Content uploaded by Mohammad Shamsul Hoq
Author content
All content in this area was uploaded by Mohammad Shamsul Hoq on Sep 27, 2021
Content may be subject to copyright.
Environmental Challenges 5 (2021) 100292
Contents lists available at ScienceDirect
Environmental Challenges
journal homepage: www.elsevier.com/locate/envc
Welfare impact of market participation: The case of rice farmers from
wetland ecosystem in Bangladesh
Mohammad Shamsul Hoq
a
,
1
,
∗
, Md. Taj Uddin
c
, Shankar Kumar Raha
b
,
Mohammad Ismail Hossain
b
a
Scientific Officer, Agricultural Economics Division, Bangladesh Agricultural Research Institute, Joydebpur 1701, Gazipur, Bangladesh
b
Department of Agribusiness and Marketing, Bangladesh Agricultural University, 2202, Mymensingh, Bangladesh
c
Department of Agricultural Economics, Bangladesh Agricultural University, 2202, Mymensingh, Bangladesh
Keywords:
Rice farmers
Market participation
Household welfare
Wetland ecosystem
Heckman model with exclusion restriction
Bangladesh
Haor ecosystem of Bangladesh plays dual roles: local economic development through agricultural income-earning
activities and preserve the environment. Smallholder haor farmers face a variety of challenges that conned their
access to the market. This study was undertaken to examine the determinants of smallholder’s market access that
plays an important role in improving their welfare. Primary data were collected from 300 haor households from
three districts, taking 100 from each district by applying a multistage random sampling technique. Heckman’s
two-stage model with exclusion restriction was used for reaching the objectives. The rst-stage probit model indi-
cated that members of an organization, extension contact, access to market information, access to credit, home to
market distance, home to haor distance, road connectivity, irrigation facilities, duration of the waterlogged, and
output price signicantly inuenced the market participation of the farmers. The OLS regression results revealed
that family size, dependency ratio, farm size, access to credit, and o-farm income signicantly inuenced the
household’s per capita consumption expenditure as a result of market participation. The policy should be triggered
on investing more in haor areas for improving rural physical infrastructure, which ensured year-round road con-
nectivity and established links with markets. Furthermore, an increase in extension contact, market information
systems, agricultural-related training, and promotion of farmer organization should be emphasized.
1. Introduction
Agricultural commercialization and farmer’s welfare mostly depend
on the smallholders’ market participation which is both a cause and ef-
fect of economic growth and development of a country ( Barrett et al.,
2012 ; Pingali, 2007 ; Reardon and Barrett, 2000 ; Timmer, 1974 ;
Von Braun, 1995 ). The dependency on agriculture in developing coun-
tries like Sub-Saharan Africa, South Asia, and the Pacic is more than
60%, while in Latin America and other high-income economies these
proportions are estimated at 18% and 4%, respectively ( Awotide et al.,
2013 ; World Bank, 2006 ). Thus, any policy change in agriculture in
both developed and developing countries directly aects the farmer’s
and farming community’s wellbeing. Agriculture is the most impor-
tant sector in Bangladesh, which contributes 13.61% of the national
gross domestic product (GDP) and the rice sub-sector alone contributes
about 70% of the total agricultural GDP ( Bangladesh Economic Re-
view BER, 2019 ; Rahman et al., 2020 ). This share of agriculture is bit
∗ Corresponding author at: Scientic Ocer, Agricultural Economics Division, Bangladesh Agricultural Research Institute, Joydebpur 1701, Gazipur, Bangladesh.
E-mail address: shamsul305@bari.gov.bd (M.S. Hoq).
1 Present address: PhD student, Department of Agribusiness and Marketing, Bangladesh Agricultural University, Mymensingh-2202.
by bit decreasing to GDP and transformed into the service and indus-
try sectors over the last decade which enables constant economic de-
velopment of the country ( Bangladesh Economic Review BER, 2019 ).
However, about 48% of the total labor force is involved in this sector
( Bangladesh Bureau of Statistics BBS, 2019 ). On the other hand, agri-
culture is the most vulnerable sector in Bangladesh due to the unfa-
vorable climate condition ( Rahman et al., 2020 ). Most of the develop-
ing countries of the world, in which particularly the farmers of Sub-
Sahara Africa and South Asia are smallholders in terms of farm size
( Kyaw et al., 2018 ; Abdullah et al., 2019 ; Mmbando et al., 2015a , 2015b ;
Ademe et al., 2017 ; Barrett, 2008 ). Bangladesh’s farm sector is also dom-
inated by smallholders of which about 60% of the farmers are function-
ally landless, and 20% of the farmers are marginal whose farm size is
less than a hectare of land ( FAO, 2008 ; Sala and Bocchi, 2014 ). Accord-
ing to Ahmed, 2017 , roughly 76% of the country’s farmers are small
and marginal, with farm sizes of up to 1.49 acres of land and depen-
dence on rice. The country’s small and marginal farmers have the least
https://doi.org/10.1016/j.envc.2021.100292
Received 24 June 2021; Received in revised form 25 August 2021; Accepted 18 September 2021
2667-0100/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
( http://creativecommons.org/licenses/by-nc-nd/4.0/ )
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
access to the market ( Sala and Bocchi, 2014 ). Many of them continue to
earn negative net income from rice farming and are unable to stop due
to their forefather’s profession, family consumption needs, and a lack
of viable alternatives ( Alamgir et al., 2020 ; Hoq et al., 2021 ). Farm-
ers have been also increasingly threatened by changing climatic events
on agricultural production and livelihoods in recent years ( Kabir et al.,
2017 ; Singha et al., 2019 ; Abdullah et al., 2021 ). However, the impact
of climate change is widely felt to be a key factor that increases the
vulnerability of agricultural production; because agriculture is highly
sensitive to climate change ( Hossain et al., 2013 ; Ruane et al., 2013 ;
IFPRI, 2013 ). According to Smith et al. (2014) , people from all over
the world are already experiencing food shortages as a result of climate
change, and global crop production is expected to decrease by 2–6%
per decade. Bangladesh, as a developing country, is not immune to the
aforementioned scenario, and the country’s northeastern regions are the
most vulnerable to severe weather extremes. The recurring and rapid oc-
currence of extreme nature-triggered disasters such as ash oods, hail-
storms, thunderstorms, droughts, embankment collapses, and changes
in upstream river discharge poses a serious threat to Bangladesh’s nat-
ural resources base life and livelihood ( Brammer, 1990 ; Jakariya and
Islam, 2017 ; Ali, 1999 ; UNDP, 2012 ). Crop production practices, eco-
nomic activities, and livelihood of the haor household are all inuenced
by ecological, geographical, and environmental factors that are distinct
from those found in other parts of the country ( Alam et al., 2010 ).
However, due to uneven rainfall in the pre-monsoon and onrush of wa-
ter from the upstream of the Asam and Meghalaya hills in India, Boro
rice cultivation in the haor areas is frequently hampered by early ash
oods ( Hossain, 2013 ; Kafy et al., 2021 ).The haor landscape is a very
low-lying river basin area of Bangladesh’s northeastern region, cover-
ing about 2.0 million hectares and accommodating about 19.4 million
people ( Sharma, 2010 ; Uddin et al., 2019 ). There are about 373 haors
located in the seven districts i.e. Sunamganj, Sylhet, Kishoreganj, Habi-
ganj, Netrakona, Maulvibazar, and Brahmanbaria in Bangladesh which
cover an area of 1.26 million hectares of which 0.68 million is under
haor ( Alam et al., 2010 ; Huda, 2004 ). Although boro rice production
in the haor region is vulnerable to a variety of nature-induced disasters
(ash oods, droughts, embankment breaches, and so on), it contributes
roughly 20% of the country’s total rice production ( Rabby et al., 2011 ).
However, due to limited market access in the haor areas, haor house-
holds do not receive a fair price for their produced rice. Because small-
holder rice farmers’ market participation is a challenge in the haor areas
due to lack of physical access, lack of roads and transportation, poor in-
frastructure, and inundation with oodwater ( Ali and Rahman, 2017 ).
As a result, the majority of farmers sell their rice at the farm gate or
at their homes. Some farmers are able to participate in the market due
to the connection with the submergible lane during the dry season and
river way connectivity during the monsoon. Thus, using the Heckman
two-stage model, this research looks at the factors that inuence mar-
ket participation and how they aect household welfare based on per
capita consumption expenditure. Several studies used this method for
estimating household welfare owing to the eect of market participa-
tion ( Abdullah et al., 2019 ; Lubungu, 2013 ; Negasa et al., 2020 ). Sub-
sequently, the background of the study is reviewed in two aspects: an
overview of rice farming, and smallholder rice farmers’ market partici-
pation and their welfare in the haor areas of Bangladesh.
2. An overview of rice farming in haor areas of Bangladesh
Bangladesh is the fourth largest rice-producing country in the world
( Singha et al., 2019 ; IRRI, 2008 ; Chowdhury and Hassan, 2013 ). About
70% of the total cropland area of the country is utilized for rice cul-
tivation. Rice is the principal food of the people of Bangladesh and
plays a vital role in their livelihood ( Chowdhury and Hassan, 2013 ;
Siddique et al., 2017 ). Rice is grown in three dierent growing sea-
sons in a year in various agro-ecological zones with dierent tempera-
tures and precipitation in Bangladesh ( Chowdhury and Hassan, 2013 ;
Rahman et al., 2017 ).These three rice growing seasons are namely
Kharif-I (Mid-March to Mid-July), Kharif-II (Mid-July to Mid-October),
and Rabi (Mid-October to Mid-March) ( Hoq et al., 2021 ). The transi-
tion period of Kharif-I (early monsoon) is between dry and wet seasons,
Kharif-II is the wet season and Rabi is the dry season ( Chowdhury and
Hassan, 2013 ; Alamgir et al., 2020 ). Farmers cultivated in three distinct
rice growing seasons namely, Aus, Aman , and Boro which represents
Kharif-I, Kharif-II , and Kharif-III respectively ( Talukder and Chile, 2014 ).
A major portion of this cultivable land in the wetland haor areas is low-
lying and ooded during monsoon every year. Rabi is the only cropping
season and lands during Kharif-I and Kharif-II seasons stay fallow due
to the inundation of oodwater. Boro rice is the only crop produced in
the Rabi season and very frequently damaged by ash oods at the time
of pre-monsoon. The pre-monsoon season is characterized by increasing
rainfall and ash oods ( Ali and Rahman, 2017 ). Even though cultivat-
ing only boro rice as a single crop and the reappearance of advanced
ash oods, the haor region alone produces about 20% of the country’s
total staple food (rice) and secures livelihood for twenty million haor
inhabitants ( Rabby et al., 2011 ). Huda (2004) mentioned that almost
80% of these haor areas are covered by boro rice, while only about 10%
area is under T. Aman rice. Moreover, hybrid rice is also grown in the
haor areas ( Das, 2004 ). Most of the peoples in the haor areas are af-
fected by the ash ood where rural households mostly depend on boro
rice-based agriculture and aquaculture for their livelihoods and many of
them are vulnerable to food insecurity ( Sayeed and Yunus, 2018 ). Rice
production has massively increased in the last decade due to techno-
logical advancements such as the introduction of new varieties and im-
proved management practices ( Rahman et al., 2017 ). However, the suc-
cess of rice production is greatly inuenced by market prices, which are
determined by the market access of rice farmers. The haor household’s
market participation is still very low due to the unfavorable ecosystem
in the haor areas. Therefore, the purpose of the study is to investigate the
factors that inuence the farmers’ decision of the rice market participa-
tion as well as the consequences of market participation on household’s
welfare.
3. Extent of smallholder rice farmer’s market participation and
their welfare
Much of the studies in the developing country of the world high-
lighted that the smallholders’ market participation in domestic and re-
gional markets remains low due to a wide range of barriers ( Kyaw et al.,
2018 ; Mmbando et al., 2015 ; Ouma et al., 2010 ) Research indicated
that a substantial fraction of the world’s households, including those
in Myanmar, Africa, Nigeria, Kenya, Tanzania, Zambia, Uganda, and
Ethiopia, experience various forms of barriers to market participation
( Awotide et al., 2013 ; Kyaw et al., 2018 ; Mmbando et al., 2015 b,
2015 a; Barrett, 2008 ; Lubungu, 2013 ; Kamara, 2004 ; Omiti et al., 2009 ;
Sebatta et al., 2014 ; Hagos et al., 2020 ). Remarkable constraints faced
by smallholder farmers in Myanmar ( Kyaw et al., 2018 ) and Tanzania
( Mmbando et al., 2015 ; Mmbando et al., 2015 ) are lack of physical ac-
cess to market, and lack of market information. Barrett (2008) identi-
ed distinguish location level constraint to participate in the market of
Eastern and Southern Africa. Poor infrastructure and weak institutions
raise transaction costs that greatly alter production and market partic-
ipation decisions in Mayanmar ( Kyaw et al., 2018 ) and Central Africa
( Ouma et al., 2010 ). Hence, Market and improved market access are crit-
ical for improving income and livelihood of rural household ( Kyaw et al.,
2018 ; Mmbando et al., 2015 ; Ouma et al., 2010 ).Empirical research
from outside Bangladesh indicates that the bulk of farmers live in re-
mote areas with insucient transportation and poor market infrastruc-
ture, making market access dicult and resulting in high transaction
2
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
costs ( Kyaw et al., 2018 ; Ouma et al., 2010 ; Makhura, 2001 ). This issue
has particular importance to the haor communities in Bangladesh where
the availability of transport and physical access is the most vital con-
cern due to the waterlogged infrastructure in haor areas of Bangladesh.
All kinds of linking roads and structures were submerged in 6 to 7
months in a year during monsoon and pre-monsoon ( BHWDB, 2012 ).
Boats are the only medium of communication in the river-way ( Ali and
Rahman, 2017 ; BHWDB, 2012 ). Infrastructures and roads are lifted from
the water surface during the Boro season, but road connectivity with
the rest of the country is severely limited due to collapsed infrastruc-
ture and a shortage of bridges and culverts ( BHWDB, 2012 ). Conse-
quently, the majority of farmers in the haor areas sell their crops at
low prices at farm gates or village markets. Their decisions to sell are
mainly inuenced by market information, output prices, and distance
to the market ( Omiti et al., 2009 ). Aside from that, the rice price fall
is another barrier for the market participation due to seasonal oversup-
ply, miller and trader dominance, and a lack of structured market in
the haor areas of Bangladesh ( Rahman et al., 2020 ; Hossain and Ver-
beke, 2010 ; Ahmed, 2020 ).Very limited farmers were able to access the
formal market (well-connected markets where large-scale agricultural
commodities buyers can participate) without having to negotiate with a
trader’s collusion, and this was dependent on the proximity of the sub-
mergible lane, which linked to the formal market. However, the welfare
of rice farmers will be achieved through increased revenue by ensur-
ing rice farmer’s market participation. Therefore, the relationship be-
tween market participation and household welfare is lacking in the haor
areas of Bangladesh. Thus, this study is undertaken to ll this gap by
analyzing the socio-economic and climatic factors of market participa-
tion and its subsequent eects on rice farmers’ welfare in haor areas of
Bangladesh. Several previous studies have used dierent indicators as a
proxy to evaluate welfare such as per capita income ( Lubungu, 2013 ;
Brown et al., 2006 ; Maertens and Swinnen, 2009 ; Maertens et al.,
2011 ), household consumption expenditure ( Abdullah et al., 2019 ;
Mmbando et al., 2015 ; Alemu, 2012 ), household poverty reduction
( Kamara, 2004 ; Maertens and Swinnen, 2009 ), and household food
security ( Smale et al., 2012 ). Unlike previous studies, this study fo-
cuses on per capita consumption expenditure ( Abdullah et al., 2019 ;
Mmbando et al., 2015 ) as a measure of household welfare, which is
a more reliable welfare indicator and less likely to measurement error
than household income. Per capita consumption expenditure was used
in this analysis as an indicator for estimating household welfare out-
comes because per capita expenditure represents the expected outcome
of a participation choice.
4. Materials and methods
4.1. Description of the study area
The haor areas are located in the North-eastern part of the
Bangladesh which is geographically excluded and ecologically vulnera-
ble ( 7FYP, 2015 ). The haor is an important wetland ecosystem, spread
over seven districts of Bangladesh like; Sunamganj, Habiganj, Moul-
vibazar, Sylhet, Kishoreganj, Netrokona and Brahmanbaria outside of
the core haor area. It’s a mosaic of wetlands, with rivers, streams, and
irrigation canals, as well as large swaths of seasonally ooded cultivated
plains and thousands of haors and beels. Three districts and two up-
azila from each district were purposively selected for the present study.
These three districts were Kishoreganj, Sunamganj and Netrokona that
were selected on the basis of core haor areas. The selected upazila were
Itna and Mithamoin from Kishoreganj district, Khaliajuri and Mohon-
ganj from Netrokona district and Derai and Salla from Sunamganj dis-
trict which were also selected on the basis of the intensive haor areas
( Fig. 1 ).
4.2. Sampling technique and sample size determination
A multistage random sampling approach was employed to select the
respondent households. The intended respondents were the vulnerable
haor household herein farmers who produce rice. In the rst stage, haor
districts were selected purposively on the basis of the largest water basin
in the haor areas of Bangladesh. In the second stage, haor uapzailas
(lower administrative units) were selected. Finally, unions (local level
administrative units) were selected and the farmers were selected ran-
domly. The following Eq. (1) was used to get the sample size for a given
population, as recommended by Kothari (2004) ;
𝑛 =
𝑍
2
𝑝.𝑞.𝑁
𝑒
2
(
𝑁 − 1
)
+ 𝑍
2
.𝑝.𝑞
(1)
Where n is the total sample size; e is the acceptable error (0.05); N is
the total population (161,832); p is the sample proportion of successes
( P = 0.5); q = 1 –p = (1 –0.5) = 0.5; and Z is the standard value for
a given 90% condence level ( Z = 1.645). The acceptable level of the
condence interval for Z score is 90, 95 and 99%.
𝑛 =
(
1 . 645
)
2
𝑋
(
0 . 5
)
𝑋
(
0 . 5
)
𝑋
(
161 , 832
)
(
0 . 05
)
2
𝑋
(
, 161 , 832 , −1
)
+
(
1 . 645
)
2
𝑋
(
0 . 5
)
𝑋
(
0 . 5
)
=
10 , 948 , 0 . 3 , 6
4 , 05 ., 25
= 270
Hence, 270 farmers were round Fig. of 300 farmers considered as a
sample for equal distribution into the six selected Upazilas. Based on
the household size of each district, 50 potential farmers were picked
at random from union level (local level administrative unit) from each
Upazila. Table 1 shows the sample distribution of the six Upazilas.
4.3. Sources of data and information
We used cross-sectional data collected from the selected study loca-
tions and performed an empirical analysis to achieve the objectives of
this study. Data were collected from September 2019 to March 2020
from the 300 haor households. This study used mainly primary data and
secondary information was only used for the comparison. Primary data
were collected from sample farmers through a pre-tested questionnaire
survey. Secondary information relating to dierent statistics of relevant
issues was gathered from various published sources (e.g. FAOStat, jour-
nal papers, research report, thesis, newspapers, etc.) through an in-depth
literature review. The statistical software Stata 14.0 (Stata Corp, College
Station, Texas 77,845 USA) was used to perform data analysis.
4.5. Analytical technique
4.5.1. Heckman’s two-step selection model
Heckman’s two-step model is used to identify the determinants of
farmer market participation and how it aects rural farmers’ wellbe-
ing. A farmer is said to be a participant in the market if he sells a
part of his output in the market. The welfare impact of the market par-
ticipants is exposed to the subsequent equation ( Awotide et al., 2013 ;
Abdullah et al., 2019 )
𝐶𝑖 = 𝑋𝑖𝜆 + 𝛾𝐷𝑖 + 𝜀𝑖 (2)
Whereas,
Ci = Per capita consumption expenditure
𝜀 i = Random normal distribution term
Di = Dummy variable representing market participation in the out-
put market (1 = market participation in market, 0 = No participation in
market). If its value is 1 then it means farmers sell any quantity in the
output market and 0 means otherwise
Xi = Vector of household and farm characteristics
The market participation decision of the farmers solely depends on
himself/herself and by deciding to participate in the market, he/she will
be self-selected to participate or not instead of random assignment. This
3
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Fig. 1. a shows the location of the study area(three districts in the haor area) in Bangladesh, and (b-d) show the survey’s two upazilas from each district.
Table 1
Sample size determination by the study area.
District name Selected upazila Total farm household (N) Selected sample household (No.)
Sunamganj Derai 39,640 50
Salla 20,229 50
Netrokona Khaliajuri 20,193 50
Mohonganj 30,860 50
Kishoreganj Itna 33,310 50
Mithamoin 17,600 50
Total 161,832 300
study assumed that farmers are risk-neutral. The market participation
can be estimated with the following index function:
𝐿𝑖 ∗= 𝑋𝑖𝛼 + 𝑉 𝑖 (3)
Li
∗
= is a later variable denoting the dierence between utility from
market participation, U
IA,
and utility from not participating in the mar-
ket, U
IN
. The farmer’s decision to participate in the market was required
on the following condition being met:
𝐿𝑖 ∗= 𝑈
𝐼𝐴
− − 𝑈
𝐼𝑁
> 0
X
i 𝛼shows the explanatory variables which aect market partici-
pation. Where V
i
is an error term. Market participation and farmer’s
welfare are expected to be interdependent, which we estimate through
Eqs. (2) and (3) . More importantly, the problem of selection bias may
occur if latent variables aect the error term of both welfare equation
( 𝜀
i
) and market participation (V
i
). As a consequence, selection bias re-
sults in a correlation between error terms of Eqs. (2) and (3) . It means
that the welfare impact of market participants may be biased due to un-
observed factors. Therefore, the estimation of Eq.(2) with the ordinary
least square (OLS) will cause biased estimates. In this study, Heckman’s
model is used to overcome this problem. Moreover, this approach si-
multaneously corrects the problem of selection bias. It is established
in the literature that Heckman two-step approach can only be used
when the correlation between the two error terms is greater than zero
( Heckman, 1979 ), and then it corrects the problem of sample selection
bias ( Homann and Kassouf, 2005 ; Siziba et al., 2011 ). This approach is
based on the restrictive assumption of normally distributive error terms
( Wooldridge, 2010 ). The rst stage (probit model) is used to nd out
what factors aect market participation (Eq.3), and the second stage
(OLS) is used to investigate the welfare eects due to market partici-
pation (Eq.2). The rst stage probit model also generates the value of
inverse mills ratio (IMR). The IMR is denoted by a symbol of 𝜆i
which
is the ratio of the ordinate of a standard normal distribution to the tail
area of the distribution ( Greene, 2003 ).
𝜆𝑖 = 𝜑 ( 𝑝 + 𝛼𝑋𝑖 ) ∕ 𝜙( 𝑝 + 𝛼𝑋𝑖 )
Where 𝜑 = standard normal density function
𝜙= standard normal distribution function
Greene (2003) explained that the IMR term corrects the problem of
selection bias. If the term 𝜆i
is statistically insignicant, then selection
4
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
bias is not a problem ( Heckman, 1979 ). Conversely, if the value of 𝜆i
is statistically signicant, then there is a signicant dierence between
the farmers who participated in the market and those that did not. The
above disparity must be taken into account when calculating the welfare
equation. The two-step Heckman’s model is as follows: in the rst step,
selection equations show whether farmers participate in the market or
not and are expressed as:
𝑃 𝐼𝑀 = 𝛼0
+ 𝛼1
𝑎𝑔𝑒 + 𝛼2
𝑒𝑑𝑢 + 𝛼3
𝑓 𝑎𝑚𝑖𝑙𝑦.𝑠𝑖𝑧𝑒 + 𝛼4
𝑑.𝑟𝑎𝑡𝑖𝑜 + 𝛼5
𝑓 𝑎𝑟𝑚.𝑠𝑖𝑧𝑒
+ 𝑎
6
𝑚𝑒𝑚𝑏𝑒𝑟.𝑜𝑟𝑔 + 𝑎
7
𝑒𝑥𝑡.𝑐 𝑜𝑛𝑡𝑎𝑐 𝑡 + 𝛼8
𝑖𝑛𝑓 𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛 + 𝛼9
𝑐𝑟𝑒𝑑𝑖𝑡
+ 𝛼10
𝑑 𝑖𝑠𝐻 𝑅 + 𝛼11
𝑑 𝑖𝑠𝐻 𝑀 + 𝛼12
𝑑 𝑖𝑠𝐻 𝐻 + 𝛼13
𝑟𝑜𝑎𝑑 𝑐 𝑜𝑛𝑛𝑒𝑐 𝑡.
+ 𝛼14
𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟 𝑡
+
𝛼15
𝑖𝑟𝑟𝑖𝑔 𝑎𝑡𝑖𝑜 𝑛
+
𝛼16
𝑤𝑎𝑡𝑒𝑟𝑙𝑜𝑔 𝑔
+ 𝛼∖17
𝑡𝑟𝑎𝑖𝑛𝑖𝑛𝑔 + 𝛼18
𝑟𝑖𝑐 𝑒.𝑖𝑛𝑐 𝑜𝑚𝑒 + 𝛼19
𝑜𝑓 𝑓 − 𝑓𝑎𝑟𝑚
+ 𝛼20
𝑜𝑢𝑡𝑝𝑢𝑡.𝑝𝑟𝑖𝑐 𝑒 + 𝛼21
𝑙𝑜𝑐 𝑎𝑡𝑖𝑜𝑛 + 𝑣
𝑖
(4)
The second step is the outcome equation which examines the eect
of market participation on the welfare of farmers and the equation is
estimated by using OLS as follows:
𝐶 = 𝛽0
+ 𝛽1
𝑎𝑔𝑒 + 𝛽2
𝑒𝑑𝑢 + 𝛽3
𝑓 𝑎𝑚𝑖𝑙𝑦.𝑠𝑖𝑧𝑒 + 𝛽4
𝑑.𝑟𝑎𝑡𝑖𝑜 + 𝛽5
𝑓 𝑎𝑟𝑚.𝑠𝑖𝑧𝑒
+ 𝛽6
𝑐𝑟𝑒𝑑𝑖𝑡 + 𝛽7
𝑑𝑖𝑠𝐻𝑅 + 𝛽8
𝑑𝑖𝑠𝐻𝑀 + 𝛽9
𝑑𝑖𝑠𝐻𝐻 + 𝛽10
𝑟𝑜𝑎𝑑𝑐 𝑜𝑛𝑛𝑒𝑐 𝑡.
+ 𝛽11
𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟 𝑡
+
𝛽12
𝑖𝑟𝑟𝑖𝑔 𝑎𝑡𝑖𝑜 𝑛
+
𝛽13
𝑤𝑎𝑡𝑒𝑟𝑙𝑜𝑔 𝑔
+ 𝛽14
𝑡𝑟𝑎𝑖𝑛𝑖𝑛𝑔 + 𝛽15
𝑟𝑖𝑐 𝑒.𝑖𝑛𝑐 𝑜𝑚𝑒 + 𝛽16
𝑜𝑓 𝑓 − 𝑓𝑎𝑟𝑚 + 𝛽17
𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛
+ 𝛽18
𝐼 𝑀 𝑅 + 𝜀𝑖. (5)
4.5.2. Model specification with exclusion restriction
The evidence of some studies demonstrates that the Heckman ap-
proach can seriously inate standard errors because of collinearity
between the correction term and also the included regressors ( Certo
et al., 2016 ; Clougherty et al., 2016 ; Mott, 1999 ; Stolzenberg and
Relles, 1997 ). The inclusion of the inverse Mills ratio (IMR) often yields
multicollinearity with the predictors, even when the correction term
(IMR) has been statistically signicant or properly implemented that
can have intense consequences on the model estimates ( Anderson, 2007 ;
Bushway et al., 2007 ). This problem is exacerbated in the absence of ex-
clusion restrictions when all the variables used to estimate the rst step
are the same as covariates in the second step ( Bushway et al., 2007 ).
Exclusion restrictions are the variables that aect the selection process
but not the substantive equation of interest meanings that the Heckman
model should include at least one variable in the rst stage that does
not appear in the second stage ( Bushway et al., 2007 ; Sartori, 2003 ).
These variables directly aect the dependent variable in the rst stage
but have no association with the dependent variable in the second stage
( Certo et al., 2016 ). The transformed predicted value (i.e. IMR) in the
rst stage that is included in the second-stage estimation correlates
strongly with the predictors in the second stage ( Bushway et al., 2007 ;
Angrist, 2001 ). As in any multiple regression model, high collinearity
yields inconsistent estimates. The best solution is to this problem is to
incorporate one or more additional predictors in the rst stage that are
then excluded in the second stage ( Certo et al., 2016 ; Anderson, 2007 ;
Bushway et al., 2007 ; Leung and Yu, 1996 ). This exclusion reduces
the problematic correlation introduced by Heckman’s correction factor
( Bushway et al., 2007 ). Some scholars concluded that a model with ex-
clusion restriction is usually stronger than a model without exclusion re-
strictions and that they proposed that researchers should use the correla-
tion between x and IMR as an indicator of exclusion restrictions strength
( Certo et al., 2016 ; Bushway et al., 2007 ; Leung and Yu, 1996 ; Kennedy,
2006 ).
5. Results
5.1. Summary statistics of the variable and its definition included in the
model
The following variables are considered as the determinant of the
market participation for the rice farmers. Table 2 shows the list of de-
pendent and independent variables used in the analysis for the study.
Two independent variables, participation in the market (1 = participa-
tion, 0 = do not participation) and per capita consumption expenditure
in BDT were used in the model. The majority of the farmers are not
participating in the market; only 38% of households can participate in
the market. The average per capita consumption expenditure is BDT
52,730.56 per year. The average age of the sample household head is
45.59 years which indicates they are still in their productive age. Simi-
larly, education is also an important parameter for market participation
and their combined education level is 4.98 years of schooling. Since
they dropped out of school after 3–5 years of education, their overall
education level remained low. The average age, dependency ratio, and
farm size of both participant and non-participant households are 5.66
years, 1.57, and 2.21 hectares respectively. Moreover, approximately
28% of household heads are members of an organization, 31% have ex-
tension contact, 46% have access to information and 49% have access
to credit facilities in case of both participant and non-participant house-
holds. The average distance from home to road, home to market, and
home to haor is about 0.83 KM, 1.08 KM, and 1.26 KM respectively.
During the dry season, approximately 68 percent of sample households
were connected to the urban area via submergible road, and only 8%
of households have their transportation, especially a boat. During the
dry season availability of irrigation water is very expensive for B oro
rice cultivation and drought is very common during the season. Only
29% of households have access to surface water for irrigation in the
vicinity of haor . On average, the land is waterlogged for an average of
5.89 months during monsoon in the haor areas of northeast Bangladesh.
The average price of paddy is BDT 15.45/kg and the average income
from the rice sale is about BDT 229,512.60 because Boro rice is the only
crop for the haor Household. About 25% of the household head have
attached with o-farm income. The existing location of inhabitation is
also largely inuencing the market participation decision of the haor
household.
5.2. Descriptive statistics of the market participant and non-participant
households
Table 3 shows the summary statistics of the model’s included vari-
ables for both participants and non-participants in the rice market, as
well as the signicance level of the mean dierence of each variable.
A signicant dierence is found between market participant and non-
participant groups in terms of per capita consumption expenditure. The
age dierences of the household head between participant and non-
participant are statistically signicant at the 5% level. It is also found
that market participant households have a larger farm size than non-
participant which is statistically signicant at the 10% level. The ex-
tent of households having a membership of an organization, extension
contact, access to market information and access to credit was higher
than a non-participant which is highly signicant at 1% level. In gen-
eral, a higher percentage of non-participant rice farmers are located far-
ther away from the nearest market than the participant. On the other
hand, those who are located far away from the nearest haor have signif-
icantly participated in the market. Road connectivity, own transporta-
tion facilities, and availability of irrigation water have signicantly in-
uences participant household to participate in the rice market than
non-participant. Agricultural training, income from rice sale, and out-
put price have a signicant impact on participant farmers than non-
participant which is statistically signicant at 1, 5, and 1% respectfully.
Surprisingly, the education of the household head, family size, and home
5
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Table 2
Summary statistics and denitions of the variables.
Variables name Denition Mean Std. Dev.
Dependent variable
Market participation 1 if the farmers participate in the market (if farmers sold their any quantity in market), 0 otherwise 0.38 0.49
Consumption expenditure Per capita expenditure on consumption (Per capita/ year) 52,730.56 27,927.43
Independent variables
Age Age of the household head 45.59 10.96
Education Years of schooling 4.98 4.02
Family size Total family members of household 5.66 2.27
Dependency ratio Ratio between total family members and active working members 1.57 0.51
Farm size Farm size in hectare 2.21 2.16
Membership of an Organization 1 if a farmer is a member of an agriculture related organization, 0 otherwise 0.28 0.45
Extension contact 1 if the farmer has contact with extension personal,0 otherwise 0.31 0.46
Market information 1 if the farmer has a mobile phone for getting market information,0 otherwise 0.46 0.50
Access to credit 1 if the farmer has access to credit, 0 otherwise 0.49 0.50
Home to road distance Distance from home to submergible road (km) 0.83 0.90
Home to market distance Distance from home to market (km) 1.08 1.04
Home to haor distance Distance from home to haor (km) 1.26 1.06
Road connectivity 1 if the road is connected with urban area, 0 otherwise 0.68 0.47
Transportation facilities
(boat) 1 if the household head has own transport facilities (boat), 0 otherwise 0.08 0.28
Irrigation 1 if the household head has access to irrigation facilities, 0 otherwise 0.29 0.45
Duration of waterlogged Number of months that land are waterlogged during monsoon 5.89 0.47
Agricultural training 1 if the farmer has participated in agricultural training, 0 otherwise 0.38 0.49
Output price The price at which each unit of paddy is sold (BDT/kg) 15.45 1.08
Income from rice HH total income from the sale of rice (BDT) 229,512.60 216,038.80
O farm income 1 if the farmers have o farm income, 0 otherwise 0.25 0.43
Location dummies Household is located in Sunamganj district is 1, Household is located in Netrokona district is 2, and
Household is located in Kishoreganj district is 3
2.00 0.82
Table 3
Descriptive statistics of the variables used in the model.
Outcome and independent variables Non-Participant ( n = 187) Participant ( n = 113) Mean dierence t-value
Consumption expenditure 50,343.960 56,680.070 − 6336.107
∗ − 1.913
Age 46.556 43.991 2.565
∗ ∗ 1.974
Education 4.717 5.407 − 0.691 − 1.443
Family size 5.722 5.558 0.164 0.606
Dependency ratio 1.592 1.539 0.053 0.861
Farm size 2.028 2.511 − 0.483
∗ − 1.883
Membership of an organization 0.080 0.602 − 0.522
∗ ∗ ∗ − 11.819
Extension contact 0.128 0.619 − 0.491
∗ ∗ ∗ − 10.318
Access to market information 0.289 0.752 − 0.463
∗ ∗ ∗ − 8.707
Access to credit 0.369 0.690 − 0.321
∗ ∗ ∗ − 5.657
Home to road distance 0.821 0.834 − 0.013 − 0.119
Home to market distance 1.149 0.962 0.187 1.509
Home to haor distance 1.013 1.681 − 0.667
∗ ∗ ∗ − 5.566
Road connectivity 0.599 0.805 − 0.206
∗ ∗ ∗ − 3.778
Transportation facilities (boat) 0.048 0.142 − 0.093
∗ ∗ ∗ − 2.867
Irrigation 0.176 0.469 − 0.293
∗ ∗ ∗ − 5.699
Duration of waterlogged 5.824 6.000 − 0.176 − 3.217
Agricultural training 0.310 0.504 − 0.194
∗ ∗ ∗ − 3.407
Income from rice 208,312.500 264,596.000 − 56,283.500
∗ ∗ − 2.201
Output price 15.096 16.036 − 0.940
∗ ∗ ∗ − 8.080
Access to o farm income 0.257 0.239 0.018 0.343
Location dummies 2.005 1.991 0.014 0.145
(
∗ ∗ ∗
,
∗ ∗
,
∗
indicates the 1%, 5% and 10% levels of signicance).
to road distance has not inuenced the market participants at a signi-
cant level.
5.3. Market participation and household welfare: Heckman model with
exclusion restriction
The Heckman two-step model was used to identify the determinants
of market participation and its consequential eects on household wel-
fare. In the rst step of the model, the dependent variable in the par-
ticipation equation is equal to 1 if the household head sold any part of
his rice in the market, and 0 otherwise. The second stage of the model
estimates the factors that aect the household’s welfare due to mar-
ket participants in terms of per capita consumption expenditure per
year. The value of Lamda ( 𝜆) was used to account for selection bias
( Abdullah et al., 2019 ), and a model of exclusion restriction was used
to eliminate the possibility of multicollinearity between the IMR and
independent variables ( Certo et al., 2016 ; Bushway et al., 2007 ).
5.3.1. Determinants of market participation decision: the first stage probit
regression
The rst stage of the Heckman model determined the factors that
aect the market participation of rice farmers. The coecient estimates
of the inuencing factors and their probability value are presented
in Table 4 . A good number of variables have the expected sign and
were statistically signicant. Age is an important determinant of mar-
ket participation. Generally, younger farmers are risk-takers and prot-
oriented and this nding conformity with Randela et al. (2008) stated
that younger farmers are innovative and they understand the need of
the day, and be aware of the benets of commercialization. It means
that the increase of the age will decrease the probability to partici-
6
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Table 4
Determinants of market participation: the rst stage probit estimation.
Variables Co-ecient Std. Err. z P > z
Age − 0.010 0.012 − 0.800 0.424
Education 0.044 0.033 1.340 0.179
Family size 0.003 0.057 0.050 0.959
Dependency ratio 0.203 0.253 0.800 0.422
Farm size 0.040 0.117 0.350 0.730
Membership of an organization 1.326
∗ ∗ ∗ 0.309 4.290 0.000
Extension contact 0.520
∗ 0.280 1.860 0.063
Access to market information 0.756
∗ ∗ 0.249 3.040 0.002
Access to credit 0.571
∗ ∗ 0.258 2.220 0.027
Home to road distance 0.039 0.113 0.340 0.731
Home to market distance − 0.248
∗ 0.132 − 1.880 0.060
Home to haor distance 0.514
∗ ∗ 0.166 3.100 0.002
Road connectivity 0.585
∗ 0.314 1.870 0.062
Transportation facilities 0.254 0.376 0.680 0.499
Irrigation facilities 0.664
∗ ∗ 0.268 2.470 0.013
Duration of waterlogged 0.657
∗ ∗ 0.300 2.190 0.028
Agricultural training 0.279 0.255 1.090 0.274
Income from rice 0.000 0.000 − 0.070 0.944
Access to o farm income 0.283 0.314 0.900 0.366
Output price 0.901
∗ ∗ ∗ 0.170 5.300 0.000
Location dummies Sunamganj is a base category
Netrokona 0.519 0.367 1.41 0.157
Kishoreganj 0.153 0.349 0.44 0.661
Constant − 21.178
∗ ∗ ∗ 3.868 − 5.47 0.000
(Note: Number of obs. = 300, Log likelihood = − 76.876392, LR chi2
(22) = 243.69, Prob > chi2 = 0.0000, Pseudo R2 = 0.6131;
∗ ∗ ∗
,
∗ ∗
,
∗ in-
dicates the 1%, 5% and 10% levels of signicance).
pate in the market but it is not statistically signicant in the study.
The education level had an insignicant positive impact on the prob-
ability of rice market participation. From this result, it is clear that
the increase in education increases the ability to obtain and process
information for market participation. This nding is in dissimilarity
with Ouma et al. (2010) who found that education levels had a sig-
nicantly negative impact on banana market participation in Rwanda
and Burundi. Similarly, family size and dependency ratio had a posi-
tive impact on market participation but were not statistically signi-
cant. It means that the higher the number of people in the household
is likely to the greater the dependency ratio tends to the greater will
be the possibility to participate in the market. Surprisingly farm size
was not statistically signicant but had a positive relationship with mar-
ket participation. A possible explanation for this would be that remark-
able lands remained fallow due to the extreme climate event and un-
certainty of the rice production and also price fall of the rice in the
waterlogged haor areas of Bangladesh. These results support the nd-
ings of Mmbando et al. (2015) in the case of pigeonpea market par-
ticipation in Tanzania. The coecient of the membership of an orga-
nization had a positive and signicant impact on rice market partici-
pation at the 1% level. Belonging to be a member of an organization
increases the probability of the rice farmers participating in the market
and it is a good platform for exchanging information, enabling them
to link to buyers at a lower cost and enhance bargaining power. The
coecient of extension contact is positive and statistically signicant
at the 5% level which implies market participation of the farmers is
positively related to the availability of the extension service. Similarly,
linking a farmer with extension agents, the probability of participation
of that farmer would also increase. This nding is linked with the nd-
ings of the study of Awotide et al. (2013) in Nigeria. The government
should ensure the availability of extension services and training pro-
grams so that farmers can overcome the present situation, these nd-
ings conform to Abdullah et al. (2019) . Access to market information
had a signicant and positive impact on market participation at the 5%
level of signicance. The mobile phone is important to the source of get-
ting market information in the haor areas. Households having a mobile
phone are more likely to have the probability of market participation.
Ouma et al. (2010) found that ownership of radios has a statistically in-
signicant impact on market participation, which is the controversy of
our study. This is possibly due to communication assets are less useful in
accessing market information. The coecient of access to credit is sta-
tistically signicant at the 5% level, which indicates there is a positive
relationship between the availability of credit and market participation.
It is also justied by Ouma et al. (2010) that access to formal credit
enhances access to production assets, which would inuence the pro-
duction of a marketable surplus. Distance from home to the road had a
positive insignicant relationship to the market participation, meaning
that an increase in time taken to reach the nearest road increases the
probability of market participation. The reason behind the argument is
that all infrastructures, including roads, were submerged 6–7 months
during monsoon ( Ali and Rahman, 2017 ) and there is no signicant im-
pact of road distance on market participation. The coecient of distance
from home to the nearest market is statistically signicant and nega-
tively related to the market participation of rice farmers. It indicates
farmers who resided distance from the markets are less likely to partici-
pate in the markets, probably because of the high cost of market access.
The results are consistent with the ndings of Ouma et al. (2010) and
Asfaw et al. (2012) who found that market participation was less for
farmers who were located far away from nearest the market. Surpris-
ingly nearness to the market negatively inuences mango market par-
ticipation in Ethiopia ( Hagos et al., 2020 ). The relationship between
distances from home to haor is statistically signicant and positively
related to the market participation of rice farmers. An increase in the
distance from home to the haor water basin increases the probability of
market participation for rice farmers. The strength of this argument is
that households located in remote areas are seasoning market accessed
and raised costs associated with the marketing ( Ouma et al., 2010 ). Road
connectivity with urban areas or the nearest market positively and sig-
nicantly inuences the farmers to participate in the market at a 10%
level of the condence interval. A study on the commercialization of
smallholders in Ethiopia by Getahun (2015) noted that proximity to all-
weather roads cheers market orientation due to the possibility of reduc-
ing marketing costs. Therefore, road connectivity is a factor that posi-
tively inuenced the decision to participate in the market among small-
holder farmers. The coecient of transportation facilities is somewhat
insignicant but positively related to market participation. The argu-
ment, in that case, any kind of transportation system in the roadway is
not possible during monsoon where roads were submerged into 6 to 7
months with oodwater and boat is the only medium of transportation
at that time ( Hoq et al., 2021 ). The availability of the irrigation water is
signicantly and positively related to the market participation at a 5%
level of the condence interval, which indicates if the irrigation water
is abundant to the farmers, they get bumper production of rice, which
tends to produce marketable surplus and promote market participation.
This result disagrees with Abdullah et al. (2019) . Impulsively, the dura-
tion of the waterlogging condition positively inuences the market par-
ticipation, which was signicant at the 5% level of the condence inter-
val. It may be due to persistent communication on the waterway, which
reduces transportation costs largely and enhances market participation.
The haor inhabitants are residing in remote areas which are detached
from the mainstream of development. Therefore, they are deprived of
modern technology and production means related training, which tends
to insignicant market participation. The coecient of income from rice
is insignicant but positive with the market participation because of
crop loss due to extreme climate events such as; ash oods, oods, and
hailstorms, which are a barrier to the production of marketable surplus.
The coecient of access to o-farm income is positive and somehow
insignicant. Generally, those farmers who have a chance of o-farm
income have opportunities to participate in the market than those who
do not have this opportunity ( Abdullah et al., 2019 ). In contrary to this,
Rios et al. (2008) appealed that the relationship between o-farm in-
come and agricultural commercialization is negative. They argued that
the available time is spent on o-farm activities, and there is less time
7
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Table 5
Determinant of household welfare: the second stage OLS estimation.
Per capita consumption expenditure Coecient Std. Err. t P > t
Age − 107.448 105.860 − 1.020 0.311
Education 314.896 279.810 1.130 0.261
Family size − 5150.136
∗ ∗ ∗ 495.733 − 10.390 0.000
Dependency ratio − 3517.809
∗ 2165.338 − 1.620 0.105
Farm size 6112.062
∗ ∗ ∗ 1089.017 5.610 0.000
Access to credit − 7071.905
∗ ∗ 2328.296 − 3.040 0.003
Home to road distance − 484.816 1202.033 − 0.400 0.687
Home to market distance 849.026 1062.202 0.800 0.425
Home to haor distance 770.075 1295.800 0.590 0.553
Road connectivity 2159.284 2562.905 0.840 0.400
Transportation facilities 2693.156 3942.250 0.680 0.495
Irrigation facilities − 2146.062 2597.794 − 0.830 0.409
Duration of waterlogged − 533.353 2567.417 − 0.210 0.836
Agricultural training 956.613 2269.005 0.420 0.674
Income from rice 0.019
∗ 0.011 1.780 0.076
Access to o-farm income − 4323.157
∗ 2564.282 − 1.690 0.093
Location dummies Sunamganj is a base category
Netrokona − 3075.318 3132.190 − 0.980 0.327
Kishoreganj 6713.592
∗ ∗ 2680.201 2.500 0.013
IMR − 2313.472
∗ ∗ 1201.777 − 1.930 0.055
Constant 80,101.370
∗ ∗ ∗ 17,468.400 4.590 0.000
(Note: Number of obs = 300, Prob > F = 0.0000, R-squared = 0.6226, Adj R-
squared = 0.5911;
∗ ∗ ∗
,
∗ ∗
,
∗
indicates the 1, 5 and 10% level of signicance).
for agricultural production. The output price had a positive impact on
the market participation of rice farmers, which is signicant at the 1%
level of the signicant. Higher market prices are expected to inuences
farmers to participate more in the market. These results disagree with
Hagos et al. (2020) in Ethiopia the case of mango farmers. Household lo-
cated in the Netrokona and Kishoreganj district is positively inuenced
to participate in the rice market as compared to Sunamganj District (ref-
erence district), but the coecient is not up to the level of signicance.
Sunamganj district is characterized by poor infrastructure and relative
remoteness compared to Netrokona and Kishoreganj districts.
5.3.2. Impact of market participation on household welfare: the second
stage analysis (OLS)
The household’s welfare due to participation in the rice market is
examined using the second stage OLS model ( Table 5 ). Before estima-
tion, the model was diagnosed for its suitability by checking possible
multicollinearity problems using the VIF (Variance Ination Factor) and
correlation matrix. The acceptable value of VIF is less than the critical
value of 10 for conrmation that multicollinearity is not a major prob-
lem ( Gujarati and Porter, 2009 ). On the other hand, the acceptable value
of the correlation coecient is globally less than 0.5, indicating weak
correlations ( Mmbando et al., 2015 ). The VIF value of IMR (11.44) is
greater than the critical value of 10, which indicates there is possible
multicollinearity among IMR and independent variables (Appendix Ta-
ble A). The correlation matrix shows these coecients such as member-
ship of an organization ( − 0.6111), extension contact ( − 0.5809), access
to market information ( − 0.5629), and output price ( − 0.5583), which
are moderately correlated with the IMR (Appendix Table B). Conse-
quently, these four variables are excluded from the second stage OLS
equation and found IMR’s value statistically signicant which indicates
the presence of selection bias in the sample ( Table 5 ). Moreover, the
VIF value of IMR (2.32) is also found less than the critical value of
10 (Appendix Table C). Hence, the model justies the Heckman model
with exclusion restriction ( Certo et al., 2016 ; Bushway et al., 2007 ;
Sartori, 2003 ). The ndings in the line of Myers (1988) and Myers and
Talarico (1986) who reported that the correlation coecient value be-
tween 𝜆and crime severity is 0.9 indicates high multicollinearity, and
the model is estimated by excluding crime severity from the equation of
sentence length. The coecient of family size is statistically signicant
and negatively correlated with the consumption expenditure means that
households having large family sizes will have low welfare. If the fam-
ily size is high, more income is required to maintain the expenses of the
family members. Moreover, most of the rice output is consumed by the
family members and little quantity remains for marketable surplus. This
nding conforms to Abdullah et al. (2019) and Tufa et al. (2014) . The
coecient of the dependency ratio is somehow signicant and nega-
tively related to the per capita consumption expenditure. It indicates an
increase in the family size may lead to a high dependency ratio which
would lead to higher consumption and more expenditure. The coe-
cient of the farm size is positive and signicant means that as farm size
increases the welfare of the household also increases. This result is con-
sistent with Asfaw et al. (2012) who showed that gain in per capita
consumption expenditure due to market participation is highest from
the largest farm size. The eect of access to credit on household welfare
due to market participation is negative and signicant at a 5% level. The
possible explanation for this is that farmers have to oer a signicant
quantity of rice for the repayment of the credit. This is in contrast to
Abdullah et al. (2019) who found a signicant and positive relationship
with the consumption expenditure. The coecient of the income from
the rice is positive and signicant wielding a positive impact on house-
hold consumption expenditure as well as welfare. It means the greater
the income from rice, the greater will be the revenue, which has a signif-
icant impact on the livelihood of the households ( Abdullah et al., 2019 ).
The impact of access to o-farm income is negative and signicant to the
household’s welfare. Since available time is allocated to o-farm activi-
ties, the relationship between o-farm income and market participation
is negative, resulting in low household welfare from agricultural pro-
duction ( Sebatta et al., 2014 ). The coecient of the households located
in Kishoreganj district is statistically signicant and positively related to
household consumption expenditure as compared to Sunamganj district
(reference district) possibly because of greater access to nearest markets.
6. Discussion
The determinants of the binary decision to market participation or
not participation and the continuous outcome function of household
welfare conditional on market participation decisions were simultane-
ously analyzed using the Heckman two-stage model following exclu-
8
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
sion restriction. It is observed from the rst stage probit model that
membership of an organization, extension contact, access to market in-
formation, access to credit, home to haor distance, road connectivity,
irrigation facilities, duration of the waterlogged condition, and output
price are the determinants that positively and signicantly inuence
the farmer’s decision to participate in the market. It means that increas-
ing these determinants will boost smallholder rice market participation.
Most of these signicant determinants are also supported by the nd-
ings of Omiti et al. (2009) , Sebatta et al. (2014) , Ademe et al. (2017) ,
Kyaw et al. (2018) , and Negasa et al. (2020) in their study.
Sebatta et al. (2014) stated that a member of an organization im-
proves access to output markets and consequently increases expected
prots which are consistent with our ndings. The main reason for
this is that working in a group creates bargaining power among the
farmers and enables them to enhance expected prot. Some of the sig-
nicant determinants such as extension contact, access to market in-
formation, and access to credit are highly inuential for market par-
ticipation. Extension workers typically support new technology, mar-
ket information that improves farmer awareness and provides a vari-
ety of market opportunities, and credit facilities that assist farmers in
their production process, all of which help farmers participate in the
rice market ( Kyaw et al., 2018 ). But the availability of these services is
very tight in the study areas ( BHWDB, 2012 ). The government should
make extension services and training programs available to farmers
so that they can overcome the current situation, these conform with
Abdullah et al. (2019) . The authors found that connectivity to the sub-
merged road during the dry season positively inuenced the market par-
ticipation of haor households. Makhura (2001) reported that the state of
bad roads negatively inuenced the farmer’s access to urban markets;
hence there is an interruption to smallholder market participation and
decreased incomes. During the dry season, farmers face acute irrigation
problems due to the distance between the water source and farmland.
But most of the farmers can use abundant haor water for irrigation pur-
poses, which tends to bumper rice production as well as market partici-
pation. Kyaw et al. (2018) reported a similar result that the increase in
rice production increased revenues from the sale of surplus rice on the
market.
The results also indicate that persistent waterlogged conditions in
the haor basin help rice farmers to participate in the market by boat
at low transportation costs in comparison to those who were not con-
nected to road communication ( BHWDB, 2012 ). Rice market prices have
a signicant and positive impact on farmers’ willingness to participate
in the market. This result is in line with that of Mather et al. (2011) , who
reported that higher output prices increased the probability of market
participation by allowing households to earn more income ( Kyaw et al.,
2018 ). On the other hand, only home to market distance negatively
and signicantly inuences the household market participation decision
meaning that the decrease of the home to market distance would signif-
icantly increase the market participation ( Kyaw et al., 2018 ). This result
is in line with the ndings of Rehima and Dawit (2012) , who found that
market participation among smallholder pepper producers in Ethiopia’s
Silte and Aalaba was negatively related to distance to the market.
The results of the second step of the Heckman model (OLS) revealed
that family size, dependency ratio, access to credit, and o-farm income
have a negative and substantial impact on the household’s per capita
consumption expenditure. Cadot et al. (2006) explained that large fam-
ilies seem to have higher opportunity costs, perhaps which is reected
in the fact that they have lower per capita income from rice produc-
tion and hence less surplus to switch to the market. The relationship
between credit availability and household welfare is inverse. Farmers
may be required to sell a substantial amount of rice in exchange for
repayment of a credit that was taken during the production process.
Rios et al. (2008) and Sebatta et al. (2014) explained that the relation-
ship between o-farm income and agricultural commercialization due
to market participation is negatively related because the available time
is spent on o-farm income rather than agricultural production which
tends to low household’s welfare from agricultural production. On the
other hand, farm size and income from rice signicantly and positively
inuence the household’s per capita consumption expenditure as a re-
sult of market participation. It means the larger the farm size, the greater
the rice output, the greater will be the market orientation and the larger
will be the income, which has a positive impact on household welfare
( Kyaw et al., 2018 ). The geographical location of the household is very
important for market participation as well as household welfare. Due
to the remoteness of the Sunamganj district, households located in the
Kishoreganj district are in a weather-gage to participate in the market
and gain household’s welfare than those in Sunamganj.
7. Conclusions and policy recommendations
The purpose of this study is to identify the physical and socioeco-
nomic factors that inuence market participation and how they aect
the welfare of rural farming households in the wetland haor ecosystem
of Bangladesh. Market and improved market access in the haor areas
are critical for ooding, collapsed infrastructure, a lack of bridges and
culverts, and the absence of roads and highways, which have been key
barriers to improving rural livelihoods. The probit model results indi-
cated that haor household’s participation in the rice market is signi-
cantly inuenced by the membership of an organization, extension con-
tact, access to market information, access to credit, home to market and
haor distance, road connectivity, irrigation facilities, duration of the wa-
terlogged, and output price. The OLS results revealed that family size,
dependency ratio, farm size, access to credit, and access to o-farm in-
come signicantly inuenced the household’s per capita consumption
expenditure as a result of market participation. The ndings from the
study exposed that the farmers that participated in the market had a
higher and signicant income from rice production, per capita consump-
tion expenditure, and output price than the farmers that did not partic-
ipate in markets. Bangladesh’s government is attempting to facilitate
the creation of haor areas by enacting various policies and plans. The
government has built an all-weather road and a submergible road to
connect waterlogged haor areas with urban areas of Bangladesh, allow-
ing farmers access to the former market. Furthermore, the government
must implement eective strategies that focus on the most signicant,
important variables that increase the haor household’s market partic-
ipation. The policies aimed to improve rural physical infrastructure,
road connectivity, market information systems, smallholder asset accu-
mulation, extension linkage, and promotion of farmers’ organizations,
all of which could increase smallholder farmers’ market participation
and marketed surplus. Most importantly, the government, NGOs, and
development agencies should devise eective strategies to encourage
farmers to form producer groups in order to increase their bargaining
power, and the Department of Agricultural Extension (DAE) can play a
key role in strengthening extension linkages to promote market partici-
pation in haor areas of Bangladesh. Additionally, in order to ensure fair
prices, the government should improve the rice procurement system by
purchasing rice directly from farmers or using other methods such as
the creation of storage facilities and increasing the bargaining power
of the farmers. Therefore, the recommendation to policymakers is to
strengthen the signicant determinants of market participation as well
as government direct rice purchases from the farmers, enabling farmers
to benet from producing a marketable surplus and ensuring welfare
from rice production.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that may have aected the research
presented in this paper.
9
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Acknowledgments
The authors are grateful to the Bangladesh government and the
Bangladesh Agricultural Research Council (BARC) for funding the rst
author’s PhD study through the NATP-2 (National Agricultural Tech-
nology Program-Phase II) Project. The funders had no involvement in
the study’s design, data collection, or analysis, or in the decision to
publish the work. We would also like to express our appreciation to
the anonymous reviewers, who provide informative comments that will
help us greatly improve the manuscript. The authors are grateful to
PIU (Project Implementation Unit), NATP-Phase-II, BARC for support-
ing the rst author’s PhD fellowship program. The nal manuscript was
read and accepted by all contributors, who oered critical input. The
manuscript was based in part on the rst author’s doctoral dissertation
to Bangladesh Agricultural University-2202.
Appendix
Table A1 , Table B1 , Table C1
Table A
Variance Ination Matrix (VIF) of the independent variables used in probit
model.
Variable VIF 1/VIF
Age 1.30 0.769653
Education 1.26 0.794199
Dependency ratio 1.18 0.845460
Family size 1.25 0.801489
Farm size 5.37 0.186110
Member of an organization 2.27 0.440172
Extension contact 1.91 0.524645
Market information 1.81 0.552814
Access to credit 1.59 0.630519
Home to road distance 1.12 0.891075
Road connectivity 1.65 0.606507
Transportation facilities 1.13 0.888024
Irrigation facilities 1.38 0.725135
Duration of waterlogging condition 1.77 0.566427
Home to market distance 1.34 0.745783
Home to haor distance 2.08 0.479621
Agricultural training 1.23 0.813577
Income from rice 5.32 0.187860
Access to o-farm income 1.23 0.811081
Output price 3.54 0.282677
Location dummies
Netrokona 2.38 0.419883
Kishoreganj 1.70 0.586808
IMR 11.44 0.087392
Mean VIF 2.40
10
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Table B
Correlation matrix of the explanatory variables.
Variables Age Education Dependency
ratio
Family
size
Farm
size
Member
of an
organization
Extension
contact
Market
information
Access
to
credit
Home
to road
distance
Road
connectivity
Age 1.0000
Education − 0.1548 1.0000
Dependency ratio − 0.2350 0.0498 1.0000
Family size 0.1473 − 0.0250 0.1619 1.0000
Farm size − 0.0102 0.1349 − 0.0025 0.0834 1.0000
Member of an organization − 0.0940 0.0852 − 0.0433 − 0.0616 0.0136 1.0000
Extension contact − 0.0824 0.1202 − 0.0247 0.0094 0.1308 0.5621 1.0000
Market information − 0.0489 0.0553 − 0.0479 0.0744 − 0.0139 0.3817 0.3379 1.0000
Access to credit 0.0154 0.0754 − 0.0104 0.0146 − 0.0355 0.2136 0.3154 0.0788 1.0000
Home to road distance − 0.0141 − 0.0709 − 0.0681 0.0134 − 0.0835 0.0044 0.0529 0.1016 0.0206 1.0000
Road connectivity − 0.0220 − 0.0519 − 0.0354 0.0032 0.2394 0.0770 0.1290 0.1421 − 0.0922 − 0.0444 1.0000
Transportation facilities 0.0984 0.1068 − 0.0875 0.0504 0.0709 0.1101 0.1343 0.1068 0.0663 − 0.0343 0.1311
Irrigation facilities − 0.0018 0.0074 − 0.0369 0.1501 0.1771 0.2176 0.2710 0.2092 0.1896 − 0.0038 0.1703
Duration of waterlogging
condition
0.0832 − 0.0263 0.0067 − 0.0856 0.0857 0.2415 0.2362 0.1330 − 0.0261 − 0.0366 0.0966
Home to market distance − 0.0100 − 0.0855 0.1055 0.0276 − 0.0703 − 0.0283 − 0.0098 − 0.0237 − 0.0421 0.0753 − 0.0539
Home to haor distance − 0.0490 − 0.1221 − 0.1168 − 0.1598 0.0361 0.0639 0.0718 0.2044 0.0886 − 0.0715 0.2488
Agricultural training 0.0158 − 0.0022 − 0.0232 0.0637 0.1484 0.1867 0.2064 0.0511 0.0912 0.0658 0.1199
Income from rice 0.0031 0.1657 − 0.0162 0.0382 0.8889 0.0068 0.1669 0.0454 0.0141 − 0.0877 0.2217
Access to off-farm income − 0.0107 0.1183 0.0042 − 0.0830 − 0.1464 0.0387 0.0415 − 0.0116 − 0.1347 − 0.1369 0.0370
Output price − 0.1364 0.0568 − 0.0237 0.0448 0.0977 0.2496 0.2287 0.2241 0.0313 − 0.0200 0.0566
Location dummies 0.1213 0.0315 − 0.0637 − 0.0288 − 0.0467 0.1186 0.0264 − 0.0164 − 0.0163 − 0.0550 − 0.0524
IMR 0.1752 − 0.1454 0.0325 − 0.0059 − 0.1429 − 0.6111 − 0.5809 − 0.5629 − 0.3060 − 0.0030 − 0.3724
Transportation
facilities
Irrigation
facilities
Duration of
waterlogging
condition
Home
to market
distance
Home
to haor
distance
Agricultural
training
Income
from
rice
Access
to
off-farm
Output
price
Location
dummies
IMR
Transportation facilities 1.0000
Irrigation facilities 0.1022 1.0000
Duration of waterlogging
condition
0.0452 − 0.0243 1.0000
Home to market distance − 0.0332 − 0.0968 0.0260 1.0000
Home to haor distance − 0.0002 0.0178 0.0158 0.1017 1.0000
Agricultural training 0.0351 0.1066 0.0683 0.1009 0.1176 1.0000
Income from rice 0.0600 0.1861 0.0465 − 0.0881 0.0495 0.1396 1.0000
Access to off-farm income − 0.0627 − 0.0596 − 0.0124 − 0.0187 0.0303 − 0.1702 − 0.1041 1.0000
Output price 0.1393 0.1421 − 0.1023 − 0.0842 − 0.0271 0.0171 0.1346 − 0.0612 1.0000
Location dummies 0.0000 − 0.0542 0.2361 0.0605 0.0043 0.0756 − 0.0079 − 0.0283 − 0.2969 1.0000
IMR − 0.2236 − 0.3725 − 0.2832 0.1822 − 0.3399 − 0.2263 − 0.1739 − 0.0027 − 0.5583 0.0468 1.0000
Table C
Variance Ination Matrix (VIF) of the independent variables used in OLS model
after exclusion restriction.
Variable VIF 1/VIF
Age 1.26 0.791466
Education 1.19 0.840556
Dependency ratio 1.16 0.862448
Family size 1.19 0.837796
Farm size 5.20 0.192237
Access to credit 1.28 0.783882
Home to road distance 1.09 0.914719
Road connectivity 1.35 0.738930
Transportation facilities 1.12 0.894489
Irrigation facilities 1.30 0.769510
Duration of waterlogging condition 1.35 0.739339
Home to market distance 1.15 0.872152
Home to haor distance 1.75 0.569959
Agricultural training 1.15 0.872561
Income from rice 5.06 0.197592
Access to o-farm income 1.16 0.861312
Location dummies
Netrokona 2.05 0.487091
Kishoreganj 1.50 0.665229
IMR 2.32 0.430791
Mean VIF 1.77
References
Abdullah, Rabbi, F., Ahmed, R., Ali, S., Chandio, A.A., Ahmad, W., Ilyas, A., Ud Din, I.,
2019. Determinants of commercialization and its impact on the welfare of smallholder
rice farmers by using Heckman two-stage approach. J. Saudi Soc. Agric. Sci. 18, 224–
233. doi: 10.1016/j.jaas.2017.06.001 .
Abdullah, H.M., Ahmed, S.M., Khan, B.M., Mohana, N.T., Ahamed, T., Islam, I., 2021.
Agriculture and sheries production in a regional blending and dynamic fresh and
saline water systems in the coastal area of Bangladesh. Environ. Chall. 4, 100089.
doi: 10.1016/j.envc.2021.100089 .
Ademe, A., Legesse, B., Haji, J., Goshu, D., 2017. Smallholder farmers’ crop commercial-
ization in the highlands of Eastern Ethiopia. Rev. Agric. Appl. Econ. 20 (2), 30–37.
doi: 10.15414/raae.2017.20.02.30-37 .
Ahmed, R., 2017. Marginal farmers have least access, The Daily Star, 05 October, 2017.
Ahmed, R., 2020. Buy paddy from farmers, not rice from millers. Dhaka Tribune, May
11th, 2020.
Alam, M.S. , Quayum, M.A. , Islam, M.A. , 2010. Crop production in the haor areas of
Bangladesh: insights from farm level survey. Agriculturists 8 (2), 88–97 .
Alamgir, M.S., Furuya, J., Kobayashi, S., Mostaz, R.B., Ahmed, M.R., 2020.
Farm income, inequality, and poverty among farm families of a ood-prone
area in Bangladesh: climate change vulnerability assessment. GeoJournal
doi: 10.1007/s10708-020-10231-2 .
Alemu, Z.G. , 2012. Livelihood strategies in rural South Africa: implications for poverty
reduction. In: Proceedings of the International Association of Agricultural Economists
(IAAE) Triennial Conference, Foz do Iguacu, Brazil, pp. 1–24 18–24 August .
Ali, A. , 1999. Climate
change impacts and adaptation assessment in Bangladesh. Clim.
Res. 12, 109–116 .
Ali, M.Y. , Rahman, M. , 2017. Floods in Northeast (haor) areas of Bangladesh. HCTT co-
ordinated needs assessment (CAN), Bangladesh, pp. 1–50, April-May .
Anderson, B., 2007. Selection models and weak instruments. https://info.umkc.
edu/drbanderson/2017/06/ , 2007.
Angrist, J. , 2001. Estimation of limited dependent variable models with dummy endoge-
nous regressors: simple strategies for empirical practice. J. Bus. Econ. Stat. 19, 2–16 .
Bangladesh Economic Review (BER), 2019. GDP, savings and investment, chapter 2,
Bangladesh economic review, nance division, ministry of nance, government of
the people’s republic of
Bangladesh, June 2019, www.mof.gov.bd .
Asfaw, S., Amare, M., Lipper, L., Davis, B., 2012. Welfare eects of market participation
under transaction costs: evidence from Tanzania. Paper presented at agricultural de-
velopment economics division (ESA), Rome, Italy, January 18, 2012.
Awotide, B.A. , Diagne, A. , Awoyemi, T.T. , 2013. Agricultural technology adoption, market
participation and rural farming household’s welfare in Nigeria. In: Proceedings of the
11
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
4th International Conference of the African Association of Agricultural Economists,
Hammamet, Tunisia, pp. 22–25 September .
Bangladesh Bureau of Statistics (BBS), 2019. Statistical year book Bangladesh, Bangladesh
bureau of statistics, government of the People’s Republic of Bangladesh.
Barrett, C.B. , 2008. Smallholder market participation: concepts and evidence from eastern
and southern Africa. Food Policy 33 (4), 299–317 .
Barrett, C.B. , Bellemare, M.F. , Bachke, M. , Michelson, H. , Narayanan, S. , Walker, T. , 2012.
Smallholder participation in contract farming: comparative evidence from ve coun-
tries. World Dev. 40 (4), 715–730 .
BHWDB, 2012. Master plan of haor areas:
volume ii, main report, ministry of water re-
sources, government of Bangladesh.
Brammer, H. , 1990. Floods in Bangladesh: geographical background to the 1987 and 1988
oods. Geogr. J. 156 (1), 12–22 .
Brown, D.R. , Stephens, E.C. , Ouma, J.O. , Murithi, F.M. , Barrett, C.B. , 2006. Livelihood
Strategies in the Rural Kenyan Highlands. Cornell University, Ithaca, NY, USA .
Bushway, S., Johnson, B.D., Slocum, L.A., 2007. Is the magic still there? The use of the
Heckman two-step correction for selection bias in criminology. J. Quant. Criminol.
23, 151–178. doi: 10.1007/s10940-007-9024-4 .
cadot, O., Dutoit, L., Olarreaga, M., 2006.
How costly cs it for poor farmers to lift them-
selves out of subsistence? world bank policy research working paper, 3881.
Certo, S.T. , Busenbark, J.R. , Woo, H.-.S. , Semadeni, M. , 2016. Sample selection bias
and Heckman models in strategic management research. Strateg. Manag. J. 37 (13),
2639–2657 .
M.A.H. Chowdhury, M.S. Hassan, Hand Book of Agricultural Technology. Bangladesh Agri-
cultural Research Council, Farmgate, Dhaka. 2013, 230p.
Clougherty, J.A. , Duso, T. , Muck, J. , 2016. Correcting for self-selection based endogene-
ity in management research: review, recommendations and simulations. Organ. Res.
Methods 19 (2), 286–347 .
Chile, L.
, Talukder, D. , 2014. Characteristics of rice cultivation and rural rice market in
Bangladesh: evidence from a survey. Agric. Econ. Rev. 15 (2), 68–84 .
Das, D.C. , 2004. Experience with modern and hybrid rice varieties in plain land ecosystem
in the Mymensingh region. In: Proceedings of the on Twentieth National Workshop
on Rice Research and Extension in Bangladesh Bangladesh Rice Research Institute,
Gazipur-1701pp 19–21 .
Getahun, K. , 2015. Protability analysis and determinants of fruit tree based agroforestry
system in Wondo district. Ethiopia. Afr. J. Agric. Res. 10, 1273–1280 .
Greene, W.H. , 2003. Econometric Analysis, 5th Ed. Prentice
Hall .
Gujarati, D.N. , Porter, D.C. , 2009. Basic Econometrics, (5th Ed.) McGraw-Hill/Irwin, New
York .
Hagos, A., Dibaba, R., Bekele, A., Alemu, D., 2020. Determinants of market participation
among smallholder mango producers in assosa zone of benishangul gumuz region in
Ethiopia. Int. J. Fruit Sci. 20 (3), 323–349. doi: 10.1080/15538362.2019.1640167 .
Heckman, J.J. , 1979. Sample selection bias as a specication error. Econometrica 47 (1),
153–161 .
Homann, R. , Kassouf, A.L. , 2005. Deriving conditional and un-conditional marginal ef-
fects in logg earnings equations estimated by Heckman’s procedures. Appl. Econ. 37
(11), 1303–1311 .
Hoq, M.S., Raha, S.K.,
Hossain, M.I., 2021. Livelihood vulnerability to ood hazard: un-
derstanding from the ood-prone Haor ecosystem of Bangladesh. Environ. Manag. 67,
532–552. doi: 10.1007/s00267-021-01441-6 .
Hossain, M. , Jaim, W.M.H. , Alam, M.S. , Rahman, A.N.M.M. , 2013. Rice biodiversity in
Bangladesh: adoption, diusion and disappearance of varieties. Research and Evalu-
ation Division, Bangladesh Rural Advancement Committee, Dhaka .
Hossain, M.I. , Verbeke, W. , 2010. Evaluation of rice markets integration in Bangladesh.
Lahore J. Econ. 15 (2), 77–96 .
Hossain, M.S. , 2013. Flood Damage and Risk Assessment Model in the Haor Basin of
Bangladesh. Bangladesh University of Engineering & Technology
MS thesis .
Huda, M.K. , 2004. Experience with modern and hybrid rice varieties in haor ecosystem:
emerging technologies for sustainable rice production. In: Proceedings fot the Twen-
tieth National Workshop on Rice Research and Extension in Bangladesh Bangladesh
Rice Research Institute. Gazipur-1701, 19–21 April .
IFPRI, 2013. Agriculture and Adaptation in Bangladesh, Current and Projected Impacts of
Climate Change, Discussion Paper 01281. International Food Policy Research Insti-
tute, Washington, DC .
IRRI, 2008. International rice research institute" archived 21 May 2008 at the Wayback
Machine.
Jakariya, M., Islam, M.N., 2017. Evaluation of climate change induced vulnerability and
adaptation strategies at
Haor areas in Bangladesh by integrating GIS and DIVA model.
Model. Earth Syst. Environ. 3 (3), 1–19. doi: 10.1007/s40808-017-0378-9 .
Kabir, M.J., Alauddin, M., Crimp, S., 2017. Farm-level adaptation to climate change in
Western Bangladesh: an analysis of adaptation dynamics, protability and risks. Land
Use Policy 64 (May), 212–224. doi: 10.1016/j.landusepol.2017.02.026 .
Kafy, A.A., Faisal, A.A., Raikwar, V., Rakib, A.A., Kona, M.A., Ferdousi, J., 2021. Geospa-
tial approach for developing an integrated water resource management plan in Ra-
jshahi, Bangladesh. Environ. Chall. 4, 100139. doi: 10.1016/j.envc.2021.100139 .
Kamara, A.B. , 2004. The impact of market access on input use and agricultural
productiv-
ity: evidence from Machakos district, Kenya. Agrekon 43 (2), 202–216 .
Kennedy, P. , 2006. A Guide to Econometrics. MIT Press, Cambridge, MA .
Kothari, C.R. , 2004. Research Methodology: Methods and Techniques, (2nd Ed.) New Age
International Publishers .
Kyaw, N.N., Ahn, S., Lee, S.H., 2018. Analysis of the factors inuencing market participa-
tion among smallholder rice farmers in Magway region, central dry zone of Myanmar.
Sustainability 10 (12), 4441. doi: 10.3390/su10124441 .
Leung, S.F. , Yu, S. , 1996. On the choice between sample selection and two-part models.
J. Econom. 72 (1), 197–229 .
Lubungu, M. , 2013.
Welfare impacts of smallholder farmers’ participation in livestock
markets in Zambia. In: Proceedings of the Paper presented at the Agricultural & Ap-
plied Economics Association’s 2013 AAEA & CAES Joint Annual Meeting. Washington
DC 4–6 August .
Maertens, M. , Colen, L. , Swinnen, J.F.M. , 2011. Globalisation and poverty in Senegal: a
worst case scenario. Eur. Rev. Agric. Econ. 38 (1), 31–54 .
Maertens, M. , Swinnen, J.F.M , 2009. Trade, standards, and poverty: evidence from Sene-
gal. World Dev. 37 (1), 161–178 .
Makhura, M.T. , 2001. Overcoming transaction costs barriers to market participation of
smallholder farmers in the Northern province of South Africa. University of Pretoria,
Pretoria, pp. 1–192. A PhD thesis submitted in partial fulllment of the requirements
for the degree of in the department of agricultural economics, extension and rural
development, faculty of natural and agricultural sciences. June .
Mather, D.L. , Boughton, D. , Jayne, T.S. , 2011. Smallholder heterogeneity and maize mar-
ket participation in Southern and Eastern Africa: Implications for Investment strate-
gies to Increase marketed food staple supply. Department of Agricultural, Food, and
Resource Economics and Department of Economics, Michigan State University: East
Lansing, MI, USA .
Mmbando, F.E., Wale, E.Z., Baiyegunhi, L.J.S., 2015a. Welfare impacts of smallholder
farmers’ participation in
maize and pigeonpea markets in Tanzania. Food Secur. 7,
1211–1224. doi: 10.1007/s12571-015-0519-9 .
Mmbando, F.E., Z.Wale, E., Baiyegunhi, L.J.S., 2015b. Determinants of smallholder farm-
ers’ participation in maize and pigeonpea markets in Tanzania. Agrekon 54 (1), 96–
119. doi: 10.1080/03031853.2014.974630 .
Mott, R. , 1999. New developments in econometric methods for labor market analysis.
In: Ashenfelter, O.C., Card, D. (Eds.), Handbook of Labor Economics. North-Holland:
Amsterdam, Netherlands, pp. 1367–1397 .
Myers, M.A. , 1988. Social background and the sentencing behavior of judges. Criminology
26, 649–675 .
Myers, M.A. , Talarico, S.M. , 1986. Urban justice, rural injustice? Urbanization and its
eects on sentencing. Criminology 24, 367–391 .
Negasa, T., Mitiku, A., Bekele, Y., 2020. Determinants and level of smallholders’ anchote
market participation in Gimbi district of southwestern Ethiopia: Heckman two stage
analysis. Agric. Socio Econ. J. 20 (2), 159–166. doi: 10.21776/ub.agrise.2020.020.2.8 .
Omiti, J. , Otieno, D. , Nyanamba, T. , McCullough,
E. , 2009. Factors inuencing the inten-
sity of market participation by smallholder farmers: a case study of rural and peri-ur-
ban areas of Kenya. Afr. J. Agric. Resour. Econ. 3, 57–82 .
Ouma, E. , Jagwe, J. , Obare, G.A. , Abele, S. , 2010. Determinants of smallholder farmers’
participation in banana markets in central Africa: the role of transaction costs. Agric.
Econ. 41, 111–122 .
Pingali, P. , 2007. Agricultural growth and economic development: a view through the
globalization lens. Agric. Econ. 37, 1–12 .
Rabby, T.G. , Alam, G.M. , Mishra, P.K. , Hoque, K.E. , Nair, S.
, 2011. Dierent eco-
nomic and policy perspectives in micro population for sustainable development:
a study of the Haor livelihood in Bangladesh. Afr. J. Bus. Manag. 5 (6), 2475–
2492 .
Rahman, M.C., Pede, V., Balie, J., Pabuayon, I.M., M.Yorobe, J., Mohanty, S., 2020.
Assessing the market power of millers and
wholesalers in the Bangladesh rice sec-
tor. J. Agribus. Dev. Emerg. Econ. 11 (3), 280–295. doi: 10.1108/JADEE-04-2018-
0053 .
Rahman, M.A., Kang, S., Nagabhatla, N., et al., 2017. Impacts of temperature and rainfall
variation on rice productivity in major ecosystems of Bangladesh. Agric. Food Secur.
6 (10). doi: 10.1186/s40066-017-0089-5 .
Randela, R., Alemu, Z.G., Groenewald, J.A., 2008. Factors enhancing mar-
ket participation by small-scale cotton farmers. Agrekon 47 (4), 451–469.
doi: 10.1080/03031853.2008.9523810 .
Reardon, T. , Barrett, C.B. , 2000. Agro-industrialization, globalization, and international
development: an overview of issues, patterns, and determinants. Agric. Econ. 23,
195–205 .
Rehima, M. , Dawit,
A. , 2012. Red pepper marketing in Siltie and Alaba in SNNPRS of
Ethiopia: factors aecting households’ marketed pepper. Int. Res. J. Agric. Sci. Soil
Sci. 2 (6), 261–266 .
Rios, A.R. , Masters, W.A. , Shively, G.E. , 2008. Linkages between market participation and
productivity: results from a multi-country farm household sample. In: Proceedings
of the Paper Presented at the American Agricultural Economics Association Annual
Meeting. Orlando, Florida, USA .
Ruane, A.C., Major, D.C., Yu, W.H., Alam, M., Hussain, S.G., Khan, A.S., et al., 2013. Multi-
factor impact analysis of agricultural production in Bangladesh with climate change.
Glob. Environ. Chang.
23 (1), 338–350. doi: 10.1016/j.gloenvcha.2012.09.001 .
Sala, S., Bocchi, S., 2014. Green revolution impacts in Bangladesh: exploring adapta-
tion pathways for enhancing national food security. Clim. Dev. 6 (3), 238–255.
doi: 10.1080/17565529.2014.886988 .
Sartori, A.E. , 2003. An estimator for some binary-outcome selection models without ex-
clusion restrictions. Polit. Anal. 11 (2), 111–138 .
Sayeed, K.A. , Yunus, M.M. , 2018. Rice prices and growth, and poverty reduction in
Bangladesh, Background paper to the UNCTAD-FAO commodities and development
report. Commodity Markets, Economic Growth and Development, Food and Agricul-
ture Organization of the United Nations, Rome .
Sebatta, C., Mugisha, J., Katungi,
E., Kashaaru, A., Kyomugisha, H., 2014. Smallholder
farmer’s decision and level of participation in the potato market in Uganda. Mod.
Econ. 5, 895–906. doi: 10.4236/me.2014.58082 .
Seventh Five Year Plan (7FYP), 2015. Agriculture sub-sector: crops and Horticulture, the
seventh ve year plan (2016–20), planning commission, ministry of planning, gov-
ernment of the people Republic of Bangladesh, Dhaka, 11 November 2015.
Sharma, P.K., 2010. Scenario of haor vulnerabilities and other obstacles for sustainable
livelihood development in Nikli Upazila. J. Bangladesh Agric. Univ. 8 (2), 283–290.
doi: 10.3329/jbau.v8i2.7939 .
12
M.S. Hoq, Md.T. Uddin, S.K. Raha et al. Environmental Challenges 5 (2021) 100292
Siddique, M.A.B. , Sarkar, M.A.R. , Rahman, M.C. , Chowdhury, A. , Rahaman, M.S. , Deb, L. ,
2017. Rice farmer’s technical eciency under abiotic stresses in Bangladesh. Asian J.
Agric. Rural Dev. 7 (11), 219–232 .
Singha, M., Dong, J., Zhang, G., et al., 2019. High resolution paddy rice maps in
cloud-prone Bangladesh and Northeast India using sentinel-1 data. Sci. Data 6, 26.
doi: 10.1038/s41597-019-0036-3 .
Siziba, S. , Nyikahadzoi, K. , Diagne, A. , Fatunbi, A.O. , Adekunle, A.A. , 2011. Determinants
of cereal market participation by sub-saharan Africa smallholder farmer. J. Agril. En-
viron. Stud. 2 (1), 180–193
.
Smale, M. , Diakité, L. , Keita, N. , 2012. Millet transactions in market fairs, millet diversity
and farmer welfare in Mali. Environ. Dev. Econ. 17 (4), 1–24 .
In Smith, K.R. , Woodward, A. , Campbell-Lendrum, D. , Chadee, D.D. , Honda, Y. , Liu, Q. ,
Field, C.B. , Barros, V.R. , Dokken, D.J. , Mach, K.J. , Mastrandrea, M.D. , Bilir, T.E. , et al. ,
2014. Human health: impacts, adaptation, and co-benets. In: Climate Change 2014:
Impacts, Adaptation, and vulnerability. Part a: Global and sectoral aspects. Contri-
bution of Working Group II to the Fifth Assessment
Report of the Intergovernmental
Panel on Climate Change. Cambridge: Cambridge University Press, pp. 709–754 .
Stolzenberg, R.M. , Relles, D.A. , 1997. Tools for intuition about sample selection bias and
its correction. Am. Sociol. Rev. 62, 494–507 .
Timmer, C.P. , 1974. A model of price marketing margin in Indonesia. Food Res. Inst. Stud.
13, 145–167 .
Tufa, A. , Bekele, A. , Zemedu, L. , 2014. Determinants of smallholder commercialization of
horticultural crops in Gemechis district, West Hararghe Zone, Ethiopia. Afr. J. Agri.
Res. 9 (3), 310–319 .
Uddin, M.T. , Hossain, N. , Dhar, A.R. , 2019. Business prospects and challenges in Haor
areas of Bangladesh. J. Bangladesh Agric. Univ. 17 (1), 65–72 .
UNDP, 2012. Terminal evaluation: coastal and wetland biodiversity management at Cox’s
Bazar and Hakaluki Haor. United Nations Development Program, Dhaka .
Von Braun,
J. , 1995. Agricultural commercialization: impacts on income and nutrition
and implications for policy. Food Policy 20 (3), 187–202 .
Wooldridge, J.M. , 2010. Econometric Analysis of Cross Section and Panel Data. MIT Press .
FAO & WFP, 2008. Special report FAO/WFP: crop and food supply assess-
ment mission to Bangladesh, Rome, Italy. 28 August 2008. Retrieved from
ftp://ftp.fao.org/docrep/fao/011/ai472e/ai472e00.pdf
World Bank, 2006. World development report 2007: agriculture for development, Wash-
ington D.C.
13
