J World Aquaculture Soc - 2021 - Khan - Production Risk Technical Efficiency and Input Use Nexus Lessons From Bangladesh

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Received: 9 November 2019 Revised: 10 December 2020 Accepted: 28 December 2020

DOI: 10.1111/jwas.12767

APPLIED STUDIES

Production risk, technical efficiency, and input use


nexus: Lessons from Bangladesh aquaculture

Md. Akhtaruzzaman Khan1 | Ratna Begum2 | Rasmus Nielsen3 |


3
Ayoe Hoff
1
Department of Agricultural Finance and Banking, Bangladesh Agricultural University, Mymensingh, Bangladesh
2
Department of Agricultural Finance & Co-operatives, Bangabandhu Sheikh Mujibur Rahman Agricultural University, Gazipur,
Bangladesh
3
Department of Food and Resource Economics, University of Copenhagen, Frederiksberg C, Denmark

Correspondence
Md. Akhtaruzzaman Khan, Department of
Abstract
Agricultural Finance and Banking, Bangladesh The optimal use of resources in aquaculture is important,
Agricultural University, Mymensingh, 2202,
especially in developing countries, to obtain the highest pos-
Bangladesh.
Email: azkhan13@bau.edu.bd sible outcome from the production process to support food
security and poverty alleviation. Thus, within this study, the
Funding information
Danish International Development Agency risk, efficiency, and input-use variation in aquaculture farms
(DANIDA), Grant/Award Number: Project in Bangladesh is investigated using a flexible stochastic fron-
code 14-06KU
tier model with a risk and an inefficiency function. The
results reveal that feed, labor, and capital have positive and
significant impacts on production. In addition, an increased
fingerling density and a larger farm increase the risk,
whereas the use of feed and the capital invested have the
opposite effect. Access to extension services has a positive
effect and increases farm efficiency. An investigation of the
farm size–productivity inverse relationship reveals that this
phenomenon is not applicable to Bangladesh aquaculture. In
general, efficient farmers are large-scale farmers, who use a
lower stocking density but a higher feeding intensity,
resulting in a higher yield. On average, farmers use less labor
and feed than what is optimal. To increase efficiency and
reduce risk, it is recommended that more technical

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.
© 2021 The Authors. Journal of the World Aquaculture Society published by Wiley Periodicals LLC on behalf of World Aquaculture
Society.

J World Aquac Soc. 2021;52:57–72. wileyonlinelibrary.com/journal/jwas 57


17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
58 KHAN ET AL.

knowledge on optimal input use, extension service, and cap-


ital is made available to aquaculture farmers.

KEYWORDS
aquaculture, flexible stochastic frontier function, optimal input use,
production risk

1 | I N T RO DU CT I O N

The growth in global aquaculture production is dominated by the Asia Pacific region, which accounts for more than
89% of world production. At the same time, aquaculture plays a vital role in the countries in this region, providing
food security, employment, income, poverty alleviation, and, in a wider sense, socioeconomic stability in rural areas
(Belton, Bush, & Little, 2018; Food and Agriculture Organization, 2020). However, aquaculture is considered a rela-
tively high-risk business (Bergfjord, 2009; Dey, Bose, & Alam, 2008) because it depends on biological processes
(Asche & Bjørndal, 2011; Dey et al., 2010) and external factors, including the availability of water, temperature, and
human skill (Gaidar, 2002; Liefert & Swinnen, 2002). Thus, a number of stochastic biophysical and socioeconomic
factors influence farmer production practices and may create substantially different risk profiles among farmers.
Bangladesh is the fifth largest aquaculture producer in the world, and aquaculture production provides 56% of
the total fish production in the country, mostly originating from earthen-pond farms (Department of
Fisheries, 2017–18). In Bangladesh, the production and productivity of aquaculture has generally been increasing;
however, variation in productivity occurs from farm to farm (Rahman, Nielsen, & Khan, 2019) and within different
locations and environments (Mitra, Khan, Nielsen, & Islam, 2020; Rahman, Nielsen, Khan, & Asmild, 2019), which
may indicate substantial production risk in pond aquaculture. Production factors that are associated with risk may
lead to inefficient use of these inputs. Additional factors that may influence production efficiency include production
input intensity (Prodhan & Khan, 2018), the size of the operation, operator experience, education and training, access
to credit (Mitra, Khan, & Nielsen, 2019), adoption of new technology (Dey et al., 2013), lack of institutional service
(Khan, Roll, & Guttormsen, 2021; Mukta, Khan, Mian, & Juice, 2019), and several other management factors
(Alam, 2011; Alam, Khan, & Huq, 2012; Khan & Alam, 2003). Each of these factors is, in turn, affected by the opera-
tor's assessment of production risk (Rahman, Nielsen, Khan & Ahsan, 2021).
What occurs under the pond surface during the culture period is largely beyond the farmer's control. Therefore,
input use techniques and managerial risk judgments are very important to securing an optimal outcome. Different
inputs have different effects on production risk (Khan, Guttormsen, & Roll, 2018), and scientific input management
practices (Prodhan & Khan, 2018) are therefore important for production optimization as well as for risk control.
Although production risk and optimal input use are important issues within aquaculture, relatively few studies
have focused on Bangladesh due to the lack of reliable farm-level data. Studies on Bangladesh include Alam,
Guttormsen, and Roll (2019), who determined that production risk and inefficiency exist in tilapia farming, and Alam
and Guttormsen (2019), who examined risk perception and risk management strategies among farmers and found
that fish diseases, fish price variability, quality of feed and fingerlings, and credit constraints were perceived as the
most important sources of risk. Khan et al. (2018) found that significant production risk exists in pangas aquaculture
and that it varies with the size of the operations. Furthermore, Anwar (2011) suggested that price fluctuation, dis-
ease, and natural disasters (floods, cyclones, and heavy rains) were the primary risk factors for pangas and tilapia
farming in Bangladesh.
Inefficiency, production heterogeneity, lack of management knowledge, and lack of institutional facilities suggest
that there is substantial potential for more efficient use of input quantities and improvements in efficiency, which
are crucial within a resource-scarce country such as Bangladesh. Thus, this study aims to address the research gap
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 59

related to the optimal combination of inputs used in pond farming in Bangladesh, which will allow farmers to opti-
mize production, reduce risk, and thereby inefficiency. Furthermore, to evaluate whether economies of scale exist,
the inverse relationship phenomenon is tested, which has not been tested within pond aquaculture. Risk, efficiency,
and input use variation are examined using a flexible stochastic frontier model including a risk and an inefficiency
function. The investigation of the influence of farm size (productivity inverse relationship) is tested using a local poly-
nomial regression model. A further aim of the study is to assist policy makers in establishing best practices for
farmers to expand aquaculture production, which could benefit food security, income generation, and poverty allevi-
ation in Bangladesh.
The article is organized as follows: In the next section, a short presentation of the Bangladesh pond aquaculture
sector is provided. Thereafter, the data and the method used are presented, followed by the empirical specification
and estimation of the model. Then, the empirical results are presented and discussed before the concluding remarks.

1.1 | Bangladesh pond aquaculture

In Bangladesh, aquaculture production has increased tremendously during the last three decades and now accounts
for 56% of the total fish production (Department of Fisheries, 2017–18). The dominant production system is earthen
ponds, which account for 79% of the total production. Ponds are most often constructed by converting farmland
(rice fields) into fishponds due to the higher profitability obtained by producing fish (Khan et al., 2018). The species
that have contributed the most to this development are carp, tilapia, and pangas, increasing pond production from
2,600 kg per hectare in 2001 to 5,000 kg per hectare in 2019 (Figure 1).
The most important inputs in the Bangladesh pond aquaculture sector are feed, fingerlings, labor, and capital.
Feed costs constitute the highest operating cost (60–70% of total production costs) and are the most important
input for intensive or semi-intensive grow-out farming (Khan et al., 2018; Prodhan & Khan, 2018). Feed is assumed
to increase fish productivity, but a farmer's ability to use the optimal quantity of feed, the frequency of feed use, and
the use of quality feed may cause production variability. Fingerlings are also an important input where farmers need
to consider stocking density and quality of fingerlings. More than 800 hatcheries produce fingerlings of different
qualities; therefore, farmers need to have proper knowledge to purchase the best quality fingerlings. In addition, the
use of non-optimal stocking densities might lead to increased mortality and reduced productivity. Labor is another
important production input for farm management, monitoring, and control and consists of a significant portion of the

FIGURE 1 Pond aquaculture production and productivity trends over two decades
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
60 KHAN ET AL.

total production cost (Prodhan & Khan, 2018; Rahman, Nielsen, Khan, & Yeboah, 2020). Efficient use of labor is
expected to reduce production risk and increase farm profitability. In commercial aquaculture, investments in capital
items, such as pumps, shallow tube wells (STWs), drums, nets, houses for caretakers, boats, aerators, feeders, feeding
trays, and weight machines, are important costs that have significant effects on farm profitability (Khan, Roll, &
Guttormsen, 2021).

2 | MATERIALS AND METHODS

2.1 | Data

For this study, aquaculture farmers were selected following a three-stage procedure. First, the Mymensingh district
was selected as the primary study area because it has a long tradition of pond aquaculture production due to favor-
able land and freshwater resources, climatic conditions, access to plentiful labor, and because the feed processing
industry is mostly located in this area. Mymensingh is the largest fish-producing district, with a production of
400,000 m.t., accounting for 17% of aquaculture production (Department of Fisheries, 2017–18). Second, 3 upazilas
(subdistrict), namely, Muktagachha, Trishal, and Phulpur, were selected among the 12 upazilas because they are the
major aquaculture-producing areas of the Mymensingh district. Finally, a list of pangas aquaculture farmers was
obtained from the Upazila Fisheries Officer (UFO), and a total of 225 pangas farmers were selected using a random
sampling technique from these three upazilas. A pilot study was conducted by pretesting the questionnaire designed
for the interviews in the field through visits and consultations with 10 farmers. The final version of the questionnaire
was then used to collect production economics data and social variables from the complete sample of farmers using
face-to-face interviews.

2.2 | Analytical technique

The point of departure for estimating production risk is the conventional specification of a stochastic frontier pro-
duction (SFP) function following Aigner, Lovell, and Schmidt (1977), Schmidt (1977), and Meeusen and van den
Broeck (1977), which can be written as the following:

yi = f ðxi ; αÞ + vi −ui , ð1Þ

where yi measures the output from the ith farm (i = 1,…,N), xi is a vector of inputs applied by the ith farm, f(.) represents
the frontier production (i.e., the deterministic component of the frontier), α is a vector of functional parameters that
define the production, νi represents the noise, and ui is the inefficiency of the ith farm, that is, how far this farm is from
operating at maximum productivity. A conventional estimation of SFP assumes that the function is homoscedastic; that
is, the noise term νi is identically distributed and independent of the input level. However, when tests for
heteroscedasticity are shown to be significant, Just and Pope (1978) specify this as production risk, that is, the influ-
ence of certain controllable (input) factors on the production noise. Conversely, if heteroscedasticity tests are nonsig-
nificant, then risk is not present in production. In the case wherein production risk is present, Just and Pope (1978)
suggested a specification of the SFP that contains the following two general functions: a mean production and a vari-
ance function. The Just and Pope model allows for a differentiation of the impact of the inputs on the outputs and on
output risk and provides adequate flexibility to accommodate both positive and negative marginal risks with respect to
inputs. The general form of the production function presented by Just and Pope (1978) is given by:

yi = f ðxi ; αÞ + hðxi ; βÞνi , ð2Þ


17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 61

where f(.) is the mean production given the input levels and h(.) captures the variance function (or production risk as
a function of input), which allows for heteroscedasticity in the random noise, and v is the exogenous stochastic dis-
turbance/noise (or production shock).
Equation (2) incorporates risk in a general production model describing the mean production of a firm as a func-
tion of relevant inputs. However, this process can be taken one step further by integrating risk in the production
frontier model given in Equation (1). This step makes it possible to obtain consistent estimates of production effi-
ciency by taking into account the possible effects of risk on efficiency assessments. Kumbhakar (2002) introduced
these effects in a flexible stochastic frontier function, extending Equation (2) by introducing an additional individual
function q(.) for modeling the technical inefficiency:

y = f ðx; αÞ + hðx;βÞv − qðz; γ Þu, ð3Þ

where q(z; γ) represents the effects of farmers' socioeconomic variables (represented by the vector z) on the techni-
cal inefficiency effects. The model given in Equation (3) is applied in the present context.

2.3 | Empirical model

In this article, the model developed by Just and Pope (1978) and extended by Kumbhakar (2002) for measuring risk
and technical inefficiency is applied. Thus, three functions (the mean, the variance, and the inefficiency functions)
must be estimated within the model (cf. Equation 3).
To choose the functional form of these three functions, different tests were performed. Based on the test
results (Table 1), translog half-normal functional form found suitable for the mean production function which can be
written as follows:

X
5
1X 5 X 5   X 2
ln½f ðx; αÞ = α0 + αi lnðX i Þ + αij lnðXi Þln Xj + δd Dd + μ, ð4Þ
i=1
2 i=1 j=1 d=1

where y is the yield of fish harvested; X is an input vector composed of labor, fingerlings, feed, farm size, and capital
costs; and Dd is a dummy representing the upazilas Muktagachha and Trishal. Phulpur is selected as the base upazilas
(or subdistricts), and μ is the half normally distributed residual.
In specifying the variance function h(x; β), the Cobb–Douglas functional form is applied. Thus, the variance func-
tion is specified as follows:

X
5 X
2
ln½hðx; βÞ = β0 + βi lnðX i Þ + δd Dd + λ: ð5Þ
i=1 d=1

Finally, the inefficiency function q(z; γ) is specified with a linear functional form, which can be written as follows:

X
6 X
2
qðz; γ Þ = γ 0 + γ s Ss + δd Dd + ν, ð6Þ
s=1 d=1

where S is a vector composed of socioeconomic variables including the age of the farmer, years of attending school,
years of experience in fish culturing, training in days, credit received or not, and extension service received or not.
The last two variables are dummy variables, where no credit and no extension service received are used as bases in
the model. To test the level of significance of the model parameters, a generalized likelihood ratio (LR) test is applied.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
62 KHAN ET AL.

3 | RESULTS AND DISCUSSION

3.1 | Tested hypotheses

The choice of the functional form is important when using parametric frontier analysis. As such, several tests have
been conducted to choose the mean functional form. Tests of the null hypotheses in the functional models are
obtained using the generalized LR test. Table 1 depicts the different tests and the results. According to the tests, the
translog half-normal functional form provides a better fit compared to that of the translog truncated normal because
the statistical test fails to reject the null hypothesis.
Similar tests of different functional forms reveal that the Cobb–Douglas form is the most appropriate for the
risk/variance function and that the linear functional form is appropriate for the inefficiency function.
One aim of this article is to investigate whether production risk exists in aquaculture production in Bangladesh.
Other studies of aquaculture have shown substantial heterogeneity among farms because of differences in manage-
ment, technology, knowledge, or biophysical conditions (Asche & Tveteras, 1999; Kumbhakar, 2002; Kumbhakar &
Tveteras, 2003). As discussed earlier, production risk can be identified by testing for heteroscedasticity. Thus, the
mean production function (Equation 4) is estimated using ordinary least square (OLS) to test for heteroscedasticity.
The results suggest that the data used in the estimated model are a good fit with an R2 value of 0.92, implying that
the variation in the dependent variable is sufficiently explained by the variation in the exogenous variables. Based on
the OLS results, Breusch-Pagan and White heteroscedasticity tests were conducted. The chi-square test statistics
results are 20.83 and 77.36, respectively. As such, both tests reject the hypothesis of homoscedasticity at a signifi-
cant level of 1%, indicating that production risk is present in the selected aquaculture farming.
This result also implies that if heteroscedasticity is present in the model, then it is most likely because risks exist in pond
aquaculture farming and not because the model has been poorly specified, as discussed in Asche and Tveteras (1999).
The maximum LR test is also used to verify the presence of production risk in the variance function. The presump-
tion is that considerable risk prevails in pond aquaculture related to the intensity of input use. Thus, a stochastic frontier
function with a flexible risk specification is a more consistent representation of production than is the assumption of
homoscedasticity, that is, no risk. Furthermore, the hypothesis that there is no inefficiency related to the socioeconomic
variables is tested. However, the estimated log-LR test rejects the null hypothesis at a 1% level of significance, which
implies that inefficiency is dependent on socioeconomic variables in the pond farming system (Table 2).

3.2 | Profitability analysis

In fish farming, the cost of production plays a vital role in farmers' decision-making processes. The production cost in
pond aquaculture includes both the cost of inputs and farm maintenance. The production of aquaculture involves

TABLE 1 Efficiency test statistics for the main production function

LR test No. of Mixture χ 20.01 critical


Hypothesis statistics restrictions valuea Decision
Choice of functional form:
I. Cobb–Douglas half-normal versus translog 32.24 14 29.384 Reject
half-normal
II. Cobb–Douglas truncated versus translog 32.31 14 29.384 Reject
truncated
III. Translog half-normal versus translog 2.41 1 5.412 Accept
truncated
a
The critical values are obtained from Kodde and Palm (1986).
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 63

TABLE 2 Test of homoscedasticity and inefficiency

LR test No. of Mixture χ 20.01


Hypothesis statistics restrictions critical valuea Decision
1 Test of homoscedasticity. 20.17 7 17.755 Reject
Test of the variance function h(x; β):
H0 : β1 = … = β5 = 0, δ1 = δ2 = 0
2 Test of no inefficiency. 10.54 1 5.412 Reject
No effects of socioeconomic variables on
inefficiency:
H0 : γ 1 = … = γ 6 = 0, δ1 = δ2 = 0
a
The critical values are obtained from Kodde and Palm (1986).

various farm operations, including the reconstruction of ponds, the application of feed, seed and fertilizer, and
harvesting. The cost data were collected using the questionnaire described earlier. The per-hectare cost, returns, and
profitability of the selected farms are displayed in Table 3.
The use of labor for fish production is approximately the same in the three different subdistricts Trishal,
Muktagachha, and Phulpur, as shown in Table 3. On average, 316 man-days of labor were used per hectare, at a cost
of BDT 61,933 representing 5% of the gross cost. Good quality feed and management practices are the key factors
for increasing productivity in aquaculture farming (Aktar, Khan, Prodhan, & Mukta, 2018; Prodhan & Khan, 2018;
Sarker, Arshad, Alam, Mohamed, & Khan, 2016a). The amount of feed used is determined by the age of the fish, the
stocking density, the amount of natural food available, the size of the farm, weather conditions, and the price of feed.
Considering all areas, farmers used 27,037 kg of feed per hectare, corresponding to a cost of BDT 770,677. The feed
cost amounted to 67% of the gross cost. This is consistent with findings in Mukta et al. (2019), Prodhan and
Khan (2018), and Khan, 2012, where feed cost accounted for 70%, 77%, and 71% of total production cost,
respectively.
Farmers in the study area purchased fry (seed) for production; the fry cost was calculated based on the actual
market prices paid by the farmers. The cost of fry depends on quality, size, and availability. Farmers usually purchase
3- to 4-day-old fry, which are placed in a nursing pond and released to the main culture pond when fingerling size is
reached. The results of this study suggest that the stocking of fry/fingerlings varied among the three subdistricts. In
Trishal, the stocking density was 44,460 fry/fingerlings per hectare; in Muktagachha, it was 73,300 per hectare; and,
in Phulpur, it was 77,802 fry/fingerlings per hectare. In Trishal, the majority of the farmers follow the scientifically
recommended stocking density rate of approximately 37,000 to 50,000 fingerlings per hectare (Bangladesh Fisheries
Research Institute, 2012). Considering all areas, the stocking density was 66,213 fingerlings per hectare, with a cost
of BDT 116,952 corresponding to 10% of the gross cost.
Farmers also use fertilizer in their ponds to facilitate the growth of algae, which is the natural feed for aquacul-
ture, and to supplement feed pellets. Fish farmers apply various types of fertilizer, including urea, manure (cow-
dung), and triple superphosphate (TSP). In the study area, the farmers use cow dung that they purchased or that is
from their own livestock. On average, 746 kg of fertilizer is used for production at an estimated cost of BDT 8,420
per hectare or 1% of the gross cost. Other variable costs include lime, which helps neutralize the acidity in the soil
and prevents diseases.
Fixed costs include the lease value of ponds or land and capital costs. Capital costs include the cost of the STWs,
pumps, bamboo, aerators, drum/fishing traps, feeding trays, weight machines, boats, nets, feed equipment and stor-
age houses, guard sheds, and other equipment costs. As shown in Table 3, the per-hectare gross cost is estimated at
BDT 1181,112, BDT 1186,292, and BDT 1089,994 in Trishal, Muktagachha, and Phulpur, respectively. The average
gross cost for all farmers in the sample is BDT 1152,522. The average gross return per hectare of all farmers was
64

TABLE 3 Profitability analysis of the selected pond aquaculture farming (per hectare)

Trishal Muktagachha Phulpur All % of


(n = 75) (n = 75) (n = 75) (n = 225) gross cost
Items Quantity Cost (BDT/ha) Quantity Cost (BDT/ha) Quantity Cost (BDT/ha) Quantity Cost (BDT/ha)
(I) Labor (man-days/ha) 324 64,092 316 62,370 306 59,441 316 61,933 5.37
(II) Fry/fingerling 44,460 114,373 73,300 140,017 77,802 96,340 66,213 116,952 10.18
(number/ha)
(III) Feed (kg/ha) 31,999 793,443 30,653 782,271 23,638 736,319 27,037 770,677 66.87
(IV) Fertilizer (kg/ha) 724 7,346 620 6,849 892 11,009 746 8,420 0.73
(V) Other costs (BDT/ha) 27,716 33,091 27,691 29,531 2.56
A. Variable cost 1,006,970 1,024,598 930,802 987,513 85.68
(BDT/ha)
B. Lease value (BDT/ha) 62,212 55,019 61,113 59,448 5.16
C. Capital cost (BDT/ha) 111,931 106,674 98,079 105,560 9.16
Gross cost (A + B + C) 1,181,112 1,186,292 1,089,994 1,152,522 100
(BDT/ha)
Production (kg/ha) 17,811 18,147 16,579 17,458
Output price (BDT/ kg) 75.68 75.98 77.63 66.64
Gross return (BDT/ha) 1,347,921 1,378,729 1,287,090 1,337,913
Gross margin (BDT/ha) 340,951 354,131 534,646 411,228
Net farm income (NFI) 166,809 192,438 197,096 185,391
(BDT/ha)
Benefit/cost ratio (BCR) 1.14 1.16 1.18 1.16

Note: BDT is Bangladesh Taka (currency), 1 USD = BDT 84, “ha” indicates hectare.
KHAN ET AL.

17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 65

estimated at BDT 1337,913. The total gross margin was estimated at BDT 411,228, and on average, the net return
was BDT 185,391.
Since the data are cross sectional, the undiscounted benefit–cost ratio was estimated. Overall, the benefit–cost
ratio was 1.16, indicating that farmers earn BDT 116 when spending BDT 100. Productivity is one of the primary
indicators of any culture system where farmers always try to increase the volume of production with the given inputs
available. In this study, the production was found to be approximately 17.5 tons per hectare.

3.3 | Production risk and efficiency

The values of the outputs and all exogenous variables included in the translog half-normal functional form were
mean-corrected and normalized using the individual sample means. In Table 4, the input coefficients of the produc-
tion function are presented.
Because all variables have been standardized by their geometric means, their logged values are zero at the sam-
ple means. Therefore, the partial production elasticities of each input at the sample means can be read from the
Cobb–Douglas portion of the mean production function (Bokusheva & Hockmann, 2006).
The coefficients of feed, labor, and capital are positive and influence production significantly. The positive sign
suggests that an increase in the use of these input factors would increase production. The input of feed has the
strongest influence on the production of aquaculture, with a positive elasticity of 0.55, which suggests that produc-
tion increases 0.55% with a 1% increase in the use of feed. Labor and capital have elasticities of 0.15 and 0.13,
respectively.
The lower part of Table 4 shows that the increased use of fingerlings increases risk in farms because the variable
is positive and significant. This negative relation can be explained as follows: farmers believe that stocking more fin-
gerlings in ponds will increase production and therefore do not follow the scientifically prescribed density of finger-
lings. As a result, the level of oxygen becomes too low, and the creation of toxic by-products such as carbon dioxide
and ammonia increases, which ultimately decreases productivity and increases production risk. Tveteras (1999)
found that fingerlings had a risk-increasing effect, supporting our results; Khan (2012) also found that fingerlings had
a risk-increasing effect in fish farming in Bangladesh.
A larger farm size also increases risk. An increase in the area used for production reduces a farmer's ability to
monitor all aspects of production, and farmers may fail to discover unfavorable conditions early enough to prevent
production yield losses. Therefore, increasing production area is associated with increasing variability and risk in
aquaculture production. A similar result was also supported by Losinger (2006) and Bokusheva and Hockmann (2006),
suggesting that risk increases with farm size.
Conversely, the results suggest that increasing investment in equipment/physical production assets can reduce
risk. Investments in STWs, boats, feed storage units, security guards, water pumps, and nets provide farmers with
more control over the production process and thereby increase efficiency. Therefore, investment in these assets can
reduce the production variability among farms. Kumbhakar (2002) also found that capital investments had a risk-
reducing effect in Norwegian aquaculture production.
Although feed has a risk-reducing effect, this effect is only found to be statistically significant at the 10% level.
This result suggests that farmers can reduce risk by applying additional feed. In contrast, Tveteras (1999) found that
feed increased output risk in Norwegian salmon farming. The different results may be explained by the fact that the
majority of Bangladesh fish farmers are small scale and not financially solvent. Under these circumstances, it is diffi-
cult for small-scale producers to provide sufficient feed to maintain an optimal combination of inputs. Table 4 further
shows that labor has no significant risk-reducing effect. In the study area, the majority of the farms use family labor,
and the variability in labor use among farms and locations is relatively low. Therefore, production variability is not
significantly affected by the use of labor. This result is also found in Bokusheva and Hockmann's (2006) study of
Russian agriculture and Tveteras' (1999) study of Norwegian salmon.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
66 KHAN ET AL.

TABLE 4 Parameter estimates of the mean production and risk function

Mean production function, f(x, α) Coefficient SE


Constant 0.115*** 0.034
ln labor 0.147*** 0.053
ln fingerling 0.027 0.025
ln feed 0.551*** 0.033
ln capital cost 0.133*** 0.035
ln farm size 0.127* 0.067
ln labor × ln labor −0.152 0.270
ln fingerling × ln fingerling −0.182* 0.095
ln feed × ln feed 0.115 0.101
ln farm size × ln farm size −0.763 0.494
ln labor × ln fingerling 0.074 0.118
ln labor × ln feed −0.216** 0.089
ln fingerling × ln feed 0.091 0.073
ln capital × ln labor 0.094 0.076
ln capital × ln fingerling −0.135* 0.080
ln capital × ln feed −0.160* 0.109
ln farm size × ln labor 0.273 0.333
ln farm size × ln fingerling 0.114 0.171
ln farm size × ln feed 0.205* 0.107
ln capital × ln farm size 0.098 0.068
Trishal (1 if Trishal, 0 otherwise) −0.026 0.027
Muktagachha (1 if Muktagachha, 0 otherwise) 0.012 0.030
Risk function, h(x, β)
ln labor −0.628 0.614
ln fingerling 0.893** 0.409
ln feed −0.800* 0.435
ln capital cost/investments in physical assets −1.045** 0.487
ln farm size 2.132** 0.843
Trishal (1 if Trishal, 0 otherwise) −0.137 0.415
Muktagacha (1 if Muktagachha, 0 otherwise) 0.804** 0.389
Constant −5.305*** 0.399

Note: “***”significant at the 1% level, “**significant at the 5% level, and “*”significant at the 10% level.

Table 5 depicts the estimated parameters related to the inefficiency function. The estimated parameters have a
negative sign, as expected, because an increase in all of these variables should contribute to a reduction in ineffi-
ciency. Experience in aquaculture farming, increased education, training, access to credit, and extension services
should allow farmers to make better decisions and use inputs more efficiently, thereby reducing the inefficiencies in
farm practices. However, only one parameter (extension service) is found to be statistically significant at a 1% level
regarding inefficiency, whereas two other parameters (training and credit) are found to have a statistically significant
impact at a 10% level.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 67

TABLE 5 Parameter estimates of the inefficiency function

Variable
Inefficiency function, q(z, γ) Coefficient SE
Age (years) −0.011 0.027
Education (years of schooling) −0.042 0.055
Experience (years) −0.071 0.100
Training (days) −0.041* 0.022
Credit (1 if received, 0 otherwise) −1.154* 0.663
Extension service (1 if received, 0 otherwise) −1.220*** 0.476
Trishal −0.167 0.709
Muktagachha −0.645 0.943
Constant −1.406 1.097
Mean technical efficiency (%) 0.92 (0.057)a
Maximum (%) 0.995
Minimum (%) 0.609

Note: “***”Significant at the 1% level, “**”significant at the 5% level, and “*”significant at the 10% level.
a
Values in parentheses indicate SD.

The Department of Fisheries (DoF) of Bangladesh is primarily responsible for all extension-related services in
the aquaculture sector. Networking through the extension service is the most important factor for reducing ineffi-
ciency. Different farm management practices, knowledge of inputs, and information related to the latest technolo-
gies and markets help to improve the efficient use of inputs and modern technology adoption, which improve farm
efficiency (Mitra et al., 2020; Rahman, Nielsen, Khan, & Asmild, 2019). Farmers face different types of problems with
regard to water quality, soil quality, temperature, and disease during the culture period. To address these problems,
the majority of farmers have contact with extension workers, who are specially trained staff members of the Depart-
ment of Fisheries.
Training reduces inefficiency because in comparison to untrained farmers, trained farmers have more knowledge
of optimal conditions for fish and are more capable of handling and managing situations that could harm the produc-
tion yield. Access to credit is another important institutional factor that can reduce inefficiency among farmers.
Farmers who receive credit are more likely to buy the inputs needed to optimize production potential, including
high-quality feed and fingerlings and new equipment. All of these factors help reduce inefficiency. These findings are
confirmed by both Mitra et al. (2019) and Khan et al. (2018).
Overall, the inefficiency function analysis suggests that inefficiency is not a major problem in aquaculture farm-
ing given the existing technology. The average technical efficiency (TE) score of the aquaculture farmers was 0.92,
ranging from a minimum of 0.60 to a maximum of 1, with a standard deviation of 0.06. This result indicates that the
farmers in the study area could increase production yield by 8% without using more input if they were able to pro-
duce what the most efficient farmers produce.
The distribution of technical efficiency is displayed in Figure 2. Notably, 81% of the farms operate within a 90%
to 100% level of efficiency, and all farms operate above a level of 60%. These results are also supported by
Alam (2011).
An average TE score of 0.92 is relatively high compared to those in other studies of aquaculture production in
Asian countries (Iliyasu et al., 2014; Khan & Alam, 2003; Sharma & Leung, 2003). However, other studies conducted
in Bangladesh (Alam, 2011; Alam et al., 2012; Sarker et al. 2016b) also found high TE scores. The relatively high TE
score may be explained by the fact that the selected sample of farms is relatively homogeneous in terms of size,
technology used, training and extension services received, and access to credit.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
68 KHAN ET AL.

FIGURE 2 Distribution of technical efficiency for selected farms

F I G U R E 3 Relationship between farm size and productivity (left panel) and farm size and efficiency level (right
panel) for aquaculture farms

3.4 | The relationship of farm size, efficiency, and productivity

One of the most contentious debates in international agricultural development involves the relationship between
farm size and agricultural productivity. Sen (1964) initially suggested that productivity decreases as farm size
increases. This relationship is explained by the relative advantage of small farms that use more family labor, thereby
reducing the monitoring and supervision costs of hired labor.
This study examines whether the inverse relationship holds for aquaculture farms in Bangladesh. The relation-
ship is tested using a simple local polynomial regression line with a 95% confidence interval. The left panel of
Figure 3 shows a positive relationship between farm size and productivity, which implies that larger farms are more
productive than smaller farms. The result may be explained by the fact that in comparison to small farms, large farms
have economic, technical, management, and financial advantages in the study area, thereby enhancing productivity.
This result is supported by Roy, Sahoo, Saradhi, and Saha (2002), a study in which the same positive relationship was
found in Indian aquaculture.
The right panel of Figure 3 depicts the relationship between efficiency and farm size, showing that larger farms
are more efficient than smaller farms. This result can be explained by the fact that although all technical advantages
available to small farms are also available to large farms, the reverse is not true. Notably, large farms enjoy better
economies of scale, management, finance and marketing. These factors help reduce the cost of production and
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 69

TABLE 6 Uses of different inputs on the basis of efficiency level

Efficiency Labor (man- Fingerling Feed Production Average farm Feed conversion
level days/ha) (pieces/ha) (kg/ha) (kg/ ha) size (ha) rate (FCR)
0.951–1.00 304 58,712 36,171 24,075 0.733 1.50
0.901–0.950 289 57,973 25,490 17,485 0.470 1.48
0.851–0.900 254 63,341 21,230 13,064 0.364 1.62
0.801–0.850 311 63,264 23,581 14,659 0.360 1.61
0.701–0.800 227 59,203 19,762 10,715 0.211 1.84
≤ 0.700 262 61,273 13,676 7,998 0.223 1.71

increase the efficiency of the resources used, thereby increasing productivity for larger farms. The division of labor is
also observed in large farms but is often not possible in small farms.

3.5 | Input use and efficiency levels

This section analyzes the relationship between input level and efficiency, providing an indication of the optimal input
combination for aquaculture farmers. The amount of labor used, fingerling stocking rate, feed quantity, production,
average size, and feed conversion rate (FCR) are presented on the basis of efficiency level in Table 6. The results
show that the most efficient farm (efficiency score 0.951–1.00) uses 304 man-days of labor per hectare of farm area
and that the fingerling stocking density is approximately 58,712 fingerlings per hectare. In contrast, less efficient
farmers (efficiency score less than 0.70) use less labor than the most efficient farms.
The results also indicate that the most efficient farmers are relatively large-scale farmers who use higher feeding
intensity, resulting in a higher yield and better FCR than those of smaller, less efficient farmers. One reason for this
result might be the increasing use of industrial pelleted feeds, which are becoming increasingly important as an input
in modern commercial aquaculture farming in Bangladesh, providing essential nutrition for farmed fish species. Com-
mercial fish diets are manufactured as either floating (extruded) or sinking (pressure-pelleted) feeds. Although the
ingredients in both types of feed are similar, floating pellets are more expensive to produce but provide a better
FCR. Large-scale farmers who have sufficient capital or access to credit use floating feed; consequently, in compari-
son to small-scale farmers, they can increase production with lower FCR.
Table 7 depicts the deviation in observed inputs for different sizes of farms relative to the inputs for farms
between 0.951 and 1.0 (identified as efficient farms, thus having optimal level of inputs), cf. Table 6. Ratios above
1 in Table 7 indicate that the observed input usages are higher than the technically efficient level (overutilization),
whereas ratios below 1 indicate the underutilization of inputs. Considering all farms, the input usage in the aquacul-
ture farms in this study is 5% lower than the optimal amount of labor and 17% below the optimal use of feed; the
overall fingerling stocking rate is close to optimal. However, when examining the input levels for the different sizes
of farms, there is a significant difference between the strategies used by small-scale and large-scale farmers. The
input level of feed appears to be too low (approximately 16–26%) for farms with less than 0.8 ha. Feed is the most
important input, constituting more than 67% of costs, and small-scale farmers are faced with credit constraints.
Among the selected farmers, approximately 55% received credit; however, the farmers' access to credit is limited,
and collateral may have been a barrier in obtaining a loan. Although NGOs provide smaller loans to farmers, these
loans are often insufficient. Commercial banks also provide loans to farmers, but it is difficult for small-scale farmers
to access these loans. Thus, liquidity remains a problem for small-scale farmers, and some of the production variation
can be explained by the variation in feed supply among farmers.
The use of labor is also less than optimal. Small farms tend to use family labor, which can be difficult to account
for correctly. Another issue is that the stocking density of fingerlings appears to be too high, which might reflect a
lack of sufficient knowledge by small-scale farmers on optimal stocking densities.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
70 KHAN ET AL.

TABLE 7 Ratios of observed inputs according to farm size compared with the optimal level of inputs

Pond size (hectare)


Input ≤0.20 0.21–0.40 0.41–0.60 0.61–0.80 0.81–1.00 ≥1 All farms
Labor 0.92 0.87 0.95 0.92 1.11 1.18 0.95
Fingerling 1.04 1.09 1.06 1.05 0.80 0.74 1.01
Feed 0.77 0.84 0.74 0.80 0.96 1.06 0.83

Conversely, the largest farms have input patterns opposite to those of the smallest farms. The larger farms are
closer to using the optimal amount of feed, possibly because they have easier access to credit and loans and are able
to purchase the amount and type of feed required for optimal production. The amount of labor increases with farm
size, which might indicate that in comparison to small farms, large farms use more paid labor, which more easily
accounts for this increase in labor.

4 | CONCLUDING REMARKS

In aquaculture, differences in production outputs are observed from farm to farm and among different farm locations
based on differences in input use behaviors and socioeconomic conditions. This indicates that input-use risk exists and that
farmers do not use inputs in an optimal manner. Focusing on these issues, this article contributes to the literature by exam-
ining production risk, farm technical efficiency, and input-use variation in pond aquaculture farming. The results show that
labor, feed, and capital investments increase mean production. Furthermore, log-LR tests show that substantial production
risk exists in pond aquaculture farming. Increasing the use of feed and capital has a risk-reducing effect, whereas fingerlings
and farm size have a risk-increasing effect. The results further suggest that access to extension services, training, and credit
reduces inefficiency significantly. Whether small-scale farmers are more productive and efficient compared to larger
farmers, that is, whether the inverse relationship is applicable, is also tested. The results show the positive effects of econo-
mies of scale, and the inverse relationship is not confirmed within pond aquaculture in Bangladesh. It is also found that in
comparison to small-scale farmers, large-scale farmers (efficient farms) use a lower stocking density but have a higher feed-
ing intensity, which results in a higher yield. An average farm uses a lower amount of labor and feed than what is estimated
to be optimal, but the overall fingerling stocking rate is close to optimal.
Significant production risk exists in aquaculture farming. Therefore, action should be taken to reduce these
input-specific production risks. The overall findings indicate that there is a lack of scientific management practices
among fish farmers. Farmers could reduce risk and become more efficient by adjusting input allocation. In this case,
small-scale farmers can reduce the stocking density of fingerlings, and larger farmers can optimize input use by
reducing both labor and feed use. Farmers should communicate with local fishery extension officers at the upazila
level whenever they face a technical problem. At the overall government level, guidelines for better input use in
aquaculture through policy and service could be introduced. Since feed is the main input for commercial aquaculture,
constituting almost 67% of the total production cost, aquaculture feed policy could be introduced, placing more
emphasis on the quality and certification of feed. In addition, a policy could address the need to increase farm man-
agement skills by providing training and extension services. The Department of Fisheries (DoF) of Bangladesh has
sufficient organizational capacity for training and extension services; however, better communication with farmers
needs to be established to enhance the scientific knowledge provided to aquaculture farmers. Furthermore, research
on how a more uniform (certified) and better quality feed can be produced and controlled is essential for managers
to be able to reduce production risk and maximize farm profits.

ACKNOWLEDGMENTS
We are grateful to the aquaculture farmers of Bangladesh for sharing their valuable information. We would
like to thank the Danish International Development Agency (DANIDA) for its financial support for the project
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
KHAN ET AL. 71

“Upgrading pangas and tilapia value chains in Bangladesh,” Project Code 14-06KU, under which this research
was conducted.

ORCID
Md. Akhtaruzzaman Khan https://orcid.org/0000-0002-2968-0375
Rasmus Nielsen https://orcid.org/0000-0001-7357-6965

RE FE R ENC E S
Aigner, D. J., Lovell, C. K., & Schmidt, P. (1977). Formation and estimation of stochastic frontier production function models.
Journal of Econometrics, 6(1), 21–37.
Aktar, S. S., Khan, M. A., Prodhan, M. M., & Mukta, M. A. (2018). Farm size, productivity and efficiency nexus: The case of
pangas fish farming in Bangladesh. Journal of the Bangladesh Agricultural University, 16(3), 513–522.
Alam, F. (2011). Measuring technical, allocative and cost efficiency of pangas (Pangasius hypophthalmus: Sauvage 1878) fish
farmers of Bangladesh. Aquaculture Research, 42(10), 1487–1500.
Alam, M. A., & Guttormsen, A. G. (2019). Risk in aquaculture: Farmers' perceptions and management strategies in
Bangladesh. Aquaculture Economics & Management, 23(4), 359–381.
Alam, M. A., Guttormsen, A. G., & Roll, K. H. (2019). Production risk and technical efficiency of tilapia aquaculture in
Bangladesh. Marine Resource Economics, 34(2), 123–141. https://doi.org/10.1086/704129
Alam, M. F., Khan, M. A., & Huq, A. A. (2012). Technical efficiency in tilapia farming of Bangladesh: A stochastic frontier pro-
duction approach. Aquaculture International, 20(4), 619–634.
Anwar, J. (2011). Market study on some freshwater farmed fish: Tilapia an Pangas (Mekong River catfish). Final Report.
USAID 141.
Asche, F., & Bjørndal, T. (2011). The economics of Salmon aquaculture. Wiley-Blackwell: Chichester.
Asche, F., & Tveteras, R. (1999). Modeling production risk with a two-step procedure. Journal of Agricultural and Resource
Economics, 24(2), 424–439.
Bangladesh Fisheries Research Institute. (2012). Thai Pangas culture in pond fish culture manual. Mymensingh, Bangladesh:
BFRI.
Belton, B., Bush, S. R., & Little, D. C. (2018). Not just for the wealthy: Rethinking farmed fish consumption in the global
south. Global Food Security, 16, 85–92.
Bergfjord, O. J. (2009). Risk perception and risk management in Norwegian aquaculture. Journal of Risk Research, 12(1),
91–104.
Bokusheva, R., & Hockmann, H. (2006). Production risk and technical inefficiency in Russian agriculture. European Review of
Agricultural Economics, 33(1), 93–118.
Department of Fisheries. (2017–18). Fishery Statistical Yearbook of Bangladesh. Dhaka, Bangladesh: Department of Fisheries
(DoF), Ministry of Fisheries and Livestock.
Dey, M. M., Bose, M. L., & Alam, M. F. (2008). Recommendation domains for pond aquaculture. Country case study: Develop-
ment and status of freshwater aquaculture in Bangladesh. WorldFish Center Studies and Review No. 1872. Penang,
Malaysia: The WorldFish Center.
Dey, M. M., Kumar, P., Chen, O. L., Khan, M. A., Barik, N. K., Li, L., … Pham, N. S. (2013). Potential impact of genetically
improved carp strains in Asia. Food Policy, 43, 306–320.
Dey, M. M., Kumar, P., Paraguas, F. J., Chen, O. L., Khan, M. A., & Srichantuk, N. (2010). Performance and nature of geneti-
cally improved carp strains in Asian countries. Aquaculture Economics and Management, 14(1), 3–19.
Food and Agriculture Organization. (2020). The state of world fisheries and aquaculture—Sustainability in action. Food and agri-
culture Organization of the United Nations. Rome, Italy: FAO.
Gaidar, Y. T. (Ed.). (2002). Russian economy in 2001: Trends and outlooks (Vol. 23). Institute for the Economy in Transition:
Moscow.
Iliyasu, A., Mohamed, Z. A., Ismail, M. M., Abdullah, A. M., Kamarudin, S. M., & Mazuki, H. (2014). A review of production
frontier research in aquaculture (2001–2011). Aquaculture Economics & Management, 18(3), 221–247.
Just, R. E., & Pope, R. (1978). Stochastic representation of production functions and econometric implications. Journal of
Econometrics, 7(1), 67–86.
Khan, M. A. (2012). Efficiency, risk and Management of Fisheries Sector in Bangladesh. (PhD thesis). Thesis number: 2012:62.
UMB School of Economics and Business, Norwegian University of Life Sciences, Ås. Norway.
Khan, M. A., & Alam, M. F. (2003). Technical efficiency of the hatchery operators in fish seed production farms in two
selected areas of Bangladesh. Bangladesh Journal of Agricultural Economics, 26(1 & 2), 55–70.
Khan, M. A., Guttormsen, A., & Roll, K. H. (2018). Production risk of pangas (Pangasius hypophthalmus) fish farming. Aquacul-
ture Economics & Management, 22(2), 192–208.
17497345, 2021, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jwas.12767 by Bangladesh Hinari NPL, Wiley Online Library on [27/05/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
72 KHAN ET AL.

Khan, M. A., Roll, K. H., & Guttormsen, A. (2021). Profit efficiency of Pangas (Pangasius hypophthalmus) fish farming in
Bangladesh – The effect of farm size. Aquaculture (accepted).
Kodde, D. A., & Palm, F. C. (1986). Wald criteria for jointly testing equality and inequality restrictions. Econometrica: Journal
of the Econometric Society, 54, 1243–1248.
Kumbhakar, S. C. (2002). Specification and estimation of production risk, risk preferences and technical efficiency. American
Journal of Agricultural Economics, 8(1), 8–22.
Kumbhakar, S. C., & Tveteras, R. (2003). Risk preferences, production risk and firm heterogeneity. Scandinavian Journal of
Economics, 105(2), 275–293.
Liefert, W., & Swinnen, J. F. M. (2002). Changes in agricultural markets in transition economies. Agricultural economic report
806. Washington, DC: Economic Research Service (ERS).
Losinger, W. C. (2006). Factors influencing the variance in expected yield on catfish farms in the United States. Aquaculture
International, 14(5), 415–419.
Meeusen, W., & van den Broeck, J. (1977). Efficiency estimation from Cobb-Douglas production function with composed
error. International Economic Review, 18, 435–444.
Mitra, S., Khan, M. A., & Nielsen, R. (2019). Credit constraints and aquaculture productivity. Aquaculture Economics & Man-
agement, 23(4), 410–427.
Mitra, S., Khan, M. A., Nielsen, R., & Islam, N. (2020). Total factor productivity and technical efficiency differences of aqua-
culture farmers in Bangladesh: Do environmental characteristics matter? Journal of the World Aquaculture Society, 51(4),
918–930. https://doi.org/10.1111/jwas.12666
Mukta, M. A., Khan, M. A., Mian, M. R. U., & Juice, R. A. (2019). Is tilapia farming financially profitable and efficient? Policy
options for sustainable farming. Journal of Bangladesh Agricultural University, 17(1), 92–98.
Prodhan, M. M. H., & Khan, M. A. (2018). Management practice adoption and productivity of commercial aquaculture farms
in selected areas of Bangladesh. Journal of the Bangladesh Agricultural University, 16(1), 111–116.
Rahman, M. T., Nielsen, R., Ahsan, D., & Khan, M. A. (2021). Perceived risk and risk management strategies in pond aquacul-
ture. Marine Resource Economics, 36(1), 43–69.
Rahman, M. T., Nielsen, R., & Khan, M. A. (2019). Agglomeration externalities and technical efficiency: An empirical application to
the pond aquaculture of Pangas and tilapia in Bangladesh. Aquaculture Economics & Management, 23(2), 158–187.
Rahman, M. T., Nielsen, R., Khan, M. A., & Asmild, M. (2019). Efficiency and production environmental heterogeneity in
aquaculture: A meta-frontier DEA approach. Aquaculture, 509, 109–139.
Rahman, M. T., Nielsen, R., Khan, M. A., & Yeboah, I. A. (2020). Impact of management practices and managerial ability on
the financial performance of aquaculture farms in Bangladesh. Aquaculture Economics & Management, 24(1), 79–101.
Roy, A. K., Sahoo, K. N., Saradhi, K. P., & Saha, G. S. (2002). Farm size and aquaculture productivity relationship. Asian Fisher-
ies Science, 15, 129–134.
Sarker, M. A., Arshad, F. M., Alam, M. F., Mohamed, Z. A., & Khan, M. A. (2016a). Feed technology and production perfor-
mance of Thai koi (Anabas testudineus) in Bangladesh. Journal of Applied Aquaculture, 28(2), 76–91.
Sarker, M. A., Arshad, F. M., Alam, M. F., Mohamed, Z. A., & Khan, M. A. (2016b). Stochastic modelling of production risk
and technical efficiency of Thai koi (Anabas testudineus) farming in northern Bangladesh. Aquaculture Economics & Man-
agement, 20(2), 165–184.
Schmidt. (1977). On the statistical estimation of parametric frontier production function. Review of Economics and Statistics,
58, 238–239.
Sen, A. K. (1964). Size of holdings and productivity. The Economic Weekly Annual No. February, 323–326.
Sharma, R. K., & Leung, P. (2003). A review of production frontier analysis for aquaculture management. Aquaculture Eco-
nomics & Management, 7(10), 15–37.
Tveteras, R. (1999). Production risk and productivity growth: Some findings for Norwegian Salmon aquaculture. Journal of
Productivity Analysis, 12, 161–179.

How to cite this article: Khan MA, Begum R, Nielsen R, Hoff A. Production risk, technical efficiency, and
input use nexus: Lessons from Bangladesh aquaculture. J World Aquac Soc. 2021;52:57–72. https://doi.org/
10.1111/jwas.12767

You might also like