Impact of socio economic and environmental hazards on the technical efficiency of shrimp farms at cam thinh dong commune, cam ranh district, vietnam

ABSTRACT This study employed stochastic frontier production function approach to investigate the technical efficiency and factors affecting the technical efficiency of shrimp farmers..

MINISTRY OF EDUCATION AND TRAINING

NHA TRANG UNIVERSITY

FOLORUNSO, EWUMI AZEEZ OLATUNJI

IMPACT OF SOCIO-ECONOMIC FACTORS AND ENVIRONMENTAL HAZARDS

ON THE TECHNICAL EFFICIENCY OF SHRIMP FARMS AT CAM THINH DONG

COMMUNE, CAM RANH DISTRICT, VIETNAM

MASTER THESIS

KHANH HOA - 2018

and Climate Change Topic Allocation decision

Decision on establishing the committee:

Supervisors: Prof Arne Eide

Dr Le Kim Long

ACKNOWLEDGEMENT

I would like to thank my international supervisor, Professor Arne Eide, Norwegian college

of fishery science, University of Tromsø, Norway, for his resourceful advice, support, comments and understanding through the course of the research and writing

I would also like to thank my local supervisor, Dr Le Kim Long, department of Economics, Nha Trang University, also for his resourceful advice, supports and guidance throughout the course of the thesis

My sincere gratitude goes to the entire board of NORHED (Norwegian Program for Capacity Building in Higher Education) for making my sojourn here a successful one with their financial and moral supports

I would as well be grateful to my parent and siblings for their unending encouragement and understanding throughout the course of my study

My gratitude also goes to Cam Thinh Dong community committee for provision of comfortable atmosphere and warm support that ensues successful field survey exercise

I would also without reservation extend my utmost gratitude to the entire teaching staffs of Nha Trang University and my classmates for their supports and encouragement

Folorunso, Ewumi Azeez Olatunji

May 2018, Nha trang, Vietnam

TABLE OF CONTENT

ACKNOWLEDGEMENT ii

TABLE OF CONTENT iii

LIST OF TABLES iv

LIST OF FIGURES v

LIST OF ABBREVIATIONS vi

ABSTRACT vii

1.0INTRODUCTION 1

1.1 BACKGROUND INFORMATION 3

1.2 PROBLEM STATEMENT 5

1.3AIM 6

1.6TECHNICAL EFFICIENCY 7

1.7MEASUREMENT OF TECHNICAL EFFICIENCY 11

1.8STOCHASTIC PRODUCTION FRONTIER 13

1.9LITERATURE REVIEW 14

2.0DATA 20

2.1DATA SAMPLING METHODOLOGY 20

2.2PRIMARY SURVEY 20

2.3PRE-TESTING 20

2.4DATA COLLECTION 20

2.5DATA SAMPLE AGGREGATE 22

3.0METHOD 23

3.1DATA PROCESSING AND ANALYSIS 23

3.3EXPECTATION 23

4.0RESULT 24

4.1THE EMPIRICAL RESULT 24

4.2TEST OF HYPOTHESIS 25

4.3INEFFICIENCY MODEL 30

4.31 Socio-economic factors 30

4.32 Environmental factors 32

5.0 DISCUSSION AND CONCLUSION 36

5.1 DISCUSSION 36

5.2 CONCLUSION 40

REFERENCES 42

APPENDIX 47

LIST OF TABLES

Table 1-2: Summary statistics for some of the variables used in the model 22

Table 1-4: Stochastic frontier and Inefficiency effects result 24

Table 2-4: Hypothesis testing 25

Table 3-4: Technical efficiency distribution 27

Table 4-4: Descriptive statistics of the technical efficiency 28

Table 5-4: Descriptive statistics of inputs 29

Table 6-4: Farmers age distribution 30

Table 7-4: Educational level of the shrimp farmers 31

Table 8-4: A chart showing farm size information 31

LIST OF FIGURES

Figure 1-1: Map of Cam Ranh showing Cam Thinh Dong 6

Figure 2-1: Input-oriented measure of TE 8

Figure 3-1: Output-oriented measure of TE 9

Figure 1-4: Chart of TE in relation to numbers of farmers 29

Figure 2-4: Chart showing the majorly encountered environmental hazard in the area 32

Figure 3-4: Chart showing the number of flood experienced by farmers 33

Figure 4-4: chart showing the majorly coping strategies 34

Figure 5-4: Chart showing the different environmental impacts affecting shrimp farming at Cam Thinh Dong 35

LIST OF ABBREVIATIONS

DEA Data Envelopment Analysis

GDP Gross Domestic Product

GFDRR Global Facility for Disaster Reduction and Recovery

FAO Food and Agricultural Organisations

MLE Maximum Likelihood Estimate

OLS Ordinary Least Square

SFA Stochastic Frontier Analysis

SPA Stochastic Production Frontier

TE Technical Efficiency

ABSTRACT

This study employed stochastic frontier production function approach to investigate the technical efficiency and factors affecting the technical efficiency of shrimp farmers With the aid of questionnaire, data were collected on farmers’ cost of major inputs (labour, seed, feed and lime), socio-economic activities, age, education, experience, household size and farm size, and on environmental hazards, flood and drought experience The data was analysed using a stochastic production frontier The result obtained revealed a mean technical efficiency of approximately 58%, reflecting that there exists a great potential for improving the efficiency of the shrimp production The input variables considered in the model (feed, seed, labour and lime) were all found to be important factors for shrimp production in the area In the inefficiency model, age, education, experience, drought experience and farm size were found to be positively related to technical efficiency While flood experience was found to be positively related to technical inefficiency, drought experience was found to be negatively related to technical inefficiency This study suggests

a huge impact of pond adjustment on farmers’ cost of production, as 98% of the farmers reported to be adopting frequent pond adjustment and maintenance as an adaptation to frequent flood events On the other hand, number of drought experience a farmer has was found to enhance his technical efficiency Since all the inputs considered in this study were found to be positively related to the technical efficiency, this study therefore suggests the farmers be encouraged to increase their output by providing them the medium or platform

to learn the best input combinations in order to reduce cost while maximizing their profit Furthermore, since age, experience, and education are positively related to TE, this study suggests establishment of or restructuring of community-based organisations and extension services to create medium for interactions between the farmers in order to allow young and less experienced farmers to learn from the older and more experienced ones Lastly, government should support farmers with sea dike structures to help curb flood impacts and also provide history of drought patterns to help farmers plan against forecasted drought events

1.0 INTRODUCTION

On a global basis, annual records of natural disasters or events amount to over 300, and in

2010 it cost global economy USD 109 billion from 106,891 fatalities (Guha sapir, et al

2012) The fish farmers and their communities globally are particularly vulnerable to these disasters because of their locations, their livelihood characteristics, total high levels of exposure to natural hazards, shocks from their livelihood and impacts of climate change The impacts on economic, social and environmental structures are significant with disproportionate effects in developing countries and on vulnerable groups, as 98% of the

262 million people are affected by weather and climate change-related disasters between

2000 and 2004 who lived in developing countries, and majority of which are dependent on aquaculture and agriculture as livelihoods (FAO, 2012)

Lying in the tropical monsoon area of the north western pacific means that Vietnam is one

of the disaster-prone countries in the world, as it is affected by floods, storms, tropical depression, storm surges, whirlwinds, coastline erosion, hail rains, drought and landslides, destroying lives, assets and degrading cultural and socio-economic structures as well as

natural environments (Anh, 2016) Using national disaster database, Chin luu et al (2017)

reported that apart from Mekong delta who has the highest number of flood fatalities, south central and north central coast were the two most affected regions in flood fatalities historically based on average per province per year in the regions investigated

Aquaculture, which is perceived to be playing key role in economic growth, food security and job creations in the country has been plagued by natural disasters such as flood, flood flash and drought (Anh, 2016) The sector remains one of the major occupation of the coastal population of Vietnam, accounting for 12% of total exports (about USD2.5billion) and providing source of livelihood for about 4 million people (GFDRR, 2011), and the fact that the high concentration of human population and economic activity in coastal areas has

a heavy reliance on fishery and aquaculture sectors that account for 6.6% of Vietnam’s

GDP in 2008 (Bierbaum, et al 2010), make aquaculture an important sector of the country’s

economy Therefore, the environmental hazards and climate change impacts that befalls the productivity of the industry may arguably be found producing an effect on the entire

economy of the country Apart from the government’s effort to implement policies and strategies to help prevent further damages and to assist residents to cope with these changes, there have been limited studies that have evaluated how the adaptive strategies (autonomous and non-autonomous) have decreased the profit-maximization of the farmers and possibly forcing some farmers out of their livelihoods

De Silva et al (2009) concluded that impact of climate change on capture fisheries have

received more attention than in aquaculture and stressed the need to assess the vulnerabilities of major aquaculture farming systems and proposed appropriate mitigation and/or adaptation measures to maintain the viability of these systems Number of researches

in agriculture have recently emerged looking up the impact of climate change on the performance of agriculture outlets using technical efficiency Oyekale (2012) considered rainfall and temperature as climate change factors affecting the technical efficiency of cocoa farms in Nigeria Also in Bangladesh, both humidity and rainfall were found to produce a positive impact on the technical efficiency of rice farm while temperature produced a negative impact on the technical efficiency of the rice farm

According to FAO (2015) report on the impact of natural hazards and disasters on agriculture and food security and nutrition, one of the key findings is that there exist major data gaps on the impact of natural hazards and disasters on the agriculture sectors in developing countries In aquaculture, significance numbers of studies have been carried out

on the impact of different factors such as farm size, farmer’s experience, farmer’s age and some socio-economic factors such as household-size on the technical efficiency of that farm, but few or limited researches have looked into the impact of environmental hazards

on the technical efficiency of farms However, Auci and Vignani, (2014) in their research considered the climate change impacts on technical efficiency of fish farm outlets in Italy Rainfall and minimum temperature were considered as one of the inefficiency factors in the inefficiency model, and it was found that rainfall variable had a positive impact on the efficiency while minimum temperature reduces the efficiency of harvested production

Recently, Nguyen et al (2017) considered the impact of climate change on the technical efficiency of Pangasius species in the Mekong delta area of Vietnam, factors such as flood

effect, salt intrusion effects, farmers experience level in climate change and access to trainings, using Data Envelopment Analysis (DEA) Their conclusion accepted with the few existing literatures that farmer’s experience in climate change impacts contribute positively to the technical efficiency of their fish farms and how to deal with the impacts

In a bid to protect development investment and strengthen aquaculture resilience to disasters, there is a need to understand the particular way the sector is being affected, the magnitude at which it is affected in order to understand how to assign priorities to the disasters during the course of planning against the disasters Khanh Hoa province which is located in the south-central region of the country and was once known a major supplier of shrimp seeds in the region has been reported to have witnessed a drastic reduction in output owing to unfavourable weather conditions, (Hoang Thu Thuy, 2008; Pham Xuan Thuy, 2004; Nguyen Thi Kieu Thao, 2012) There has been little or no investigation carried out

on the impacts of these environmental hazards on the efficiency of the shrimp farms This study will investigate the impact or effects of farmers experience on climate change, flood effects, drought effects, level of education, age of the farmer, and the farm size on the technical efficiency of the fish farms

1.1 Background information

Cam Thinh dong is a commune in Cam Ranh, which is located in Khanh Hoa province, south central of Vietnam The province has a total area of 5,197 km2 and a provincial coastline that spreads 385 km featuring numbers of creek mouths, lagoons, river mouths and hundreds of islands and islets from Dailanh commune to the end of Cam Ranh Bay This province is contiguous to Phu yen province in the north and south-eastern border, Dak lak province at the west, Ninh Thuan province at the southern border and the eastern border with the south-china sea

This province was known for its influential contribution to the development of shrimp farming in Vietnam, with 1,019 farms and production of about 3.25 million ton of shrimp seeds between 1995 and 2000, when it accounted for 40.8% of total shrimp production in the country (Hoang Thu Thuy and Kim anh 2008) Favourable indicators of temperature, humidity and rainfall were described as climatic factors that had produced great influence

on the development of Aquaculture in this province (Hoang Thu Thuy, 2008) According

to Pham Xuan Thuy (2004), Shrimp production reduced drastically in Khanh Hoa province between 2001 and 2002 despite the increase in farming area, which has been on the increase since 1999, and increased to 53.2 km2 from 49.57 km2 in 2001 This is not in parallel to production, as the latter fell short from 7,452 tons in 2001 to 6,275 tons in 2002 The author described temperature and rainfall as the major factors that are behind any success of shrimp farming in the province, therefore, extreme local fluctuation of the two weather factors in this year was reported to be responsible for the reduction (Pham Xuan Thuy, 2004) This went ahead to force farmers out of their livelihood and subsequently dealt a blow to the production of fish seeds; as the number of farms has gone to decrease from 1,282 in 2003

to 507 in 2006; and fingerlings production decreased from 69,531 tons in 2004 to 41,410 tons Specifically in Cam Ranh, number of farms decreased from 630 in 2003 to 120 in

2006 and production fell to 9,470 in 2006 from 38,430 tons in 2003 Also in Nicole Portley (2016) report, Khanh Hoa province aquaculture production rose to be listed among the top five exporter of shrimp in 2010 with a production of 15,912 tons, and gone on to decrease from 2011 to 2014, where it only produced 11,540 tons The weather factors such as unsteady distribution of rainfall, has forestalled the potential success of aquaculture in Khanh Hoa province (Hoang Xuan Huy, 2009) and the small reservoirs of underground water in the province only get to supply for the small scale production in the coastal areas

Furthermore, Nguyen & Fisher, (2014) from their research stated that production area in Khanh Hoa province decreased from 2012 to 2013 and the output started decreasing from

2012 to 2014 10,788 tons produced in 2012 was the highest in the whole of the three years, when 8,850 tons and 7,912 tons were recorded respectively for the year 2013 and 2014 Factors such as disease epidemics, unfavourable weather conditions and environmental pollution were the stated causes of the reduced production

The environmental condition of this region, which was considered to be of great advantage

to the successive aquaculture production in the region, has been savaging the aquaculture operational activities recently Hoang Thu Thuy (2009) reported the air temperature of the major aquaculture-producing districts in his research to produce the evidence of the

decreasing production in the area Nha trang has a highest temperature of 370C, and Cam Ranh has 39.30C The evidence of the impact of the environmental distress, which is probably brought by climate change, continues with another record of low performance of lobster farming, owing to sudden increase in temperature Lobster, which normally survives between 26- 290C in the summer and about 22-270C in winter, has been reported to be facing low growth and mass mortality of the juveniles The increase of 3-50C reported in Cam ranh district causes mortality of the juvenile of this species and subsequently causing

a low output (Nguyen Thi Kieu Thao, 2012)

1.2 Problem statement

As indicated above, environmental stresses from environmental variation could lead to the low output, forcing farmers out of their livelihood while reducing the profits of the existing ones The current reductions in output of shrimp production reported by Hoang Thu Thuy (2009); Pham Xuan Thuy (2004) and Nguyen Thi Kieu Thao, (2012) in this region has in a way or the other been attributed to unstable environmental conditions and previous studies

in this region has concentrated on the investigation of how socio-economic factors such as age, level of education, farm-size, household size affect the farmers technical efficiency, as observed in Hoan Thu Thuy (2008), who compared the performance of shrimp farms in

Cam Ranh, Nha Trang and Ninh Hoa districts Also, Nguyen et al (2017) recently

investigated the impact of environmental hazards (flood and salt water intrusion) along with

the common factors negatively affecting technical efficiency of Pangasius farms in the

Mekong delta region of Vietnam, and it was found that there was a positive effect on the technical efficiency by farmers’ education level and experience in flooding and salt water intrusion Farms affected by salt-water intrusion had a lower scale of efficiency as they reduce stocking rate and frequency In this regard, there may be unresearched or univestigated environmental factors or hazards affecting the technical efficiency of shrimp farms in some areas in the country, as this result shows that environmental hazard could also be hindering the optimal prioritisation of shrimp farming in Cam Thinh Dong, and could be responsible for the reported output reduction in the region

1.3 Aim

This study therefore intends to investigate the relative technical efficiency of the shrimp farms in the region using stochastic production frontier by considering the input, output and inefficiency factors The study will further investigate the impact of the environmental hazards (drought and flood) along with socio-economic factors (age, education, experience, and farm size) on the technical efficiency of the shrimp farms in this commune Hopefully, the result obtained will help recognize the contributions of these factors on the economic performance of the shrimp farms in this community Recognizing and establishing the magnitude of the effects of these factors on the shrimp productivity will help farmers and policy makers to make headway on assigning or allocating priorities to these effects when planning or preparing for adaptation and/or mitigation techniques

1.5 Study site

Cam Thinh Dong commune is located in the south-western part of Cam Ranh, one of the major aquaculture districts in Khanh Hoa province, which is located in the south central region of Vietnam with more than 300 km of coastline running from 11040’ to 12050’ Northern latitudes This commune has four villages including, Hiep thanh, My thanh, Hon quy, and Hoa diem village

Figure 1: Map of Cam Ranh showing Cam Thinh Dong

1.6 Technical Efficiency

Farell (1957), who defined technical efficiency as a reflection of the ability of a firm to obtain maximum outputs from a given set of inputs, was reported as the first to study the modern technical efficiency building on the foundation of the initial work by Debreu and Koopmans, (1951) An efficiency can be input-oriented or output oriented In the former, a target point that maximizes the proportional reduction of inputs or produces a given level

of output from an optimal combination of outputs is formed and the latter finds a target point which maximizes the proportional augmentation of outputs or produces the optimal output from a given set of inputs

Farell (1957) further proposed technical efficiency and allocative efficiency as two components of efficiency after using a simple model with two inputs and single output under constant return to scale In his seminar paper, he stated that the observed input-per-unit of output values for firms would be above or beyond the unit of isoquant curve In his explanation, he assumes a firm uses X1 and X2 inputs to produce Y output, in such a way that the points defined by the input-per-unit of output ratios (X1/Y, X2/Y), are above the unit isoquant curve The unit isoquant curve defines the input-per-unit of output ratios associated with the most efficient use of the inputs to produce the output involved Technical inefficiency of the firm was considered to be the deviation of observed input-per-unit of output ratios from the unit isoquant So, if given isocost line AA' and isoquant curve KK', at S when a farmer has used certain quantities of inputs to produce output (y), technical efficiency is commonly measured by the ratio TE = OT/OS, 0<TE≤1

Figure 2: Input-oriented measure of TE S is inefficient input combination to produce a unit of output Y (line TS measured the amount which the input would have to be reduced to assume technical efficiency), X 1 and X 2 are inputs, Y is output, T is technically efficient point, T' is economically efficient point, AA' is isocost line, R is optimal input quantity and line KK' is isoquant curve (efficient input combinations)

Using Farell’s model of constant return to scale, the isoquant curve KK' captures the minimum combination of inputs per unit of output needed to produce a unit of output In this framework, every input combination along the isoquant curve are considered to be technically efficient, while input combinations above the curve such as S are considered inefficient because the input combination employed is beyond what is needed to produce

a unit of output For this producer using input combination S, the distance TS measures the technical inefficiency because this distance represents the amount of inputs which all inputs can be divided without reducing the amounts of output Ratio TS/OS measures the technical inefficiency associated to input combination at point S while the technical efficiency of the producer under this analysis (1-TS/OS) would be given by the ratio OT/OS TE value ranges between zero and one and a firm that is fully technically efficient has its technical efficiency score to one and vice versa

Isocost line AA' reflects input price ratio which can be assumed from a particular

segment RT, allocative inefficiency can be measured as SR/OR This ratio is deduced from the cost reduction that a producer would be able to reach if he intends to move from T, a technically efficient but allocatively inefficient point, to T', a technical and allocatively efficient point Therefore, allocative efficiency (AE) that characterizes the producer at point

S is given as the ratio OR/OT Farell (1957) defined what he referred to as the overall efficiency (Economic efficiency) as a multiplicative interaction of both technical and allocative efficiency

EE = TE ×AE = OT/OS ×OR/OT = OR/OS

Figure 3: Output-oriented measure of TE Q 1 and Q 2 are outputs, X is input, T is technically efficient point, T’ is economically efficient point, S is inefficient output combination, line AA' is isorevenue line, and line KK' is production frontier, where all output combinations are efficient

The second is output-oriented measure that tends to increase output proportionally with the same level of input, also under the assumption of constant return to scale This approach implies maximum radial expansion in all outputs that is feasible with given technology (Murillo-Zamorano, 2004) In the illustration above, assume a farm produces two outputs from a single input If the input quantity is held constant at a particular level, the technologies can be represented in two dimensions (Murillo-Zamorano, 2004) The farm

on the production frontier KK' are technically efficient and the farm operating below the curve such as S is inefficient While T is technically efficient point, higher revenue could

be achieved by producing at T', where marginal rate of transformation is equal to the price ratio or slope of the isorevenue line AA' Distance ST is the measure of technical

inefficiency or the amount of output that could be increased without increasing the inputs Therefore, technical efficiency is the ratio of OS/OT, the ratio by which returns may be increased without affecting the inputs

If input and output prices are available, an isorevenue line AA' could be drawn and allocative efficiency (AE) = OT/OR, will indicate the reduction in production cost that would occur if production were to occur at the allocatively and technically efficient point T' Though, technical efficiency may be 100% obtainable at T, but technical efficiency would be obtained at point T' with lowest cost of input As obtained in input-oriented measure, overall efficiency or economic efficiency is the multiplicative interaction of technical efficiency and allocative efficiency

EE = TE ×AE = OS/OT × OT/OR = OS/OR

TE, AE and EE scores are bounded by zero (totally inefficient) and one (totally efficient)

In summary, the relationship that exist between an observed production and the best practice production that a farm is able to put to use defines the level of technical efficiency

at which such farm operates (Hai au, 2009)

For the purpose of this study, the efficiency of the farms will be estimated using oriented measure of technical efficiency, this is because, firstly, the research is concentrating only on one output (shrimp) and secondly, because, mostly, the farmers only have the ability and capacity to manipulate their inputs and they have less control over output The production technology with the input-oriented measure of technical inefficiency can be expressed as:

Yi = f (Xi × Øi), i = 1…….N

Where Yi is the scalar output of each shrimp farm, Xi is the vector of inputs (Feed, seed, lime, labour) used to produce an output for a growing season Ø is scalar technical efficiency, calculated as Xe

ij/Xji ≤ 1, Xe

ij is input-vector in efficiency units Input-oriented technical inefficiency is measured as 1- Ø While i represent the farms, j is the rate at which all inputs could be reduced without reducing output One-stage approach that express

inefficiency effects (Ui) as an explicit function of a vector of firm-specific variables and a random errors was employed This model specification may be expressed below:

Yit = χit β + (V it – U it )

Where Yit is the log of the production of a specific farm (i) and at a specific time (t)

Χi is a K X 1 vector of input quantities transformation of a specific farm i

βi is a vector of unknown parameter

Vi andUi are the components of error term, Vi accounts for measurement error and other random factors that are beyond the control of the farms and the Ui is the inefficiency Uit represent technical cost efficiency and it must be positive

Uit = /Uit/, where Uit ⁓ N (0,σ2 )

V it are random variables that are assumed to be iid (independently and identically

distributed) Vit ⁓ N (0,σ2 ) and it is independent of Uit

1.7 Measurement of Technical efficiency

The approach of measuring technical efficiency was reported to have first been proposed

by Farell (1957) Building on the work of Farell (1957) were Charnes, Cooper, and Rhodes, (1978) on the specific research on efficiency measured for production units The work receives contribution from Banker in 1984, which resulted into the proposal of two approaches for the measurement of technical efficiency Non-frontier and frontier approaches were developed Non-frontier approach measures technical efficiency by comparing the actual output with the standard frontier estimated from the experimental data While this approach helps to separate and examine the conventional and non-

conventional inputs, it’s too expensive to conduct experimental study and the experimental result may not portray the real situation in production (Hai au, 2009)

Stochastic production frontier (Frontier approach) has been described as the appropriate tool for economic sectors where randomness is important tool in the production system Frontier approach describes the maximum output that can be produced from any combination of input by an efficient farm Data Envelopment Analysis (DEA) and stochastic production frontier are methods that are considered the best approaches for measuring technical efficiency of aquaculture farms (Xuan Huy, 2009)

Xuan Huy (2009) in her research referred to frontier and non-frontier approaches as parametric and non-parametric The latter is treated as a random variable due to the existence of exogenous factors that affect the relationship that exists between the output and the input, leading to the estimation of stochastic frontier that gives the expected values

of output using the conditional level of inputs

The non-parametric approach makes use of linear computer programming that produces stochastic production using piece-wise linear deterministic frontiers According to Xuan Huy (2009), this approach neither impose functional forms nor do they take into account the randomness that exist in the parametric approach Therefore, they are less prone to misspecification and are not subsequently subject to the problems of assuming an underlying distribution about error terms

However, technical efficiency estimated when the data generating process is not characterized as a full-frontier deterministic production model is negatively biased due to the feature of DEA that allows the largest random frontier shock in the data to determine the production frontier estimate (Sengupta, 1985) Also, the primary criticism of DEA approach is that measurement errors can have a large influence upon the shape and positioning of the estimated frontier In other to account for the presence of measurement error in production in the specification and estimation of frontier production functions, Aigner, Lovell, and Schmidt, (1977), and (Meeusen and van den Broeck, (1977)

independently proposed the frontier production function (Battese, et al.1996)

1.8 Stochastic production frontier

Stochastic production frontier has been described as a parametric and econometric approach that construct a production function based on average values of the observed data (Hai au, 2009) It maps a production frontier, finds the locus of maximum outputs that are associated with a given input levels and further estimates the farm-specific technical efficiency as a

deviation from the fitted frontier (Singh, et al 2009) It was reported to have first been

proposed by Aigner, Lovell and Schmidt, and Meeusen and Van den Broeck in 1977 It was described to have original specification that involved a production function that is specified for cross-sectional data which had an error term that had two components, one to be take account of the random effects and the other to take account of the technical inefficiency According to Färe and Lovell, (1978)), among the methods commonly used for measuring technical efficiency, stochastic production frontier has been considered as the most appropriate for assessing the technical efficiency of agricultural farms in the tropics because

it takes into account the measurement error and other stochastic factors such as weather condition and diseases which often influences the data This approach also takes into account the stochastic variation which is important if the output is affected by random noise

However, it requires specific functional forms such as Cobb-Douglas, translog or quadratic functions in order to estimate the production function In order to separate the inefficient component from stochastic component, it’s important to have some distributional assumptions This method cannot be applied to production functions with multiple outputs

The traditional approach to investigate the relationship between technical efficiency and socio-economic variables is estimating the stochastic production frontier at the first phase, providing the basis for measuring farm-level technical efficiency while the second phase estimates technical efficiency as a function of the various attributes of the farms in the sample using separate two-limit tobbit equations (Kehar singh, 2009) This two-staged approach includes a first stage that specifies and estimate the stochastic frontier production function and the prediction of the technical inefficiency effects with the assumption that these inefficiency are identically distributed and the second stage involves the specification

of a regression model for the predicted technical inefficiency effects that contradicts the

assumption of identically distributed inefficiency effort in the stochastic frontier However,

in the recent working paper of Coelli, (1996) on stochastic production frontier, the pronounced author described this two-stage approach as being inconsistent regarding the independence of the inefficiency and therefore, he suggested the approach may not provide estimates that are as efficient as a single-stage approach estimation method proposed by Reifschneider and Stevenson, (1991) The latter proposed stochastic frontier models in which the inefficiency model are expressed as an explicit function of a vector of firm specific variables and a random error

It was argued that the socio-economic variables should be incorporated into the production frontier model because such variables are capable of producing significant influence on the production efficiency Therefore, the simplicity, computational flexibility and ability to examine the various farm-specific variables on technical efficiency in an economically consistent manner has made the technical inefficient model Battese and Coelli, (1995) popular among other models (Kehar singh, 2009)

1.9 Literature review

Numbers of literatures have explored the use of technical efficiency in agricultural farms

to investigate the factors that are having the positive or negative impact on the production capacity of farms In 2008, Kareem and Dipeolu in Nigeria investigated the factors affecting production in 85 fish farms using stochastic production frontier approach based on the cobb-Doulas production The technical inefficiency function used were experience of the farmer, age, education level and size of the household while the input factors used included labour, time, fingerlings, feed and other materials In the result obtained, the mean technical efficiency of earthen ponds and concrete tank were 0.88-0.89 with no significant difference The experience of the farmers were found to have negative impact on the technical inefficiency of the concrete pond

Also, Islam, Tai, and Kusairi (2016) used farm level data in five regions in peninsula Malaysia to investigate technical efficiency of brackish water fish cage farms were also investigated The study considered socio-economic factors such as age, education, and

experience and production cycle as inefficient factors that might be affecting the technical efficiency The authors reported a mean technical efficiency of 0.37, a result that drove the authors’ suggestion that potentials for increasing fish production exists in the entire area of peninsula Malaysia The result of the technically inefficiency model was reported to show that farmers experience and the number of production cycles were negatively influencing the efficiency of finfish cage culture in the study areas Two important conclusions drawn from the results by the authors were that efficiency of fish cage culture can be improved through increase of production cycles and also that increasing the capacity building in the practice will increase production

In Bangladesh, in a two-output approach that involves a polyculture of prawn and carp, the technical efficiency was measured using Data Envelopment Analysis (DEA) with a cross-section of 105 farms The inputs considered included, fingerlings, labour, organic fertilizers and feed Pond size was found to have a positive impact on technical and cost efficiency, and a negative relationship was found to exist between pond size and allocative efficiency, feed application and technical, allocative and cost efficiency of the farms at the region However, Misraa & Misra (2014) research on establishing the systematic difference on the technical efficiency of fish farms of different size-classes categorized on various socio-economic conditions was reported to show that while experience, ownership and sole proprietorship were found to be important determinants of technical efficiency, pond size and education were reported to show no significant relationship with the technical efficiency

In order to compare the effects of technical inefficiency factors on different levels of production (extensive, intensive and semi-intensive aquaculture), Nguyen and Fisher, (2014) investigated the effects of pollution on intensive, extensive and semi-intensive shrimp farms in Mekong river delta in Vietnam Aside the comparisms of the efficiency of the different practices, the study also compared the efficiency of downstream farms and upstream farms using group frontier and meta-frontier analysis in a sample of 292 farms

In their conclusion, they reported that extensive fish farms were more efficient than intensive and semi-intensive shrimp farms which could be as a result of low cost of

operation They further reported that upstream farms were more efficient than downstream farms in the region, siting pollution as the major environmental variable that influenced the efficiency of the downstream region In their elaborate discussion, while assessing the impact of socio-economic factors, experience was found to be negative in relation to the technical efficiency of intensive farms and positively related to the TE of the extensive farms, while in the meta-frontier result, age of household head, farm size and experience were found to be positive on the technical efficiency of the extensive farms

In reference to numbers of literatures in the past, experience has seems to be an important socio-economic factor that contributes positively to the increasing yield of agricultural

firms However, this was not the case in the Kehar Singh et al (2009) research when the

authors assessed the level of technical efficiency and its determinants of small-scale fish production in the west Tripura district of the state of Tripura, India The authors used both one-stage and two-stage procedures of stochastic production frontier approach to analyze the determinants The study reported that TE ranges between 0.21 and 0.96, and the mean technical efficiency is 0.66 The authors further confirmed Battese and Coelli (1995) reported that one-stage procedure of stochastic production frontier gives reliable estimates

of the coefficients of stochastic frontier production function than that of the two-stage procedure and that Cobb-Douglas functional form is more dependable than that of translog form under the faming conditions in the west Tripura district of Tripura state In the technical inefficiency model, the authors found that there is a correlation positive between age and experience of the farmers and a negative correlation between the age of a farmer and education level as majority of the farmers were having a primary education or lower However, the author reported that the experience of farmers do not show positivity to the quantity of output Farmers’ age were reported to influence the practices either directly or indirectly through labour, management and knowledge as young and middle-aged farmers were more willing to adopt new technologies and old farmers were reported to be conservative and risk aversive

Aside these type of researches that investigate the factors that produce significant effect on the production efficiency of the system using set of economic inputs, aggregate cost and

output, there seems to be scanty literatures that consider looking into the impact of climate change factors on the productivity of agricultural farms However, Amin and Yacob, (2012) considered the empirical analyses of the impact of climate change and climate variability on the production efficiency and incomes at farm level in Kansas, Australia Climate variables, temperature and rainfall were considered in an approach that assumes that climate variabilities affect the technical efficiency of farms and then farm incomes These variables were included in technical inefficiency model and used along with another model, stochastic production frontier model which considered dependent variables such as farm products, inputs, capital, labour, purchased inputs, time and precipitation The third model was fixed effect model that includes quadratic terms of weather variables, farm full effects, vector of observed determinants of farm income which are time varying The result showed that climate variability produced significant effects on mean output elasticities with respect to inputs, return to scale and technical efficiency It was reported that purchased inputs showed more responsiveness to climate variability than capital and labour The authors reported that 30 years climate projections from this research showed that farm incomes will increase with a modest increase in mean maximum temperatures and decrease with a modest increase in mean minimum temperatures, with the combination showing a modest decline in average farm income that ranges from 0.2%-0.5%

In 2014, Auci and Vignani investigated the efficiency of Agricultural farms in Italian regions during the period of 2000-2010, when there was an observation climate change impact was on the increase Using stochastic production frontier approach, the effects of production inputs such as labour, physical and human capital, from inefficient meteorological factors such as temperature, and rainfall was analysed The main objective

of the work was to analyze the economic impacts of climate change on the agricultural sector in Italian regions The analysis was concluded by ranking of the Italian regions on the basis of these estimated technical efficiencies The result obtained showed that variability of rainfall had a positive impact on the production capacity while the average minimum temperature produced a significant reducing effect on the quantity of harvested production

In Chile, Roco et al (2017) analysed the impact of climate change adaptation on

productivity for annual crops of 265 farms in the central part of the country using stochastic production frontier approach In order to measure climate change adaptation, set of 14 practices were adopted in three different specifications; binary variable, count, and index, which represent decision, intensity and adaptation quality respectively These variables were used in three different stochastic production frontier models The authors suggested that the use of adaptive strategies had a significant and positive impact on productivity The practice with the highest impact on productivity was irrigation improvement Empirically, they concluded that there is a relevance of climate change on farmers’ productivity and it

is believed that this should facilitate discussions regarding the need to implement adaptation measures

Recently in Vietnam, impact of climate change on the technical efficiency of striped catfish

was carried out by Nguyen, et al (2017) in the Mekong Delta The study aimed at

identifying the major factors that affect the technical efficiency of Pangasius farms, while exploring the relationship that exist between those factors and the impact climate change

has or would have on Pangasius farming DEA-bootstrapping running in the

R-environment was described to be used to annul the overestimation of the technical efficiency because the empirical sample is usually a fraction of the entire population of the decision-making units; a problem that has been established with DEA approach The method employed multiple inputs to produce multiple outputs The result showed that the well-educated farmers and the more efficient farmers have perceived the impact of climate change and have defined means of dealing with the influences Salinity intrusion was found

to have reduced the scale of operation of the farmers located at the down-stream regions while the farmers’ located at flood-prone areas of the upstream and midstream regions have larger scale of operation This success of the upstream and midstream may be attributed to the higher level of education of the farmers and their huge experience at dealing with impact

of climate change as reported by the data It may also be attributed to the fact that Pangasius farming originated from this region and the farmers may possess quite better technical know-how of the best operational practices However, the higher TE found in the

downstream was reported to have been due to the lower cost of production and use of small areas of lands for farming operations, which facilitated the authors’ conclusion that in order

to increase technical efficiency, inputs cost should be reduced

Apart from the thesis research of Xuan Huy (2009) that was conducted on the comparisms

of technical efficiencies of Khanh Hoa districts using only socio-economic data, on labour, machines, pond area and depth and activities in a two-outputs (shrimp and fish) and five-inputs DEA approach, there seems to be no literature that considers the impact of the environmental stresses on the efficiency of the shrimp farms in the region Therefore, this study will seize an initiative of investigating the factors (including environmental and socio-economic factors) that are influencing the optimal efficiency of shrimp farms at Cam Thinh Dong, which is located in one of the one-time largest producer of shrimp’s district The study will measure the technical efficiency using the stochastic production frontier approach of Battese and Coelli (1995), investigating the variables of the technical inefficiencies that is probably deviating the practice from its efficient position To investigate the inefficiency factors that are influencing the farms’ efficiency, the study will consider, age, education, experience, farm size and farmers experience on flood and drought in the region as inefficiency variables

2.0 DATA

2.1 Data sampling methodology

The survey exercise was carried out using 50 questionnaires The farmers were selected with the help of the community committee based on the farmer’s activeness in the livelihood, willingness to answer the questions in the questionnaire and those that were perceived to be true representative of shrimp farmers in the region in order to limit or avoid sampling error

2.2 Primary survey

Prior to data collection, primary survey was carried out to observe the aquaculture activities

at the selected study area, Cam Thinh dong commune, Cam Ranh district The primary survey revealed the majorly cultured species in the area is shrimp, and it also revealed the mostly adopted culture systems The pre-survey exercise was carried out with the help of a Vietnamese mate for ease of communication, and during the course of the pre-survey, chances were taken to interview and discuss with the farmers, and also met with the necessary authority for the approval of the research in the area

2.3 Pre-testing

A pre-testing exercise was conducted to measure the level of viability of the questionnaire for the research Five households were selected randomly for the test and the pre-testing exercise was also accompanied with interviews The pre-testing exercise revealed the necessary variables that should be included in the questionnaire and corrections were made

on the questionnaire to suit the objectives of the research

2.4 Data collection

Haven made correction to the pre-tested questionnaire from the pre-survey exercise, questionnaire was structured and main survey was carried out 50 households were selected using the mentioned criteria for the main survey exercise and cross-sectional data were collected on household characteristics, farmers’ major cost and income (only on last crop), experience on environmental impacts and adaptation strategies The breakdown of these are listed in the following list of information collected:

1 Household characteristics: Age, gender, level of education, years of experience and number of households

2 Labour (Own and hired): Numbers of labours, total working hours and amount of money paid to labours per hour

3 Output: Weight and value of harvested shrimps (only last crop)

4 Cost of inputs: Weight (kilogram) and values of shrimp seeds, feed lime and units

of electricity

5 Environmental impacts on farming operation: Using ranking from 1 to 5, with 1 indicating the “No impact” and 5 indicating “Strong impact”, impact of flood, drought, and sea water intrusion were investigated on the shrimp farm operation

6 Experience on environmental hazards: Farmers were asked the rough estimation of numbers of flood, drought and salt-water intrusion that they experienced in their farming operation in the past

7 Adaptive strategies: Using ranking from 1 to 5, with 1 indicating “Never adopted strategy” and 5 indicating “mostly used strategy”, information were collected on pond adjustment, increasing irrigation, integrated farming, Insurance, polyculture, switching to brackish species culture, and finding off-farm job

Below are the variables used in the study:

 Production function

Y Total output of last crop (Kg)

X1 Labour used (hours)

X2 Shrimp seeds (1000indv.)

Z6 Number of drought events experienced

2.5 Data sample aggregate

Table 1: Summary statistics for some of the variables used in the model

3.0 METHOD

3.1 Data processing and analysis

The processed data was analysed using frontier 4.1 software developed for stochastic production frontier by Coelli, (1996) The software was used to estimate technical efficiency of each farm, the mean technical efficiency of the entire farms and the proposed inefficiency variables that are influencing the efficiency of the farms

How the software estimated the technical efficiency according to Coelli, (1996)

 The software first estimated the ordinary least squares (OLS) All β estimators with the exception of the intercept were unbiased

 The software then conducted a two-phase grid search of ɤ, with the β parameters (except β0) set to the OLS values and the β0 and σ2 parameters adjusted according to the created OLS formula presented in (Coelli, 1995) In this grid search, other parameters such as δ and µ are set to zero

 Using the values selected from the grid search, the final maximum likelihood estimates were obtained in an iterative procedure using those selected values

3.3 Expectation

Many of the studies carried out on the investigation of factors affecting the efficiency has been centered on socio-economic factors The few that have investigated environmental factors are in agriculture, some of which investigate temperature and rainfall, Auci and Vignani (2014); Mugera and Zereyenus (2012) Researches are still unclear on the impact

of environmental hazards on the technical efficiency of shrimp farms However, the association of environmental factors to continuous reduction of shrimp output in this region

by Pham Xuan Thuy (2004), Nguyen Thi Kieu Thao, (2012), and Dang Hoang Xuan Huy

(2009); coupled with recent report from Nguyen et al (2017) study on the impact of salt

water intrusion and flood on Pangasius farming in Mekong river which revealed that salt water intrusion derailed the operation of farms in flood prone area may have given some element of hope that the investigated environmental factors in this study may be producing

a sounding effect on the farmers technical efficiency

4.0 RESULT

4.1 The Empirical Result

According to table 1, some of the variables are positive while others are negative While 50% of the variables are statistically significant at 1%, 5% or 10%, others are statistically insignificant at the three levels

Table 1: Stochastic frontier and Inefficiency effects result

Coefficient Standard error T-ratio

Stochastic frontier

Variable

Constant 5.1646436 *** 0.49665524 10.398851 β1 (Labour) 0.15062554 0.099934406 1.5072441

LR (Observe) = 31.275713

LR (Given) = 11.911 (At 95% significance level)

Note: *, **, and *** indicate 10%, 5% and 1% significance level respectively

The elasticities of output with respect to inputs are given by the MLE estimates In contrast

to labour, which is insignificant at 1%, 5% and 10%, the output elasticities for feed, seed

and lime are significant at 1% and 5% significant level Feed is the highest, followed by lime and seed

Whereas, in the inefficient model of the MLE estimates, apart from δ5, which is positive, other variables are negative The negative and statistically significant sign of the coefficient

of education indicates that shrimp farmers with higher level of education are more efficient than farmers with lower level of education

Gamma (ɤ) measures the total variation of output from the frontier that attributed to the existence of random noise or inefficiency

4.2 Test of hypothesis

According to Battese and Coelli (1995), the technical inefficiency model can only be estimated if the technical inefficiency effects (Ui) are stochastic and have particular distributional properties Therefore, it is of interest to test two null hypothesis expressed in the table below:

Table 2: Hypothesis testing

to a traditional average response function in which the explanatory variables in the technical

inefficiency model are also included in the production function (Sharma, et al 2009) To

test this a generalized likelihood test was carried out

λ = -2 (ln (L(H0)) –ln(L(H1)),

Where, L (H0) (-38.135) is the value of the log-likelihood function of the frontier under the null hypothesis and L (H1) (-22.497) is the log-likelihood function of the frontier under the alternative hypothesis The λ value gotten from this equation is 31.276 and this was compared with the critical value (2.706) which was gotten from Kodde and Palm, (1986) table under 1 restriction at 95% confidence level Since the observed value is greater than the critical value, the null hypothesis was rejected and alternative hypothesis was accepted

Hypothesis 2: The null hypothesis of this test examines that technical inefficiency effects are not present, hence the inefficient variables and the variance parameter are equal to zero This was calculated using the same formula, but it is required that all δ parameters and variance parameter ɤ are equal to zero The degree of freedom is the number of parameters (including the variance) that are restricted The null hypothesis was also rejected as the critical value (13.401) is less than the observed critical value (39.022), indicating that inefficiency effects are presents