Mangrove and production risk in aquaculture in mekong river delta, vietnam

ABSTRACT Utilizing survey data in aquaculture activities in 2014 from the Mekong river Vietnam, this paper aims to examine the impact of mangrove forests on profit and profit variability

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

MANGROVE AND PRODUCTION RISK IN AQUACULTURE IN THE MEKONG RIVER

DELTA, VIETNAM

BY

DO HUU LUAT

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, DECEMBER 2015

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

MANGROVE AND PRODUCTION RISK IN

AQUACULTURE IN THE MEKONG RIVER

DELTA, VIETNAM

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

DO HUU LUAT

Academic Supervisor:

TRUONG DANG THUY

HO CHI MINH CITY, DECEMBER 2015

DECLARATION

This declaration is to certify that this thesis entitled “Mangrove and Production risk in

aquaculture in the Mekong river delta, Vietnam”, which is submitted to fulfill the

requirements for the degree of Master of Art in Development Economics to the Vietnam – The Netherlands Programme (VNP), constitutes only my original work only All materials used in this thesis have been acknowledged and cited properly following the Programme’s standards

DO HUU LUAT

“Success is no accident It is hard work, perseverance, learning, studying, sacrifice and

most of all, love of what you are doing or learning to do.” – Pele

This thesis has been finished thanks to supporting and motivation from many people

First of all, I am proud of me and would like to give my first thank to myself Two years ago, I did not believe that I could complete this course as well as a scientific thesis I have overcome myself and struggled a lot in order to do this

Secondly, the authors are grateful to Economy and Environment Program for Southeast Asia (EEPSEA) for funding data collection

Thirdly, I would like to give the sincerest thank to my supervisor - Dr Truong Dang Thuy, who has given a precious opportunity to experience a study in reality Besides,

he allowed me to utilize his data to serve this thesis and gave invaluable suggestions to

me Moreover, I am grateful to all lecturers VNP and staffs VNP who have helped and taught me salutary knowledge over two years

Fourthly, I want to spend the best wishes to all my friends at VNP They have been accompanying me in the journey of learning and studying at VNP, this will be my unforgettable experience

Finally, I devote my thesis to my parents who always support and give me incentive regardless of what the way I choose

DO HUU LUAT

ABSTRACT

Utilizing survey data in aquaculture activities in 2014 from the Mekong river Vietnam, this paper aims to examine the impact of mangrove forests on profit and profit variability in extensive and semi-intensive aquaculture farms (mostly shrimp farms) The Just-Pope framework for a stochastic short-run profit function is applied to examine the impacts of inputs on both the deterministic component and the stochastic component of profit The most crucial characteristics of mangrove forests such as the area of mangrove forests in farm, the density of mangrove trees per 100 square meter, and the area of mangrove forests within 500, 1000, and 2000, are utilized in this paper The main estimation method is the Maximum likelihood (ML) estimator for the log-likelihood function employed to investigate the relationship between mangrove forests and profit as well as its variability Apart from the ML estimator, other estimation methods (including FGLS, Robust S.E, and SUR) are also employed to test robustness

delta-of the regression results The results show robust evidences that mangrove forests have negative effect and variance-reducing effect on profit in extensive and semi-intensive aquaculture farms From these results, it implies that when converting more mangrove into water surface area, farmers earn higher profit at higher risk, and that a risk-averse farmer will plant more mangrove forests in farm than the risk-neutral farmer

Keywords: Mangrove forests, Production risk, Aquaculture, Profit function

TABLE OF CONTENT

ABSTRACT i

ABBREVIATIONS iv

LIST OF FIGURES v

LIST OF TABLES vii

CHAPTER 11 INTRODUCTION 1

1.1 Problem statement 1

1.2 Research objective 4

1.3 Research questions 4

1.4 Scope of the paper 4

1.5 Structure of this thesis 4

CHAPTER 25 LITERATURE REVIEW 5

2.1 The ecological functions of mangrove forests 5

2.2 The economic value of mangrove systems 9

2.3 The theory of production risk 12

2.4 Empirical studies 14

2.4.1 The impact of mangrove forests on production process 15

2.4.2 Empirical studies about production risk 16

CHAPTER 3 RESEARCH METHODS 22

3.1 Overview of the Mekong river delta 22

3.2 Analytical framework 24

3.3 Neoclassical economic theory 24

3.3.1 Two approaches in production economics 24

3.3.2 A dual approach - The profit function 26

3.4 Research methods 30

3.4.1 Model specification 30

3.4.2 Data collection method 35

3.4.3 Variable description 36

3.4.4 Estimation methods 39

CHAPTER 4 RESEARCH RESULTS 50

4.1 Descriptive statistic 50

4.2 Bivariate analysis 55

4.2.1 Mangrove forests versus profit 55

4.2.2 Output and input prices versus profit 59

4.2.3 Fixed inputs versus profit 61

4.2.4 Operator’s management ability versus profit 62

4.3 Estimation results 64

4.3.1 Testing for production risk 64

4.3.2 Regression results for the effect of mangroves on profit per square meter 66

4.3.3 Regression results for the impact of mangrove forests on the profit variability in aquaculture production 73

CHAPTER 5 CONCLUSION AND POLICY IMPLICATION 80

5.1 Conclusion remarks 80

5.2 Policy implication 82

5.3 Limitations and further research 82

REFERENCES 84

APPENDIX 91

Appendix A: Questionnaire 91

Appendix B: Test for the presence of heteroskedasticity 94

Appendix C: Correlation matrix and testing for multi-collinearity 101

Appendix D: Technical efficiency 102

ABBREVIATIONS

LIST OF FIGURES

Figure 1.1 Land use in the Mekong River delta, 2007 2

Figure 2.1 The ecological functions of mangrove forests are in seafood production 6

Figure 2.2 The mixed mangrove-aquaculture farming systems 8

Figure 3.1 Profit function with respect to output price 29

Figure 4.1 The distribution of observed aquaculture farms 50

Figure 4.2 Scatter diagram of profit per square meter on the ratio of mangrove area in farms 55

Figure 4.3 Scatter diagram of profit per square meter on the density of mangrove trees in farms 56

Figure 4.4 Scatter diagram of profit per square meter on the area of mangroves in 500 meter 57

Figure 4.5 Scatter diagram of profit per square meter on the area of mangroves in 2000 meter 57

Figure 4.6 Scatter diagram of profit per square meter on output price 58

Figure 4.7 Scatter diagram of profit per square meter on chemical price 59

Figure 4.8 Scatter diagram of profit per square meter on fry price 59

Figure 4.9 Scatter diagram of profit per square meter on total area 60

Figure 4.10 Scatter diagram of profit per square meter on family working hours 61

Figure 4.11 Scatter diagram of profit per square meter on age of household head 62

Figure 4.12 Scatter diagram of profit per square meter on Operator’s schooling years 62

Figure 4.13 Scatter diagram of production risk on the ratio of mangrove area in farm 73

Figure 4.14 Scatter diagram of production risk on the density of mangrove trees in farm 74

Figure 4.15 Scatter diagram of production risk on the area of mangroves in radius 500 meter 75 Figure 4.16 Scatter diagram of production risk on the area of mangroves in radius 2000 75

LIST OF TABLES

Table 2.1 Total economic value of a mangrove resource 11

Table 3.1 The primal and dual approach 23

Table 3.2 Data matrix 33

Table 4.1 Summary of the sample data 51

Table 4.2 Descriptive statistics of the variables in the sample data 53

Table 4.3 Heteroskedasticity tests with the models 65

Table 4.4 Regression results of mangrove forests and profit per a square meter (MLE) 67

Table 4.5 Regression results of mangrove forests and profit per square meter (SUR) 69

Table 4.6 Regression results of mangrove forests and profit per square meter (FGLS and Robust S.E) 72

Table 4.7 The impact of mangrove forests on the variability of profit (Production risk) 76

Table 4.8 The impact of mangrove forests on the profit variability using other methods (FGLS, Robust S.E, and SUR) 79

CHAPTER 1 INTRODUCTION 1.1 Problem statement

Mangrove forests are primary ecosystems along coastlines, riverbanks in the tropical and subtropical regions of the world They provide protection to deal with extreme climate problems such as storms, floods, and tsunamis About 30 mangrove trees per

100 square meters with the depth of 100 meters may prevent the damage of a tsunami

up to 90 percent (Hiraishi and Harada, 2003) Based on biodiversity, mangroves also provide good habitats as well as nutrients to flora and faunal species, for instance birds, monkeys, and snakes Moreover, mangrove forests can alleviate erosion of riverbank shoreline and alleviate the rising of sea level with allowable and reasonable cost (e.g Khail, 2008; Paola, 2012) In addition, mangrove forests contribute a significant part to income of households who live nearby mangrove forests Barbier (2007) found that people could earn $12,392 per hectare of mangrove forest, economic annual value, from wood products, fishery and non-wood products (e.g honey, nipa palm)

Nowadays, the area of mangrove forests have been significantly shrunk worldwide since the mid-twentieth century Specifically, over one-fifth of mangrove forests have

been lost since 1995 (Spalding, Kainuma, & Collins, 2010), and most of mangrove

forests decrease occur in Southeast Asia and Latin America For example, 70 percent

of the mangrove forest was diminished in the period from 1920 to 1990 in Philippines, while this rate in Malaysia was around 17 percent during the period of 1965-1985 (WRI 1996) The area of mangrove forests in 1993 have existed about 54 percent in comparison with this one in 1975 in Thailand (Sathirathai, 1998)

In Vietnam, the rapid reduction of mangrove forests occurred in the last century Do (2005) showed that the mangrove forests area was as much as 400,000 ha in Viet Nam

in 1940s, but this area was reduced to only 170,000 ha in 2010 (Phan and Quan, 2012) There are many reasons for the decrease in mangrove forest areas such as conversing

of forest land to economic activities, harvesting timber products, and increasing in

population; whereas the major factor is the expansion of aquaculture ponds into mangrove forests (Spalding, Blasco, and Field., 1997; Lewis et al., 2003) Besides, the deforestation of mangrove forests has significantly increased in recent years in South Asia due to the use and scale of forest products of local users (Giri et al., 2011)

Figure 1.1 Land use in the Mekong River delta, 2007

shrimp farming or crops cultivation However, the area of mangrove forests allotted to households was over-exploited because of the poor enforcement of regulations in Vietnam This over-exploitation has brought some serious problems to social as well as household’s welfare, for instance floods, hurricanes, and storms Therefore, the existence of mangrove forests in production area has contributed to household’s welfare and helped to alleviate the damage caused by natural disasters in production process

Researchers have recognized the importance of mangrove forests and the reality of deforestation Hence, most of studies have focused on calculating of the value of mangrove forests or finding a good way to reduce mangrove deforestation Furthermore, some studies attempt to investigate the behavior of households towards the conversion of some area of mangrove forests into other land uses Nevertheless, whether the existence of mangrove forests have effects on household production activities in the area of aquaculture or agriculture is still an open question

Therefore, this paper aims to investigate the effects on household production in aquafarming of mangrove forests in the Mekong river delta through estimating the profit function for various types of aquaculture comprising extensive, intensive and semi-intensive culture The proposed hypothesis is that mangrove forests will improve

or mitigate risks in aquaculture production through reducing cost of water treatment, mitigating damage of climate changes, etc Since mangrove forests are often planted inside and outside extensive and semi-intensive aquaculture farms (including mangrove forests and surface water), this thesis concentrates on analyzing the impacts of mangrove forests on profit and profit variability in these farms As a result, the study will provide information local farmers and authorities about the advantage of mangrove forests and propose recommendation to policy makers these kinds of aquaculture often contain a certain area of mangrove forest either into or around farms

1.2 Research objective

This paper aims to examine the effects of mangrove forests on profit and profit variability (production risk) in extensive and semi-intensive aquaculture farms (mostly shrimp farms) in Mekong river delta, Vietnam

1.4 Scope of the paper

This study will address the above research questions using cross sectional data from

205 extensive and semi-intensive aquaculture farms (mostly shrimp farms) in 8 locations in 6 provinces (Ben Tre, Ca Mau, Bac Lieu, Kien Giang, Soc Trang, and Tra Vinh) in 2014

1.5 Structure of this thesis

The paper consists of five chapters Chapter 1 introduces the problem and states the aims and objectives of the paper In Chapter 2, the relevant theories about the ecological services of mangrove forest and the theory of production risk will be reviewed Moreover, the previous empirical studies related to the effects of mangrove forests on production yield as well as the variability of output will be presented After that, data collection and methodology utilized in this study presented in Chapter 3 In Chapter 4, the empirical results will be interpreted and discussed Finally, in Chapter 5, based on the research results, the paper will draw conclusions and recommends suitable policies

for the conservation of mangrove forests

CHAPTER 2 LITERATURE REVIEW

This chapter briefly reviews theories about the ecological functions of mangrove forests and their role in the production process Additionally, the theory of production risk is also reviewed in this part To demonstrate these theories, this paper also represents empirical studies which are analyzed the impacts on economic activities of mangrove forests and estimated the effects of inputs on production risk

2.1 The ecological functions of mangrove forests

Mangroves are diverse huge and pervasive categories of trees and shrubs that live in the tropical and subtropical regions of the world They can easily adapt to difficult environmental conditions and are one of important plants in the ecosystem that have brought a high productivity “Mangroves provide a wide range of ecological services like protection against floods and hurricanes, reduction of shoreline and riverbank erosion, maintenance of biodiversity, etc.” (Rönnbäck, 1999) These functions help maintain economic activities in coastal areas and in the tropical region such as the Southeast Asia Besides, mangrove ecosystems provide local economic activities with natural products which are directly harvested, for example wood, aquatic products, and birds

In terms of aquaculture production, the mangroves support some diverse services for production activities through the mechanisms in Figure 2.1 Above all, mangrove forests play an important role in alleviating the turbulences of environmental conditions which can destroy aquaculture production in coastal areas such as floods, hurricanes, and storms

Secondly, mangrove forests help to sustain the quality of water This function helps to decrease levels of pollutants, mitigate variation in salinity and turbidity, etc Larsson et

al (1994) and Kautsky et al (1997)imply that the existence of mangrove forests in semi-intensive farms for shrimp production is necessary to maintain the water quality

Mangrove forests are also a safe habitat for predators as well as a good habitat for

mollusk species to seek food through this tangled root system (Nagelkerken et al.,

2008) In addition, mangrove forests provide nutrients to the coral reef and flora

communities This root system also reduces the flow of tidal water, causing the

deposition of sediment which may filter out and treat toxins It is easy to see that the

value of wastewater treatment via mangrove forests is better than the expense to set up

a new wastewater treatment system Hence, the integrated shrimp-mangrove farming is

being encouraged in recent years

Figure 2.1 The ecological functions of mangrove forests are in seafood production

Source: Rönnbäck (1999)

Moreover, mangroves can produce and supply food inputs to aquatic organisms in

aquaculture activities (Nagelkerken et al., 2008; Hong and San, 1993) The first, organic

material and nutrients may be directly served as a food input to fauna species in the

mixed shrimp-mangrove farms (e.g extensive and semi-intensive aquaculture)

Furthermore, they can be exported to specialized production areas (e.g intensive

aquaculture) Larsson et al (1994) found that the area of semi-intensive farm covers

Water quality maintenance

Flood and erosion control Nutrient assimilation Sediment trapping

Food input

Detritus

Fishmeal Broodstock

Seed

approximately 25% of the area of mangrove forests, whereas it provides about 70% the quantity of shrimp food in compared with a semi-intensive farm without mangroves And the rest of shrimp food is from the bacterial and fungal films on mangrove leaf detritus The last one is fish and mollusk species producing from mangrove forest systems can be used as a direct input food or a component in formulated foods in aquaculture

Finally, the natural seed production in mangrove forests perhaps express the crucial relationship between mangroves and aquatic organism farming The developing of juvenile fishes and shrimps in mangrove system can be explained by reasons such as the wealth nutrition, diminished water force caused by the shallow water ecosystems, and sophisticated tangible composition in mangroves (Beck et al., 2001) These features can affect natural hatcheries for fish and shrimp cultivation, extend density, and maintain the level of growth as well as the survival of seed Furthermore, the availability

of seed is low in mangrove forests can reduce the productivity of aquatic organism cultivation (Menzel, 1991) Kautsky et al., (2000) claimed that mangrove forests have directly affected on the productivity and sustainability of shrimp cultivation Consequently, these advantages of mangrove forests can affect the mangrove forest conservation behavior

Apart from the advantages of mangrove forests, many empirical studies have shown evidences of adverse effects of mangrove forests on aquaculture In reality, three systems which combines mangroves and aquaculture are integrated, associated, and separated mangrove-aquaculture farming systems (described in Figure 2.2) Johnston

et al., (2000a and 2000b) with technical evidences from the mixed mangrove-shrimp systems proposes that pond design, poor water quality and management technique are crucial factors in declining shrimp outputs in those systems Moreover, water quality and biodiversity in ponds have been affected directly from mangrove forests

In particular, the leaf-litter fall have crucial role to survival and growth of shrimps (Johnston et al., 2000b), and each species of mangrove brings different effects to

aquatic organisms in farms (e.g Basak et al., 1998; Tran and Yakupitiyage, 2005) First, dissolve oxygen was consumed significantly in decomposition process of leaf litter The decomposition rate of leaf litter also varies with species and different environmental conditions, and this happens faster in ponds than on lands (e.g Ashton, Hogarth and Ormond, 1999; Dick and Osunkoya, 2000) Second, the high of leaf litter loadings in ponds have contributed negative effects to aquatic organisms via reduced water quality, sediment quality, and body weight of shrimp (Fitzgerald, 2000 cited by Tran and Yakupitiyage 2005) In this way, it leads to decrease of natural food production in aquatic ponds (Lee, 1999) In ponds without aeration, aquatic organisms even died within 2 days if the loading rate of leaves is more than 0.5 g DM L-1 (Tran and Yakupitiyage, 2005) Finally, the low levels of dissolve oxygen in ponds can cause physiological inhibition which leads to low productivity in extensive farming In fact, survey on water quality in the mixed shrimp-mangrove forests systems found that the level of dissolve oxygen is low and ranges from 0,3 to 3,9 ppm (Roijackers and Nga, 2002)

Figure 2.2 The mixed mangrove-aquaculture farming systems, from left to right: a) the

integrated mangrove-aquaculture farming; b) the associated mangrove-aquaculture farming; c) the separated mangrove-aquaculture farming

Source: Bosma et al., 2014

In fact, compounds extracted from mangrove trees will have negative effects on survival and growth of aquatic organisms In particular, tannic acid extracted from mangrove leaves is the main poison which causes profit from aquiculture to reduce, especially Rhizophora leaves (Inoue et al., 1999 cited by Primavera 2000) According

to Madhu and Madhu (1997, cited by Tran and Yakupitiyage 2005), besides, aquatic organisms (e.g shrimp and crab) have been seriously damaged by compounds extracted from other aspects of mangroves trees such as root, bark and stems

In addition, the age of mangrove trees, the density of mangroves, and the mangroves coverage in farm may induce damage to aquatic organisms Binh, Phillips, and Demaine (1997) implies that the age of mangrove trees in farms which is older than 7 years will decrease shrimp profit owing to less nutrients producing when the trees are older Meanwhile, high-density of mangrove forests decrease fish yield due to creating a good habitat to attract predators (e.g birds and snakes) and reducing of plankton and benthic algae (Burbridge and Koesobiono, 1984 cited by Primavera 2000)

To sum up, the ecological functions of mangrove forests mayhap bring both advantages and disadvantages to household and social welfare Despite the opposition of many farmers to mangrove forests, there were compelling evidences suggest that mangroves forests play important role to deal with the unexpected changes of natural conditions which can damage production means (e.g ponds, seeds) Nevertheless, in terms of economics, the role of mangrove forests in aquaculture has still existed many shortcomings Therefore, specified mangrove-aquaculture models will bring various economic benefits to farmers

2.2 The economic value of mangrove systems

Based on the ecological functions of mangrove forests, the economic value of mangrove forests is classified as use values (direct, indirect and option values) and non-use values (Bann, 1998)

Direct use values are directly extracted from the natural environment (i.e mangroves)

to serve the production process and consumption These values can be calculated as the increased value in production They consist of consumptive and non-consumptive values Consumptive values are the values of extracted natural resources (e.g timber, firewood, fishing), whereas non-consumptive values that are defined as the services of mangroves (e.g tourism, education, scientific research) Direct use of mangroves can also be divided into commercial and non-commercial activities Non-commercial activities are always crucial for necessary needs for local households (Bann, 1998)

Indirect use values are benefits from the ecological functions of mangroves which support and protect to economic activities As mentioned previously, the main ecological services of mangroves in aquaculture activities are primarily determined as indirect use values Mangrove forests can control and protect against the impacts of floods, hurricanes too With the tangle of mangrove roots that could sustain and improve soil, mangroves may stabilize shoreline and decrease erosion of coastlines Besides, the pollution control function of mangroves could reduce and recycle pollutants and human waste, then maintain the water quality The economic value of pollution treatment service of mangrove forests has been estimated at approximately US$6700 per ha (Costanza et al., 2006) Moreover, mangroves could produce natural products as seed, broodstock, food, and nutrient; these products seem to support to the production process The indirect use value of an ecological function can be estimated

by the production or consumption surplus In production function, this value may be reflected in changes in productivity (Bann, 1998) Furthermore, travel cost, hedonic pricing, and contingent valuation methods are employed to estimate the indirect use values of mangrove functions

Option values relate to the future use value of mangroves which can be evaluated and selected by individuals On the other hand, people can reserve those resources for next generations

Table 2.1 Total economic value of a mangrove resource

Direct value  Timber, firewood, woodchips, charcoal

 Groundwater recharge and discharge

 Flood and flow control

 Human waste and pollutants storage and recycling

 Biodiversity maintenance

 Migration habitat provision

 Nursery and breeding ground for fish

 Nutrient retention

 Coral reef maintenance and protection

 Saline water intrusion prevention Option value  Future use as per direct and indirect value

Non-use

value

 Cultural and aesthetic

 Spiritual and religious

Source: Bann (1998)

From Table 2.1, the ecological services of mangrove forests in shrimp farming can be evaluated by an indirect approach The indirect value of mangrove in aquaframing are proved through protection and support of mangrove forests Particularly, mangrove will provide food, seed and nutrient; and facilitate waste disposal in pond, maintaining the good environment for shrimp Moreover, mangrove may reduce production risk which primarily comes from the damage of natural disasters which affect the shrimp production and survival In fact, the worth of these services is principally non-marketed, and the valuation methods for these service will be present in Chapter 3, and theoretical background for this issue is provided in the next section

2.3 The theory of production risk

Production risk may be caused by many reasons like natural disasters, human mistakes, misapply technologies, etc This may lead to a variability of output or revenue target in production process Hence, production risk have attracted significant attention from researchers and policy makers recently Most researchers cope with production risk by employing Just and Pope (1978) framework In the paper of Just and Pope (1978), they introduced eight postulates which satisfy the utilization of specifications for stochastic production function incorporating risk Several postulates of Just and Pope (1978) force restrictions on the mean function are similar to the usual deterministic function Other postulates are flexible conditions for an output variance function The key significance

in their specification is marginal risk in input use may be positive, zero or negative Namely, inputs are permitted to be either risk-increasing or risk-reducing As a result, they imply shortcomings on popular production specifications in production studies such as 𝑦 = 𝑓(𝑥)𝑒𝜀 with 𝜀 is a stochastic disturbance (𝐸(𝜀) = 0; 𝑉(𝜀) > 0) For such specifications, there are often positive in marginal risk Besides, for an additive specification 𝑦 = 𝑓(𝑥) + 𝜀, marginal risk is frequently zero According to Just and Pope (1979), these shortcomings of popular production functions can be illustrated by the Cobb-Douglas function:

𝑦 = 𝑓(𝑥)𝑒𝜀

marginal effect of input use on marginal productivity variability depend on the effect

of input on variability of output These marginal effects are verified now:

where u is the error term, 𝑣𝑎𝑟(𝑢) = ℎ(𝑥)𝜎𝜀2, and mean of zero According to the theory

of heteroskedasticity, results in estimated parameters are still consistent but not efficient due to bias in estimated standard errors To settle this problem, empirical studies have employed some estimation methods to estimate the mean function and risk function together such as Maximum Likelihood, SUR, or FGLS (e.g Just and Pope (1978), (1979); Wan, Griffiths, and Anderson (1992); Hurd (1994); Traxler, Falck-Zepeda, and Ortiz-Monasterio (1995); Ache and Tveteras (1999); Saha, Havenner, and Talpaz (1997); Kumbhakar (1993 and 2002)) These empirical studies will be discussed hereafter in detail

2.4 Empirical studies

This section includes two parts-one reviews some empirical studies of the impact of mangrove forests on production process and the other summarizes empirical papers related to production risk

2.4.1 The impact of mangrove forests on production process

This part briefly reviews empirical papers of the impact of mangroves on aquaculture production In fact, the studies of this issue in the economic field are limited Most of the papers have investigated the impact of mangrove deforestation on offshore fisheries adopting a production function approach (e.g Sathirathai, 1998), or they make an evaluation of the ecological functions of mangroves in supporting the shrimp fishery using the dynamic approach to production function analysis (Barbier and Strand, 1998)

In terms of aquaculture activities, the impact of mangroves on household’s production has been studied from a biological perspective

2.4.1.1 Integrated shrimp‐mangrove farming systems in the Mekong delta of Vietnam

(Binh et al, 1997)

To investigate the integrated shrimp-mangrove system in Ngoc Hien district, Binh et al (1997) used a sample of 161 households (living on the west and east coast) who operate shrimp farming and directly took part in the culture The authors collected primary and secondary data about economy, environment and technology from the interview survey

at Ngoc Hien district (the Forestry Fisheries Department, the Provincial Fisheries Department, the Provincial Forestry Department, and the State Forestry Fisheries Enterprises) They find that the negative effects of mangrove density, pond age, and pH

on shrimp yields existed on the east coast, but the same did not occur on the west coast

In terms of economics, farming systems with about 30%-50% mangrove area in a pond will give the highest return On the other hand, a pond with the cleared mangroves will give the lowest annual return These results imply that the integrated mangrove-shrimp farming system will get a better economic return, if mangrove forests are maintained

2.4.1.2 Effect of an integrated mangrove-aquaculture system on aquacultural health

(Peng et al, 2009)

Based on experiment method, Peng et al (2009) tried to find out whether the traditional method or integrated mangrove aquaculture farming is better They classified nine experimental aquaculture ponds and one control pond without mangrove forests, then

all ponds fed two fish species in the first year and additional crab species in the following year Through that approach, they can observe for a period of three years the differences between experimental ponds and control pond in terms of water quality, production of aquatic organisms, and growth of mangroves in integrated mangrove-aquaculture farming systems Their findings imply that water quality of experimental

ponds is better than the control ponds while Aegiceras corniculatum is the best type for

improving water quality compared with the rest In general, the experimental ponds had higher yield more than the control pond with 19% in the first year Particularly, the

experimental ponds planted with Kandelia obovata and Aegiceras corniculatum brought a higher seafood yield than the experimental pond planted with Sonneratia

caseolaris Moreover, they found that if an aquaculture pond had 15% of its area planted

with Aegiceras corniculatum, the pollutants in production process would be reduced

and aquaculture production quality would be improved

2.4.2 Empirical studies about production risk

2.4.2.1 A three - step estimation procedure in Hurd (1994)

In Hurd’s study, the effects of the adoption of new and changing technologies (e.g integrated pest management) and production inputs on output and output variability in agriculture were analyzed through the possible effects of inputs on the first two moments (i.e mean and variance) of the yield distribution The Just and Pope framework for a stochastic production function is applied to investigate these effects Besides, to obtain consistent and efficient parameters, a feasible generalized least square is employed The estimation procedure includes three steps The first step is to estimate consistent estimators of 𝛼̂ 𝑎𝑛𝑑 𝑢̂ in equation (1) Then, consistent estimators

of 𝛽̂ will be obtained through regression of the square estimated residual on inputs in equation (2) Finally, after estimating the error covariance matrix, a consistent and efficient parameter (𝛼̂∗)will be obtained in formulation (3) These equations are described below:

𝑦 = 𝑓(𝑥, 𝛼) + ℎ12(𝑥, 𝛽)𝜀 (1) With 𝜀 = 𝑁(0, 𝜎2), and 𝐸(𝜀𝑖𝜀𝑗) = 0

ln(𝑢̂2) = ln(ℎ(𝑥, 𝛽)) + ln(𝜀̂2) = 𝑥′𝛽 + 𝑣 (2)

𝛼∗ = (𝑥′Ω̂−1𝑥)−1𝑥′Ω̂−1𝑦 (3)

Using a cross-sectional data for San Joaquin Valley cotton producers, the author finds that soil quality and nitrogen have a positive effect on output Moreover, the reflected variables for integrated pest management have only four variables (including crop rotation, frequency of field monitoring, the use of an independent pest control advisor, and the use of biological preserves) which contribute to yield considerably The paper also implies that almost production inputs do not contribute to yield variability However, a farm advisor contact will reduce risk in production process Besides, the author estimated the value of marginal product of inputs that have statistically significant effects on yield and analyzed the elasticity of inputs on yield and yield variability computed at the average value of inputs and yield

There is still the limitation in this paper Particularly, using FGLS method, the functional form of risk function must be correctly assumed to obtain an efficient estimator In this paper, the multiplicative form for heteroskedasticity is applied because of its property for positive predicted variances To overcome this limitation, Asche and Tveterås (1999) used a heteroskedasticity-consistent covariance matrix instead of FGLS estimator Besides, Harvey (1976) and Saha et al (1997) implies that the estimated parameters in FGLS estimator are consistent and less efficient than in ML estimator These papers are discussed in the following sections

2.4.2.2 A two-step estimation procedure in Asche & Tveterås (1999)

Based on the Just and Pope postulates, Asche & Tveterås (1999) explained and estimated production risk through a two-step procedure The inconstant variance of the model (i.e heteroskedasticity) may be representative of production risk if the Just-Pope

postulates are satisfied Thus, tests for heteroskedasticity were analyzed first After that, the mean and risk function are estimated again separately if heteroskedasticity is detected, which is an advantage to model specification for each function in more detail From an econometric viewpoint, when heteroskedasticity is present, the result of OLS estimators will be inefficient To solve this problem, some empirical studies applied an FGLS and ML estimator, but its assumption involves an accurate functional form for the variance function to obtain efficient parameters In contrast, a White’s heteroskedasticity-consistent covariance matrix does not assume any functional specification of the risk function Moreover, a White’s heteroskedasticity-consistent covariance matrix will be more efficiency in large sample

In this paper, the authors also suggest to use the additive interaction between a flexible functional form and the error term

Asche & Tveterås apply an empirical study to Norwegian Salmon Aquaculture Using

an unbalanced panel data in the period of 1985-1993, the authors investigate the effects

of inputs on output and output variability by using a flexible functional form (e.g a quadratic function) However, it is difficult to interpret estimated parameters meaningfully, so the elasticity of marginal inputs is utilized to imply the research result The mean and variance function are specified below:

𝑦𝑖𝑡 = 𝛼0+ ∑ 𝛼𝑘𝑥𝑘,𝑖𝑡

𝑘

+ 0.5 ∑ ∑ 𝛼𝑗𝑘𝑥𝑗,𝑖𝑡𝑥𝑘,𝑖𝑡

𝑘 𝑗

+ 𝜇𝑖

+ 𝑢𝑖𝑡

𝑉(𝑢𝑖𝑡) = exp (∑ 𝛽𝑘𝑥𝑘,𝑖𝑡

𝑘

+ 0.5 ∑ ∑ 𝛽𝑘𝑗𝑥𝑘,𝑖𝑡𝑥𝑗,𝑖𝑡

𝑗 𝑘

2.4.2.3 A Maximum Likelihood estimator for production risk in Isik and Devadoss

(2006)

Isik and Devadoss (2006) attempt to examine the effects of climate changes on the mean, variance and covariance of crop outputs In addition, using simulation data, the authors find out whether climate in the future will affect to these factors of crop outputs According to the theory model for production risk (i.e the Just and Pope framework),

an econometric model was built to evaluate those effects The econometric model is developed on the maximum likelihood method, and it is presented in the equations below:

𝑦𝑖𝑡 = 𝑓(𝑥𝑖𝑡, 𝛼) + ℎ12(𝑥𝑖𝑡, 𝛽)𝜔𝑖𝑡 (1)

𝑦𝑖𝑡 = 𝑓(𝑥𝑖𝑡, 𝛼) + 𝑢𝑖𝑡 (2)

Where x is the vector of climate variables (precipitation and temperature); 𝜔 is

production shocks and assumed 𝜔𝑖𝑡 = 𝑁(0,1); 𝑢𝑖𝑡 = ℎ12(𝑥𝑖𝑡, 𝛽)𝜔𝑖𝑡 with 𝑣𝑎𝑟(𝑢𝑖𝑡) =ℎ(𝑥𝑖𝑡, 𝛽)𝜎2

According to Saha et al (1997) and Huang (2004), the log-likelihood function is described below to estimate the maximum likelihood coefficients of 𝛽 and 𝛼:

in the stage of 1939-2001, while the data of potato outputs was collected in four districts

in the period from 1949 to 2001, and the sugar beet output was collected in three districts in the stage of 1975-2001

The empirical results find that the precipitation has a negative effect on the mean wheat output in both a quadratic and a linear functional form, whereas the temperature has only a negative effect on the mean wheat output in a linear functional form In contrast, the negative effect exerts on the mean barley output of the precipitation in a linear functional form, and in both functional forms with the effects of the temperature In terms of the mean potato and sugar beets yield, the estimated coefficient of the precipitation on the mean potato yield are statistically significant negative effect, but

no statistical significance is found in the mean sugar beets yield On the other hand, the temperature has a positive significant effect on the mean potato and sugar beets yield

in a linear functional form

The trend has a positive effect on the variability across different crop outputs The negative coefficients of the precipitation and temperature on the variability of wheat,

barley and sugar beets were estimated, these imply that the increase in the precipitation will reduce the output variability In other words, there is a risk-decreasing input Nevertheless, on the potato yield variability, the elasticity of precipitation and temperature are risk-increasing effects

This paper also took into account the effects of climate change on the covariance among crops (i.e wheat and potato; potato and barley; wheat and barley) The estimated coefficient of the precipitation on the covariance across different crops is a negative sign This implies that the covariance between crops will decrease when the level of precipitation rises Although the temperature level has a positive effect on the covariance among crops, the estimated parameters are not statistically significant

Based on forecast climate change data in Hadley Center (2003), the authors investigated the impact on the mean output, output variance and covariance of two long-run climate change scenarios (i.e the range of 2025-2034 and 2090-2099) in both a quadratic and linear functional form The results implies that the long-term climate change could raise the mean wheat and potato outputs and reduce the mean barley and sugar beets outputs

In the variance function, the long-run climate change is a risk-decreasing effect for the variability of wheat and barley output, but its effect is risk-increasing for the variability

of potato and sugar beets yield The percentage change in the mean output and output variability is different among crops in two climate change projections On the other hand, the wheat and barley rotation would increase in the output covariance, but decease significantly in the rest rotation

CHAPTER 3 RESEARCH METHODS

This chapter includes four parts Firstly, an overview of the Mekong river delta in Viet Nam is presented Secondly, analytical framework is shown in Section 3.2 The paper reviews the basic issues of production economics that are how firms decide on output supply and input uses at the optimal level The primal and dual approach introduced in Section 3.3 can be used to address those issues Finally, the methods for estimating impact on aquaculture activities as well as its production risk of mangrove forest systems are setup in the following section Also, data collection method is represented

in that part

3.1 Overview of the Mekong river delta

The Mekong river delta, which is known as Southwestern region of Vietnam, is the area

in downstream of the Mekong River It is approximately 40,000 square kilometers in area; and the administrative boundary is defined by 13 provinces and 1 city The population of the whole Mekong Delta is about 18 millions, accounting for 20.5 percent

of the national population Population density is not equally allocated between urban and rural areas, specifically population in rural areas accounts for 78.7 percent and 21.3 percent in urban areas It is high rural population density that provides good conditions

to develop agriculture and fishery As a result, this plays an important role in socioeconomic development in the Mekong Delta In detail, GDP in this area made up 47.8 percent the national GDP The GDP growth rate in the Mekong Delta averaged 11.5 percent from 2001 – 2010

The Mekong Delta is the largest farming and fishing in Vietnam, accounts for 70 percent of total farming area of the whole country Every year, this area provides over

52 percent fishery quantities In 2014, the aquaculture in the area is around 800,000 ha, its production gains over 2.4 million ton, including 1.2 million ton of Tra catfish, 400,000 ton of shrimp; and this supplies demand of processing and exporting industry There are many methods used in aquafarming, but chiefly they are extensive, intensive

and semi-intensive The intensive is based on mainly formulated feed The farming system has been well developed such as supplying and draining water automatically, fitting up all facilities and so on; thus, it is easy to manage and operate Although intensive cultivation brings high profit, the cost is quite high and waste water which is not treated well will lead to water pollution The semi-intensive culture is based on the mixture of formulated and natural feed; but input water depends on tide-water And the last one is extensive culture which is specified by a large farming area and natural food Nonetheless, the density of shrimp is low because of depending on natural breed The semi-intensive and extensive cultures are often known as the mixed mangrove – aquaculture farming due to the creation of unique ecological environment and more biological efficiency than the others

The total mangrove area in coastal area and the river mouth is approximately 100,000ha, distributed in Ca Mau (58,285 ha), Bac Lieu (4,142 ha), Soc Trang (2,943 ha), Tra Vinh (8,582 ha), Ben Tre (7,153 ha), Kien Giang (322 ha), Long An (400 ha) Based on the statistics of Forestry Institution, Mekong Delta has 98 species of

mangrove (e.g Rhizophora apiculata, Kandelia obovata, Sonneratia caseolaris,

Avicenmaalba), whereas the rate of Rhizophora apiculata is higher than the other

species (Forest Inventory and Planning Institute, 2001) Recently, mangrove forests have significantly been reduced because of conversion to agriculture and aquaculture Particularly, in the period of 1980 – 1995, about 72,825 ha of mangrove area was decreased; and manual average rate of loss is about 4,855 ha (approximately 5 percent per year) These mangrove forests losses lead to instability of ecological environment

in the area

Currently, policies of the allocation and contraction of mangrove areas to households

in the south of Vietnam have been conducted To be specific, an area of 4-8 hectare in which about 70% for mangrove forest conservation is allotted to each household (Johnston et al., 1999) Households are assigned mangrove and allowed to clear the rest part of the mangrove into housing and production (aquatic organisms) Under the poor

enforcement of regulations, the area of mangrove forests were converted into extensive and semi-intensive aquaculture farms more than what they allowed However, consequences of over-exploited mangroves have not been informed clearly and thoroughly to households In more detail, over-exploitation may affect not only the defense capability of mangrove forests when confronting natural disasters or overcoming of human mistakes, increasing production risk, but also profitability in aquaculture production

Owing to the importance of mangrove forests in ecological systems, this paper focuses

on objective of analyzing the influence on aquaculture activities of mangrove forests in the Mekong river delta Specifically, aquaculture under the extensive and semi-intensive methods is investigated in this study

3.2 Analytical framework

The Just-Pope framework for a stochastic production model presented in Chapter 2 is applied to estimate the effects of mangrove forests on profit as well as the profit variability (production risk) in extensive and semi-intensive aquaculture farms (mostly shrimp farms) in the Mekong river delta Profit function for the model is estimated with input prices, output price, and fixed inputs in the short-run under two different functional forms – one is a linear functional form and another is a quadratic functional form To investigate this effect, the paper will apply estimation procedures such as FGLS, Robust standard error, SUR, and MLE for the log-likelihood function Then, the research results of each method will are compared to between two functional forms

3.3 Neoclassical economic theory

3.3.1 Two approaches in production economics

According to the Neo-classical theories of production, we can say that basic issues in production economics are how to calculate input uses as well as output supply at the optimal point with the changes of their prices According to production duality,

economists may solve those issues through a primal approach or dual approach presented in Table 3.1

Table 3.1 The primal and dual approach

is more convenient to avoid those issues as well as estimating the impact on welfare of

Factor 1

Factor 2

Factor 1

Factor 2

the changes in the economic conditions Particularly, this second approach, which is to adapt the changes of outside elements, for example input and output prices, natural factors (e.g environment, climate), is based on neoclassical duality theory to help producers make decisions in the production process Duality theory implies that the cost minimization and profit maximization principles can generally determine the optimum behavior of producers in production process In other words, the output supply and input uses are directly calculated by a cost or profit function Based on its advantages, the dual approach will be employed to answer research questions in this paper Furthermore, from the econometric viewpoint, the profit function permits output levels

to fluctuate endogenously in corresponding to the price of inputs and outputs, whereas output have been treated as an exogenous variable in the cost function (Alpay et al, 2002)

3.3.2 A dual approach - The profit function

This sector gives a brief presentation about the properties of profit function and the specification of restricted profit function

3.3.2.1 The properties of unrestricted profit function

The definition of the profit function is decided below:

be generated from the determined cost function or the determined profit function

First and foremost, the premise of the producer has an input-regular production with a potential yield f(x) Each input will produce positive output level y with y in f(x), the input price as well as output price information must be available because the assumption that the producer operates in a competitive market

Underlying the properties of cost and revenue functions, the properties of the profit function including five properties are determined Overall, those properties were completely derived from the hypothesis of profit maximization Firstly, one of properties claims that firm’s profit in the long run is never negative (𝜋 ≥ 0) This is absolutely suitable in reality when firms may be willing to accept losses in short run and stop all production activities if this situation still continues in long run

Secondly, the profit function is non-decreasing in p with fixed input prices w, it can be

explained that when the output price increases, firm’s profit will raise, ceteris paribus

With a positive output price p and p’ (p’ > p), the premise of profit is maximized at an output price p in response with (y, x) After that, output price rises to p’, but firm still decides to product at (y, x) It is easy to see that the increase in profit is showed below,

so the profit at p’ will not decrease:

𝑝′𝑦 − 𝑤𝑥 ≥ 𝑝𝑦 − 𝑤𝑥

⟺ (𝑝′− 𝑝)𝑦 ≥ 0

Thirdly, the profit function is non-increasing in w with fixed output price p, it implies

that firm’s profit will be reduced when input prices increase, other things equals To

demonstrate that, let w’ ≥ w Suppose at w, x firm’s maximum profit function is 𝜋(𝑝; 𝑤) Under an input price w’ and a corresponding quantity input x’, then the profit function

is 𝜋′(𝑝; 𝑤′) Obviously, the input quantity x’ is not chosen at input price w though it is available in (3.1) Besides, as w’ ≥ w, then w’x’ ≥ wx’ in (3.2) These inequalities will

be showed below in more detail:

𝑝𝑓(𝑥) − 𝑤𝑥 ≥ 𝑝𝑓(𝑥′) − 𝑤𝑥′ (3.1)

𝑝𝑓(𝑥′) − 𝑤′𝑥′ ≤ 𝑝𝑓(𝑥′) − 𝑤𝑥′ (3.2) Combining (3.1) and (3.2), that property is proved:

𝑝𝑓(𝑥′) − 𝑤′𝑥′ ≤ 𝑝𝑓(𝑥′) − 𝑤𝑥′ ≤ 𝑝𝑓(𝑥) − 𝑤𝑥

⟹ 𝜋(𝑝; 𝑤′) ≤ 𝜋(𝑝; 𝑤)

Fourthly, if the profit is maximized at rigorously positive input price vector w and

output price vector p, and if the producible output bundle is constant Then, if all input and output price vector are multiplied by a scalar t (t >0), revealing the total profit with

a constant producible output will be multiplied by t This property points out that the

profit function is linearly homogenous of degree one This property is described below:

𝜋(𝑡𝑤, 𝑡𝑝) = 𝑡𝜋(𝑤, 𝑝) (𝑡 > 0)

Fifthly, the property of the profit function is convex in (w, p), which ensured assumed profit maximization Let y, y’, y’’ maximize profit at p, p’, p’’ respectively The property is proved by Varian (1992) and presented below:

𝜋(𝑝′′) = 𝑝′′𝑦′′ = (𝑡𝑝 + (1 − 𝑡)𝑝′)𝑦′′ = 𝑡𝑝𝑦′′+ (1 − 𝑡)𝑝′𝑦′′ (3.3)

𝑡𝑝𝑦′′ ≤ 𝑡𝑝𝑦 = 𝑡𝜋(𝑝) (3.4)

(1 − 𝑡)𝑝′𝑦′′≤ (1 − 𝑡) = (1 − 𝑡)𝜋(𝑝′) (3.5) Substitute (3.5) and (3.4) to (3.3), the property is demonstrated:

𝜋(𝑝′′) ≤ 𝑡𝜋(𝑝) + (1 − 𝑡)𝜋(𝑝′) Moreover, this property can be easily explained by the intuitional result via Figure 3.1 Assumed that, the profit maximization production plan (y*, x*) is at (p*, w*) Presume that, the output price raises, but the firm still decides to practice the same production strategy (y*, x*) The profit function is called “passive profit function”, and it is illustrated as a straight line 𝜋′(𝑝) = 𝑝𝑦∗− 𝑤∗𝑥∗ From that, firms will determine an optimal policy for production plan as least as large as a passive policy (i.e 𝜋(𝑝) ≥

𝜋′(𝑝)) showed in Figure 3.1 Thus, a convex function in 𝜋(𝑝) is required

Figure 3.1 Profit function with respect to output price

Source: Varian (1992)

In addition, conditional factor demand for input i and output supply j can be calculated

by taking partial derivatives of profit function with respect to price of input i and output

3.3.2.2 The restricted profit function

Unless all inputs and outputs are variable, the unrestricted profit function is defined and specified in the long run However, with some inputs or outputs are fixed, then the profit function is a restricted profit function (or a short-run profit function) Corresponding to fixed inputs or outputs, we will take different restricted profit functions If all outputs are fixed, the restricted profit function is a cost function On the other hand, if all inputs are fixed, the restricted profit function is a revenue function The restricted (short-run) profit function can be written as below: