea tree density in Phu Tho province is 470 trees to 500 trees a sao.

Therefore, an increase in tea tree density reduces productivity (Table 5.2).

In the inefficiency model, age, gender and education are statistically significant, while the other variables (household category, harvest times and harvest methods) are not significant. The positive coefficient of age implies that older producers obtain a lower technical efficiency and this reflects that technical efficiency of tea depends on technological advances rather than experience of producers. Specifically, if the age of producers increases by a year, then technical inefficiency of tea increases by 2.8 percent, ceteris paribus. The coefficient of gender is positive which implies that technical efficiency of female-head households is higher than that of male-head households. This result may be explained by the carefulness and sensitivity of women in tea production. In contrast, education has a negative relationship with technical inefficiency of tea and this suggests that households which have a head with a higher education level gain higher technical efficiency compared to their counterparts. This supports the importance of knowledge in using and adopting technological advances in tea production. Currently, most contracted farmers have only graduated from secondary school and therefore improvement of their education is necessary to enhance technical efficiency of tea in Phu Tho province (Table 5.2).

Technical efficiency

Technical efficiency of tea production of contracted farmers is presented in Table 5.3.

Table 5.3. Technical efficiency of tea production of contracted farmers

Levels of technical efficiency Number % <50% 55 33.54 50-60% 17 10.37 60-70% 24 14.63 70-80% 28 17.07 80-90% 33 20.12 >90% 7 4.27 TOTAL 164 100.00 Average technical efficiency 0.599 Min technical efficiency 0.109 Max technical efficiency 0.947 Source: Author’s calculation, 2018

Average technical efficiency of 0.599 implies that technical inefficiency of tea production of contracted households is 40.1 percent. Maximum technical efficiency is

94.7 percent and minimum technical efficiency is 10.9 percent. A third of households

(or 33.5 percent) have low technical efficiency of below 50 percent, followed by 20.1 percent with 80–90 percent efficiency level, 17 percent with 70–80 percent efficiency level, 14.6 percent with 60–70 percent efficiency level, and 10.3 percent with 50–60 percent efficiency level. Seven households have technical efficiency of more than 90 percent (Table 5.3).

5.3.2 Tea productivity and technical efficiency of independent farmers

Tea productivity

Maximum likelihood estimates of the stochastic frontier model and inefficiency model for tea production of independent farmers are presented in Table 5.4.

Table 5.4. Maximum likelihood estimates of the stochastic frontier model and inefficiency model for independent farmers

Variables Coefficient Standard Error Z P>[z]

Stochastic model LnLabour 0.054 0.127 0.43 0.668 LnLand size for tea 0.715*** 0.063 11.22 0.000 LnCapital 0.133*** 0.031 4.26 0.000 LnNPK 0.047 0.044 1.06 0.291 LnPesticide 0.003 0.036 0.10 0.920 LnDensity -0.388* 0.206 -1.88 0.061 Constant 8.316*** 1.280 6.49 0.000 Inefficiency model Age 0.025* 0.013 1.94 0.052 Gender 0.478 0.312 1.53 0.125 Education -0.029 0.055 -0.54 0.591 Household category 0.278 0.283 0.98 0.325 Harvest times -0.260* 0.149 -1.75 0.081 Harvest methods -0.224 0.734 -0.31 0.759 Constant -0.080 1.247 -0.06 0.949 Noise variation ( v) 0.337 Inefficiency variation ( u) 0.920 2 2 2 Total error varianceδ ( v u ) 0.961 Signal-to-noise ratio ( δ u/ v) 2.731 Observations δ = δ + δ 194 LogLikelihood λ = δ δ -179.94 Prob > chi2 0.00 LR test of u = 0 chibar2(01) = 10.89 Prob >= chibar2 = 0.00 Source: Author’s calculation, 2018 Note: *** andδ * mean statistical significance at 1% and 10%, respectively

Similar to contracted farmers, the value of lambda of estimation for independent households is greater than 1 (λ = 2.731) reflecting that the majority of variation in tea productivity is dominated by inefficiency effects (Table 5.4).

A similar hypothesis is tested to examine the existence of technical inefficiency in the model. In this case, the p-value is less than the predetermined value (0.00 < 0.05)

and therefore it is concluded that technical inefficiency exists in the model for independent farmers (Table 5.4).

In the stochastic model, land size, capital and tree density are statistically significant, while the other variables (labour, NPK and pesticide cost) are not significant. Similar to contracted households, land is a key input for tea production and therefore tea productivity may be increased if independent farmers have more land area.

Capital also positively influences tea productivity and this implies that an increase in capital increases tea productivity. Currently, the average capital for tea production owned by independent households is VND11 million (US$498) and therefore they need to borrow more capital from credit organisations. However, tree density has a negative relationship with tea productivity which implies that if producers grow more tea trees in an area, then tea productivity tends to decrease. Currently, the average number of tea trees planted by independent farmers is 513 trees per sao and this number is slightly higher than the density recommended by experts of 470–500 trees per sao. Hence, an increase of tea tree density reduces productivity (Table 5.4).

In the inefficiency model, age of household head and harvest times are statistically significant, while the other variables (gender, education, household category and harvest methods) are not significant. The positive coefficient of age suggests that older producers obtain a lower technical efficiency and this implies that technical efficiency of tea depends on technological advances rather than experience of producers. Specifically, if the age of producers increases by a year, then technical inefficiency of tea increases by 2.5 percent, ceteris paribus. In contrast, harvest time negatively affects technical inefficiency which reflects that if harvest times increase, then technical efficiency also increases. Currently, the average number of harvests

carried out by independent farmers is 5.3 times per year. According to the recommendations of experts, growers may harvest leaf tea from nine to ten times a year and the gap between two harvest times should be from seven to ten days. Thus, technical efficiency of tea can be increased if farmers harvest leaf tea more frequently

(Table 5.4).

Technical efficiency

Technical efficiency of tea production of independent farmers is presented in Table 5.5.

Table 5.5. Technical efficiency of tea production of independent farmers

Levels of technical efficiency Number % <50% 74 38.14 50-60% 30 15.46 60-70% 22 11.34 70-80% 49 25.26 80-90% 18 9.28 >90% 1 0.52 TOTAL 194 100.00 Average technical efficiency 0.551 Min technical efficiency 0.127 Max technical efficiency 0.914 Source: Author’s calculation, 2018

Average technical efficiency of 0.551 implies that technical inefficiency of tea production of independent households is 44.9 percent. Maximum technical efficiency is

91.4 percent and minimum technical efficiency is 12.7 percent. Over one third of households (38.1 percent) have low technical efficiency below 50 percent, followed by

25.2 percent at 70–80 percent efficiency level, 15.4 percent at 50–60 percent efficiency level, 11.3 percent at 60–70 percent efficiency level, and 9.2 percent at 80–90 percent

efficiency level. Only one household has technical efficiency at more than 90 percent

(Table 5.5).

5.4 Discussion

5.4.1 Factors affecting tea productivity

The influence of determinants on tea productivity varies in different models.

Land size is statistically significant and positively impacts tea productivity of both contracted and independent farmers. This result is consistent with conclusions of

Saigenji and Zeller (2009), Ngoc et al. (2014), Hong and Yabe (2015) and Basnayake and Gunaratne (2002). Results imply the importance of land size to tea production involved by small farm households in exporting tea countries like Sri Lanka and

Vietnam.

Pesticide costs have a positive relationship with tea productivity of contracted households. This outcome is consistent with arguments of Saigenji and Zeller (2009) and Ngoc et al. (2014). This result may be explained by advantages of contracted farmers in terms of purchasing pesticides supplied by contractors. Specifically, in a certain contract, production inputs such as fertilizers and pesticides are often supplied to farmers by contractors at a lower price compared to the open market and in some cases, farmers can pay later after harvesting tea. Thus, overwhelming use of pesticides has become a preferable method chosen by contracted households to enhance tea productivity in Vietnam. However, pesticide costs are not significant in the model for

independent households and this result is consistent with conclusions of Hong and Yabe

(2015) and Basnayake and Gunaratne (2002).

Capital is statistically significant and has a positive impact on tea productivity of independent households. This finding is relevant to the conclusions of Saigenji and

Zeller (2009), but Hong and Yabe (2015) argued that capital is not significant. Results present the importance of capital to tea production of independent households who need a larger amount of capital to purchase fertilisers, pesticides, and harvested machines.

Tree density has a negative influence on tea productivity of both contracted and independent farmers. Currently, the number of tea trees planted by farmers in Phu Tho province ranges from 513 to 604 trees per sao, while experts recommend that the density should be 470–500 trees per sao and therefore tea productivity tends to decrease if farmers grow more tea trees in an area.

Family labour is not statistically significant and this result is consistent with conclusions of Hong and Yabe (2015) and Basnayake and Gunaratne (2002). This result implies a negligible role of household labour in boosting tea productivity in Phu Tho province, Vietnam. Instead, tea productivity tends to rely on other resources such as land and capital much more than labour. However, Saigenji and Zeller (2009) and Ngoc et al. (2014) concluded that labour has a positive influence on tea productivity.

Different outcomes can be interpreted by differences in calculating labour units. For instance, our research computes the number of labours in households, while Saigenji and Zeller (2009) calculated the number of days, which labours spend for tea production, and Ngoc et al. (2014) computed labour costs of households for tea production.

The quantity of mixed fertiliser (NPK) is also not statistically significant and this conclusion is consistent with the result of Basnayake and Gunaratne (2002). This result present a negligible role of mixed fertiliser in terms of enhancing tea productivity.

5.4.2 Factors affecting technical efficiency of tea production

Age of household head is statistically significant and has a positive effect on technical inefficiency of tea for both contracted and independent households and this implies that older producers obtain a lower technical efficiency. This result reflects the importance of advanced technologies in tea production rather than experience of growers. The finding is consistent with the conclusions of Dube and Guveya (2014). According to the

Department of Agriculture and Rural Development in Phu Tho province, by 2015, about

71.5 percent of planted area of tea in this province is cultivated by new varieties such as

LDP1, LDP2, Phuc Van Tien, and Kim Tuyen. Further, in the same time, the planted area of safe tea was 3,600 hectares, with 1,640 hectares of planted area certified by the

Rainforest Alliance method of the European Union. Therefore, technological progress has played an important role much more than experience in terms of fostering technical efficiency of tea production.

Gender of household head has a positive relationship with technical inefficiency of tea for contracted farmers and this suggests that technical efficiency of tea of female- headed households is higher than that of male-headed households. Tea is a perennial crop, which requests a sensitive and careful production and harvest and therefore characteristic of women are often consistent with these requirements much more than men. However, Ngoc et al. (2014) claimed that gender is not significant. Different

results may be explained by differences in sample sizes. Specifically, 358 tea farmers in

Phu Tho province are chosen to interview in our study, while the sample size of Ngoc et al. (2014) was 47 farmers in Thai Nguyen and Phu Tho provinces, Vietnam.

Education positively affects technical efficiency of tea for contracted households and this suggests that farmers have a chance to obtain higher technical efficiency if they have a higher education level. Farmers, who have a higher education level, are often chosen to participate in a contract of tea production since they can adopt technological advances based on the requirements given by contractors. However, Ngoc et al. (2014) found that education is not statistically significant.

Number of harvest times has a positive effect on technical efficiency for independent farmers and this implies that producers may gain a higher technical efficiency if they increase the number of harvest times a year. Our results indicate that number of harvest times of tea growers in Phu Tho province is nearly 5.4 times annually. However, according to recommendations of tea scientists, in Northern midlands and mountainous area, Vietnam, farmers can harvest tea from 20 to 30 times a year after tea trees are four years and above and the distance between two closest harvest times varies from seven to ten days.

According to the estimation, the technical efficiency of tea for contracted farmers is 59.9 percent, while independent households have technical efficiency of 55.1 percent. The 4.8 percent higher technical efficiency of contracted farmers can be explained by the advantages to contracted farmers from participating in a contract. For example, contracted producers purchase fertilisers and pesticides with adequate quality and reasonable prices, and receive free technical assistance provided by contractors.

The technical efficiency of tea production varies in different studies. Saigenji and Zeller (2009) showed that the average technical efficiency of tea in Moc Chau district, Son La province, Vietnam for contracted households (with state-owned) was 69 percent, followed by contracted households (with private/cooperative) at 58 percent, and non-contracted households (47 percent). The average technical efficiency of tea in

Tuyen Quang, Thai Nguyen and Phu Tho provinces, Vietnam estimated by Phu and

Nguyen (2014) was 32 percent. The average technical efficiency of tea in the Northern mountainous region, Vietnam calculated by Hong and Yabe (2015) was 89.6 percent.

Basnayake and Gunaratne (2002) indicated that the average technical efficiency of tea in the Mid Country Wet Zone, Sri Lanka was 61.06 percent (for the Cobb-Douglas model) and 83.10 percent (for translog model). Baten et al. (2010) demonstrated that the average technical efficiency of tea in Bangladesh was 59 percent. Dube and Guveya

(2014) concluded that the technical efficiency of tea in Chipinge district, Zimbabwe was

79 percent. The results of technical efficiency of tea in Phu Tho province are lower than in other studies, except for research conducted by Phu and Nguyen (2014). Different results in tea production between studies can be explained by differences in selecting research sites, durations, and variables in the model. For example, Baten et al. (2010) estimated technical efficiency of tea production in Bangladesh using a time-series data over the period 1990_2004, and temperature, rainfall, and Herfindahl_Hinrschman Index are chosen to assess technical efficiency of tea production in this country. However, we assess technical efficiency of tea production in Phu Tho province, Vietnam using a cross-sectional data gathered between March and June 2016. Moreover, age of household head, gender, education, household category, harvest times, and harvest methods are chosen to estimate technical efficiency of tea production.

100

5.5 Chapter conclusions

This chapter examined the effects of factors on tea productivity and technical efficiency of tea production in Phu Tho province. The influence of determinants on tea productivity varied in different models. For instance, land size is statistically significant and positively impacts tea productivity of both contracted and independent farmers because in Phu Tho province, farmers often cultivate tea trees in vast hills and extensive production is preferred to choose by farm households rather than intensive production due to lower initial costs and investment. Hence, tea productivity relies on land size much more than adoption of high technologies in tea cultivation. Pesticide costs have a positive relationship with tea productivity of contracted households due to advantages of contract households in purchasing pesticides provided by contractors. Capital is statistically significant and has a positive impact on tea productivity of independent households because they need a larger amount of capital to purchase fertilisers, pesticides, and harvest machines. Tree density has a negative influence on tea productivity of both contracted and independent farmers because the number of tea trees is planted by farmers in Phu Tho province is higher than the recommendation of tea scientists by 30–100 trees and therefore tea trees do not have enough spaces for development and this leads to a lower productivity. Moreover, family labour and the quantity of mixed fertiliser (NPK) are not statistically significant.

In terms of factors affecting technical efficiency of tea, age of household head is statistically significant and has a positive effect on technical inefficiency of tea for both contracted and independent households and this implies that older producers obtain a lower technical efficiency. Gender has a positive relationship with technical inefficiency

101

of tea for contracted farmers and this suggests that technical efficiency of tea of female- headed households is higher than that of male-headed households. This result can be interpreted by sensitivity and carefulness of women. In addition, results showed that education positively affects technical efficiency of tea for contracted households and this suggests that farmers can obtain higher technical efficiency if they have a higher education level. Clearly, education assists farmers to have a better adoption of new techniques in tea production. Number of harvest times has a positive impact on technical efficiency for independent farmers and this implies that producers may gain a higher technical efficiency if they increase the number of harvest times.

Further, results showed that technical efficiency of contracted farmers is 4.8 percent higher than their independent counterparts. However, the majority of variation in tea productivity of both contracted and independent farmers is dominated by inefficiency effects and therefore reduction of inefficiency effects is one of essential methods to improve tea productivity.

102

CHAPTER 6: EFFECTS OF CONTRACT FARMING ON TEA

PRODUCTIVITY, INCOME AND POVERTY OF FARMERS

6.0 Introduction

This chapter examines the effect of contract farming on tea productivity, income and poverty of tea farmers in Phu Tho province. This is important because the literature review suggested that contract farming improves tea productivity and farmers’ income.

Propensity score matching is used to carry out this analysis with six steps: estimation of determinants affecting contract participation; demonstration of common support region; selection of a matching algorithm; assessment of the match quality; estimation of average treatment effects on the treated (ATT); and implementation of sensitivity analysis. Moreover, a logit model is applied to estimate determinants on poverty of tea producers in Phu Tho province.

6.1 Overview of analytical results

This section provides a brief overview of the analytical results (see chapter 4 on methodology for details).

ATT = E(Y Y / P = 1) (6.1)

1i 0i i Equation 6.1 −(which is the same as equation 4.11 under methodology) is used

for propensity score matching which estimates effects of contract farming on tea

productivity and farmer income in this chapter. Two sets of variables (see details in

Tables 4.4 and 4.5 of chapter 4) are chosen to run propensity score matching for tea

103

productivity and farmer income because factors affecting tea productivity and farmer

income are different. Results in estimation of effects of contract farming on tea

productivity and farmer income are presented in Table 6.9.

Ln = + + + = + Ɛ (6.2) Pi 1−Pi 0 1 1 𝑘𝑘 𝑖𝑖𝑖𝑖 𝑖𝑖 Equation� � 6.2 (which𝛽𝛽 𝛽𝛽 is𝑋𝑋 the same⋯ 𝛽𝛽 as𝑋𝑋 equation𝑋𝑋′ β4.13 under methodology) is used for the Logit model which estimates determinants on poverty of tea farmers in this chapter.

Demographical variables (age and gender) and socio-economic variables (labour, land size, capital, tea productivity, and leaf tea price) are chosen to estimate determinants on poverty of tea farmers because these present potential effects on poverty of tea producers. Results in determinants on poverty of tea farmers are presented in Table

6.11.

6.2 Effects of contract farming on tea productivity and farmer income

6.2.1 Estimation of determinants affecting contract participation (propensity scores estimation)

The logistic model is used to estimate determinants affecting contract participation for tea productivity and farmer income.

104

Table 6.1. Logistic model of factors determining contract participation for tea productivity

Variables Coefficient Standard z P>[z] Marginal Error effect Dependent variable: contract participation (1 for contract participation and 0 for otherwise) Age 0.014 0.013 1.05 0.293 0.002 Gender 1.655*** 0.465 3.56 0.000 0.290*** Household category -1.614*** 0.453 -3.56 0.000 -0.283*** Family labour 0.245* 0.126 1.93 0.053 0.043** Land size for tea 0.018** 0.008 2.13 0.033 0.003** Tea tree age -0.039*** 0.014 -2.69 0.007 -0.007*** Tree density 0.006*** 0.001 5.17 0.000 0.001*** Constant -6.013*** 1.173 -5.12 0.000 Number of observations 358 LR chi2(7) 118.32 Prob > chi2 0.000 Pseudo R2 0.239 Log likelihood -187.72 Source: Author’s calculation, 2018 Note: ***, ** and * mean statistical significance at 1%, 5%, and 10%, respectively

The value of pseudo R squared of 0.239 implies that 23.9 percent of variation in contract participation is explained by independent variables in the model. The p-value of 0.000 and log likelihood of -187.72 reflect the fitness of the model (Table 6.1).

All independent variables are statistically significant, except age of household head. Gender, family labour, land size for tea, and tree density have positive relationships with contract participation, while household category and tea age negatively influence contract participation. Specifically, the likelihood of contract participation of male-headed households is higher than that of female-headed households by 29 percent, ceteris paribus. In farm households in rural areas of Vietnam, major decisions are often implemented by men. Moreover, male-headed households tend to accept risks more than female-headed households (Table 6.1).

105

Family labour and land size are key resources for tea production. Hence, if the labour and land size of a household increase by one person and a sao, then the probability of contract participation rises by 4.3 percent and 0.3 percent, respectively, ceteris paribus. If tree density increases by a tree per sao, then the likelihood of contract participation increases by 0.1 percent, ceteris paribus. This implies that households who grow more tea trees in an area prefer to participate in a contract more than their counterparts (Table 6.1).

The probability of contract participation of non-poor households is higher than that of poor households by 28.3 percent, ceteris paribus. Non-poor households tend to participate in a contract more than poor households since the number of labourers and amount of land and capital of non-poor households are greater than those of their counterparts. If the average age of tea trees increases by a year, then the likelihood of contract participation decreases by 0.7 percent. The average age of tea trees in the survey sample is 21 years. Tree age has a negative relationship with contract participation because growers believe that old tea trees often produce a lower yield

(Table 6.1).

106

Table 6.2. Logistic model of factors determining contract participation for farmer income

Variables Coefficient Standard z P>[z] Marginal Error effect Dependent variable: contract participation (1 for contract participation and 0 for otherwise) Age -0.006 0.013 -0.49 0.621 -0.001 Gender 1.887*** 0.520 3.62 0.000 0.339*** Household category -1.888*** 0.456 -4.14 0.000 -0.339*** Family labour 0.409*** 0.124 3.29 0.001 0.073*** Land size for tea 0.025*** 0.008 2.88 0.004 0.004*** Tea yield -0.017 0.048 -0.35 0.724 -0.003 Price 0.527*** 0.102 5.14 0.000 0.094*** Constant -4.745*** 1.029 -4.61 0.000 Number of observations 358 LR chi2(7) 113.52 Prob > chi2 0.000 Pseudo R2 0.229 Log likelihood -190.12 Source: Author’s calculation, 2018 Note: *** means statistical significance at 1%

The value of pseudo R squared of 0.229 implies that 22.9 percent of variation in contract participation is explained by independent variables in the model. The p-value of 0.000 and log likelihood of -187.72 suggest the fitness of the model (Table 6.2).

All independent variables are statistically significant, except age and tea yield.

Gender, family labour, land size for tea and tea prices have positive relationships with contract participation, while household category negatively affects contract participation. Specifically, the likelihood of contract participation of male-headed households is 33.9 percent higher than that of female-headed households, ceteris paribus. In farm households in rural areas of Vietnam, important activities are often decided by men. In addition, male-headed households tend to accept risks more than female-headed households (Table 6.2).

107

Family labour and land size are key resources for tea production. Hence, if labour increases by one person and land size of households increase by a sao, then the probability of contract participation increases by 7.3 percent and 0.4 percent, respectively, ceteris paribus. If tea prices rise by VND1,000 per kg, then the likelihood of contract participation increases by 9.4 percent, ceteris paribus. This is expected because when participating in a contract, farmers often expect to gain a higher price for tea (Table 6.2).

The probability of contract participation of non-poor households is 33.9 percent higher than poor households, ceteris paribus. Non-poor households tend to participate in a contract more than poor households since non-poor households have more labourers, land and capital than their counterparts (Table 6.2).

6.2.2 Demonstration of the common support region

The purpose of this step is demonstration of the common support region.

Table 6.3. Distribution of estimated propensity scores

Categories Observations Min Mean Max SD 1. For tea productivity Contracted farmers 164 0.062 0.630 0.948 0.229 Independent farmers 194 0.006 0.312 0.950 0.234 Overall 358 0.005 0.450 0.950 0.280 2. For farmer income Contracted farmers 164 0.00982 0.61242 0.99678 0.19164 Independent farmers 194 0.00345 0.32764 0.94611 0.24211 Overall 358 0.00345 0.45810 0.99678 0.26200 Source: Author’s calculation, 2018 Note: SD means standard deviation

For tea productivity, the overall estimated propensity scores are between 0.005 and 0.950. For contracted farmers, the propensity scores vary between 0.062 and 0.948,

108

while the propensity scores range between 0.006 and 0.950 for independent households.

This implies that the region of common support lies between 0.062 and 0.950 (Table

6.3). Therefore, outliers which lie below and above this range need to be dropped. Out of 358 households, 24 observations (1 contracted farmer and 23 independent farmers) were dropped from the analysis since their propensity scores lie outside the region of common support. Thus, 334 observations are used to predict the impact of contract participation on tea productivity for this study.

For farmer income, the overall estimated propensity scores are between 0.00345 and 0.99678. Amongst contracted farmers, the propensity scores vary between 0.00982 and 0.99678, while the propensity scores range between 0.00345 and 0.94611 for independent households. This implies that the region of common support is between

0.00982 and 0.94611 (Table 6.3). Thus, outliers which lie below and above this range need to be dropped. Out of 358 households, 8 observations (7 contracted farmers and 1 independent farmer) were dropped from the analysis since their propensity scores lie outside the region of common support. Hence, 350 observations were used to predict the impact of contract participation on income of farmers.

6.2.3 Selection of a matching algorithm

Several criteria, such as number of explanatory variables with insignificant mean difference between the matched groups of contracted and independent farmers, pseudo

R2 and matched sample size, are tested to choose appropriate matching algorithms. The nearest neighbour matching (NNM) with replacement and Kernel matching (KM) are used to select the most appropriate matching method.

109

Table 6.4. Matching performance of matching methods for tea productivity

Matching methods Number of explanatory Pseudo Matched variables with insignificant R2 sample size mean between the match of contracted and independent farmers Nearest neighbour matching NNM(1) 5 0.033 334 NNM(2) 5 0.029 334 NNM(3) 7 0.018 334 NNM(4) 7 0.011 334 NNM(5) 7 0.007 334 Kernel matching KM(0.1) 6 0.011 334 KM(0.25) 6 0.017 334 KM(0.5) 1 0.067 334 Source: Author’s calculation, 2018 Note: numbers in parentheses denote the number of neighbour estimators in the NNM and bandwidth in the KM

Table 6.5. Matching performance of matching methods for farmer income

Matching methods Number of explanatory Pseudo Matched variables with insignificant R2 sample size mean between the match of contracted and independent farmers Nearest neighbour matching NNM(1) 4 0.042 350 NNM(2) 4 0.031 350 NNM(3) 5 0.026 350 NNM(4) 5 0.026 350 NNM(5) 5 0.023 350 Kernel matching KM(0.1) 7 0.010 350 KM(0.25) 7 0.009 350 KM(0.5) 3 0.052 350 Source: Author’s calculation, 2018 Note: parentheses are the number of neighbour estimators in the NNM and bandwidth in the KM

Matching methods are chosen if these methods have all explanatory variables with insignificant mean between the match of contracted and independent farmers, the lowest pseudo R2, and the largest matched sample size (Tsadik et al., 2015; and Mulatu et al., 2017). 110

In this case, the most appropriate matching method for tea productivity is

NNM(5) because it has all insignificant explanatory variables, the lowest pseudo R2

(0.007), and the largest matched sample size. Thus, NNM(5) is selected to estimate the impacts of contract farming on tea productivity (Table 6.4).

In terms of farmer income, the most appropriate is KM(0.25) because it has all insignificant explanatory variables, the lowest R2 (0.005), and the largest matched sample size. Therefore, only KM(0.25) is selected to assess the impact of contract farming on the income of tea farmers (Table 6.5).

6.2.4 Assessment of the match quality

To examine the match quality of NNM(5) and KM(0.25), means of the propensity score and explanatory variables are estimated and compared before and after matching.

Table 6.6. Propensity scores and covariate balancing for the NNM(5)

Variables Sample Mean % bias reduction T-test Treated Control % bias |bias| t p-value Pscore Unmatched 0.631 0.351 125.3 11.45 0.000 Matched 0.631 0.623 3.7 97.0 0.35 0.729 Age Unmatched 48.399 48.754 -3.7 -0.33 0.739 Matched 48.399 47.666 7.5 -106.0 0.65 0.518 Gender Unmatched 0.957 0.847 37.3 3.39 0.001 Matched 0.957 0.938 6.3 83.1 0.75 0.456 Household category Unmatched 0.036 0.169 -44.6 -4.04 0.000 Matched 0.036 0.039 -0.8 98.2 -0.12 0.908 Family labour Unmatched 3.018 2.526 47.1 4.30 0.000 Matched 3.018 3.078 -5.8 87.8 -0.46 0.642 Land size for tea Unmatched 22.715 17.139 36.1 3.31 0.001 Matched 22.715 22.28 2.8 92.2 0.24 0.807 Tea tree age Unmatched 17.791 22.257 -47.8 -4.37 0.000 Matched 17.791 18.622 -8.9 81.4 -0.86 0.390 Tree density Unmatched 604.36 516.9 83.8 7.63 0.000 Matched 604.36 606.64 -2.2 97.4 -0.18 0.858 Source: Author’s calculation, 2018

111

Table 6.7. Propensity scores and covariate balancing for the KM(0.25)

Variables Sample Mean % bias reduction T-test Treated Control % bias |bias| t p-value Pscore Unmatched 0.594 0.329 125.4 11.49 0.000 Matched 0.594 0.555 18.5 85.2 1.93 0.055 Age Unmatched 48.573 48.865 -3.0 -0.28 0.781 Matched 48.573 48.048 5.4 -79.7 0.48 0.635 Gender Unmatched 0.955 0.792 50.4 4.56 0.000 Matched 0.955 0.951 1.2 97.6 0.16 0.869 Household category Unmatched 0.044 0.227 -55.3 -4.99 0.000 Matched 0.044 0.047 -1.0 98.2 -0.14 0.887 Family labour Unmatched 3.025 2.554 45.9 4.25 0.000 Matched 3.025 2.896 12.5 72.7 0.98 0.326 Land size for tea Unmatched 22.301 16.521 41.7 3.90 0.000 Matched 22.301 20.14 15.6 62.6 1.27 0.206 Tea yield Unmatched 5.241 5.440 -7.4 -0.66 0.509 Matched 5.241 5.104 5.1 31.0 0.54 0.589 Price of tea Unmatched 3.834 3.028 53.5 4.82 0.000 Matched 3.834 3.731 6.8 87.3 0.71 0.478 Source: Author’s calculation, 2018

Before matching, all covariates exhibit statistically significant differences, except age. However, after matching, covariates are balanced. The insignificant likelihood ratio tests of all covariates imply that the matching procedure is able to balance the characteristics in the treated group (contracted farmers) and the control group (independent farmers) (Tables 6.6 and 6.7).

Table 6.8. Quality of matching methods before and after matching

Matching Before matching After matching Total bias methods Pseudo LRchi2 Mean Pseudo LRchi2 Mean reduction R2 (p-value) bias R2 (p-value) bias (%) NNM(5) 0.227 105.00 53.2 0.008 3.65 4.8 50 (0.000) (0.887) KM(0.25) 0.228 109.92 47.8 0.017 7.30 8.3 67 (0.000) (0.504) Source: Author’s calculation, 2018

Low pseudo R2 and the insignificant likelihood ratio tests reflect that both treatment and control groups have the same distribution in covariates after matching,

112

and matching methods can balance characteristics of the treated and control groups

(Chege et al., 2015; and Huluka and Negatu, 2016).

After matching, pseudo R2 dropped from 0.227 to 0.008, the likelihood ratio is insignificant (0.887), and total bias decreased by 50 percent for NNM(5), while for

KM(0.25), pseudo R2 dropped from 0.228 to 0.017, the likelihood ratio is insignificant

(0.504), and total bias decreased by 67 percent (Table 6.8).

6.2.5 Estimation of average treatment effects on the treated (ATT) for tea productivity and farmer income

Table 6.9. Estimation of average treatment effects on the treated for tea productivity and farmer income

Matching Outcome Unmatched Difference Standard T-stat method variables and ATT Error NNM(5) Tea Unmatched 4781.76 902.66 5.30 productivity ATT 3996.92 1272.38 3.14 KM(0.25) Household Unmatched 21668.942 7235.06 2.99 income ATT 16809.027 8103.31 2.07 KM(0.25) Income per Unmatched 5134.811 1666.52 3.08 capita ATT 4174.666 1868.80 2.23 KM(0.25) Income per Unmatched 6689.763 2131.94 3.14 labourer ATT 5003.825 2399.60 2.09 Source: Author’s calculation, 2018

The average treatment effects on the treated value is 3996 which implies that tea productivity of contracted households is higher than that of independent households by

3,996 kg/year (or 3.99 tonnes/year) (Table 6.9).

The income of contracted households is VND16.8 million a year (US$720/year) higher than that of their counterparts. The annual income per capita and income per

113

labourer of contracted farmers are higher than those of independent farmers by VND4.1 million (US$175) and VND5 million (US$214), respectively (Table 6.9).

6.2.6 Implementation of sensitivity analysis

The Rosenbaum bounds (rbounds test) is used to test the presence of hidden bias due to unobserved covariates for NNM(5) and KM(0.25). The results show that the p-critical values of all outcome variables estimated at various level of critical values of gamma are significant, which implies that important covariates affecting the participation and outcome variables are considered. The ATT is insensitive to unobserved selection bias and accordingly, we can conclude that there is a pure effect of contract participation on tea productivity and farmer income (see details in Tables B1 and B2 of the appendix B).

6.3 Influences of determinants on poverty of tea farmers

In Vietnam, poor households in rural areas are households that have average capita income less than or equal to VND700,000 per month (US$31/capita/month), based on criteria for ‘poor’ for the period 2016–2020 in Decision number 59/2015/QD-TTg of the

Prime Minister of Vietnam.

Table 6.10. Number of poor and non-poor households for tea producers surveyed

Household Contracted farmers Independent farmers (n = 164) (n = 194) Number % Number % Poor households 139 84.76 165 85.05 Non-poor households 25 15.24 29 14.95 Total surveyed 164 100.00 194 100.00 Source: Author’s calculation, 2018

114

Table 6.10 presents the number of poor and non-poor households for tea producers. This is necessary to estimate influences of determinants on poverty of tea producers in Phu Tho province. The majority of tea producers surveyed in Phu Tho province are poor, with about 85 percent of both contracted and independent farmers considered poor households (Table 6.10).

Table 6.11. Determinants of poverty of tea producers in Phu Tho province

Variable Coefficient Standard z P-value Marginal Error effects (dy/dx) Age -0.010 0.02 -0.50 0.620 -0.000 Gender -0.309 0.77 -0.40 0.688 -0.024 Labour 0.151 0.19 0.79 0.429 0.012 Land for tea -0.092*** 0.01 -4.97 0.000 -0.007*** Capital 0.000** 0.00 2.16 0.031 5.02e-06** Tea productivity -0.000*** 0.00 -2.74 0.006 -7.83e-06*** Leaf tea price -0.229*** 0.08 -2.59 0.010 -0.018*** Contract -1.169*** 0.45 -2.57 0.010 -0.094*** Constant 6.057*** 1.36 4.42 0.000 Dependent variable: Poverty (1 for poor household and 0 for otherwise) Number of obs. 358 LR chi2 (8) 98.88 Prob > chi2 0.000 Pseudo R2 0.337 Log likelihood -97.12 Source: Author’s calculation, 2018 Notes: numbers in parentheses denote standard errors *** and ** mean statistical significance at 1% and 5%, respectively

Land for tea, tea productivity and tea price have positive impacts on poverty reduction for tea producers in Phu Tho province. However, the probability of poor farmers participating in contract farming is lower than that of non-poor households and this implies that contract farming of tea production has played a less importance in terms of improving the livelihood for the poor in Phu Tho province (Table 6.11).

6.4 Discussion

115

6.4.1 Effects of contract farming on tea productivity

There are a few studies on influences of contract farming on tea productivity. Saigenji and Zeller (2009) concluded that contracted households have higher technical efficiency of tea than non-contracted ones, but the authors did not provide specific evidence on the impact of contract farming on tea productivity. Ngoc et al. (2014) argued that there is no statistically significant difference in technical coefficients for both contracted and independent farmers in tea production. The results here for Phu Tho province demonstrated that the difference in tea productivity between contracted and independent households is 3.9 tonnes a year.

6.4.2 Influences of contract farming on income of farmers

Income of contracted households is VND16.8 million a year (US$720/year) higher than their counterparts. The annual income per capita and income per labourer of contracted farmers are higher than those of independent farmers by VND4.1 million (US$175) and

VND5 million (US$214), respectively. In 2016, the average income per capita in

Vietnam and Phu Tho province accounted for US$2,215 and US$1,365, respectively

(General Statistics Office of Viet Nam, 2017; and Phu Tho Statistics Office, 2017). The difference in the annual income per capita between contracted and independent households is US$175 and this is equivalent to 7.9 percent compared to the average income per capita of the whole country and 12.8 percent compared to the average income per capita of Phu Tho province in 2016. Therefore, due to small differences in

116

income between the two farmer groups, contract farming has play a negligible role in improving the income of farm households.

Previous studies demonstrated that contract farming has a positive impact on the income of tea farmers, but some authors did not provide evidence to support this assessment. For example, Oanh et al. (2016) concluded that worker tea farmers in Phu

Tho province obtained higher turnover and value added compared to partial contracted and independent households. Manjunatha et al. (2016) argued that contract farming in

India increases the income of contracted households because it improves market access and provides better use of resources and better management of technology. In contrast,

Ngoc et al. (2014) claimed that contract farming was not an effective instrument to increase farmers’ income in tea production in Thai Nguyen and Phu Tho provinces,

Vietnam.

Saigenji and Zeller (2009) studied contract farming of tea production in Son La province, Vietnam and found that the income of contracted households is higher than that of non-contracted households by US$128 a year. Thus, the results in income differences between contracted and independent households in Phu Tho province are higher than those found in Saigenji and Zeller (2009). Possible reasons for this include different technical efficiency of tea production and income variations. Specifically, our results address that technical efficiency of tea production for contracted and independent households are 59.9 percent and 55.1 percent, respectively. Saigenji and

Zeller (2009) concluded that contracted farmers gain technical efficiency of tea production at 58 percent, and 47 percent for non-contracted households. Further, income of tea farmers is different due to effects of inflation rates and variations of fertilisers, pesticides, and tea prices in different periods.

117

6.4.3 Effects of determinants on poverty of tea farmers

Land size for tea plantation, tea productivity and leaf tea price contribute to poverty reduction in Phu Tho province. The results on the effects of land size on poverty reduction are consistent with Rotich et al. (2017a) on tea farmers in Konoin sub-county,

Kenya. However, age of household heads is not significant in Phu Tho province, while

Rotich et al. (2017b) argued that age had a negative relationship with poverty gap of tea farmers in Konoin sub-county, Kenya and this implies that older farmers have a lower poverty. The reason is older household heads have a better access to land and resources compared to younger household heads who are more likely affected by land fragmentation. In addition, Rotich et al. (2017b) found that gender is not significant in

Konoin sub-county, Kenya and this is consistent with our findings. Thus, policies implemented in Phu Tho province should focus on solving bottlenecks in land for tea cultivation and improving tea productivity as well as boosting leaf tea price to achieve poverty reduction.

6.5 Chapter conclusions

This chapter evaluated the impacts of contract farming on tea productivity, farmers’ income and poverty of tea farmers in Phu Tho province. The difference in tea productivity between contracted and independent households is 3.9 tonnes a year because of a higher technical efficiency of tea production and the advantages of contracted households compared to their counterparts in accessing adequate inputs like fertilisers and pesticides and technical assistance provided by contractors.

118

The income of contracted households is higher than that of their counterparts by

VND16.8 million a year (US$720/year). The annual income per capita and income per labourer of contracted farmers are higher than those of independent farmers by VND4.1 million (US$175) and VND5 million (US$214), respectively. The difference in the annual income per capita between contracted and independent households is equivalent to 7.9 percent compared to the average income per capita of the whole country and 12.8 percent compared to the average income per capita of Phu Tho province in 2016.

Therefore, due to small differences in income between the two farmer groups, contract farming has played a negligible role in improving the income of farm households as well as reducing poverty for tea farmers in Phu Tho province.

Finally, policies implemented by Phu Tho province should focus on solving challenges in land for tea cultivation and improving tea productivity as well as boosting leaf tea price to achieve poverty reduction.

119

CHAPTER 7: THE RELATIONSHIP BETWEEN ECONOMIC GROWTH, TEA

EXPORTS AND POVERTY IN VIETNAM

7.0 Introduction

This chapter attempts to examine the causal relationship between economic growth, tea exports and poverty in Vietnam for the last four decades (1977–2016) by using the vector autoregressive (VAR) model. This is addressing the research objective of examining the causal relationship between economic growth, tea exports and poverty in

Vietnam.

7.1 Overview of analytical results

This section provides a brief overview of the analytical results (see chapter 4 on methodology for details).

= + + + (7.1)

𝑡𝑡 1 𝑡𝑡−1 𝑝𝑝 𝑡𝑡−𝑝𝑝 𝑡𝑡 Equation𝑌𝑌 𝐴𝐴 𝑌𝑌 7.1 (which⋯ 𝐴𝐴 is 𝑌𝑌the sameƐ as equation 4.14 under methodology) is used for the VAR model which examines the relationship between economic growth, tea exports and poverty in Vietnam in this chapter. Results are presented in Table C1 of the appendix C.

7.2 Economic growth, tea exports and poverty in Vietnam: An overview

120

Table 7.1. Characteristics of economic growth, tea exports and poverty in Vietnam (1977_2016)

Items Unit Mean SD Min Max GDP million US$ 50012.76 61155.66 1884.65 205276.2 Export value of tea thousand US$ 64800.68 61518.14 7400 224847 Proportion of the % 0.12 0.23 0 0.78 population living below the poverty line Poverty gap % 0.04 0.09 0 0.36 Source: Author’s calculation, 2018 Notes: SD denotes standard deviation Poverty gap is a mean distance below the poverty line, which provides an estimate of the amount of resources needed to eliminate poverty in a given social group.

The average annual GDP of Vietnam was about US$50 billion over the four decades

(1977–2016), while the value of tea exported from Vietnam was US$64.8 million a year, on average, for the same period. In 2017, Vietnam is ranked as the seven-largest producer and the fifth-largest exporter of tea in the world. According to the Ministry of

Agriculture and Rural Development of Vietnam, by the end of 2017, the export volume of tea was 140,000 tonnes, turnover of tea was US$229 million and the price of exported tea was US$1,659 per tonne, on average. About 80 percent of Vietnamese tea is exported to the international market. On average, 12 percent of the population lived in a household below the poverty line, and the poverty gap was 4 percent (Table 7.1).

7.3 The relationship between economic growth, tea exports and poverty in Vietnam

This section investigates the causal relationship between economic growth, tea exports and poverty in Vietnam between 1977 and 2016 by using the VAR model. The procedure of a VAR model includes six steps: (1) performing the unit root test; (2)

121

determining lag length; (3) estimating the VAR model; (4) testing the Granger causality;

(5) checking the stability of eigenvalues; and (6) implementing the Johansen test for co- integration.

7.3.1 Implementation of the unit root test

The unit root test is performed to examine the stationarity of the time series variables

(Adeola and Ikpesu, 2016; and Shadab, 2018). In this research, the Augmented Dickey-

Fuller test is used to check the stationarity of GDP, export value of tea, proportion of the population living below the poverty line and poverty gap with the hypothesis as follows:

Null hypothesis (H0): The variables contain a unit root

Alternative hypothesis (Ha): The variables do not contain a unit root

122

Table 7.2. Augmented Dickey-Fuller test for the unit root

Variables Level 1st difference 2nd difference LnGDP T-statistic: 0.35 T-statistic: -0.00 T-statistic: 0.31 P-value: 0.97 P-value: 0.95 P-value: 0.97 Critical values: Critical values: Critical values: 1% level: -3.65 1% level: -3.66 1% level: -3.66 5% level: -2.96 5% level: -2.96 5% level: -2.96 10% level: -2.61 10% level: -2.61 10% level: -2.61 LnExport value of tea T-statistic: -0.86 T-statistic: -0.86 T-statistic: -0.91 P-value: 0.79 P-value: 0.80 P-value: 0.78 Critical values: Critical values: Critical values: 1% level: -3.65 1% level: -3.66 1% level: -3.66 5% level: -2.96 5% level: -2.96 5% level: -2.96 10% level: -2.61 10% level: -2.61 10% level: -2.61 LnProportion of the T-statistic: -3.40 T-statistic: -2.49 T-statistic: -2.13 population living below P-value: 0.01 P-value: 0.11 P-value: 0.23 the poverty line Critical values: Critical values: Critical values: 1% level: -3.65 1% level: -3.66 1% level: -3.66 5% level: -2.96 5% level: -2.96 5% level: -2.96 10% level: -2.61 10% level: -2.61 10% level: -2.61 LnPoverty gap T-statistic: -4.25 T-statistic: -2.98 T-statistic: -1.97 P-value: 0.00 P-value: 0.03 P-value: 0.29 Critical values: Critical values: Critical values: 1% level: -3.65 1% level: -3.66 1% level: -3.66 5% level: -2.96 5% level: -2.96 5% level: -2.96 10% level: -2.61 10% level: -2.61 10% level: -2.61 Source: Author’s calculation, 2018

Results show the null hypothesis cannot be rejected because p-values of all variables are greater than critical values at 1%, 5%, and 10%, respectively and these suggest that variables exhibit a unit root (Table 7.2).

7.3.2 Determination of the lag length

The objective of this step is to identify the optimal lag for the VAR model. If the lag used is too short, then the residual of the regression will not show the white noise

123

process and as the result, the actual error would not be accurately estimated by the model (Suharsono et al., 2017).

Table 7.3. Selection of the lag length

Lag LL LR df p FPE AIC HQIC SBIC 0 -134.69 0.02 7.70 7.76 7.88 1 -23.41 222.57 16 0.000 0.00 2.41 2.71 3.29 2 14.32 75.48 16 0.000 0.00 1.20 1.75 2.78 3 63.75 98.85 16 0.000 7.0e-06 -0.65 0.14 1.63* 4 92.27 57.04* 16 0.000 4.2e-06* -1.34* -0.30* 1.64 Endogenous: LnGDP, LnExport value of tea, LnProportion of the population living below the poverty line, LnPoverty gap Exogenous: Constant Number of observations = 36 Source: Author’s calculation, 2018 Notes: * denotes lag order selected by the criterion; LL is log likelihood values; LR is sequential modified LR test statistics; FPE is final prediction error; AIC is Akaike information criterion; SC is Schwarz information criterion; HQIC is Hannan-Quinn information criterion; and SBIC is Schwarz’s Bayesian information criterion.

As seen in Table 7.3, results suggest that the optimal lag length in this case is four lags because this value is recommended by FPE, AIC and HQIC indicators, while three lags is only recommended by the SBIC indicator. Therefore, a value of four lags was chosen to run the VAR model in the next step.

7.3.3 Estimation of the VAR model

The export value of tea negatively affects GDP, meaning that if the export value of tea increases, then GDP decreases. Results showed that GDP and poverty gap have negative impacts on proportion of the population living below the poverty line, and proportion of the population living below the poverty line has a positive relationship with poverty gap

(see details in Table C1 of the Appendix C).

124

7.3.4 Testing the Granger causality

The purpose of the Granger causality is to evaluate the predictive capacity of a single variable on other variables (Musunuru, 2017). This approach can be used to test the non-zero correlation between the error processes of the cause and effect variables

(Gudeta et al., 2017). In this study, hypotheses are tested as follows:

Testing the relationship between GDP and other variables:

Null hypothesis (H0): GDP does not cause export value of tea, proportion of the population living below the poverty line and poverty gap

Alternative hypothesis (Ha): GDP causes export value of tea, proportion of the population living below the poverty line and poverty gap

Testing the relationship between export value of tea and other variables:

Null hypothesis (H0): Export value of tea does not cause GDP, proportion of the population living below the poverty line and poverty gap

Alternative hypothesis (Ha): Export value of tea causes GDP, proportion of the population living below the poverty line and poverty gap

Testing the relationship between proportion of the population living below the poverty line and other variables:

Null hypothesis (H0): Proportion of the population living below the poverty line does not cause GDP, export value of tea and poverty gap

Alternative hypothesis (Ha): Proportion of the population living below the poverty line causes GDP, export value of tea and poverty gap

Testing the relationship between poverty gap and other variables:

125

Null hypothesis (H0): Poverty gap does not cause GDP, export value of tea and proportion of the population living below the poverty line

Alternative hypothesis (Ha): Poverty gap causes GDP, export value of tea and proportion of the population living below the poverty line

Table 7.4. Results of the Granger causality Wald test

Directional relationship Probability Conclusion GDP Tea export 0.28 > 0.05 Accept H0 GDP Proportion of poverty line 0.93 > 0.05 Accept H0 GDP Poverty gap 0.88 > 0.05 Accept H0 Tea export GDP 0.21 > 0.05 Accept H0 Tea export Proportion of poverty line 0.29 > 0.05 Accept H0 Tea export Poverty gap 0.36 > 0.05 Accept H0 Proportion of poverty line GDP 0.47 > 0.05 Accept H0 Proportion of poverty line Tea export 0.98 > 0.05 Accept H0 Proportion of poverty line Poverty gap 0.01 < 0.05 Reject H0 Poverty gap GDP 0.51 > 0.05 Accept H0 Poverty gap Tea export 0.98 > 0.05 Accept H0 Poverty gap Proportion of poverty line 0.00 < 0.05 Reject H0 Source: Author’s calculation, 2018

There is a directional relationship running from the proportion of population living below the poverty line to the poverty gap and vice versa, from the poverty gap to the proportion of population living below the poverty line (Table 7.4).

7.3.5 Examination of eigenvalue stability

The goal of this analysis is to check the stability of the eigenvalues in the VAR model.

All the eigenvalues lie inside the unit circle and it is concluded that the VAR model satisfies the stability condition (Figure 7.1).

126

Figure 7.1. Eigenvalue stability condition

Roots of the companion matrix 1 .5 0 Imaginary -.5 -1

-1 -.5 0 .5 1 Real

Source: Author’s calculation, 2018

7.3.6 Performance of the Johansen co-integration test

The Johansen co-integration test is performed to examine the long-run relationship between variables. If variables are co-integrated, it suggests that there is a long-term relationship between variables (Musunuru, 2017; and Gudeta et al., 2017).

The hypothesis to be tested is as follows:

Null hypothesis (H0): There is no co-integration among variables

Alternative hypothesis (Ha): There is co-integration among variables

127

In this study, the Johansen co-integration test is carried out by trace statistic test.

Trace test is a likelihood-ratio-type test, which operates under different assumptions in the deterministic part of the data generation process (Lutkepohl et al., 2001).

Table 7.5. Results of trace statistic in the Johansen co-integration test

Maximum rank LL Eigenvalue Trace 5% critical 1% critical statistic value value 0 52.71 79.13 47.21 54.46 1 80.69 0.788 23.16*1*5 29.68 35.65 2 87.48 0.314 9.57 15.41 20.04 3 92.18 0.229 0.18 3.76 6.65 4 92.27 0.005 Source: Author’s calculation, 2018 Notes: *1 and *5 denote the number of co-integrations (ranks) chosen to accept the null hypothesis at 1% and 5% critical values, respectively

As seen in Table 7.5, the null hypothesis cannot be rejected in the rank one (one co-integration) because the trace statistic is less than both the 1% critical value (23.16 <

35.65) and 5% critical value (23.16 < 29.68), suggesting there is a co-integration among variables (Table 7.5).

7.4 Discussion

In terms of the relationship between economic growth, tea exports and poverty in the whole of Vietnam, the export value of tea negatively affects GDP in the short run.

Results showed that GDP and poverty gap have negative influences on the proportion of the population living below the poverty line, and the proportion of the population living below the poverty line has a positive relationship with poverty gap. Further, there is a causal relationship between GDP, export value of tea, proportion of the population living below the poverty line, and poverty gap of Vietnam in the long run. According to

128

the Ministry of Industry and Trade of Vietnam, in 2017, the total value of exports was

US$214 billion, of which the export value of tea was US$229 million. Tea only contributes about 0.1 percent to the total export value of Vietnam, suggesting that the economic growth of Vietnam relies on other exported agricultural commodities much more than tea products. For example, in the first eight months of 2018, the export value of timber products was US$5.6 billion, followed by aquaculture (US$5.5 billion), vegetables and fruits (US$2.7 billion), coffee (US$2.5 billion), and rice and cashew nut

(US$2.2 billion each). Hence, the export value of tea has a negative relationship with economic growth of Vietnam in the short run. GDP and poverty gap have negative impacts on the proportion of the population living below the poverty line, and these results reflect that Vietnam has reduced poverty in recent years. The results reflect that if GDP increases, then the income of households will increase and consequently, the proportion of households with consumption or income per capita below the poverty line decreases. Moreover, if the poverty gap is wider, then the proportion of the population below the poverty line is deeper and this implies that the proportion of households with consumption or income per capita below the poverty line declines. By contrast, if the proportion of households with consumption or income per capita below the poverty line increases, then the mean distance below the poverty line also increases and this reflects a positive relationship between the proportion of the population below the poverty line and poverty gap.

There is no relationship between the export value of tea and poverty reduction in

Vietnam. In contrast, Balat et al. (2009) argued that trade positively affects poverty reduction in Uganda because it assists farmers to access markets of developed countries where have high prices for exported agricultural products such as coffee, tea, cotton,

129

and fruits. However, export opportunities may not be realizable if complementary domestic factors are not available. The findings indicated that tea exports had a negative impact on economic growth in the short run, while Oyakhilomen and Zibah (2014) found that agricultural production positively affected economic growth in Nigeria.

Different results may be explained by different contributions of tea export to economic growth in Vietnam and agricultural production to economic growth in Nigeria. For instance, in 2017, the export value of tea only contributed about 0.1 percent to the total export value of Vietnam. By contrast, in 2011, about 40 percent of Nigerian GDP came from the agricultural sector. Further, the results showed that economic growth causes poverty reduction in the short run in Vietnam, while Nindi and Odhiambo (2015) concluded that economic growth does not cause poverty reduction in both the short run and long run in Swaziland. Due to the high level of income inequality in Swaziland, not surprisingly, economic growth alone is unable to achieve the target in poverty reduction.

7.5 Chapter conclusions

This chapter investigated the causal relationship between economic growth, tea exports and poverty in Vietnam. The export value of tea negatively affects GDP in the short run.

Results showed that GDP and the poverty gap have negative impacts on the proportion of the population below the poverty line. The proportion of the population below the poverty line also has a positive relationship with the poverty gap. Further, there is a directional relationship running from the proportion of population living under the poverty line to the poverty gap and vice versa.

130

These results show that economic growth has made an important contribution to poverty reduction in Vietnam. However, tea exports have had a negligible effect on economic growth and poverty reduction in Vietnam. Therefore, if the government promotes tea exports as a tool of economic growth and poverty reduction in Vietnam, then it has not been as successful as their expectations.

131

CHAPTER 8: CONCLUSIONS AND POLICY IMPLICATIONS

This chapter summarises the outcomes of the research and also recommends policies to the Government of Vietnam and Phu Tho province in order to encourage and support contract farming and achieve the target in poverty reduction.

8.1 Conclusions

The research investigated factors affecting tea production in Phu Tho province; estimated the impacts of contract farming on tea productivity, farmers’ income and poverty of tea farmers in Phu Tho province; and investigated the relationship between economic growth, tea exports and poverty in Vietnam.

In order to explore the research objectives, the quantitative approach has been employed to analyse cross-sectional and time series data. Specifically, the stochastic frontier model is used to estimate technical efficiency of tea production in Phu Tho province. The propensity score matching is employed to assess impacts of contract participation on tea productivity and farmer’s income in Phu Tho province. The logit model is used to investigate determinants affecting poverty of tea farmers, and the vector autoregressive model is applied to examine the relationship between economic growth, tea exports, and poverty in Vietnam.

According to the estimation, technical efficiency of tea for contracted farmers is higher than that of their counterparts by 4.8 percent. The results also demonstrated that contracted households produce 3.9 tonnes of tea a year more than independent households because of the better technical efficiency and the advantages in accessing

132

adequate inputs like fertilisers and pesticides and technical assistance provided by contractors. In Phu Tho province, tea productivity of farmers is dominated by inefficiency effects and therefore reduction of inefficiency effects plays an important role in fostering tea productivity.

The income of contracted households is VND16.8 million a year (US$720/year) higher than independent households. The annual income per capita and income per labourer of contracted farmers are higher than those of independent farmers by VND4.1 million (US$175) and VND5 million (US$214), respectively. However, due to small differences in income between the two farmer groups, contract farming has been negligible in reducing poverty for tea farmers in Phu Tho province.

Results support the importance of land size, tea productivity and leaf tea price for poverty reduction in this province. Thus, policies carried out by Phu Tho province should concentrate on solving challenges in land for tea cultivation, improving tea productivity as well as enhancing leaf tea price.

Furthermore, the export value of tea negatively affects GDP in the short run.

Results also indicated that GDP and the poverty gap have negative impacts on the proportion of the population living below the poverty line and this proportion has a positive relationship with the poverty gap. In addition, there is a causal relationship between GDP, export value of tea, proportion of the population living below the poverty line and the poverty gap in Vietnam in the long run. These results show the important role of economic growth in reducing poverty in Vietnam. However, tea exports have played a much less important role in economic growth and poverty reduction in

Vietnam because tea exports account for a very small proportion of the total value of exports and therefore the economic growth of Vietnam has relied on other agricultural

133

products such as aquaculture, vegetables and fruits, coffee, rice and cashew nut rather than tea.

The research has contributions to knowledge and empirical aspects. First, it provides information on factors affecting tea productivity and technical efficiency of tea production in Phu Tho province under contract regimes. Second, by exploring in-depth the effect of contract farming on tea productivity and incomes of farmers as well as assessing determinants of poverty of tea producers in this region, it contributes to knowledge for improved policy development and delivery in economic development, specifically the design and effectiveness of the Vietnamese government’s NTPHEPR in

Phu Tho province. Lastly, it re-examines the role of tea exports in economic growth and poverty reduction in Vietnam by investigating the relationship between tea exports, economic growth, and poverty between 1977 and 2016 using a time-series data.

8.2 Policy implications

Policies are recommended to the Government, Phu Tho province, enterprises and tea growers to increase contract farming and achieve the target in poverty reduction.

8.2.1 The Government

First, the Land Law of 2013 should be implemented to encourage parcel consolidation, concentration and accumulation of agricultural land. For example, the State promotes allocation of agricultural land to households and individuals to use for the long term.

Formalities in land lease and land transformation need to be reduced to create a

134

transparent market for land transactions. For instance, the State allows farm households to manage and use agricultural land for up to 50 years and as a result, farmers are able to invest in their production for the long term, especially in tea production. According to the survey conducted by the Vietnam Business Council for Sustainable Development

(VBCSD) in 2014, 43.28 percent of tea households had problems of a shortage of land and 35.7 percent of respondents faced problems of soil degradation and erosion

(VBCSD, 2015). Further, our results also address that land size has a positive relationship with tea productivity for both contracted and independent farmers in Phu

Tho province. Based on results, policies implemented by the State should focus on solving issues in parcel consolidation, concentration and accumulation of agricultural land to assist tea farmers in terms of enhancing tea productivity.

Second, investment in science and technology in agriculture should be promoted. Leaf tea production is associated with certification in sustainable development and food safety such as Vietnamese Good Agricultural Practice, UTZ and

Rainforest Alliance to improve the yield, quality and competitive capacity of tea products. For instance, in Vietnam, 41.03 percent of tea companies said their existing technologies were worse compared to the world level and 8.57 percent of respondents reported they had challenges in accessing new foreign technologies (VBCSD, 2015).

For tea processing at the household level, traditional and modern methods for tea processing are integrated to make high quality products. In 2014, 50.85 percent of tea households in the VBCSD survey claimed they had difficulties in production machines and equipment (VBCSD, 2015). In addition, processing controls also need to be enforced to guide applications of standard systems to meet requirements in hygiene and food safety. Households should be helped to assure product quality with strict penalties

135

imposed on processors who produce and sell either unsafe or poor quality products. Our results indicate that education positively affects technical efficiency of tea production for contracted households, while harvest times has a positive impact on technical efficiency of tea production for independent farmers. Based on results, training programs in agricultural extension should be facilitated to educate tea farmers in terms of improving technical efficiency of tea production.

Third, regulations in the agreement should be clarified specifically by the State.

In Vietnam, the agreement relation between parties is regulated in the Law of

Commerce in 2005 and the Civil Code Law in 2015. However, these regulations are generous and very hard to apply in tea production since agricultural production often has a longer cycle than industry and services, and it has a higher risk due to a heavy dependence on the weather. For example, articles 52 and 53 in the Law of Commerce in

2005 refer to price determination, but these do not demonstrate specific methods for determining prices in the agreement. Similarly, the Civil Code Law in 2015 specifies the agreement in terms of civil transactions and therefore it does not clarify rights and duties of each party in the agreement. Consequently, implementation of these regulations in the reality is difficult for parties. Thus, regulations in the agreement should be specifically regulated and clarified by the State in order to enhance success of the agreement.

Fourth, national planning in the tea sector should be modified to meet requirements in tea production, processing and marketing. Before implementing the

Decision number 124/QD-TTg of the Prime Minister on the plan of development of the agricultural sector toward 2020 and vision toward 2030 issued on 2 February 2012, planning in the tea sector often focused only on production and this led to an imbalance

136

in producing, processing and marketing tea products. Instead, planning in the tea sector needs to recognise the commodity approach. Tea production zones should be designed and constructed in association with processing plants and markets to establish value chains that connect producers, processors and marketers.

Fifth, the State should assist enterprises and farmers in searching for markets for tea products. For example, in 2014, nearly 59 percent of tea households in the VBCSD survey said they had difficulties in selling products (VBCSD, 2015). Thus, tea production zones should be established in association with processing plants and export enterprises. Moreover, distribution and marketing networks for tea products need to be re-organised and completed to establish supply chains for the tea sector. Market reports and bulletins on tea as a commodity should be supplied to both tea enterprises and farmers to provide current market information on tea products at both domestic and international levels. Lastly, the capacity of tea farmers to access markets may be facilitated through implementing training courses in agricultural extension and marketing.

Finally, impacts of tea exports on economic growth and poverty reduction in the whole country should be carefully re-examined. Our results demonstrate that tea exports have had negligible effects on economic growth and poverty reduction in Vietnam over the period 1977_2016. Actually, in 2017, the export value of tea only contributed about

0.1 percent to the total export value of Vietnam suggesting that the economic growth of this country relies on other exported agricultural commodities such as timber products, aquaculture, vegetables and fruits, coffee, rice, and cashew nut much more than tea products. Clearly, these agricultural products present comparative advantages in terms of boosting the economic growth compared to tea production. However, tea production

137

should be promoted in locals, where have adequate climate and soils like Phu Tho province since this crop creates employment and improve income for farm households in the rural communities.

8.2.2 Phu Tho province

First, Phu Tho province should pay more attention to the role of tea production in the program of poverty reduction because tea production generates employment and improves income for small farm households in rural areas of this province. Land for tea cultivation, tea productivity and leaf tea price have positive influences on poverty reduction for tea farmers. Therefore, consistent policies implemented by the province should focus on dealing with challenges in land size for tea cultivation, improving tea productivity as well as boosting leaf tea prices to reduce poverty for tea producers.

Second, encouragement of domestic and foreign enterprises to invest more in tea production, processing and exportation in order to facilitate coordination between businesses and farmers in tea production. Policies in land and taxes should be transparent along with improving infrastructure such as electrical and irrigation systems and rural roads. Tea enterprises need much more land to build up production and processing zones for tea products. However, businesses often face problems in administrative formalities at the provincial level because their projects need to be approved by the Department of Planning and Investment, and the People’s Committee of Province.

Lastly, negotiation capacity of farmers in the contract of tea production should be facilitated to improve their benefits in the tea value chain. In terms of involving

138

transactions with contractors, farmers often face challenges because of their constraints in knowledge and negotiation capacity. In addition, our research indicates that most of tea households in Phu Tho province are small farmers who have cultivated tea in the planted area of 0.7 hectares, on average. Scatter and small-scale production of tea farmers along with a negligible role of the farmer associations lead to their weaker negotiations in terms of bargaining the contract and as a result, farmers often must be accepted requirements given by contractors. Hence, the voice of tea farmers may be improved by consolidating the role of farmer associations and farmer groups.

8.2.3 Enterprises and tea farmers

Enterprises and farmers are key actors in contract farming of tea production in Phu Tho province. Thus, these stakeholders determine either the success or failure of contract farming in tea production. Clearly, the most important theme is each party needs to recognise the benefits of participating in contract farming of tea production.

First, both firms and producers need to clearly specify the relationship in the contract. To reduce risk and uncertainties in contract farming, the relationship between the two parties should be a partnership relationship rather than either a competitive relationship or exploitative relationship. A partnership relationship helps ensure truth between parties in contract farming. Furthermore, this relationship also encourages establishing a win-win model in contract farming because it reduces breach of the contract, which occurs when a party pulls benefits from and pushes risk and uncertainties to another party.

139

Second, the contents of the contract should be constructed through discussions between enterprises and tea growers. The contract needs to clarify the rights and duties of each party in the contract. In reality, the contents of a contract between firms and tea farmers are often simple and, as a consequence, it is very hard to demonstrate the exact rights and responsibilities of each party and resolve dispute and conflict between two parties if these occur. To deal with this issue, the contract should include ten main elements as follows: (1) duration of the contract, specified from signed date to delivery of products and payment; (2) product quantity, indicating volume of products and measured units such as kilograms, tonnes and so on; (3) product quality, indicating colours, shapes, tannin content and so on of products; (4) prices, specifying the manner of price computation, price negotiation and so on; (5) location of cultivation, indicating places where farmers grow tea; (6) production techniques, indicating varieties, land preparation, irrigation, fertiliser and pesticide application, and harvesting used for tea production; (7) payments, indicating mean, duration and location of payments); (8) insurance, specifying insurance levels for tea production; (9) risk and uncertainty sharing, specifying the responsibilities of each party when risk and uncertainty such as natural disasters, diseases, market changes and price changes occur in tea production; and (10) referee terms, indicating methods and legal systems applied to resolve dispute and conflict between parties in the contract.

Third, enterprises should sign the contract with farmers through intermediate organisations such as farmer associations and cooperatives. This means there are three signatures on a contract: the firm, the farmer and the third party. This approach assists enterprises to reduce their transaction costs compared to making an individual contract with every household. Moreover, the presence of intermediate parties encourages

140

farmers to consider their duties in the contract since farmers are members of these organisations.

Fourth, the contents of a contract should be consistent with the conditions and nature of the contracting parties. For example, enterprises write elements in a contract and they often have requirements relating to land, capital and production techniques to select appropriate producers. However, most tea producers are small farmers and therefore it is very difficult for them to meet these strict conditions. This means that farm households who do not have the minimum land size and capital will not be chosen to participate in a contract. This can lead to social conflict in rural communities and adverse selection because experienced farmers may be overlooked in this selection process.

Fifth, a tea farming information system should be adopted since it can improve productivity, service delivery and profitability, and facilitate communication between the processed factory and tea growers (Munene and Kasamani, 2017).

Lastly, it is noted that contract farming of tea production is not appropriate for all communities and tea farmers. Contracts should recognise the natural and socio- economic characteristics of local communities. In addition, some farmers are unable to sign a contract because they do not meet requirements specified by contractors. Contract farming assists tea farmers to access international markets and improve specialisation in production in a transitioning economy like Vietnam. While the economic effects have been shown in the positive relationship between contract farming and tea productivity and income of farmers, the social and environmental effects of contract farming have been ignored in contract farming research. Therefore, it is important to carry out further

141

research on the social and environmental effects of contract farming in tea production in

Vietnam.

8.3 Research limitations and direction for further studies

The research has some limitations. First, tea is an industrial crop which is mainly cultivated in three areas of Vietnam: the Northern midlands and mountainous area,

North Central and Central coastal area, and Central Highlands. Hence, research outcomes could make more contribution to policy implications if study sites included all three areas, instead of focusing only on one province. Second, policies related to the influences of contract farming on tea productivity and farmers’ income and effects of tea production on poverty reduction are most appropriate for areas which have similar natural and socio-economic characteristics as Phu Tho province. Third, due to constraints in time, human and financial resources, primary data for this research is gathered from a cross sectional survey in Phu Tho province, between March and June

2016 and as a consequence, it is very difficult to assess impacts of contract farming on tea productivity and farmers’ income along with time variation. Lastly, the research faces challenges in constructing policies in contract farming of tea production due to the shortage of official documentations in contract farming of tea production in Vietnam, especially in the provincial level. Therefore, further research should be carried out to address these limitations.

142

REFERENCES

ActionAid (2015). Contract farming and out-grower schemes: Appropriate development

models to tackle poverty and hunger? Policy Discussion Paper, March 2015.

Adekanmbi, D. B. (2017). Comparison of probit and logit models for binary response

variable with applications to birth data in South-Western, Nigeria. American

Journal of Mathematics and Statistics, 7(5), 199-208.

Adeola, O., and Ikpesu, F. (2016). An empirical investigation of the impact of bank

lending on agricultural output in Nigeria: A vector autoregressive (VAR)

approach. The Journal of Developing Areas, 50(6), 89-103.

Ajao, A. O., and Oyedele, G. A. (2013). Economic efficiency of contract farming in

Oyo state: Experience from British American Tobacco Company. International

Journal of AgriScience, 3(9), 699-707.

Ajibefun, I. A., Daramola, A. G., and Falusi, A. O. (2006). Technical efficiency of small

scale farmers: An application of the stochastic frontier production function to

rural and urban farmers in Ondo State, Nigeria. International Economic Journal,

20(1), 87-107.

Alchian, A. A. (1965). Some economics of property rights. II Politico, 30(4), 816-829.

Ali, A., and Abdulai, A. (2010). The adoption of genetically modified cotton and

poverty reduction in Pakistan. Journal of Agricultural Economics, 61(1), 175-

192.

Ali, A., Sharif, M., Mahmood, K., and Akmal, N. (2013). Determinants of cherry

production and marketing in Pakistan: A propensity score matching approach.

Agricultural Economics Review, 14(1), 45-58.

143

Anderson, T. L., and Hill, P. J. (1991). The race for property rights. Journal of Law and

Economics 33, 177-197.

Arrow, K. (1969). The organisation of economic activity: Issues pertinent to the choice

of market versus non-market allocation. In the Analysis of and Evaluation of

Public Expenditure: The PPB System. Vol. 1. US Joint Economic Committee, 1st

Session, 59-73.

Asian Development Bank (2004). The value chain for tea in Vietnam: Prospects for

participation of the poor. Making Markets Work Better for the Poor. Discussion

Paper No 01, November.

Bac, H. V., Nanseki, T., and Chomei, Y. (2017). Factors influencing tea farmers’

decision to adopt Vietnamese good agricultural practices in Northern Vietnam.

Journal of Agricultural Economics and Development, 6(2), 012-020.

Bajari, P., and Tadelis, S. (2001). Incentive versus transaction costs: A theory of

procurement contracts. The RAND Journal of Economics 32, 387-407.

Baker, J. L. (2000). Evaluating the impact of development projects on poverty: A

handbook for practitioners. World Bank Publications.

Balat, J., Brambilla, I., and Porto, G. (2009). Realizing the gains from trade: Export

crops, marketing costs, and poverty. Journal of International Economics, 78, 21-

31.

Barzel, Y. (1997). Economic analysis of property rights (2nd Edn.). Cambridge

University Press, Cambridge.

Basnayake, B. M. J. K., and Gunaratne, L. H. P. (2002). Estimation of technical

efficiency and it’s determinants in the tea small holding sector in the Mid

144

Country Wet Zone of Sri Lanka. Sri Lankan Journal of Agricultural Economics,

4(1), 137-150.

Baten, A., Kamil, A. A., and Haque, M. A. (2010). Productive efficiency of tea

industry: A stochastic frontier approach. African Journal of Biotechnology,

9(25), 3808-3816.

Baumann, P. (2000). Equity and efficiency in contract farming schemes: The experience

of agricultural tree crops. Working Paper 139, October 2000. Overseas

Development Institute, London.

Becerril, J., and Abdulai, A. (2010). The impact of improved maize varieties on poverty

in Mexico: A propensity score-matching approach. World Development, 38(7),

1024-1035.

Begum, I. A., Alam, M. J., Buysee, J., Frija, A., and Huylenbroeck, G. H. (2012).

Contract farmer and poultry farm efficiency in Bangladesh: A data envelopment

analysis. Applied Economics 44, 3737-3747.

Begum, M. E. A., Hossain, M. I., and Papanagiotou, E. (2013). Technical efficiency of

shrimp farming in Bangladesh: An application of the stochastic production

frontier approach. Journal of the World Aquaculture Society, 44(5), 641-654.

Bellemare, M. F. (2012). As you sow, so shall you reap: The welfare impacts of contract

farming. World Development, 40(7), 1418-1434.

Boland, M. A., Golden, B. B., and Tsoodle, L. J. (2008). Agency theory issues in the

food processing industry. Journal of Agricultural and Applied Economics, 40(2),

623-634.

Bolwig, S., Gibbon, P., and Jones, S. (2009). The economics of smallholder organic

contract farming in tropical Africa. World Development, 37(6), 1094-1104.

145

Bravo-Ureta, B. E., Greene, W., and Solís, D. (2012). Technical efficiency analysis

correcting for biases from observed and unobserved variables: An application to

a natural resource management project. Empirical Economics, 43(1), 55-72.

Brousseau, E., and Glachant, J. M. (2002). The economics of contracts and the renewal

of economics. The economics of contracts. Cambridge University Press,

Cambridge, pp. 3-31.

Buccola, S. T., and French, B. C. (1981). Portfolio analysis of contracting strategies for

farmer marketing cooperatives. Agricultural Experiment Station, Oregon State

University, Corvallis.

Bülow, D. V., and Sorensen, A. (1993). Gender and contract farming: tea outgrower

schemes in Kenya. Review of African Political Economy, 56, 38-52.

Cahyadi, E. R., and Waibel, H. (2016). Contract farming and vulnerability to poverty

among oil palm smallholders in Indonesia. The Journal of Development Studies,

52(5), 681-695.

Cakmakyapan, S., and Goktas, A. (2013). A comparison of binary logit and probit

models with a simulation study. Journal of Social and Economic Statistics, 2(1),

1-17.

Carson, S. J., Madhok, A., and Wu, T. (2006). Uncertainty, opportunism, and

governance: The effects of volatility and ambiguity on formal and relational

contracting. Academy of Management Journal, 49(5), 1058-1077.

Chang, K. (2015). World tea production and trade: Current and future development.

FAO Intergovernmental Group on Tea. Trade and Markets Division. Food and

Agriculture Organization of the United Nations, Rome.

146

Chege, J. W., Nyikal, R. A., Mburu, J., and Murrithi, B. W. (2015). Impact of export

horticulture farming on per capita calorie intake of smallholder farmers in

Eastern and Central provinces in Kenya. International Journal of Food and

Agricultural Economics, 3(4), 65-81.

Chiona, S., Kalinda, T., and Tembo, G. (2014). Stochastic frontier analysis of the

technical efficiency of smallholder maize farmers in Central province, Zambia.

Journal of Agricultural Science, 6(10), 108-118.

Chirwa, W. E., and Mwafongo, W. K. (1998). Stochastic production functions and

technical efficiency of farmers in Southern Malawi. Working Paper, No.

WC/04/98.

Choudhury, V. (1997). Strategic choices in the development of interorganizational

information systems. Information Systems Research, 8(1), 1-24.

Chuyen, P. T. H. (2016). Poverty reduction in Viet Nam: Outstanding achievements and

solutions to overcome limitations in the near future [online]. Retrieved from

http://english.tapchicongsan.org.vn/Home/Culture-Society/2016/622/Poverty-

reduction-in-Viet-Nam-Outstanding-achievements-and-solutions-to-overcome-

limitations.aspx on 9 May 2018.

Coase, R. H. (1937). The nature of the firm. In G. J. Stigler and K. E. Boulding (Eds.).

Readings in price theory. Irwin, Homewood, III, pp. 386-405.

Coase, R. H. (1959). The federal communications commission. Journal of Law and

Economics 2, 1-40.

Coase, R. H. (1960). The problem of social cost. Journal of Law and Economics 3, 1-

44.

147

Coelli, T. J., Rao, D. S. P., O'Donnell, C. J., and Battese, G. E. (2005). An introduction

to efficiency and productivity analysis. Springer Science & Business Media.

Cook, M., and Barry, P. (2004). Organizational economics in the food, agribusiness, and

agricultural sectors. American Journal of Agricultural Economics, 86(3), 740-

743.

Daniel, W. W., and Cross, C. L. (2013). Biostatistics: A foundation for analysis in the

health sciences. Tenth Edition. John Wiley & Sons, Inc, 2013.

Dedehouanou, S. F., Swinnen, J., and Maertens, M. (2013). Does contracting make

farmers happy? Evidence from Senegal. Review of Income and Wealth, 59,

S138-S160.

Dehejia, R. H., and Wahba, S. (2002). Propensity score-matching methods for

nonexperimental causal studies. Review of Economics and Statistics, 84(1), 151-

161.

Department of Agriculture and Rural Development in Phu Tho province (2015). Report

on tea production in 2015 and the direction for tea development in 2016. Phu

Tho province, Viet Nam.

Dhillon, S. S., and Singh, N. (2006). Contract farming in Punjab: An analysis of

problems, challenges and opportunities. Pakistan Economic and Social Review.

XLIV(1), 19-38.

Dube, L., and Guveya, E. (2014). Productivity analysis of smallholder out-grower tea

(Camellia Sinensis) farming in Chipinge district of Zimbabwe. Journal of

Agriculture Economics and Rural Development, 2(4), 128-139.

Dube, L., and Mugwagwa, K. E. (2017). The impact of contract farming on smallholder

tobacco farmers’ household incomes: A case study of Makoni district,

148

Manicaland province, Zimbabwe. Scholars Journal of Agriculture and

Veterinary Sciences, 4(2), 79-85.

Eaton, C., and Shepherd, A. W. (2001). Contract farming: partnerships for growth. FAO

Agricultural Services Bulletin 145. Food and Agriculture Organization of the

United Nations, Rome.

Eisenhardt, K. M. (1989). Agency theory: An assessment and review. The Academy of

Management Review, 14(1), 57-74.

FAO (Food and Agriculture Organization) (2016). Socio-economic factors and

implications for the livelihood of tea smallholders. Intergovernmental Group on

Tea. Twenty-second Session. Naivasha, Kenya, 25–27 May.

FAO (Food and Agriculture Organization) (2018a). Producer price of tea in selected

countries [online]. Retrieved from http://www.fao.org/faostat/en/#data/PP on 27

March 2018.

FAO (Food and Agriculture Organization) (2018b). Tea exports and imports of the

world [online]. Retrieved from http://www.fao.org/faostat/en/#data/TP on 29

March 2018.

FAO (Food and Agriculture Organization) (2018). Crops and livestock products

[online]. Retrieved from http://www.fao.org/faostat/en/#data/TP on 15 October

2018.

Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal

Statistical Society, 120(3), 253-290.

Furuboth, E. G., and Pejovich, S. (1972). Property rights and economic theory: A

survey of recent literature. Journal of Economic Literature 10, 1137-1162.

149

Garrido, M. M., Kelley, A. S., Paris, J., Roza, K., Meier, D. E., Morrison, R. S., and

Aldridge, M. D. (2014). Methods for constructing and assessing propensity

scores. Health Services Research, 49(5), 1701-1720.

Gautam, S., Schreinemachers, P., and Uddin, M. N. (2017). Impact of training vegetable

farmers in Bangladesh in integrated pest management (IPM). Crop Protection

102, 161-169.

Gemici, S., Rojewski, J. W., and Lee, H. (2012). Use of propensity score matching for

training research with observational data. International Journal of Training

Research, 10(3), 219-232.

General Statistics Office of Viet Nam (2015). Statistical Yearbook of Vietnam 2014.

Statistical Publishing House, Ha Noi.

General Statistics Office of Viet Nam (2016). Statistical Yearbook of Vietnam 2015.

Statistical Publishing House, Ha Noi.

General Statistics Office of Viet Nam (2017). Statistical Yearbook of Vietnam 2016.

Statistical Publishing House, Ha Noi.

Gibbons, R. (2005). Four formal theories of the firm? Journal of Economic Behavior &

Organization 58, 200-245.

Gill, K. K. (2001). Diversiﬁcation of agriculture and women employment in Punjab.

The Indian Journal of Labour Economics, 44(2), 259-267.

Girma, J., and Gardebroek, C. (2014). The impact of contracts on organic honey

producers’ incomes in Southwestern Ethiopia. Forest Policy and Economics 50,

259-268.

Glover, D. (1994). Contract farming and commercialization of agriculture in developing

countries. In Von Braun, J. and Kennedy, E. (Eds.). Agricultural

150

commercialization, economic development and nutrition. Johns Hopkins, pp.

166-175.

Glover, D., and Ghee, L. T. (1992). Contract farming in Southeast Asia: Three country

studies. Institute for Advanced Studies. Monograph Series: SM No. 5.

University of Malaya, Kuala Lumpur.

Greene, W. (2010). A stochastic frontier model with correction for sample selection.

Journal of Productivity Analysis, 34(1), 15-24.

Gregory, T. (2011). Transaction cost economics and directions for relational governance

research. Proceedings of the Southern Association for Information Systems

Conference, Atlanta, GA, USA, 25–26 March.

Grossman, S., and Hart, O. (1986). The costs and benefits of ownership: A theory of

vertical and lateral integration. Journal of Political Economy 94, 691-719.

Gudeta, D. O., Arero, B. G., and Goshu, A. T. (2017). Vector autoregressive modelling

of some economic growth indicators of Ethiopia. American Journal of

Economics, 7(1), 46-62.

Guo, H., Jolly, R. W., and Zhu, J. (2007). Contract farming in China: Perspectives of

farm households and agribusiness firms. Comparative Economic Studies, 49(2),

285-312.

Hamilton, N. D. (2008). Agricultural contracting: A US perspective and issues for India

to consider. In A. Gulat, P. K. Joshi and M. Landes (Eds.). Contract farming in

India: A resource book.

Harryman, W. R. (1994). Production contracts, farm economics: Facts and opinions.

Department of Agricultural Economics, College of Agriculture, University of

Illinois at Urbana-Champaign, CES, 94(4), 1-4.

151

Hart, O. (1995). Firms, contracts, and financial structure. Clarendon Press, Oxford.

Hart, O., and Moore, J. (1990). Property rights and the nature of the firm. Journal of

Political Economy 98, 1119-1158.

Hassan, S., and Ahmad, B. (2005). Stochastic frontier production function, application

and hypothesis testing. International Journal of Agriculture and Biology, 7(3),

427-430.

Heckman, J. J. (1979). Statistical models for discrete panel data. Department of

Economics and Graduate School of Business, University of Chicago, Chicago,

IL.

Heckman, J. J., Ichimura, H., and Todd, P. (1998). Matching as an econometric

evaluation estimator. The Review of Economic Studies, 65(2), 261-294.

Hernandez, M. A. (2011). Contract farming: Implications and perspectives for

smallholders. Workshop on Climate Change and Market Access, Maputo, 17

August.

Hong, N. B., and Yabe, M. (2015). Technical efficiency analysis of tea production in

the Northern Mountainous Region of Vietnam. Global Journal of Science

Frontier Research: D Agriculture and Veterinary, 15(1), 30-42.

Hoppe, R. A., Green, R., Banker, D., Kalbacher, J. Z., and Bentley, S. E. (1996).

Structural and financial characteristics of U.S. farms. 18th Annual Family Farm

Report to Congress. Agricultural Information Bulletin No. 728. United States

Department of Agriculture, Economic Research Service, October 1996.

Hoppe, R. A., Tames, J., Perry, J. E., Penni, K., Judith, S. E., James, R. T., Robert, G.

C., Ron, D., and Monke, J. (2001). Structural and financial characteristics of

U.S. farms. Family Farm Report. Economic Research Service, United States

152

Department of Agriculture, Agricultural Information Bulletin No. 768, May

2001.

Huluka, A. T., and Negatu, W. (2016). The impacts of farmer field school training on

knowledge and farm technology adoption: Evidence from smallholder maize

farmers in Oromia, Ethiopia. Journal of Economics and Public Finance, 2(1), 1-

21.

Igweoscar, O. (2014). Effect of contract farming on productivity and welfare of

cassava-based farmers in South Eastern Nigeria. European Journal of Business

and Management, 6(7), 334-339.

Jensen, M. C., and Meckling, W. H. (1976). Theory of the firm: Managerial behavior,

agency costs and ownership structure. Journal of Financial Economics 3, 305-

360.

Kalimangasi, N. N., Kihombo, A., and Kalimangasi, N. (2014). Contribution of contract

cocoa production on improving livelihood of smallholder farmers. International

Journal of Scientific and Research Publications, 4(10), 1-10.

Key, N. (2013). Production contracts and farm business growth and survival. Journal of

Agricultural and Applied Economics, 45(2), 277-293.

Kherallah, M., and Kirsten, J. (2001). The new institutional economics: Applications for

agricultural policy research in developing countries. MSSD Discussion Paper

No. 41. Market and Structural Studies Division. International Food Policy

Research Institute, Washington DC, USA.

Khoi, D. (2014). Contract farming in Vietnam: Policy issues and challenges. Workshop

on the Legal Dimension of Contract Farming, Bangkok, 26 September.

153

Khoi, N. V., Lan, C. H., and Huong, T. L. (2015). Vietnam tea industry: An analysis

from value chain approach. International Journal of Managing Value and

Supply Chains (IJMVSC), 6(3), 1-15.

Kim, J., and Mahoney, J. T. (2005). Property right theory, transaction costs theory, and

agency theory: An organizational economics approach to strategic management.

Managerial and Decision Economics 26, 223-242.

Kohls, R. L., and Uhl, J. N. (1985). Marketing of agricultural product. Sixth Edition,

MPC, Collier MacMillan Publication, London.

Kristen, J., and Sartorius, K. (2002). Linking agribusiness and small-scale farmers: is

there a new role for contract farming. Development Southern Africa, 19(4), 504-

529.

Kumar, A., Roy, D., Joshi, P. K., Tripathi, G., and Adhikari, R. P. (2016). Impact of

contract farming on profits and yield of smallholder farms in Nepal: An

evidence from lentil cultivation. Selected Paper Prepared for Presentation at the

2016 Agricultural & Applied Economics Association Annual Meeting, Boston,

Massachusetts, 31 July – 2 August.

Larsen, K. (2008). Economic consequences of collaborative arrangements in the

agricultural firm. Doctoral Thesis Submitted to Swedish University of

Agricultural Sciences, Uppsala.

Leuven, E., and Sianesi, B. (2003). psmatch2. STATA module to perform full

Mahalanobis and propensity score matching, common support graphing, and

covariate imbalance testing, version, 1(3).

154

Lütkepohl, H., Saikkonen, P., and Trenkler, C. (2001). Maximum eigenvalue versus

trace tests for the cointegrating rank of a VAR process. The Econometrics

Journal, 4(2), 287-310.

MacNeil, I. R. (1974). The many futures of contracts. Southern California Law Review,

47, 691-816.

Maertens, M., and Velde, K. V. (2017). Contract-farming in staple food chains: the case

of rice in Benin. World Development, 95, 73-87.

Manjunatha, A. V. (2016). Present status and prospects of contract farming in India.

International Journal of Agriculture Sciences, 8(7), 1072-1075.

Martins, R., Serra, F. R., Leite, A. D. S., Ferreira, M. P., and Li, D. (2010). Transaction

cost theory influence in strategy research: A review through a bibliometric study

in leading journals. Working Paper No. 61/2010.

Masten, S. E. (2000). Contractual choice. In B. Boukaert and G. DeGeest (Eds.).

Encyclopedia of law and economics. Edward Elgar Publishing and the

University of Ghent.

Matthews, R. C. O. (1986). The economics of institutions and the sources of growth.

Economic Journal 96, 903-918.

Meshesha, J. G. (2011). Impact of contract farming on household income of smallholder

farmers: the case of organic honey production in South West Ethiopia, Sheka

zone. Thesis submitted to the Wageningen University and Research Centre in

Partial Fulfillment of the Requirements for the Master of Science Degree in

Organic Agriculture.

155

Minh, N. K., and Long, G. T. (2009). Efficiency estimates for the agricultural

production in Vietnam: A comparison of parametric and non-parametric

approaches. Agricultural Economics Review, 10(2), 62-78.

Minot, N. W. (1986). Contract farming and its effects on small farmers in less

developed countries. Working Paper No. 31. Department of Agricultural

Economics, Michigan State University.

Minot, N. (2007). Contract farming in developing countries: Patterns, impacts, and

policy implications. Case study #6-3 of the program: “Food Policy for

Developing Countries: The Role of Government in the Global Food System”,

Cornel University.

Minot, N., and Sawyer, B. (2016). Contract farming in developing countries: Theory,

practice, and policy implications. Innovation for Inclusive Value Chain

Development: Successes and Challenges, International Food Policy Research

Institute, Washington, DC, 127-155.

Mishra, A. K., Kumar, A., Joshi, P. K., and D'Souza, A. (2018). Impact of contract

farming on yield, costs and profitability in low‐value crop: evidence from a low‐

income country. Australian Journal of Agricultural and Resource Economics,

62(4), 589-607.

Mishra, A. K., Kumar, A., Joshi, P. K., D'Souza, A., and Tripathi, G. (2018). How can

organic rice be a boon to smallholders? Evidence from contract farming in India.

Food Policy, 75, 147-157.

Miyata, S., Minot, N., and Hu, D. (2009). Impact of contract farming on income:

Linking small farmers, packers, and supermarkets in China. World Development,

37(11), 1781-1790.

156

Morrison, P. S., Murray, W. E., and Ngidang, D. (2006). Promoting indigenous

entrepreneurship through small-scale contract farming: The poultry sector in

Sarawak, Malaysia. Singapore Journal of Tropical Geography, 27(2), 191-206.

Mpeta, D., Kuzilwa, J. A., and Sebyiga, B. (2018). Analysis of contract farming effects

on efficiency and productivity of small-scale sunflower farmers in Tanzania – a

propensity score method approach. Acta Scientiarum Polonorum, 17(1), 75-84.

Mulatu, G., Haji, J., Legesse, B., and Ketema, M. (2017). Impact of participation in

vegetables contract farming on households’ income in the Central Rift valley of

Ethiopia. American Journal of Rural Development, 5(4), 90-96.

Munene, N. K., and Kasamani, S. B. (2017). Analysis and design of a tea farming

information system. International Research Journal of Engineering and

Technology (IRJET), 04(12), 1846-1859.

Musunuru, N. (2017). Causal relationships between grain, meat prices and exchange

rates. International Journal of Food and Agricultural Economics, 5(4), 1-10.

Mwambi, M. M., Oduol, J., Mshenga, P., and Saidi, M. (2016). Does contract farming

improve smallholder income? The case of avocado farmers in Kenya. Journal of

Agribusiness in Developing and Emerging Economies, 6(1), 2-20.

Narayanan, S. (2014). Profits from participation in high value agriculture: Evidence of

heterogeneous benefits in contract farming schemes in Southern India. Food

Policy, 44, 142-157.

Nghia, D. T. (2009). Transition to organic tea production in Thai Nguyen province,

Vietnam: Economic and environmental impacts. EEPSEA Research Report

2008-RR8. Economy and Environment Program for Southeast Asia.

157

Ngoc, N. T. B., Bac, H. V., and Huyen, N. T. (2014). Comparative analysis of non-

contract and contract farmers in tea sector in Vietnam: The case study in Thai

Nguyen and Phu Tho provinces. Journal of Science & Technology, 119(05),

143-149.

Nhan, T. Q., Takeuchi, I., and Hoang, D. V. (2013). Rice contract farming – The

potential key to improve rice growers’ income: A farm level study in An Giang

province. Journal of Science & Development, 11(7), 1062-1072.

Nindi, A. G., and Odhiambo, N. M. (2015). Poverty and economic growth in Swaziland:

An empirical investigation. Managing Global Transitions International

Research Journal, 13(1), 59-74.

North, D. C. (1990). Institutions, institutional change, and economic performance.

Cambridge University Press, Cambridge.

North, D. C. (2004). Understanding the process of economic change. Princeton

University Press, Princeton.

Oanh, L. T. K., Nga, B. T., and Lebailly, P. (2016). Tea production between contract

and non contract farmers in Phu Tho Province of Vietnam. Scholars Journal of

Agriculture and Veterinary Sciences, 3(2), 117-122.

O’Looney, J. A. (1998). Outsourcing state and local government services: Decision-

making strategies and management methods. Quorum Books, Westport, CT.

Otsuka, K., Nakano, Y., and Takahashi, K. (2016). Contract farming in developed and

developing countries. Annual Review of Resource Economics, 8, 353-376.

Oyakhilomen, O., and Zibah, R. G. (2014). Agricultural production and economic

growth in Nigeria: Implication for rural poverty alleviation. Quarterly Journal of

International Agriculture, 53(3), 207-223.

158

Parirenyatwa, K. T., and Mago, S. (2014). Evolution and development of contract

farming in Zimbabwe: a reflection for agribusiness. Mediterranean Journal of

Social Sciences, 5(20), 237-244.

Pasour, E. C. (1983). Rent seeking: Some conceptual problems and implications. The

Review of Austrian Economics, 1(1), 123-143.

Paul, C. J. M., Nehring, R., and Banker, D. (2004). Productivity, economies, and

efficiency in U.S. Agriculture: A look at contracts. American Journal of

Agricultural Economics, 86(5), 1308-1314.

Paulson, N. D., Katchova, A. L., and Lence, S. H. (2010). An empirical analysis of the

determinants of marketing contract structures for corn and soybeans. Journal of

Agricultural & Food Industrial Organization, 8(1), 1-23.

Pejovich, S. (1982). Karl Marx, property rights school and the process of social change.

Kyklos 35, 383-407.

Pejovich, S. (1995). Economic analysis of institutions and systems. Kluwer Academic,

Boston.

Pfaff, B. (2008). VAR, SVAR and SVEC models: Implementation within R package

vars. Journal of Statistical Software, 27(4), 1-32.

Phu, N. V., and Nguyen, T. T. (2014). Agricultural extension and efficiency of tea

production in Northeastern Vietnam. Document de Travail No. 2014-11, June

2014. Bureau d'Economie Théorique et Appliquée, UDS, Strasbourg.

Phu Tho Statistics Office (2017). Statistical Handbook of Phu Tho 2016.

Phu Tho Statistics Office (2018). Statistical Handbook of Phu Tho 2017.

159

Pouncgchompu, S., Hajime, K., and Chantanop, S. (2016). The characteristics of

contract farming for farm smallholder: The case of Thailand and Lao PDR. Asia-

Pacific Social Science Review, 16(2), 85-93.

Prowse, M. (2012). Contract farming in developing countries – A review. Institute of

Development Policy and Management, University of Antwerp.

Ragasa, C., Lambrecht, I., and Kufoalor, D. S. (2018). Limitations of contract farming

as a pro-poor strategy: The case of maize outgrower schemes in Upper West

Ghana. World Development 102, 30-56.

Rehber, E. (2007). Contract farming: Theory and practice. The ICFAI University Press.

Rosenbaum, P. R., and Rubin, D. B. (1983). The central role of the propensity score in

observational studies for causal effects. Biometrika, 70(1), 41-55.

Ross, S. A. (1973). The economic theory of agency: The principal’s problem. American

Economic Review, 63(2), 134-139.

Rotich, J. K., Odhiambo, M., and Ngeno, V. (2017a). Effect of land characteristics on

poverty levels among tea farming households in Konoin sub-county, Kenya.

European Journal of Business and Management, 9(35), 28-38.

Rotich, J. K., Odhiambo, M., and Ngeno, V. (2017b). Effect of tea farming household

characteristics on poverty levels among tea farmers in Konoin sub-county,

Kenya. Global Journal of Commerce & Management Perspective, 6(3), 34-44.

Roy, E. P. (1963). Contract farming - USA. The Interstate Printers and Publishers, Inc.

Danville, Illinois, USA.

Saenger, C., Torero, M., and Qaim, M. (2014). Impact of third-party contract

enforcement in agricultural markets - A field experiment in Vietnam. American

Journal of Agricultural Economics, 96(4), 1220-1238.

160

Saigenji, Y., and Zeller, M. (2009). Effect of contract farming on productivity and

income of small holders: The case of tea production in North-western Vietnam.

Contributed Paper prepared for presentation at the International Association of

Agricultural Economists Conference, Beijing, China, 16–22 August.

Schreinemachers, P., Wu, M. H., Uddin, M. N., Ahmad, S., and Hanson, P. (2016).

Farmer training in off-season vegetables: Effects on income and pesticide use in

Bangladesh. Food Policy 61, 132-140.

Shadab, S. (2018). Tourism and economic growth in the United Arab Emirates: A

Granger causality approach. IOSR Journal of Business and Management (IOSR-

JBM), 20(4), 1-6.

Shantha, A., Ali, B. G. H., and Bandara, R. A. G. (2012). Efficiency and managerial

ability of paddy farming under minor irrigation conditions: A frontier production

function approach. Journal of Agricultural Sciences, 7(3), 145-158.

Sharma, G. (2017). Pros and cons of different sampling techniques. International

Journal of Applied Research, 3(7), 749-752.

Silva, C. A. B. D. (2005). The growing role of contract farming in agri-food systems

development: Drivers, theory and practice. Agricultural Management, Marketing

and Finance Service. FAO, Rome, July.

Simmons, P. (2002). Overview of smallholder contract farming in developing countries.

Graduate School of Agricultural and Resource Economics. University of New

England, Armidale.

Simmons, P., Winters, P., and Patrick, I. (2005). An analysis of contract farming in East

Java, Bali, and Lombok, Indonesia. Agricultural Economics, 33(3), 513-552.

161

Singh, S. (2000). Contract farming for agricultural diversification in the Indian Punjab:

A study of performance and problems. Indian Journal of Agricultural

Economics, 35(3), 241-261.

Sokchea, A., and Culas, R. J. (2015). Impact of contract farming with farmer

organizations on farmers’ income: A case study of Reasmey Stung Sen

Agricultural Development Cooperative in Cambodia. Australasian Agribusiness

Review, 23(1), 1-11.

Song, M., and Montoya-Weiss, M. M. (2001). The effect of perceived technological

uncertainty on Japanese New Product Development. Academy of Management

Journal, 44(1), 61-80.

STATISTA (2018). Annual per capita tea consumption of the world wide in 2016

[online]. Retrieved from https://www.statista.com/statistics/507950/global-per-

capita-tea-consumption-by-country/ on 10 May 2018.

Suharsono, A., Aziza, A., and Pramesti, W. (2017). Comparison of vector

autoregressive (VAR) and vector error correction models (VECM) for index of

ASEAN stock price, in AIP Conference Proceedings 1913, 1, pp. 020032-1-

020032-9. AIP Publishing. International Conference and Workshop on

Mathematical Analysis and its Applications.

Swain, B. B. (2013). An analysis on impact of contract farming on farm productivity

and efficiency: the case of hybrid paddy seed cultivation in Andhra Pradesh.

Indian Journal of Agricultural Economics, 68(3), 477.

Swinnen, J. F. M., Deconninck, K., Vandemoortele, T., and Vandeplas, A. (2015).

Quality Standards, Value Chains, and International Development: Economic

and Political Theory. Cambridge, UK: Cambridge University Press.

162

Swinnen, J., and Maertens, M. (2007). Globalization, privatization, and vertical

coordination in food value chains in developing and transition countries.

Agricultural Economics, 37(1), 89-102.

The People’s Committee of Phu Tho province (2012). Decision number 3321/QD-

UBND in 30 November 2012 of the People’s Committee of Phu Tho province

on the plan of development of tea production in Phu Tho province toward 2020.

The Prime Minister of Viet Nam (2008). Decision number 99/2008/QD-TTg of the

Prime Minister in 14 July 2008 on the development of socio-economics of Phu

Tho province toward 2020.

Tirole, J. (1999). Incomplete contracts: Where do we stand? Econometrica, 67, 741-81.

Ton, G., Vellema, W., Desiere, S., Weituschat, S., and D'Haese, M. (2018). Contract

farming for improving smallholder incomes: What can we learn from

effectiveness studies? World Development, 104, 46-64.

Tongchure, S., and Hoang, N. (2013). Cassava smallholders’ participation in contract

farming in Nakhon Ratchasima province, Thailand. Journal of Social and

Development Sciences, 4(7), 332-338.

Trang, N. T. T. (2014). Legal policies and regulations governing contract farming in

Vietnam. Unidroit Working Group for the preparation of a legal guide on

contract farming. Fourth Meeting, Rome, 17–22 November.

Trifković, N. (2014). Governance strategies and welfare effects: vertical integration and

contracts in the catfish sector in Vietnam. Journal of Development Studies,

50(7), 949-961.

163

Tripathi, R. S., Singh, R., and Singh, S. (2005). Contract farming in potato production:

An alternative for managing risk and uncertainty. Agricultural Economics

Research Review, 18(1), 47-60.

Tsadik, E. T., Tamir, B., and Sahle, Z. (2015). The impact of village poultry technology

adoption on the livelihood of smallholder farmers in the Central Oromia Region,

Ethiopia. Journal of Animal Production Advances, 5(11), 807-819.

UNCTAD (2016). Tea: An INFOCOMM commodity profile. UNCTAD Trust Fund on

Market Information on Agricultural Commodities. United Nations Conference

on Trade and Development. New York and Geneva.

Vavra, P. (2009). Role, usage and motivation for contracting in agriculture. OECD

Food, Agriculture and Fisheries Working Papers, No. 16, OECD Publishing.

Vietnam Business Council for Sustainable Development (VBCSD) (2015). Report on:

Competitiveness of tea, coffee & rubber sectors. Ha Noi.

Wang, H., Moustier, P., and Loc, N. T. T. (2014). Economic impact of direct marketing

and contracts: The case of safe vegetable chains in Northern Vietnam. Food

Policy 47, 13-23.

Wang, H. H., Wang, Y., and Delgado, M. S. (2014). The transition to modern

agriculture: Contract farming in developing economies. American Journal of

Agricultural Economics, 96(5), 1257-1271.

Warning, M., and Key, N. (2002). The social performance and distributional

consequences of contract farming: an equilibrium analysis of the Arachide de

Bouche Program in Senegal. World Development, 30(2), 255-263.

Watts, M. J. (1994). Life under contact: Contract farming, agrarian restructuring and

flexible accumulation. In P. D. Little and M. J. Watts (Eds.). Living under

164

contract: Contract farming and agrarian transformation in Sub-Saharan Africa.

The University of Wisconsin Press, Madison, Wisconsin, USA, pp. 21-70.

Williamson, O. E. (1975). Markets and hierarchies: Analysis and antitrust implications.

Free Press, New York.

Williamson, O. E. (1979). Transaction-cost economics: the governance of contractual

relations. Journal of Law and Economics, 22(2), 233-261.

Williamson, O. E. (1981). The economics of organization: The transaction cost

approach. The American Journal of Sociology, 87(3), 548-577.

Williamson, O. E. (1985). The economic institutions of capitalism. Free Press-

McMillan, New York.

Williamson, O. E. (1996). The mechanisms of governance. Oxford University Press,

Oxford.

Winters, P., Simmons, P., and Patrick, I. (2005). Evaluation of a hybrid seed contract

between smallholders and a multinational company in East Java, Indonesia. The

Journal of Development Studies, 41(1), 62-89.

World Bank (2018). World Development Indicators. GDP of Vietnam [online].

Retrieved from

http://databank.worldbank.org/data/reports.aspx?source=2&series=NY.GDP.MK

TP.CD&country= on 15 October 2018.

Wu, S. Y. (2006). Contract theory and agricultural policy analysis: A discussion and

survey of recent developments. The Australian Journal of Agricultural and

Resource Economics 50, 490-509.

Wu, X. L. (1992). A comparison of tobit and OLS estimates of Japanese peanut import

demand. Journal of Agricultural Economics, 43(1), 38-42.

165

Young, L. M., and Hobbs, J. E. (2002). Vertical linkages in agri-food supply chains:

Changing roles for producers, commodity groups, and government policy.

Review of Agricultural Economics, 24(2), 428-441.

Ze-ying, H., Ying, X. U., Di, Z., Chen, W., and Ji-min, W. (2018). One size fits all?

Contract farming among broiler producers in China. Journal of Integrative

Agriculture, 17(2), 473-482.

166

APPENDICES

Appendix A (Chapter 3)

Law and regulations on contracts and contract farming

The Vietnam National Assembly, the Government of Vietnam and the Ministry of

Agriculture and Rural Development have enacted laws and regulations on contracts and contract farming in recent years.

The 2005 Civil Code, approved by Vietnam National Assembly, is an important policy initiative to protect the rights of contracting parties. The Code states that the entry of parties into a civil contract must follow these principles: (1) freedom to participate in the contract, provided that it is not contrary to law and social ethics; and

(2) voluntariness, equality, goodwill, cooperation, honesty and good faith. While this policy is not specific to contract farmers, its terms have implications for these farmers.

The Law on Commerce was approved by the Vietnam National Assembly in

2005. The form of contract for purchase and sale of goods is referred to as article 24, section 1, chapter 2 of this Law. A contract for purchase and sale of goods that is determined by a certain transaction, can be prepared by either the oral or writing. A law stipulates that certain types of contracts for purchase and sale of goods must be made in writing.

The Law on Commercial Arbitration was approved by the Vietnam National

Assembly in 2010. Article 4, chapter 1 of this Law presents principles for dispute resolution by arbitration as follows: (1) arbitrators must respect the agreement of the parties if it does not breach prohibitions and is not contrary to social norms; (2)

167

arbitrators must be independent, objective and impartial and must comply with the provisions of law; (3) parties in dispute will have equal rights and obligations; (4) dispute resolution by arbitration will be implemented in private, unless otherwise agreed by the parties; and (5) an arbitration award will be final.

Decision 62/2013/QD-TTg was issued on 25 October 2013 by the Prime

Minister on the regulation of support policies of the State in facilitating coordination among producers, cooperatives and enterprises in production, processing and marketing of agricultural commodities based on large-scale production. This decision explains contents and conditions of support from the State for enterprises and producers. For example, support of the State for enterprises and producers is as follows: (1) enterprises and producers are exempted from fees and rent for land in terms of constructing processing factories, stores and houses for workers; (2) enterprises and producers are prioritised if participating in export contracts and food procurement programs of the government; (3) enterprises can obtain partial financial support for constructing transport, irrigation and electrical systems; (4) producers are able to receive support of up to 30 percent in the first year and 20 percent in the second year of total expenditures for plant protection; (5) enterprises, farmer organisations and farmers can get support of up to 50 percent, 50 percent and 100 percent, respectively for technical training costs for a times; (6) farmers can receive financial support of up to 30 percent of the seed costs of the first crop; and (7) farmers receive up to 100 percent financial support for storage in enterprises for a maximum of three months.

Circular 15 was issued on 29 April 2014 by the Ministry of Agriculture and

Rural Development. This circular aims to guide the implementation and application of

Decision 62 of the Prime Minister on encouraging coordination between enterprises and

168

producers in production, processing and marketing of agricultural products. In addition, this circular also provides a sample for contracts in production and marketing of agricultural products.

Policies and regulations in the tea sector

Decision number 69/2007/QD-TTg was issued on 18 May 2007 by the Prime Minister on the development of the processing industry for agricultural products in industrialisation and modernisation until 2010 and directions toward 2020. The specific goal of tea production is to increase the proportion of fresh tea processed by industrial methods to more than 80 percent by 2020, with 50 percent black tea and 50 percent green tea.

Decision number 23/QD-TTg was issued on 6 January 2010 by the Prime

Minister on the development of trade in rural areas in the period 2010 to 2015 and directions toward 2020. Based on this decision, by 2020, the proportion of agricultural products sold through contracts should be 45–50 percent; and all communes across the whole country should have markets based on the requirement of the new rural program.

Decision number 124/QD-TTg was issued on 2 February 2012 by the Prime

Minister on the plan of development of the agricultural sector toward 2020 and vision toward 2030. According to this decision, by 2020, the planted area of tea should reach

140,000 hectares, with 7,000 hectares in the mountainous region in northern Vietnam.

Tea should be produced and processed based on the criteria of safe tea that meets food safety requirements. Furthermore, new varieties of tea should be researched and cultivated to obtain higher productivity and better quality. Investment and improvement

169

of tea processing factories should achieve processing capacity of 840,000 tonnes of fresh tea annually and the volume of processed tea should reach 270,000 tonnes, with 55 percent black tea and 45 percent green tea. By 2020, the exported price of Vietnamese tea should be equivalent to average global prices.

Directive number 87/2001/CT-BNN was approved by the Ministry of

Agriculture and Rural Development (MARD) on 5 September 2001 on management of chemicals and pesticides use and application in tea trees. This directive aims to guide chemical and pesticide use based on guidelines of MARD as follows: (1) chemicals and pesticides registered for tea trees; (2) chemicals and pesticides which do not contain organic chlorine; and (3) isolated duration of at least 7 days for pesticides and 14 days for chemicals. Secondly, business and advertisements of chemicals and pesticides for tea trees, which are not in the portfolio of the MARD, are prohibited. Lastly, use of pesticides for tea trees made from biological sources is encouraged.

Decision number 824/QD-BNN-TT was approved by the MARD on 16 April

2012 on the development of the horticultural sector until 2020 and vision toward 2030.

According to this decision, by 2020, the planted area of tea should reach 140,000 hectares with production of leaf tea of 1 million tonnes. The exported volume of leaf tea is expected to reach 130,000 tonnes in 2020. Production of safe tea should be enhanced to assure food safety and new varieties of tea, which have higher yield and better quality, should be cultivated. The decision plans major zones for tea production, including mountainous regions in the Northern, the Central Highland, and the North

Central zones.

170

Decision number 3321/QD-UBND was approved by People’s Committee of Phu

Tho province on 30 November 2012 on the plan of development of tea production in

Phu Tho province toward 2020. According to this decision, by 2020, the planted area for tea in Phu Tho province is expected to remain at 15,500 hectares, and the planted area for safe tea will be expanded in different periods. By 2020, new varieties for tea should be more than 80 percent, average yield should reach 11 tonnes per hectare and production of leaf tea should reach 160,000 tonnes; 6,500 hectares of planted area for tea should satisfy the requirements of Vietnamese Good Agricultural Practice with average yield of leaf tea of 12.5 tonnes per hectare; 80–100 percent of tea processing units should have contracts with producers, 100 percent of tea processing units must improve processing technologies to assure food safety and 70 percent of tea processing enterprises have to establish and adopt Hazard Analysis and Critical Control Points and

International Organization for Standardization practices.

171

Appendix B (Chapter 6)

Figure B1. Distribution of propensity scores for tea productivity

0 .2 .4 .6 .8 1 Propensity Score

Untreated Treated

Source: Author’s calculation, 2018

The density distribution of the propensity scores for contracted and independent farmers is illustrated in Figure A1. The y-axis denotes the frequency of the propensity score distribution. The upper-half of the graph presents the propensity score distribution for the treated individuals (contracted farmers), while the bottom-half refers to untreated individuals (independent farmers). The distribution of propensity scores shows there is a high chance of obtaining a good match because the propensity score distribution is skewed to the left for contracted households and to the right for independent ones

(Figure B1).

172

Figure B2. Distribution of propensity scores for farmer income

0 .2 .4 .6 .8 1 Propensity Score

Untreated Treated

Source: Author’s calculation, 2018

The density distribution of the propensity scores for contracted and independent farmers is illustrated in Figure B1. The y-axis denotes the frequency of the propensity score distribution. The upper-half of the graph presents the propensity score distribution for the treated individuals (contracted farmers), while the bottom-half refers to untreated individuals (independent farmers). The distribution of propensity scores shows a high chance of obtaining a good match because the propensity score distribution is skewed to the left for contracted households and to the right for independent ones (Figure B2).

173

Table B1. Results of the sensitivity analysis for hidden bias for NNM(5)

Gamma Sig+ Sig- 1 0 0 2 0 0 3 0 0 4 1.2e-15 0 5 7.0e-13 0 6 5.0e-11 0 7 1.1e-09 0 8 1.1e-08 0 9 6.5e-08 0 10 2.7e-07 0 Source: Author’s calculation, 2018 Note: Gamma presents log odds of differential assignment due to unobserved factors Sig+ means upper bound significance level Sig- means lower bound significance level

Table B2. Results of the sensitivity analysis for hidden bias for KM(0.25)

Gamma Sig+ Sig- 1 0 0 2 0 0 3 0 0 4 2.2e-16 0 5 2.1e-13 0 6 1.8e-11 0 7 4.4e-10 0 8 5.0e-09 0 9 3.3e-08 0 10 1.5e-07 0 Source: Author’s calculation, 2018 Note: Gamma presents log odds of differential assignment due to unobserved factors Sig+ means upper bound significance level Sig- means lower bound significance level

174

Appendix C (Chapter 7)

Figure C1. GDP of Vietnam (1977_2016)

250000.00

200000.00

150000.00

million US$ 100000.00

50000.00

0.00 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Source: World Bank, 2018

Figure C2. Export value of tea of Vietnam (1977_2016)

250000

200000

150000

thousand US$ thousand 100000

50000

0 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Source: FAO, 2018

175

Table C1. Estimation of the VAR model

Variables Coefficient Standard Error t P-value LnGDP LnGDP L1 1.432*** 0.10 13.18 0.000 L2 -0.481** 0.20 -2.38 0.028 L3 -0.087 0.21 -0.41 0.688 L4 0.200* 0.98 2.04 0.055 LnExport value of tea L1 -0.145** 0.06 -2.17 0.043 L2 0.019 0.06 0.28 0.785 L3 0.042 0.06 0.62 0.542 L4 0.013 0.06 0.21 0.835 LnProportion of the population living below the poverty line L1 -0.207 0.41 -0.50 0.621 L2 -0.059 0.19 -0.30 0.769 L3 -0.053 0.22 -0.24 0.813 L4 0.112 0.46 0.24 0.813 LnPoverty gap L1 0.134 0.28 0.47 0.645 L2 0.053 0.13 0.39 0.703 L3 0.046 0.14 0.32 0.749 L4 -0.045 0.29 -0.16 0.878 Constant 0.305 0.34 0.88 0.390 LnExport value of tea LnGDP L1 0.549 0.33 1.66 0.114 L2 -0.067 0.61 -0.11 0.915 L3 -0.398 0.65 -0.61 0.549 L4 0.306 0.30 1.02 0.320 LnExport value of tea L1 0.331 0.20 1.62 0.121 L2 0.089 0.21 0.42 0.677 L3 -0.102 0.20 -0.49 0.632 L4 0.307 0.20 1.53 0.142 LnProportion of the population living below the poverty line L1 0.704 1.25 0.56 0.582 L2 0.273 0.60 0.45 0.659 L3 0.920 0.68 1.34 0.195 L4 -1.133 1.43 -0.79 0.439 LnPoverty gap L1 -0.519 0.87 -0.59 0.561 L2 -0.164 0.41 -0.39 0.700 L3 -0.413 0.44 -0.94 0.361 L4 0.713 0.90 0.79 0.439 Constant 0.229 1.06 0.22 0.831

176

Table C1 (Continued)

LnProportion of the population living below the poverty line LnGDP L1 -1.511* 0.84 -1.80 0.088 L2 2.330 1.56 1.49 0.153 L3 -1.506 1.65 -0.91 0.373 L4 0.201 0.76 0.26 0.795 LnExport value of tea L1 -0.144 0.51 -0.28 0.784 L2 -0.021 0.53 -0.04 0.968 L3 0.068 0.53 0.13 0.898 L4 0.183 0.50 0.36 0.722 LnProportion of the population living below the poverty line L1 6.901** 3.18 2.16 0.043 L2 5.149*** 1.54 3.34 0.003 L3 3.724** 1.73 2.14 0.045 L4 -4.054 3.63 -1.12 0.279 LnPoverty gap L1 -4.603* 2.22 -2.07 0.052 L2 -3.434*** 1.06 -3.23 0.004 L3 -2.500** 1.12 -2.23 0.038 L4 1.907 2.28 0.83 0.414 Constant 0.973 2.68 0.36 0.721 LnPoverty gap LnGDP L1 -1.971 1.19 -1.65 0.114 L2 2.933 2.22 1.32 0.202 L3 -1.958 2.34 -0.84 0.413 L4 0.187 1.07 0.17 0.864 LnExport value of tea L1 -0.124 0.73 -0.17 0.867 L2 -0.039 0.76 -0.05 0.959 L3 0.146 0.75 0.19 0.848 L4 0.296 0.72 0.41 0.685 LnProportion of the population living below the poverty line L1 10.475** 4.52 2.32 0.032 L2 8.192*** 2.18 3.75 0.001 L3 4.263 2.46 1.73 0.100 L4 -5.914 5.15 -1.15 0.265 LnPoverty gap L1 -7.020** 3.14 -2.23 0.038 L2 -5.439*** 1.50 -3.61 0.002 L3 -2.923* 1.58 -1.84 0.081 L4 2.774 3.24 0.86 0.402 Constant 0.481 3.81 0.13 0.901 Source: Author’s calculation, 2018 Notes: L1, L2, L3, and L4 denote the number of lags ***, **, and * denote statistical significance at 1%, 5%, 10%, respectively

177