Land Use Policy 109 (2021) 105601

Contents lists available at ScienceDirect

Land Use Policy

journal homepage: www.elsevier.com/locate/landusepol

An economic analysis of agricultural adaptation to climate change impacts

in Sri Lanka: An endogenous switching regression analysis
Kanesh Suresh a, b, Uttam Khanal c, Clevo Wilson a, Shunsuke Managi d, *, Annette Quayle a,
Samithamby Santhirakumar e
a
Queensland University of Technology, Brisbane, Australia
b
Eastern University, Sri Lanka
c
Agriculture Victoria, Department of Jobs, Precincts and Regions, Victoria, Australia
d
Urban Institute and Department of Civil Engineering, Kyushu University, Japan
e
South Eastern University, Sri Lanka

A R T I C L E I N F O A B S T R A C T

Jel classification: Climate change and its impact on the agricultural sector in developing economies is a matter of considerable
Q10 academic and political debate. This paper examines the impact of climate change and variability on Sri Lankan
Q15 agriculture and identifies the potential adaptation practices and their impacts on rice productivity. More spe­
Q50
cifically, this study investigates how farming households’ decisions to adapt to climate change affects agricul­
Q56
tural productivity in the Batticaloa district of Sri Lanka. The data were collected through a primary survey of 238
Keywords:
farming households. We employ a simultaneous equations model with endogenous switching to investigate the
Climate change
Adaptation
differing effects of adaptation on adapters and non-adapters. The findings show that most farmers perceive there
Agriculture are adverse climate change impacts on their agricultural production. We also find that farmers’ adaptation
Rice production measures substantially boost rice yields. Moreover, it is shown that such strategies would benefit both adapters
Sri Lanka and non-adapters. Farmers’ access to climate related information, education, membership in farmer organiza­
tions and size of plots are found to play a key role in the adaptation process. Overall, this study provides
empirical evidence of the positive impact of adaptations on food productivity and farm income, thus suggesting
the need for policy interventions that enhance farmers uptake of strategies against climate change impacts.

1. Introduction than 2 ha who are at particular risk from the climatic change (Khanal
et al., 2021). The issue of climate change has, therefore, led to the rapid
In recent years climate change has become a major global concern growth of literature on the impacts of climate on the economy and
and in particular for developing countries where there are profound livelihood of smallholder farming communities (Balasubramanya and
effects on the agricultural sector (Wunder et al., 2018; Kurukulasuriya Stifel, 2020).
and Mendelsohn, 2007). Climate change impacts (2014–2015) on Several studies show that sensitivity to climate indicates that it is
developing countries’ agriculture sectors account for 23% ($93 billion) having and will have a negative impact on agricultural output (Rose­
of the sectors’ total losses and damage (FAO, 2015). According to the nzweig and Parry, 1994; Mendelsohn and Dinar, 1999; Tang and Hailu,
Food and Agriculture Organization (FAO), of these losses in 2017, crop 2020). That is a product of warmer future temperatures, changing
and livestock production loss driven by flood accounted for 37% and rainfall patterns and increased frequency and/or severity of extreme
drought 19% in developing countries. The relevance of these losses lies weather events all of which are forecast to reduce average crop yield
in the agriculture sector’s prominent role in these countries’ develop­ (Weerasekara et al., 2021a, 2021b; Khanal et al., 2021). This in turn
ment and its status for many as the backbone of the economy in terms of produces greater volatility in yields and poses a challenge globally in
self-subsistence, poverty alleviation and employment generation terms of food security for smallholder farmers in developing countries
(Duong and Thanh, 2019; Khanal et al., 2018; Koirala et al., 2016). (Wheeler and Von Braun, 2013). The effect of climate variability on rice
Global figures indicate that there are 475 million farmers cultivating less production in the Philippines accounted for 82.4 per cent of the total rice

* Corresponding author.
E-mail address: [email protected] (S. Managi).

https://doi.org/10.1016/j.landusepol.2021.105601
Received 25 November 2020; Received in revised form 13 May 2021; Accepted 8 June 2021
Available online 15 June 2021
0264-8377/© 2021 Elsevier Ltd. All rights reserved.
K. Suresh et al. Land Use Policy 109 (2021) 105601

losses between 1970 and 1990 (Lansigan et al., 2000). Beyond the direct new economic models that incorporate social heterogeneity and
effects on agricultural output, climate change can also increase price different policy mechanism are needed to address these issues sur­
volatility in low-income countries (Haile et al., 2017). Rural incomes, rounding climate change adaptation. This study fills a gap in the liter­
investments, and development trajectories can therefore, directly and ature by measuring the economic benefit of climate change adaptation
indirectly, suffer due to climate change (Nelson et al., 2010). by rice farmers in Sri Lanka. Understanding the needs of farmers
According to the Central Bank of Sri Lanka (CBSL) in 2018 the involved in climate change adaptation practices and its impact on their
contribution of the agriculture sector to the economy was 7.5% of GDP agricultural output plays a crucial role in food security and
and accounted for 27% of direct employment. 72 per cent of the popu­ self-sufficiency in many developing countries. However, the impact of
lation is indirectly involved in agriculture related activities for their climate change on agriculture and autonomous adaptation measures
livelihood (Gunasena, 2008). Approximately 4.1 per cent of the total taken by farmers has not been well researched in Sri Lanka. This paper
population have incomes under the official poverty line with 4.3% of the contributes to the limited quantitative research measuring the potential
rural population and 8.8% of those working on agricultural estates being benefits of adaptation and the opportunity cost of non-adaptation in
the most vulnerable communities in the agricultural sector (CBSL, smallholder farming in developing countries by employing an endoge­
2018). nous switching regression analysis.
The increased occurrence of extreme rainfall events due to climate
change, droughts, and floods, has become a common climatic feature of 2. Material and methods
Sri Lanka during recent decades. In 2009, Sri Lanka’s Disaster Man­
agement Centre (DMC) estimated that over 29 million people had been 2.1. Rice production in Sri Lanka
affected by such natural disasters over the 34 year period 1974–2007
producing an average of over 850,000 people affected each year (4.6% According to the latest statistics (DCS, 2020) Sri Lanka has sown
of the total population). The total crop area affected by these hazards is around 2986150 acres of rice land (28% of the total arable land area), a
estimated at over 800,000 ha over this period (DMC, 2009). majority of which is located in the dry zone employing approximately
The impact of climate change on food production and food security is 1.2 million farmers (70% rural). Most of Sri Lanka’s paddy (rice) culti­
evidenced by lower rice yields in Asia (FAO, 2015). Sri Lanka’s agri­ vation occurs under irrigation (major irrigation accounts for 45%; minor
culture outside the plantation sector is dominated by small landholder irrigation accounts for 25%) and only 30% is grown under rain-fed
farmers. According to the Department of Census and Statistics (DCS) Sri conditions. The total production of rice was approximately 3380,780
Lanka (2020), more than 90% of farmers in Sri Lanka have less than 2 ha metric ton (DCS, 2020). There are two cultivation seasons - Maha and
of cultivatable land. Hence, a substantial portion of their production is Yala. The Maha season accounts for most cultivation which is rainfed.
consumed domestically and output from the farms therefore has a There is a large variability in the area cultivated as there is in yield (see,
sizeable impact on domestic food security. Rice is the staple food for 20 Fig. 1). Furthermore, there is the impact of extreme weather events. For
million Sri Lankans with per capita rice consumption per person being instance, as a result of severe drought in 2017, farmers were able to
about 107 kg per annum (USDA, 2020). As much as 95% of the total cultivate only a third of the usual rice hectarage. Similarly, yield was
demand for rice is produced in the country. This has been highlighted in drastically reduced due to the dry spell.
an FAO study (2015) which identifies Sri Lanka as one of the food se­ The Batticaloa district is located in the Eastern Province of Sri Lanka
curity hotspots in the Asia-Pacific region. Large adverse impacts on where agriculture and fishing are the major livelihoods. The district is
productivity, especially among smallholders who depend on farm pro­ one of the top five rice producers in the island with cultivation in excess
ductivity for their livelihood and subsistence, can lead to a rise in of 61,321 ha annually. Most of the agricultural activities in Batticaloa
poverty levels (World Bank, 2008; Wang et al., 2014; Zhou et al., 2017). are rain-fed since the total irrigable area of the district is less than 35% of
Recent studies (Jawid and Khadjavi, 2019; Khanal et al., 2018; Di the total cultivatable extent.
Falco et al., 2011) suggest that adaptation reduces the negative impact The Sri Lankan government has been involved in agricultural
of global warming on crop yields. The adaptation mechanisms adopted development in a number of ways since independence and has partic­
by developing country farmers - such as changing crop mixes and har­ ularly targeted the smallholder farmers with generous support (e.g.,
vesting times, selection of differing crop varieties, extent of land culti­ subsidised fertilizers, seeds and free water). Furthermore, land reform
vated, and changing institutional arrangements (Bairagi et al., 2020). was undertaken in the 1970s and 1980s to assist landless people and
Reilly et al. (2003) - suggest that farmers do adopt technologies (irri­ ensure equity and efficiency for smallholder dealings. Under the 1972
gation, grain drying and production choices) that reduce yield losses. land reform act the maximum extent of land which could be under the
Nevertheless, poor countries are more vulnerable and have lower direct ownership of any person was 10 ha and such land was to be
adaptive capacities to climate change (Valizadeh et al., 2014). This exclusively used for rice cultivation. The purpose of such a policy was to
study therefore aims to contribute to improving understanding of local support productivity, employment, equity and social justice. In addition,
smallholder farmers regarding the opportunities and scope of measures
for adaptation to climate change and to developing an awareness of the
underlying factors that contribute to the successful adaptation. Several
studies reveal that while agricultural producers are exposed to the ex­
tremes of climate variability not all are vulnerable to the same degree.
Differences in vulnerabilities are closely related to a variety of social,
economic and political conditions. Hence, it is vital to investigate the
heterogeneity of climate change adaptation practices by smallholder
farmers in developing countries such as Sri Lanka.
In the past, climate’s impact on agriculture has often been over­
looked with only a few studies measuring the effects of climate change
on food production in developing countries. However, quantitative es­
timates of climate impacts have greatly improved over the past decade
(Khanal et al., 2021; Pearce et al., 1996; McCarthy et al., 2001; Men­
delsohn et al., 2006). Nevertheless, there is little known regarding the
risk consequences associated with climate adaptation by vulnerable
farming communities and their associated cost. Wider adaptation and Fig. 1. Trends in rice cultivated areas and yields for Sri Lanka.

2
K. Suresh et al. Land Use Policy 109 (2021) 105601

the government of Sri Lanka has invested in research and development 2.2. Study area
(R&D) in the agricultural sector (22 million PPP US$ in 2009) which is
overseen by the Sri Lanka Council for Agricultural Research Policy This study was conducted in the Batticaloa district of Sri Lanka (see,
(SLCARP) and 13 government, and 7 higher education agencies. How­ Fig. 2). Rice is a major cultivating crop in the district: the average annual
ever, the country has not yet achieved the desired outcome for these sown area is around 86,916 ha with average yields of 2088 kg per
policies due to the impacts of climate change (Ministry of Agriculture, hectare (DCS, 2020). The district is comprising of 14 divisional secre­
2021). Hence, there is an urgent need to better understand the impact of tariat (DS) divisions, 10 of which have rice cultivation. This study
climate change on smallholder farming in Sri Lanka. selected four DS divisions in the Batticaloa district based on the intensity
and the extent of rice production. They were Koralai Pattu South, Ear­
avurpattu, Manmunai West and Porativupattu. The four DS divisions are

Fig. 2. Map of Batticaloa district and sample DS divisions.

3
K. Suresh et al. Land Use Policy 109 (2021) 105601

located in different geographical locations and rice is the major crop Table 1
cultivated in each division. A survey was administered among 250 Variable and descriptive statistics of the sample.
farmers, producing 238 usable responses - a 95% response rate. We used Output Mean Std. Min Max
a simple random sampling method to select the survey sample and deviation
interviewed farm households on their fields. Farmers who had been Production in kg 2713.843 1017.703 1087.262 5436.313
cultivating over the past year were identified as sample respondents Inputs
prior to executing the survey questionnaires. The farm household Land in hectare 5.213 4.246 1.011 14.568
questionnaire took around 40 min on average to complete and informed Labour in man days 30.0460 23.261 10.038 86.486
Fertilizer applied in kg 192.866 51.194 108.726 444.789
consent was obtained from the respondents before administering each Seed used in kg 219.865 41.175 103.416 339.769
survey instrument. Tractor usage hours/ 9.885 13.515 1.7413 14.532
hectare
2.3. Data Age 51.64 10.749 29 78
Education (years of 8.17 3.348 4 13
schooling)
Data on the inputs and outputs used in rice production, climate Migration (dummy, 1 = 0.630 0.492
change adaptation and socio-demographic characteristics are obtained yes) if any member of
from the household survey. The selection of variables was carried out HH migrated more than
3 months.
following the available literature assessing agricultural productivity and
Off-farm work (dummy, 0.710 0.455
efficiency (Di Falco et al., 2011; Rahman and Rahman, 2009; Zhang yes=1) only HH
et al., 2011; Mishra et al., 2015; Gedara et al., 2012; Khanal et al., 2018). agriculture as income
The output variable is measured by the total value of rice production in source
Sri Lankan rupees (Rs). We collected the real volume of rice production Household size 6.82 2.798 1 11
Market distance in km 10.508 6.024 0.5 30
through the household survey. However, the output price was obtained
Training (dummy, yes=1) 0.810 0.284
through focus group discussions in each village. Therefore, our HH received
assumption is that the output price is equal for all the farmers in a given agricultural training
village. The output includes the value of both sold quantities and that within last 5 years
kept for family consumption. We collected data on actual rice produc­ Seed sources (dummy, 0.180 0.389
yes=1) own seeds
tion in kilograms. The selected inputs for this study are land area under Hybrid varieties (dummy, 0.460 0.530
rice farming measured in hectares, the number of labourers used in rice yes=1)
production quantified in man-days, the quantity of fertilizer used Varieties: number of 2.360 1.903 1 5
measured in kilograms and capital input measured in Sri Lanka Rupees varieties cultivated
Distance to extension 8.205 5.501 0.5 25
(Rs). Capital input includes the expenditure on irrigation, pesticides,
services in km
seeds sources, distance to market, training and loan facilities. As can be Plots cultivated 2.29 3.548 1 6
seen from descriptive statistics in Table 1, these input variables and the Affected by drought or 0.66 0.474
output variable have high standard deviations. In the ESR specifications, flood in the last five
we used the log transformation for these variables to reduce the extrema years (dummy, 1 = yes)
Any family members 0.69 0.219
and to curtail the effects of outliers. associated with
This data is used to inform the central part of this study which ex­ agricultural related
plores climate related variables such as the perception of climatic con­ institutions (dummy,
ditions, perceptions of the impact of climate change, the nature of 1 = yes)
Access to credit (dummy, 0.44 0.498
adaptation practices and information on climatic changes.
1 = yes)
Receives information on 0.47 0.319
3. Empirical model climate change
(dummy, 1 = yes)
3.1. Endogenous switching model
proposed a considerable number of key elements of adaptation. Deressa
We employ an endogenous switching regression (ESR) technique in
et al. (2011) find that the level of higher education of farm households
modelling the adoption of climate change adaptations and the resultant
ensured ready accessibility to information regarding technological im­
impact on rice yield under the assumption that farmers choose between
provements. Accordingly, it is found that household heads are more
adaptation and non-adaptation. The ESR model is a natural extension of
willing to practice adaptation approaches to climatic change if they have
classical experiments which allow for testing of assumptions concerning
a higher level of education (Deressa et al., 2011). As well, household size
the exogeneity of treatment effects derived from survey data. Therefore,
is found to positively affect the uptake of improved agricultural tech­
we choose the method which allows us to treat two different groups
nologies. This can be explained in terms of the increased labour force
(adaptors and non-adopters) in our study. It is assumed in this study that
provided by larger families which acts as an incentive to uptake adap­
farm household will embrace climatic change adaptation measures if
tations (Croppenstedt et al., 2003). Moreover, given large households
they derive a net gains from adoption. We follow Khanal et al. (2018) in
are typically subject to a low opportunity cost of labour, they are further
modelling the climatic change adaptation choices and the resultant ef­
incentivised to carry out adaptations (Hassan and Nhemachena, 2008).
fect on rice yields. In the first step, Mi represents a binary variable which
Climatic information received by farm extension officials has been
equals 1 for farm households which adopted at least one adaptation
shown to facilitate farmers’ decisions relating to the timing of employing
against the impact of climate change and 0 if not. α represents a vector of
innovations and farmers’ adaptive strategies for climatic risk. In this
parameters to be estimated. η represents the error term which has a
context, Masud et al. (2017) revealed the existence of a positive corre­
mean of zero. The variance represented by σ2η, encapsulates the mea­
lation between farmer information openings and the agricultural
surement errors and variables which are not observed by the researcher
extension facilities and climate change adaptation. Abdulai and Huff­
but are known to farmers. Smallholder farming and demographic attri­
man (2014) find that knowledge of the possibility of implementing
butes that affect farm households’ choices to embrace adaptation mea­
sustainable farm management practises and technologies is increased
sures to the climate resilience approaches are signified by the vector Z.
where farmers derive such information from farmers’ organizations and
Researchers studying farmers’ climatic risk adaptation have

4
K. Suresh et al. Land Use Policy 109 (2021) 105601

social networks. As well, acquisition of climate change information denoted as the difference between Eqs. (6) and (8):
raises the likelihood that farmers will use different crop varieties to
support minimising the effects of climatic risk (Deressa et al., 2011). TU = E [YKi | Mi = 0] – E [YMi | Mi = 0] = XKi (βK − βM) + (σKη −
σMη)λKi (10)
In the case of farmers who are resource-poor, where family members
take up non-farm activities the financial burden is lessened and provides
the opportunity to use productivity increasing inputs. The income from
such non-agricultural activities is found to step up adaptation including 3.2. Summary statistics
the planting of trees, adjustment of planting dates and the use irrigation.
Some scholars, however, argue that involvement in non-agricultural Table 1 presents descriptive statistics for the surveyed households.
activities can hinder involvement in cultivation activities (Abdulai and The average rice production was 2713 kg per hectare. This is largely
Huffman, 2014). An Ethiopian Nile basin study shows credit access consistent (2066 kg) with the rice yield statistics of the Department of
positively and significant affects employment of climate change strate­ Census and Statistics (2019). On average, the annual land size for sown
gies such as soil conservation activities, planting date adjustments and rice is 5 ha per farm and fertilizer usage is 229 kg. The average age of the
use of irrigation (Deressa et al., 2011). Similarly, Hassan and Nhe­ farmers is 51years and the average length of schooling was 8 years.
machena (2008) reported a positive and significant association between
access to credit and the adoption of adaptation measures.
4. Results and discussion
In this study, a wide number of household characteristics are
included. They are, family size, education, age, ownership level of farm
4.1. Perception of climate change
assets and whether involved in non-agricultural activities. Included
among agricultural related attributes are number of cultivated parcels
Based on the farm household survey data from the respondents, this
and the number of drought and flood affected parcels. For the adaptation
section briefly summarizes farmers’ perception of climate change and
to climatic risk model, YMi implies the net gain farmer i draws from
the strategies they prefer, and which have trickled down by way of
adaptation and YKi the net gain from not adapting. A small-scale
climate induced shocks. The surveyed farm households were asked
farmer’s choice of adaptation measures thus will depend on whether
questions about their observations of temperature, rainfall and natural
the net returns obtained remain larger compared to the benefits from not
disaster occurrences over a last 5 years.
executing the adaptation (Abdulai and Huffman, 2014). In other words,
It is clear that local farm households have well noticed there is a
YMi > YKi.
substantial climate variability over the course of their cultivation (see,
As noted, farmers’ perception of adaptation net benefits are not
Table 2). It is revealed from the discussion with farm households that
revealed to the researcher. What is revealed to the researcher carrying
they have been experiencing continuous unprecedently adverse weather
out the survey is farmer characteristics and adaptation attributes. The
condition (drought and floods) over recent decades. All the respondents
relevance in assessing adaptation strategies and its adoption is acquisi­
have observed there is a considerable increase in temperature which has
tion of relevant information including market and credit constraints. In
caused a negative impact on their agricultural output. The results
terms of climatic risk information, the way in which regional climatic
revealed that about 99% of the respondents observed decreasing rainfall
events are changing and how climatic risk adaptation trickles down are
over the last 5 years. Overall, most of the farmers interviewed professed
of key importance.
to have experienced highly unpredictable weather conditions which has
In contrast to the expected rice output of small holder farm adapters,
affected the full range of their farming practices.
we use an endogenous switching model in which:
The study identifies an increasing intensity of drought, shortages of
Mi* = Ziα + ηi with Mi = 1 if Mi* > 0 and 0 otherwise (1) water, pest and crop diseases as major challenges perceived by farming
households. (see, Table 3). Most depend on rain-fed agriculture for
Y = f (M, X, β) + ε (2) which the major season is Maha (north-east monsoon) for rice cultiva­
tion. Almost half of the respondents perceived increased flooding, the
Regime 1: YMi = XMiβM + εMi if Mi = 1 (3)
appearance of invasive species and soil degradation as being due to
Regime 2: YKi = XKiβK + εKi if Mi = 0 (4) climate change.

E [YMi | Mi = 1] = XMiβM + σMηλMi (5)

4.2. Adaptation practices
E [YKi | Mi = 0] = XKiβK + σKηλKi (6)
The results of Table 4 show that most farm households (60%) culti­
From the above equations, the anticipated output of the adapter - if vate flood and drought tolerant varieties and use pest-resistant varieties
the farm household had decided not to involve in an adapting process -
and the anticipated output of the non-adapter should the farm house­
Table 2
hold have decided to take on the adaptation process – can be presented Perception of climatic condition.
follows:
Perception of climatic Increase (%) Decrease (%) Stayed the
E [YKi | Mi = 1] = XMiβK + σKηλMi (7) condition n = 238 n = 238 same (%)
n¼ 238
and. Temperature during 100% – –
summer season
E [YMi | Mi = 0] = XKiβM + σMηλKi (8) Temperature during 100% – –
winter season
The difference between Eqs. (5) and (7) is the variation in rice yield Length of the summer 100% – –
due to climate change adaption outcomes which indicates the mean season
treatment effect (TT): Length of the winter 97.1% 1.3% 1.7%
season
TT = E [YMi | Ai = 1] – E [YMi | Mi = 1] = XMi (βM − βK) + (σMη − Precipitation (rainfall) 1.3% 36.4% 62.3%
σKη)λMi (9) Frequency (and intensity) 18.4% 40.2% 41.4%
of heavy rainfall
In the same way, the effect of the treatment (adaptation) on the Unpredictability of 96.2% 1.3% 0.8%
weather
untreated (non-adapted) (TU) for households not adapting can be

5
K. Suresh et al. Land Use Policy 109 (2021) 105601

Table 3 Table 5
Perception of the impacts of climate change. Endogenous switching regression results for adaptation and impact of adapta­
Type of impacts % of the sample population affected
tion in rice yield.
n¼ 238 Variables Adaptation n¼ 238 Rice yield (log) n¼ 238
Increase in intensity and duration of 66.9% Adapters Non-adapters
drought
Shortage of irrigation water 66.1% Land (log) -0.183 0.510c 0.441a
Increase flooding 55.6% (0.201) (0.193) (0.231)
Crop pests and diseases 65.7% Labour (log) -0.194 0.622c 0.581a
Introduction of invasive species/more 51.5% (0.281) (0.258) (0.320)
weeds Fertilizer (log) -0.073 0.133 0.090
Soil degradation 47.7% (0.158) (0.159) (0.173)
Yield decline 31.8% Seed (log) 0.190 0.374a 0.253
(0.179) (0.203) (0.173)
Age -0.009 -0.005 -0.001
(0.008) (0.008) (0.0112)
Table 4 Education 0.055a 0.005 0.037
Farmers’ preferences for various alternative adaptation practices. (0.028) (0.026) (0.038)
Family size 0.012 0.033b 0.064
Adaptation practices % of households adapted n¼ 238 (0.035) (0.014) (0.040)
Extension -0.021 0.003 -0.002
Grow diverse varieties 41.8%
(0.015) (0.016) (0.017)
Grow drought tolerant varieties 62.9%
Rice plots 0.058b -0.028 -0.248a
Grow flood tolerant varieties 64.0%
(0.023) (0.020) (0.148)
Grow disease/pest resistant varieties 59.4%
Drought/flood 0.184c -0.136 -0.095
Grow short duration varieties 37.2%
(0.044) (0.177) (0.216)
Grow less water intensive varieties 9.6%
Off-farm income -0.229 0.272 0.505c
Change planting location of varieties 3.3%
(0.210) (0.186) (0.189)
Change planting/harvesting time 0.4%
Membership 0.430c 0.368c 0.439
Change planting methods 0.4%
(0.146) (0.097) (0.712)
Increase/improve irrigation 0.4%
Loan 0.101 -0.102 0.138
Increase rainwater harvesting 0.4%
(0.172) (0.164) (0.205)
Reduce tillage 19.2%
Climate information 0.438c
Increase number of weedings 50.6%
(0.212)
Improve/increase farmyard manure use 19.7%
Constant 4.010c 2.713b 3.011c
Use more pesticides 39.7%
(1.324) (1.014) (0.747)
Switching to non-rice crops 2.9%
σ 1.063c 0.854c
(0.078) (0.076)
ρ -0.866c -0.107
to better adapt to changes in climate. This is confirmed by other studies (0.043) (0.591)
which show the increasing need to undertake natural disaster adaptation
measures (for example, see, Weerasekara et al., 2021a, 2021b; Zhang Endogenous switching regression model
Number of obs = 238
and Managi, 2020). Surprisingly, relatively few farm households were
Wald chi2 (13) = 95.09
involved in changing planting and harvesting times, cultivation
Log likelihood = - 441.65928
methods, rainwater harvesting and improving irrigation. Only 9.6% of Prob > chi2 = 0.0000
farm households grew less water-intensive varieties - a very low per­ a
Significant at the 10% level;
centage compared to other counties such as Bangladesh and Nepal. The b
Significant at the 5% level;
most interesting finding was that only 2.9% of farm households were c
Significant at the 1% level
involved in crop switching to cope with climate change. Thus, these
findings broadly indicate that Sri Lankan farmers were disinclined to adapt new technologies. Growing different varieties is shown to depend
change their traditional cropping patterns (which they had been prac­ on plots size. Farmers who have large sized plots are willing to grow
ticing for generations) to address the effects of climate change. different varieties and tend to adapt faster than those with only a few
Table 5 shows the estimates of the endogenous switching regression smaller plots. This result is consistent with Xin and Tao (2020) study
model estimated by full information maximum likelihood (see, Di Falco which showed higher outputs were being produced by combating
et al., 2011). The second, third and fourth columns represent the esti­ climate risk in China through rotation and/or intercropping.
mated coefficients of the selection Eq. (1) on adapting or not adopting to Natural disasters (flood/drought) are positively correlated with
climate change in rice production activities for farm households which adaptation to climate change. When disaster occurrences increase over
adapted or did not adapt to climate change respectively. time, that leads to a greater use of climate change adaptation strategies
It can be seen from our findings (see, Table 5) that education has a among farm households. This research supports the findings of Khanal
positive and significant impact on adaptation to climate change. That is, et al. (2018) that farmers affected by flood and drought are more likely
households with a higher level of education are shown to be more likely to employ climate adaptation strategies than households not affected. In
to adopt adaptation practices in response to climate change. Similar addition, farm households which were a member of an organization
findings were also reported by Deressa et al. (2011), Rahman and Alam were found to be more likely to adapt to climate change. In the country
(2016) and Khanal et al. (2018). It can be expected then that education specific context of Sri Lanka, several state organizations and NGOs
plays a major role in modernizing farming practices and in the adoption provide farmer organizations with training and awareness programmes
of new climate strategies such as crop switching, cultivating more on better farm practices and adaptation to climate change.
water-resistant varieties and changing cultivation periods. As can be expected, information on climate change has a positive and
Another important finding is that households which own a larger significant impact on climate change adaptation. Underlying this
number of rice plots were found to be more likely to adapt through the assumption is an acknowledgment that appropriate information on the
use of different practices in response to climate change. This finding is farming process is vital throughout the production year. Our finding is
consistent with several studies (Fosu-Mensah et al., 2012; Khanal et al., consistent with the study of Menghistu et al. (2020) who found that
2018). The number of plots cultivated also influences the decision to access information is a key factor in any adaptation process. Rice

6
K. Suresh et al. Land Use Policy 109 (2021) 105601

farming in particular needs a considerable amount of water before Table 7

cultivation and during the growing season. Hence, information on Impact of adaptation in monetary terms.
temperature, rainfall and cultivation season could minimize the nega­ Sub sample Decision stage Treatment Total
tive impacts on productivity. n¼ 238 effects opportunity
To adapt Not to adapt
The covariance term ρ for the adapters is statistically significant. This cost
indicates that self-selection occurs in adaptation. Thus, adaptation to Household RS. 141, RS. RS. 25,939. RS. 6,
climate change may not have the same effect on the non-adapters if they which 799.68 115,860.285 395 199,515.40
choose to adapt. Furthermore, the differences in the coefficients of the adopted US $ 894.72 $731.05 US $ 163.67 US $ 39117.36
Household RS. RS. RS. RS.
rice production equation between farm households that adopted and did which did 26,154.035 117,492.21 8661.825 2070,176.18
not adapt illustrate the presence of heterogeneity in the sample [Table 5, not adopt US $796 US $ 741.35 US $ 54.65 US $ 13062.28
column (3) and (4)]. Inputs such as land, labour and seeds are signifi­
cantly associated with an increase in rice yields of farm households that
adapt to climate change (Kabubo-Mariara, and Mulwa, 2019). However, adapt to climate change perform better than the non-adapters in terms of
labour and off-farm income appear to significantly affect rice yields of average rice productivity which is significantly higher than non-
farm households that did not adapt. Both the estimated coefficients of adapters. A number of studies have highlighted that tropical regions
the correlation term ρ are not significantly different from zero (see, may face severe food shortages due to a climate change induced
Table 5). This indicates that if farmers receive climatic information, they reduction of yields. This would cause a rise in poverty levels given the
are more likely to adapt their practices to such information. lack of livelihood opportunities for those involved in the increasingly
vulnerable farming sector. Adaptation to climate change is therefore
timely - delaying this process is shown to result in very significant
4.3. Treatment effects of the impact of climate change adaptation
financial losses to farmers and lead to food insecurity. The adverse
impact of climate change on agricultural production could, however, be
The most striking result to emerge from endogenous switching
minimized by applying suitable adaptation strategies.
regression-based treatment effects (see, Table 6) is that farmers adopting
The study also analyses the various adaptation practices used by
climate adjustment measures have a positive and significant impact on
smallholder farmers. It indicates that there is no single tool which can be
rice yields. The expected quantity produced per hectare by farm
used for climate change adaptation. Instead, a holistic approach in the
households that adopted is about 3151 kg per hectare but only 2611 kg
adaptation process is shown to be the key to any successful climate
for those who did not consider adaptation. Hence, climate change
change adaptation. This study reveals that education is one of the major
adaptation increased the expected rice yield by 576 kg. Whereas farmers
drivers in mitigating climate impacts. Other drivers include provision of
who did not adopt improvements through the adaptation process would
information on climate change and membership of a farming organi­
have increased outputs by only 192 kg.
zation. Farmers associated with farming organization are shown to be
more likely to adopt practices in response to climate change. Heightened
4.4. Financial benefits of adaptation local perceptions regarding climate change derived from active partic­
ipation in such organizations would therefore provide better outcomes
It is interesting to note the effect of adaptation in monetary terms. for farmers in terms of higher productivity.
Table 7 show that farmers involved in the adaptation process receive An effective climate change adaptation process depends on a number
significant financial benefits (US$ 39,117) compared with farmers who of factors: effective institutions and governance, innovation, investment
did not (US$ 13,062) per annum. This study therefore confirms that in infrastructure and practicing a sustainable livelihood. These represent
there is a cost (opportunity cost) farmer have to pay if they do not adopt the central pillars of a successful regime of adaptation to climate change.
new practices in response to climate change. The findings have signifi­ Adaptation options for agriculture include technological support in
cant implications for policymakers given the magnitude of overall pro­ modern farming practices, enhancing smallholder farmers’ access to
duction losses due to climate factors and the lack of awareness of credit and improving market access and the profit level. The most
smallholder farmers in Sri Lanka and other developing countries. common phenomenon in many developing countries is a lack of access
to credit for cultivation (Linh et al., 2019; Latawiec et al., 2017; Seck
5. Conclusion and policy implications et al., 2010). Financial institutions have imposed a number of re­
strictions which reduce access to credit to farmers in developing coun­
This paper examines the nature of smallholder farmers’ actions to tries (Reardon et al., 2009; Ukwuaba et al., 2021). Farming households
combat the adverse impacts of climate change. In doing so, we assess in such countries can, therefore, face a severe debt trap due to adverse
how expected changes in climate are likely to affect rice farming and rice climatic condition. Hence, government and financial institutions should
yields. Climate variability impacts on agricultural production globally consider financial policies which ensure loans to farmers are not usuri­
and at the micro-level of farm households - impacts which have been ous. Farmers’ financial losses in developing countries can represent a
observed in terms of reduced agricultural output. Our study shows that a significant percentage of their profits due to weak institutional ar­
majority of smallholder farmers perceive that climate factors signifi­ rangements and excessive profits derived by middlemen in the
cantly impact their agricultural outputs. We show that farmers who

Table 6
Impact of adaptation on expected rice yield.
Sub-samples n¼ 238 Decision stage Treatment effects

To adopt Not to adopt

Household which adapted (a) 3151.104 Kg. (114.772) (c) 2574.673 Kg. (103.213) TT=576.431*** Kg. (133.376)
Household which did not adapt (d) 2803.423 Kg. (175.74) (b) 2610.938 Kg. (126.407) TU=212.485*** Kg. (118.461)

Standard errors in parenthesis. ***Significant at the 1% level

(a) TT: the effects of the treatment (adaptation) on the treated (farm households that adopted)

(a) TU: the effects of the treatment (adaptation) on the treated (farm households that did not adopt)

7
K. Suresh et al. Land Use Policy 109 (2021) 105601

marketing process. Jawid, A., Khadjavi, M., 2019. Adaptation to climate change in Afghanistan: evidence on
the impact of external interventions. Econ. Anal. Policy 64, 64–82.
Moreover, developing countries with a low level of economic
Kabubo-Mariara, J., Mulwa, R., 2019. Adaptation to climate change and climate
diversification depend substantially on primary agricultural products. It variability and its implications for household food security in Kenya. Food Secur. 11
is these countries in which agricultural output depends critically on (6), 1289–1304.
favourable climate conditions. From a policy perspective, it is thus vital Khanal, U., Wilson, C., Hoang, V.N., Lee, B., 2018. Farmers’ adaptation to climate
change, its determinants and impacts on rice yield in Nepal. Ecol. Econ. 144,
to pay far more attention to climate change adaptation measures which 139–147.
can make an important contribution to ensuring the implementation of Khanal, U., Wilson, C., Rahman, S., Lee, B.L., Hoang, V.N., 2021. Smallholder farmers’
the United Nations Sustainable Development Goals in developing adaptation to climate change and its potential contribution to UN’s sustainable
development goals of zero hunger and no poverty. J. Clean. Prod. 281, 124999.
countries. In particular, supporting sustainability and capabilities of Koirala, K.H., Mishra, A., Mohanty, S., 2016. Impact of land ownership on productivity
smallholder farmers through capacity building, providing crop insur­ and efficiency of rice farmers: the case of the Philippines. Land Use Policy 50,
ance and continue monitoring and guidance, would minimize the shocks 371–378.
Kurukulasuriya, P., & Mendelsohn, R. (2007). A Ricardian analysis of the impact of
of climate change. climate change on African cropland. The World Bank. Retrieved July 2019 from
https://elibrary.worldbank.org/doi/pdf/10.1596/1813–9450-4305 on 10.08.2020.
CRediT authorship contribution statement Lansigan, F.P., De los Santos, W.L., Coladilla, J.O., 2000. Agronomic impacts of climate
variability on rice production in the Philippines. Agric. Ecosyst. Environ. 82 (1–3),
129–137.
Manuscript title: An economic analysis of agricultural adaptation to Latawiec, A.E., Strassburg, B.B., Silva, D., Alves-Pinto, H.N., Feltran-Barbieri, R.,
climate change impacts in Sri Lanka. All authors certify that they have Castro, A., Beduschi, F., 2017. Improving land management in Brazil: a perspective
from producers. Agric. Ecosyst. Environ. 240, 276–286.
participated sufficiently in the work to take responsibility for the con­
Linh, T.N., Long, H.T., Chi, L.V., Tam, L.T., Lebailly, P., 2019. Access to rural credit
tent, including participation in the concept, design, analysis, writing, or markets in developing countries, the case of Vietnam: a literature review.
revision of the manuscript. Furthermore, each author certifies that this Sustainability 11 (5), 1468.
material or similar material has not been and will not be submitted to or Masud, M.M., Azam, M.N., Mohiuddin, M., Banna, H., Akhtar, R., Alam, A.F., Begum, H.,
2017. Adaptation barriers and strategies towards climate change: challenges in the
published in any other publication before its appearance in Land Use agricultural sector. J. Clean. Prod. 156, 698–706.
Policy. Conception and design of study: KS, UK, CW, SM, AQ, SS. McCarthy, J.J., Canziani, O.F., Leary, N.A., Dokken, D.J., & White, K.S. (Eds.). (2001).
Acquisition of data: KS, UK, CW, SM, AQ, SS. Analysis and/or inter­ Climate change 2001: impacts, adaptation, and vulnerability: contribution of
Working Group II to the third assessment report of the Intergovernmental Panel on
pretation of data: KS, UK, CW, SM, AQ, SS. Drafting the manuscript: KS, Climate Change (2). Cambridge University Press.
UK, CW, SM, AQ, SS. Revising the manuscript critically for important Mendelsohn, R., Dinar, A., 1999. Climate change, agriculture, and developing countries:
intellectual content: KS, UK, CW, SM, AQ, SS. does adaptation matter. World Bank Res. Obs. 14 (2), 277–293.
Mendelsohn, R., Dinar, A., Williams, L., 2006. The distributional impact of climate
change on rich and poor countries. Environ. Dev. Econ. 11 (2), 159–178.
References Menghistu, H.T., Abraha, A.Z., Tesfay, G., Mawcha, G.T., 2020. Determinant factors of
climate change adaptation by pastoral/agro-pastoral communities and smallholder
Abdulai, A., Huffman, W., 2014. The adoption and impact of soil and water conservation farmers in sub-Saharan Africa. Int. J. Clim. Change Strateg. Manag. 12 (3), 305–321.
technology: an endogenous switching regression application. Land Econ. 90 (1), Ministry of Agriculture (2021), National Agriculture Policy of Sri Lanka, MOE. Retrieved
26–43. from http://www.agrimin.gov.lk/web/images/pdf/Policy/2021.03.27%20-%
Bairagi, S., Mishra, A.K., Durand-Morat, A., 2020. Climate risk management strategies 20NAP/NAP%20Web%20Version%20-%2026%20March%202021.pdf. on
and food security: evidence from Cambodian rice farmers. Food Policy 95, 101935. 20.04.2021.
Balasubramanya, S., Stifel, D., 2020. Viewpoint: water, agriculture & poverty in an era of Mishra, V., Ganguly, A.R., Nijssen, B., Lettenmaier, D.P., 2015. Changes in observed
climate change: Why do we know so little? Food Policy 93, 101905. climate extremes in global urban areas. Environ. Res. Lett. 10 (2), 024005.
CBSL, Annual Report, (2018). Central Bank of Sri Lanka. Retrieved July 2019 from Nelson, R., Kokic, P., Crimp, S., Martin, P., Meinke, H., Howden, S.M., Nidumolu, U.,
〈https://www.cbsl.gov.lk/en/publications/economic-and-financial-reports/annual 2010. The vulnerability of Australian rural communities to climate variability and
-reports/annual-report-2018〉. change: Part II—Integrating impacts with adaptive capacity. Environ. Sci. Policy 13
Croppenstedt, A., Demeke, M., Meschi, M.M., 2003. Technology adoption in the presence (1), 18–27.
of constraints: the case of fertilizer demand in Ethiopia. Rev. Dev. Econ. 7 (1), 58–70. Pearce, D.W., Cline, W.R., Achanta, A.N., Fankhauser, S., Pachauri, R.K., Tol, R.S.,
Department of Census and Statistics, Sri Lanka (2020), Paddy extent sown and harvested, Vellinga, P., 1996. The social costs of climate change: greenhouse damage and the
average yield and production by district, 2020, Department of Census and Statistics, benefits of control. Climate change 1995: Economic and Social Dimensions of
DCS, Colombo, Sri Lanka. Retrieved from 〈http://www.statistics.gov.lk/Agricultur Climate Change, 179–224. Cambridge University, Press, United Kingdom.
e/StaticalInformation/rubpaddy〉. Rahman, M., Alam, K., 2016. Forest dependent indigenous communities’ perception and
Deressa, T.T., Hassan, R.M., Ringler, C., 2011. Perception of and adaptation to climate adaptation to climate change through local knowledge in the protected area—A
change by farmers in the Nile basin of Ethiopia. J. Agric. Sci. 149 (1), 23–31. Bangladesh case study. Climate 4 (1), 12. https://doi.org/10.3390/cli4010012.
Disaster Management Centre (2009). Sri Lanka National Report on Disaster Risk, Poverty Rahman, S., Rahman, M., 2009. Impact of land fragmentation and resource ownership on
and Human Development Relationship, DMC, United Nation development productivity and efficiency: the case of rice producers in Bangladesh. Land Use
Programme (UNDP) in Sri Lanka. 〈https://www.preventionweb.net/english/hyogo/ Policy 26 (1), 95–103.
gar/background-papers/documents/Chap3/Asia-overview/Sri-Lanka-DRAFT-mar Reardon, T., Barrett, C.B., Berdegué, J.A., Swinnen, J.F., 2009. Agrifood industry
ch-09.pdf〉. transformation and small farmers in developing countries. World Dev. 37 (11),
Di Falco, S., Veronesi, M., Yesuf, M., 2011. Does adaptation to climate change provide 1717–1727.
food security? A micro-perspective from Ethiopia. Am. J. Agric. Econ. 93 (3), Reilly, J., Tubiello, F., McCarl, B., Abler, D., Darwin, R., Fuglie, K., Mearns, L., 2003. US
829–846. agriculture and climate change: new results. Clim. Change 57 (1–2), 43–67.
Food Agriculture Organisation (2015). Regional Overview of Food Insecurity Asia and Rosenzweig, C., Parry, M.L., 1994. Potential impact of climate change on world food
the Pacific, Food and Agricultural organizations of United Nations FAO. Retrieved supply. Nature 367 (6459), 133–138.
July 2019 from 〈http://www.fao.org/3/CA0950EN/ca0950en.pdf〉. Seck, P.A., Tollens, E., Wopereis, M.C., Diagne, A., Bamba, I., 2010. Rising trends and
Duong, P. B, Thanh, P.T, 2019. Adoption and effects of modern rice varieties in Vietnam: variability of rice prices: threats and opportunities for sub-Saharan Africa. Food
Micro-econometric analysis of household surveys. Econ. Anal. Policy 64, 282–292. Policy 35 (5), 403–411.
Fosu-Mensah, B.Y., Vlek, P.L., MacCarthy, D.S., 2012. Farmers’ perception and Tang, K., Hailu, A., 2020. Smallholder farms’ adaptation to the impacts of climate
adaptation to climate change: a case study of Sekyedumase district in Ghana. change: evidence from China’s Loess Plateau. Land Use Policy 91, 104353.
Environ. Dev. Sustain. 14 (4), 495–505. Ukwuaba, Ikenna Charles, Zechariahs Benapugha Owutuamor, and Cynthia C. Ogbu.
Gedara, K.M., Wilson, C., Pascoe, S., Robinson, T., 2012. Factors affecting technical "Assessment of agricultural credit sources and accessibility in Nigeria." Review of
efficiency of rice farmers in village reservoir irrigation systems of Sri Lanka. J. Agric. Agricultural and Applied Economics (RAAE) 23.1340–2021-010 (2021): 3–11.
Econ. 63 (3), 627–638. United States Department of Agriculture (2020), Grain and Feed Annual Report, 2020
Gunasena, H.P.M., 2008. Strategies for increasing food production in Sri Lanka: some USDA, Global Agricultural Information Network, New Delhi pp 2–16. Retrieved from
suggestions for government’s food production drive. Econ. Rev. (Sri Lanka) 34 https://apps.fas.usda.gov/newgainapi/api/Report/DownloadReportByFileName?
(7&8), 28–31. fileName=Grain%20and%20Feed%20Annual_New%20Delhi_Sri%20Lanka_03–27-
Haile, M.G., Wossen, T., Tesfaye, K., von Braun, J., 2017. Impact of climate change, 2020 on 26.04.2021.
weather extremes, and price risk on global food supply. Econ. Disasters Clim. Change Valizadeh, J., Ziaei, S.M., Mazloumzadeh, S.M., 2014. Assessing climate change impacts
1 (1), 55–75. on wheat production (a case study). J. Saudi Soc. Agric. Sci. 13 (2), 107–115.
Hassan, R.M., Nhemachena, C., 2008. Determinants of African farmers’ strategies for Wang, J., Huang, J., Zhang, L., Li, Y., 2014. Impacts of climate change on net crop
adapting to climate change: multinomial choice analysis. Afr. J. Agric. Resour. Econ. revenue in North and South China. China Agric. Econ. Rev. 6 (3), 358–378.
2 (1), 83–104.

8
K. Suresh et al. Land Use Policy 109 (2021) 105601

Weerasekara, S., Wilson, C., Lee, B., Hoang, V.N., 2021a. Impact of natural disasters on Wunder, S., Noack, F., Angelsen, A., 2018. Climate, crops, and forests: a pan-tropical
the efficiency of agricultural production: an exemplar from rice farming in Sri Lanka. analysis of household income generation. Environ. Dev. Econ. 23 (3), 279–297.
Clim. Dev. 1–14. Xin, Y., Tao, F., 2020. Developing climate-smart agricultural systems in the North China
Weerasekara, S., Wilson, C., Lee, B., Hoang, V.N., Managi, S., Rajapaksa, D., 2021b. The Plain. Agric. Ecosyst. Environ. 291, 106791.
impacts of climate induced disasters on the economy: winners and losers in Sri Zhang, D., Managi, S., 2020. Financial development, natural disasters, and economics of
Lanka. Ecol. Econ. 185, 107043. the Pacific small island states. Econ. Anal. Policy 66, 168–181.
Wheeler, T., Von Braun, J., 2013. Climate change impacts on global food security. Zhou, D., Shah, T., Ali, S., Ahmad, W., Din, I.U., Ilyas, A., 2017. Factors affecting
Science 341 (6145), 508–513. household food security in rural northern hinterland of Pakistan. J. Saudi Soc. Agric.
World Bank, World Development Report (2008). Agriculture for Development. Retrieved Sci. 18 (2), 201–210.
from 〈https://openknowledge.worldbank.org/bitstream/handle/10986/5990/WDR
%202008%20-%20English.pdf?sequence=3&isAllowed=y on 02.06.2020〉.