The relationship between the agricultural trade and productivity in vietnam case

27 Table 4.2: Annual growth rate of quantity of selected crops 33 Table 4.5: The mechanization rate in agricultural production activities 36 Table 4.6: Labor force in Vietnam by resident

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM – NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE RELATIONSHIP BETWEEN AGRICULTURAL TRADE AND PRODUCTIVITY

IN VIETNAM CASE

A thesis submitted in partial fulfillment of requirements for degree of Master of

Arts in Development Economics

By NGUYEN HOANG DIEP

Academic supervisor

Dr TRAN TIEN KHAI

HO CHI MINH CITY, JANUARY 2015

ACKNOWLEDGEMENT

So many persons that I want to say thanks to them I write these acknowledgments to express my gratitude to them who helped me directly and indirectly to finish this thesis

First of all, Dr Tran Tien Khai who is my supervisor, he helped me to complete my ideas Furthermore, he also gave me advices to writing logical for this thesis, and already gave his hands when I needed

The other person that is Dr Tran Khanh Nam, who gave me advice from beginning when I wrote the thesis design In addition, he provided me the VARHS data set to apply in my thesis

Important to me is to thank the Vietnam-Netherlands program and all of people and lecturers who supported and gave me their knowledge

Finally yet importantly, with all my respect for my parents who gave me a change to study in this program They have encouraged me to continue my studying

ABBREVIATION

Statistical Database

ABSTRACT

This thesis tries to find out whether trade has an effect on agricultural productivity in Vietnam case There are two models to estimate relationship between trade and productivity including national-level model and farm-level model The time series data of nine agricultural commodities, which is taken from FAOSTAT, is used in national-level model Farm-level model uses cross section data from VARHS 2010 including 1449 farm household in nine provinces In both models, yield refers as agricultural productivity The both models support for a strong positive relationship between trade and yield The result also indicates that the land, irrigation, human capital, cost of production might be necessary for improve productivity

Key words: Agricultural productivity, international trade, cross-section, time series

TABLE OF CONTENT

Page

ACKNOWLEDGEMENT 2

ABBREVIATION 3

ABSTRACT 4

TABLE OF CONTENT 5

LIST OF TABLE 7

LIST OF FIGURE 8

Chapter 1: INTRODUCTION 9

1.1 Problem statement 9

1.2 Research objective 10

1.3 Thesis structure 11

Chapter 2: LITERATURE REVIEW 12

2.1 Theory 12

2.1.1 Agricultural productivity 12

2.1.2 Agricultural trade 15

2.2 Empirical study 17

2.3 Conceptual framework 19

Chapter 3: RESEARCH METHODOLOGY 21

3.1 Agricultural productivity measurement 21

3.2 The tradability index 22

3.3 Empirical model 23

3.3.1 The national level model 23

3.3.2 The farm level model 24

3.4 Data 25

3.5 Analytic method 28

3.5.1 Test for multicollinearity 29

3.5.2 Test for heteroscedastiscity 30

3.5.3 Test for autocorrelation 30

3.5.4 Test for stationary 31

Chapter 4: RESULT AND DISCUSSION 32

4.1 Overview Vietnam agricultural trade 32

4.1.1 Agricultural production in Vietnam 32

4.1.2 Land 34

4.1.3 Irrigation 35

4.1.4 Fertilizer 35

4.1.5 Farm machinery 36

4.1.6 Human capital 36

4.1.7 Trade 37

4.1.7.1 Export 37

4.1.7.2 Import 38

4.2 Econometric result 39

4.2.1 The product tradability index and national level response 39

4.2.2 The farm tradability index and farm level response 42

Chapter 5: CONCLUSION 47

5.1 Conclusion 47

5.2 Future research 48

REFERENCE 49

APPENDIX 53

LIST OF TABLE

Page Table 3.1: The description of each variable and expected sign 27

Table 4.2: Annual growth rate of quantity of selected crops 33

Table 4.5: The mechanization rate in agricultural production activities 36 Table 4.6: Labor force in Vietnam by resident region in selected years

Table 4.11: Summary statistics of variables in farm level model 42 Table 4.12: The correlation among coefficients in farm level model 43 Table 4.13: The correlation result of each variable (VIF, TOL=1/VIF) 43 Table 4.14: Production function result of the farm-level analyses

(dependent variable = yield of an important crop in a farm) 44

LIST OF FIGURE

Page Figure 2.1: The linkage between agricultural productivity and

Figure 4.2: The land of annual and perennial crops 1990-2012 in Vietnam 34 Figure 4.3: Export value of main agricultural commodities since 1990-2011 38 Figure 4.4: The import value of main agricultural commodities in Vietnam

Figure 4.5: The relationship between maize yield and tradable index over time 42

Chapter 1 INTRODUCTION

1.1 Problem statement

Trade helps people, regions, and countries exchange what they have and what they need In food security side, world population might be reach at 9 billion people in 2050; the question is asked how to meet this great demand of food with limited producers, scare land and water resources The answer is productivity, and open trade may encourage farmers to increase quantity of agricultural products to meet a requirement of food in the world

The core of role of agriculture can summary in three issues: “raising productivity, providing market, and generating saving for economic diversification” (Johnson, 2009) However, traditional farmers have suspicions about commercialization process that generate new demand, output or more competition

An increase agricultural productivity has attracted many economists studying about role of it in development economics in years (Matsuyama, 1992; Machicado et al., 2008) In addition, agricultural productivity has an essential role in industrialization and development economics That means country improves its productivity with using less labors in agriculture, and access labors in agriculture transfer into manufacturing Furthermore, when producers in agriculture sector increase their incomes thanks to efficient production, demand for manufacturing increases to meet more demand On the other hand, some argue that agricultural productivity has a negative relationship with industrialization Field (1978) and Wright (1979) indicate labor force can be a main fight between manufacturing and agriculture sector because of comparative advantage When agriculture has low productivity, manufacturing has an abundant supply of labor with cheap wage

Matsuyama (1991) shows explanation about these conflicting debates may be relative to opened economy He debates that in closed economy an increase agricultural productivity might make agricultural labors shift to manufacturing and contribute to economic growth However, in open economy “high productivity and output in the agriculture sector may, without offsetting changes in relative price,

squeeze out the manufacturing sector and the economy will de-industrialize over time, and in some case, achieve a lower welfare level.”

From Doi Moi 1980, Vietnam agriculture has improved from production and especially exporting (Nguyen, 1998) Agriculture sector shifts from self-sufficient to commercialization to supply both domestic and export markets Vietnam became the second biggest rice exporter in the world, and production of coffee, pepper, rubber, cashew nut, fruit and vegetable have increased in quantity and export value During 1980-1990, agricultural export value increased from 339 US$ million to 2,404 US$ million, and total value of trade increased 3.89 times Quantity of export rice went rapidly up from 1 million tons in 1990 to 3 million tons in 1997 Coffee increased from 90,000 tons to 404,000 tons, cashew increased 27,400 tons to 99,000 tons When Vietnam joined World Trade Organization (WTO), the agricultural export coffee went

up 1,256,400 tons, and 178,500 tons for cashew nut, respectively These evidences may support for advantage effects of opening trade in agricultural Vietnam Furthermore, the question is asked that how trade affects productivity in individual farm household

Correspondingly, this study tends to analyze the linkage between agricultural trade and productivity in Vietnam, contributes to debate above and tries to answer whether international trade (import and export) in agricultural commodities is related

to agricultural productivity at national and farm level This paper applies the product tradable index measurement to represent international trade In order to analysis of productivity, country and farm level analyses are implemented with time series and cross-section analysis in Vietnam

1.2 Research objective

This thesis concentrates on studying and evaluating effect of trade on agricultural productivity in Vietnam Some questions, which this thesis tries to answer: whether international trade increases agricultural productivity in Vietnam?

In order to answer these questions, this thesis uses two different levels of analysis from overview to detail to understand clearly how trade influences agricultural productivity

i) To evaluate the effects of international trade on some main commodities’

productivity at national level, and

ii) To analyze the relationship between trade and agricultural productivity at

farm household level

1.3 Thesis structure

Thesis includes five chapters This chapter introduces the effects of trade on agricultural productivity, and research objectives for this thesis Chapter 2 provides the review of theories and empirical studies relative to effects of trade on agricultural productivity Chapter 3 describes the measurement of tradability index, and other control variables may need in farm level model in detail The econometric model for national level and farm level are also introduced in this chapter Data for each model and method to estimate each model will be provided In chapter 4, the overview of Vietnam agriculture will be represented The next in this chapter, it shows the results

of two models, and discusses how trade influence agricultural productivity in Vietnam Basing on outcome in chapter 4, chapter 5 summarizes these conclusions and gives the possible future research can support for this thesis

Chapter 2 LITERATURE REVIEW

to GDP at that country The challenge for increase agricultural production is the scarce resources such as lack of water, uncultivated land Therefore, yield refers as an increasing quantity with same cultivated land, combination of effective inputs, applying new technology that may solve this problem (Mundlak, Butzer, & Larson, 2008)

2.1.1.2 Factors affect agricultural productivity

There are several methods to increase agricultural productivity Firstly, increasing output and inputs as well, however, proportionate of increasing output is larger than inputs Secondly, an increase output with constant inputs Thirdly, decreasing output and inputs with inputs decreasing more Finally, decrease inputs with remain output (Adewuyi, 2006) Using effective inputs lead to increase output method requires technical progress and inputs quality For example, applying new technology in production, investment in machinery, applying new technical method, irrigation system, use of fertilizer and pesticides, etc must be considered to increase productivity The recent discuss papers emphasize in role of technology and its changes over time (Mundlak, 1992)

The concept of technical efficiency rises due to scarce inputs in agricultural production The question of technical efficiency is that how farmers utilize given inputs to increase output According to Farrell (1957) divided productive efficiency

into its technical, allocative and scale components The level of technical efficiency is measure as a gap between actual and optimal potential production Allocative (price) efficiency refers as the capacity of farmers to choose inputs in minimize the cost of production Scale efficiency is defined that an increasing productivity due to an increasing farm size Both parametric and non-parametric methods are used to measure productivity efficiency, while the stochastic frontier model is used widely

Odhiambo & Nyangito (2003) had a review of factors determine productivity involving resources inputs, fertilizer use, market access, extension services, farm size, biophysical factors, and land tenure While the market access ability of farmers must

be considered in agricultural process The commercialization in agriculture influences

on productivity through specialization and intensification This point of view will be discussed in next part in detail to get the idea of effects of trade on agricultural productivity

2.1.1.3 Agricultural productivity measurement

Agricultural productivity is measure in different ways, and it can determine in physical term or value term In economics, agricultural productivity is described as the ratio of outputs to inputs/land and uses labor productivity or yield to measure productivity that also called the partial measures productivity Dharmasiri (2009) applies the Average Productivity Index (API), which examines the different views of productivity consist of productivity of land, labor and capital In many researches, they use total factor productivity (TFP) as a measurement of agricultural productivity (Teweldemedhin& Van Schalkwyk, 2010)

Land is an important factor in agriculture Productivity of land can increase through increasing of seeds, fertilizer, chemical pesticides, and labor Farmers can diversify crops in agricultural land to increase productivity Labor plays an essential role in livelihood of people relative to agricultural production It often is measured by hours/days of work needed to produce a unit of product Labor productivity is described as the total output per unit of labor Capital (purchase of land, investment in land, drainage, irrigation system, seeds, agricultural equipment, etc.) is a priority factor to increase agricultural productivity Human capital also is concern in measure

productivity Correspondingly, agricultural productivity is impacted by physical, socio-economic and technology (Dharmasiri, 2009)

The partial productivity is measured as output divided by a single input; therefore, this measurement has many formulas depending on a particular input such

as labor productivity, capital productivity, and land productivity The commonly used partial measure is output per unit of land or crop yield for short Crop yield is usually used as a comparison among locations/countries or among periods However, the partial measurement has some limitations The measurement is meaningful if other factors are unchanged Because land and labor productivity may increase by increasing of other factors such as fertilizers, tractors, technology, management of water, human capital, etc Therefore, the multifactor productivity, which is also called the Total factor productivity (TFP) is applied in order to solve these problems

TFP is usually used to measure agricultural productivity The traditional measurement of TFP assumes the output is technical efficiency, by contrast, the recent approaches allow inefficiency of productivity There are four methods to estimate TFP: estimation of aggregate production function, TPF index, Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis (SFA)

Aggregate production function approach assumes that there is no technical inefficiency The growth of TFP only includes technological change (Solow, 1957) Scale change may be included in some researches The technological change is estimated by adding time trend variable into aggregate production function or using the growth accounting method (Solow, 1957) Scale change is estimated as the sum of elasticities of inputs Aggregate production function is popular usage, however, it do not reflect the technical efficiency change

TFP index was introduced by Hicks (1961) and Moorsteen (1961) It measures

as ratio of growth rate of output to growth rate of total inputs Measuring TFP index is simple and does not need a complex estimation technique In addition, the TFP index needs the output and input prices, which usually is not available in most countries

DEA method applies the linear technique to estimate the production frontier based on dataset This method does not require any information about output and input

prices or any a specific form of production function However, it depends on a combination of efficient production, hence, it is sensitive to outliers

SFA is a method of estimating the stochastic production function in which the error term is decomposed into random noise and technical inefficiency Darku, Malla

& Tran (2012) applied a SFA approach and decomposed TFP into scale effects, technical efficiency change and technical change

2.1.2 Agricultural trade

In general, economists state that the integration into world market will make an increased productivity of firm through increasing competition between countries/industries, market share, innovation and technology, and spillover (Wong, 2009) When country joins world market, it has to face the competition from other countries, which may be higher quality of goods with cheaper prices To respond more competition, firms have less competitive capacity may not survive in competitive market However, domestic firms/industries can get benefits from lower tariff barriers such as they can access higher technology from higher developed countries with lower price, lower price of material of production processes, the higher skilled labors from abroad Moreover, firms/industries with higher capacity to compete in world market, they have an incentive to export their goods with higher prices

David Ricardo introduced the comparative advantage theory comparing to the absolute advantage theory in 1817 The theory answers the question why private individuals or firms engage in trade, governments favor it and economists defend it A person has a comparative advantage at producing something if he can produce at a lower cost in comparison with another However, we consider the case that two countries produce two products The question is how to know which country should specify which product if one country have a lower cost at both products The answer is

“opportunity cost” which refers as the cost of a good we give up to produce another product The country has a comparative advantage in one product which has a lower opportunity cost Accordingly comparative advantage, if farmers increase their productivity, their comparative advantage in agricultural production increase Therefore, farmers in that region can export more these products into the world market

The challenges effecting farmer’s decisions are the price of inputs, price of output, markets, and how to increase productivity International trade will contribute

to balance agricultural price in both parties For the importing country, opening trade will push the market price lower and domestic consumers gain from free-trade By contrast, domestic producers lose in free-trade because the price of products decrease

as well as quantity supplied For the exporting country, the shift from no trade to free trade increases the price Therefore, the producers make the profit from selling product with higher price, and consumers lose well-being On the other hand, international trade will bring the benefit for domestic country such as advanced technology, better inputs quality, and managerial skills (Wong, 2006)

The project for agriculture commercialization and trade (PACT) in Nepal wants

to help farmers access to new markets and build a strategic to improve productivity and quality PACT’s target is improving of competitive ability of farmers and agribusiness within selected commodities value chain The consequences of four years implementation PACT was increasing volume, sales, and productivity of selected goods

Term of international trade is usually measured as nominal imports plus exports relative to nominal GDP (Alcala & Ciccone, 2004) They debated that the nominal value will make a misleading about effects of trade Therefore, the “real value” is considered to measure international trade that is imports plus exports in exchange rate relative to GDP in purchasing power parity The “real value” is implemented due to eliminate the distortion of different prices in different regions

There are many studies about the role of international trade in development economics as general and productivity as particular Consequently, many papers show the positive relationship between international trade and productivity Some researchers use cross-country or time-series analysis to estimate the trade and productivity linkage (Teweldemedhin& Van Schalkwyk, 2010) While another use specific country data for analyzing international trade and productivity Most of paper uses the total factor productivity (TFP) to measure the level of productivity (Shevtsova, 2010); however, some others use openness as trade measurement (Alcala

& Ciccone, 2004; Wong, 2006) According to paper of Fleming & Abler (2012), they

applied the definition of a product-specific trade exposure index as measurement of trade on agricultural products in particular year in Chile case The trade exposure index is calculated import plus export quantity divided agricultural production

2.2 Empirical study

Alcala and Ciccone (2004) emphasize that the real openness which refers as import plus export relative to purchasing power parity GDP is better than openness only to measure trade because the real openness reflects the real value of non-tradable goods in different countries As a result, paper proves that the effect of international trade on productivity in cross-countries is positive significant and economically The paper also provides evidence that the productivity have a positive significant with size

of country when international trade accounts

Sara A Wong (2009) investigates how trade effects on manufacturing productivity at that country, especially at Ecuador when country opens their market to global market The study emphasizes on how export-import sector responds to opened trade Consequently, the result of study found that trade openness positively affects productivity in export-oriented manufacturing industries in Ecuador

Shevtsova (2010) uses the micro-data to estimate the relationship between export and productivity at firm-level in manufacturing and services sectors during period from 2000 to 2005 The study tests two hypotheses First is self-selection effect which estimates productivity effect before entering export market and result shows that the firm with higher TFP has higher incentive to enter export market Second is learning by exporting effect which estimate productivity effect in entry export-market period and result is the same with first hypothesis; however, some industries show that

no productivity gain after entry export-market

For the agriculture sector, the efficiency of international trade may bring benefits for increasing agricultural productivity Barbara Coello (2007) analyses the impact of international trade on export of agricultural commodities in rural Vietnam during 1993-1998 The study uses price of commodities as trade measurement on local prices, and then it affects on household income, productivity and profits As a result, international trade affects farmers in different way depending on product

specification and geographical location such as in Southern region of Vietnam tradable agricultural goods are more sensitive to international prices

Hassine et al (2010) recently imply Computable general equilibrium (CGE) model to estimate the relationship between trade liberalization, agricultural growth/productivity, and poverty alleviation The study tends to examine the effects of trade openness on agricultural productivity and analysis how agricultural productivity contributes to poverty reduction in Tunisia

“Does agricultural trade affect productivity?” is the title that Fleming & Abler (2012) used to estimate whether international trade influences agricultural productivity The commodity trade exposure index, which is applies in cross-section model with 70,000 individual household extracted from Chile agricultural census, refers as the proportion of imports and export to total agricultural production In addition, they analysis two groups: one focuses on traditional products, another focuses on traditional and non-traditional products In addition, they used the production function to show the relationship between trade and yield The results support for hypothesis of trade effecting yield However, the effect of trade on households planting traditional and non-traditional crops is larger than households planting only traditional crops

Nirodha et al (2013) provides an evidence of effects of trade liberalization policies on agricultural production growth in Sri Lanka This paper applies the Ordinary Least Square and multiple regression models with series data from 1960 to

2010 The analysis divides into two periods before and after liberalization: 1960-1977 and 1977-2010 The econometric result indicates that the trade openness, investment are statistical significant positive with agricultural growth and eventually lead to increase agricultural productivity in Sri Lanka

The question is how trade liberalization can impact to poverty through increasing agricultural productivity that is the main purpose of Nadia’s paper (2008) The paper provides the link between trade and agricultural productivity, and then estimates the link between agricultural productivity and poverty reduction It applies panel data for Mediterranean regions, and uses the talent class stochastic frontier model to estimate Technical Efficiency and Total Factor Productivity The data

consists of 36 agricultural commodity and inputs variables such as labor, fertilizer, agricultural land, irrigation, and machinery The results indicate that trade openness has a positive influence to agricultural productivity though technology and help to reduce poverty

There are many researches that evaluated the effects of agricultural trade on poverty reduction through increase income Nadia et al (2010) also provide an evidence of linkage between trade and productivity, and then they applied a Computable General Equilibrium (CGE) in order to associate the trade-productivity linkage with poverty in Tunisia The results support an increasing agricultural productivity thanks to opening trade through the transfer of technology from advanced countries It also provides the inverse impact of policies on agricultural productivity The contribution of other control variables such as education, farm size, technology, political stability, control of corruption, and government effectiveness must be consider in agricultural productivity growth

2.3 Conceptual framework

At the household level, the factors effect directly on agricultural productivity including farmer characteristic (education, age, gender, experience, health, etc.), farming characteristics (farm size, type of land, tenure, etc.), inputs (fertilizers, pesticides, capital of investment, seed, human capital), technical process, market access, biophysical factors (soil type, climate, rainfall, etc.) Many researches determines factors effect on productivity such as land, capital, and labor (Yair et al 1997; Majeed and Afaf 2001); human labor, farmyard manure, chemical fertilizers, water for irrigation, transportation, electricity, diesel fuel, and machinery (Banaeian & Zangeneh, 2011) Seed, fertilizers, manure, human labor, animal labor, machine, pesticides, irrigation and land (Elumalai, 2011)

As mention above, the method to increase productivity is an increase output with reduce/given level of inputs And technology change may a result of effective inputs usage

At the macro level, international trade affects the price of products The export country will obtain more profit when sells its product with higher price in international markets Therefore, producers have an incentive to increase their product

though increasing productivity In agricultural sector, price of agricultural products is influenced by trade, farmers are encouraged to improve productivity Farmers apply new technique to use resource inputs efficiently such as mechanization, new technology for increasing quality of inputs While international trade openness an opportunity for domestic country though technology transfer Figure 2.1 describes the linkage among factors can influence agricultural productivity based on the theories and empirical studies which are discussed above

Figure 2.1: The linkage between agricultural productivity and international trade

Agricultural productivity

Inputs use

International Trade Price

Outputs

Technology

Technical efficiency Allocative efficiency Scale efficiency

Chapter 3 RESEARCH METHODOLOGY

Many studies apply diversified approaches to evaluate effects of trade on agricultural at many respects of economics and society Generally, most of studies use Total Factor Productivity (TFP) as to measure productivity in agricultural sector (Teweldemedhin & Van Schalkwyk, 2010) However, this study considers crop yields represented agricultural productivity in Vietnam Crop yields are not perfect to measure productivity but it reflects productivity and available variables The reason why crop yield is measured as agricultural productivity will be present in below section

Trade variable commonly measures as nominal imports plus exports relative to nominal GDP or also called “openness” However, if we use nominal variables when measure trade, it will make misleading in evaluation Therefore, “real openness” replaces “nominal openness” to estimate international trade This study uses the same method to estimate capacity of agricultural goods trading Thus, trade variable is described as the share of imports plus exports in the total production of a specific commodity, referred to as a product tradable index (TI)

In order to overall view about the effect of trade on agricultural productivity, this study provides two models at national and farm levels For national level, model tries to find out how trade effects on each commodity in 1961-2010 For farm level, model goes into detail how farms respond effects of trade through agricultural yields For each model, this study uses a specific trade variable, which will be described in detail in next section

3.1 Agricultural productivity measurement

Agricultural productivity is usually measured as a ratio of output quantities to input quantities In general, there are two main types to measure agricultural productivity, distinguished by their handling of inputs: partial productivity and aggregate productivity The ratio of output to a single production factor (input) is called the partial productivity; in other hand, the ratio of output to all production factors is called the aggregate or multifactor productivity

The partial productivity is measured as output divided by a single input; therefore, this measurement has many formulas depending on a particular input such

as labor productivity, capital productivity, and land productivity The commonly used partial measure is output per unit of land or crop yield for short Crop yield is usually used as a comparison among locations/countries or among periods However, the partial measurement has some limitations The measurement is meaningful if other factors are unchanged Because land and labor productivity may increase by increasing of other factors such as fertilizers, tractors, technology, management of water, human capital, etc Therefore, the multifactor productivity, which is also called the Total factor productivity (TFP) is applied in order to solve these problems

TFP is a ratio of an index of agricultural output to an index of aggregate agricultural inputs However, TFP differs from Partial productivity in that they use a value-weighted sum of output and input to measure index of output or inputs The TFP includes other factors may effect on productivity as land, labor, fertilizer, pesticide, physical capital, irrigation, etc Although TFP has an advantage in measuring productivity, TFP is much more complex than partial productivity Some researchers used yield to measure productivity (Fleming & Abler, 2010; Banaeian & Zangeneh, 2011; Dharmasiri, 2009) Yield is not perfect to represent agricultural productivity, however the underlying data is often easily available in over years Moreover, farmers in Vietnam trend to increase land of agricultural crops which have

a high price in order to increase their profit For example, in 1999 the land for pepper was 15,000 hectare and increased 45,000 hectare in 2003 Maize increases from 730,000 hectare to 1,156,000 hectare during 2000-2012 Therefore, yield is used to estimate agricultural productivity in Vietnam for this thesis In addition, in farm level model the other variables as irrigation, human capital, labor, cost of production, machinery are used as the control variables

3.2 The tradability Index

In general, the tradability index is measured as formula below

𝑇𝐼𝑖𝑗 = (𝐸𝑥𝑝𝑜𝑟𝑡𝑖𝑗 + 𝐼𝑚𝑝𝑜𝑟𝑡𝑖𝑗)/𝑇𝑜𝑡𝑎𝑙 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛𝑖𝑗 (1)

Where 𝑇𝐼𝑖𝑗 is the tradability index of commodity i in year j, and TI is calculated as the summation of exports and imports quantity over the total local

production of commodity i in year j The result responds as zero in calculation; it means that commodity consume in the country TI is infinite when the products are imported completely into the country

This thesis uses two models at different level to evaluate the effects of trade on agricultural productivity: national level, and farm level At national level, the TI is used as equation (1) and bases on the FAOSTAT database to calculate the TI for each commodity

At farm level, trade variable is denoted as FTI for Farm tradability index and is calculated by dividing the total sale for the total production quantity

3.3.1 The national level model

National-level analysis is used in this thesis to examine whether international trade affects agricultural productivity in general This analysis tries to prove the hypothesis that is: the country faces the higher the tradable index in agricultural commodity, its yield is higher over time To examine the correlation between trade and agricultural productivity FAOSTAT database is used in this examination

𝑌𝑖 = 𝛼0+ 𝛼1𝑇𝐼𝑖 + 𝑒 (3)

Where Yi is yield of commodity i, 𝑇𝐼𝑖 is tradable index of commodity i, and e

is an error term In this model, TI is estimated following equation (1) The analysis concentrates on nine commodities: rice, pepper, vegetable, cassava, cashew nuts, tea, fruit, maize, and coffee Because these commodities are traded the most in Vietnam For example, in 2011 rice exported 7.72 million tons over 27.15 million tons of total production Philippines, Malaysia, and Indonesia are the main imported markets of

Vietnam This model is estimated for each commodities during the period 1961-2010

in order to understand that the changes of yield may come from trade or not

3.3.2 Farm-level model

Farm-level analysis has the same purpose with national-level analysis, however, it expresses in individual producer Farm model analysis will base on production function 𝑌 = 𝑓(𝑋1, 𝑋2, 𝑋3, … , 𝑋𝑛) Where Y is a quantity of output and Xs are vectors of inputs In addition, production function can be express as the relationship between one output and one input or multiple inputs A production function do not have a specific form of function, it is depend on what combination of inputs produce which output There are many form of production function, however the Cobb-Douglas production function is frequently used in estimate productivity The Cobb-Douglas model was represented firstly in 1928 by Charles Cobb and Paul

Douglas in the journal American Economic Review The study tested the model of the

growth of American economy during 1899 – 1992 The model showed the production output depended only two inputs which are capital and labor, while there are many other factors effect on production output

This farm level model applies a Cobb-Douglas production function to express the relationship between dependence and independence variables Y= 𝑎𝑋1𝑏𝑋2𝑐… while

a is total factor productivity; b, c are the output elasticity of inputs, respectively The Cobb-Douglas function can be expressed as a translog function:

ln(𝑌) = ln(𝑎) + 𝑏 ln(𝑋1) + 𝑐𝑙𝑛(𝑋2) … The production function is constant return to scale when b + c + = 1, meaning that the double of using inputs lead to double output as well, b + c + > 1 is increasing return to scale, and b + c + < is decreasing return to scale

I replace Y as crop yield, and X1, X2, consist of irrigation, labor, machinery, amount of production capital, sex, education In addition, the objective of this analysis

is that farm’s productivity depends on trade Hence, the farm tradable index will be added in equation (2) and model assumes to follow a constant return to scale, we have formula as below:

ln(𝑌𝐿𝐷) = 𝛽0 + 𝛽1(𝐹𝑇𝐼) + 𝐵2𝐿𝑛(𝑇𝐿𝑎𝑛𝑑) + 𝛽3𝐼𝑅 + 𝛽4ln(𝐿) + 𝛽5(𝑑𝑀) +

𝛽6𝑙𝑛(𝐾) + 𝛽7(𝑑𝑆𝐸𝑋) + 𝛽8𝐿𝑛(𝐸𝐷𝑈) + 𝑒 (4)

Where YLD is yield of an important crop, FTI is farm tradable index, Tland is total cultivated land of important crop, IR is farm area covered by irrigation, L is a number of household member relative to agricultural production, dM is a dummy variable of presence of use of machinery, K is amount of production capital, dSEX is the sex of farmer, EDU is the average level of education of farmers who join in agricultural production, e is error term These variables will be described in table 3.1 involving their units, sources, expected sign, and definition

The second model represents a household level and uses dataset extracted from The Vietnam Access to Resources Household Survey 2010 (VARHS) VARHS is implemented by the University of Copenhagen, CIEM, ILSSA, and CAP-IPSARD This survey supplies detailed information of agricultural households It consists of questions as in VHLSS and additional questions about land, agriculture, income, spending, assets, investments, market linkages, etc VARHS has implemented since

2006, 2008, 2010 and 2012 For this study, VARHS 2010 is used in farm model In general, VARHS 2010 reports 3,202 household’s data with 2,200 panel households at twelve provinces including Ha Tay, Lao Cai and Phu Tho, Lai Chau, Dien Bien, Nghe

An, Quang Nam, Khanh Hoa, Dak Lak, Dak Nong, Lam Dong, Long An Because farms may not produce one crop but many crops, the farm model uses yield of a specific commodity, which has highest proportion of yield among crops in one household In collecting data process, household sample size remains 1450

observations at 9 provinces (Dak Lak, Dien Bien, Ha Tay, Khanh Hoa, Lai Chau, Lao Cai, Nghe An, Phu Tho, Quang Nam) because I eliminated observations what are not available data

As I said above, total land of main crop and quantity of main crop are extracted from VARHS 2010 in order to calculate the yield variable for farm analysis Firstly, I calculated total land of each crop and then use quantity of each crop divided to total land of each crop that I had kilogram per square meter of each crop, converted it to tone per ha I have yield of each crop Secondly, I chose the highest proportion of yield among crops in one household, I had yield variable of main crop in one household

Farm tradable index was calculated by sale quantity divided to production of main crop The sale quantity and production can be taken out VARHS 2010: what was the total quantity of output produced? And what was the total quantity sold or bartered?

For irrigation variable, I measured it as the ratio of total irrigated land to total land I summarized data from questions in VARHS: is this plot irrigated? What is the total area of this plot? Then I calculated the total irrigation land and total area land, apply the formula I had irrigation variable The calculating process a household land may include many types of land such as: annual crops land, perennial crops land, forestry land, house with garden, residential land, grass land, and fish and shrimp pond I just took a sum of annual crops land, perennial crops land and house with garden as total land relative to crop production In addition, the irrigated land also depends on those kinds of land

In employment part of VARHS data, I extracted information: Are there members of household who participate in household production related to agriculture, forestry, and aquaculture? Therefore, I calculated how many people join household agricultural activities The labor variable was built up

For the dummy of presence machinery in household production, I extracted information from question: how many of following items does your household have? There includes machines for agricultural production such as: rice milling machine, grain harvesting machine, pesticide prayers, tractor Denoting the number 1 in order to present available machines, and 0 for no machine in crop production

In farm model, this thesis considers sex of farm head and level of education of member in household relative to agricultural production in their household Sex available is presented as dummy variable with 1 denoted as male and 0 denoted as female While education variable is measured as the average level of education of total agricultural labors in one household

The amount of production capital is calculated from the table of question about the amount of using inputs for total production, rice, and maize I had collected the main crop for individual household, basing on main crop I calculated as following The household plants one crop or main crop is rice/maize that is easy to collect production capital However, if household plants more than two crops and main crop

is not rice/maize, I divide two cases One is that household has rice or/and maize, and other crop, I get total amount investment for production minus amount of investment

of rice/maize One is that a household plants more than two crops, which does not have rice/maize, I decided to eliminate this household out of data The amount of production capital includes amount of investment relative to production main crop Table 3.1 provide a summary of variables, their unit, source, and expected sign

Table 3.1: The description of each variable and expected sign

sign

Y i Yield of commodity i (rice,

maize, etc.)

Tons/ha FAOSTAT

Tland Total cultivated land area M2 VARHS

EDU The average level of education

of household labors who join in agricultural production

VARHS

2010

+

3.5 Analytic method

The thesis divides into two separate analyses, one bases on time series analysis

to find out the correlation between international trade and productivity (crops yield) of some agricultural commodities in Vietnam over times Another one shows the relationship between trade and yield at the household level basing on cross section analysis Therefore, this thesis uses two different ways to run two regression