climate change adaptation: factors influencing chinese smallholder farmers' perceived...
TRANSCRIPT
ORIGINAL ARTICLE
Climate change adaptation: factors influencing Chinesesmallholder farmers’ perceived self-efficacy and adaptation intent
Morey Burnham1• Zhao Ma2
Received: 2 June 2015 / Accepted: 20 April 2016
� Springer-Verlag Berlin Heidelberg 2016
Abstract Understanding how individuals perceive their
ability to adapt to climate change is critical to under-
standing adaptation decision-making. Drawing on a survey
of 483 smallholder farmer households in the Loess Plateau
region of China, we examine the factors that shape
smallholder farmer perceptions of their ability to adapt to
climate change and their stated intent to do so. We apply a
proportional odds ordered logistic regression model to
identify the role that determinants of adaptive capacity play
in shaping smallholders’ perceived self-efficacy and
adaptation intent. Our study provides further evidence that
self-efficacy beliefs are a strong, positive predictor of
adaptation intent. Our study suggests that human capital,
information and technology, material resources and
infrastructure, wealth and financial capital, and institutions
and entitlements all play an important role in shaping
smallholder perceived self-efficacy, while state-society
dependencies may reduce smallholder perceived self-effi-
cacy. In addition, our study suggests that perceiving cli-
mate change risks and impacts do not necessarily lead to an
intention to adapt. Overall, our findings highlight the
importance of incorporating both the objective determi-
nants of smallholders’ adaptive capacity and their subjec-
tive perceptions of these objective determinants into future
climate change adaptation programs and policies in order
to facilitate adaptive actions. Identifying factors that cause
individuals to have a low estimation of their adaptive
ability may allow planned adaptation interventions to
address these perceived limitations and encourage adaptive
behavior.
Keywords Adaptive capacity � Climate change � Climate
perception � China � Smallholder farmers
Introduction
Smallholder farmers in the Global South are among the
most vulnerable groups to climate change (Morton 2007).
‘‘Double exposure’’ to global environmental change and
globalization (Leichenko and O’Brien 2008), and low
levels of adaptive capacity caused by on-the-ground social
conditions, such as inequality, poverty, and poor planning,
(Ribot 2009) will likely exacerbate many of the difficulties
smallholders already face. As such, researchers have
increasingly investigated the role that adaptation can play
in mitigating the impacts of climate change on smallholder
lives and livelihoods. Recent scholarly calls to shift vul-
nerability research toward the determinants of adaptive
capacity have highlighted that many of the barriers to
successful adaptation are the outcomes of social processes
(Jones and Boyd 2011). This has led to much recent
research theorizing and empirical examination of the social
determinants of adaptive capacity at various levels, from
individuals to households to countries.
Editor: Erica Smithwick.
Electronic supplementary material The online version of thisarticle (doi:10.1007/s10113-016-0975-6) contains supplementarymaterial, which is available to authorized users.
& Morey Burnham
1 Department of Environmental Studies, State University of
New York College of Environmental Science and Forestry,
106 Marshall Hall, 1 Forestry Drive, Syracuse, NY 13210,
USA
2 Department of Forestry and Natural Resources, Purdue
University, 195 Marsteller Street, West Lafayette, IN 47907,
USA
123
Reg Environ Change
DOI 10.1007/s10113-016-0975-6
In order to provide conceptual clarity, we first discuss
several key definitions. Multiple competing definitions of the
term adaptation exist (Burnham and Ma 2015). For the
purposes of this study, we define adaptation to climate
change as actions in a given social–ecological system
taken by an individual, community, or institution ‘‘in
response to actual and expected impacts of climate change in
the context of interacting non-climatic changes.’’ (Moser and
Ekstrom 2010). We define adaptive capacity as the ability of
an individual, community, or institution to prepare for and/or
adapt to stresses ex ante or react to stresses ex post (Smit and
Pilifosova 2001; Brooks and Adger 2005; Engle 2011).
Adaptive capacity is linked to the concept of vulnerability
(Adger and Vincent 2005). The vulnerability of a system is
the product of both its exposure and sensitivity to biophysical
and social stressors, perturbations, and processes, and in part
arises from its capacity to adapt (Turner et al. 2003).
Researchers have argued that understanding the constraints
to the adaptive capacity of an individual, community, or
institution can help remove barriers that prevent them from
engaging in autonomous adaptation, facilitate the develop-
ment of planned adaptation projects, and reduce the vul-
nerability of the individual, community or institution (Adger
2003; Smit and Pilifosova 2001). Following Klein et al.
(2014), we define an adaptation constraint as ‘‘a factor or
process that makes adaptation planning and implementation
more difficult’’ (p. 906). Adaptation constraints are syn-
onymous with terms ‘‘barrier’’ and ‘‘obstacle’’ that are fre-
quently used in the adaptation literature (Klein et al. 2014).
Constraints to adaptive capacity can be both objective
and subjective (Grothmann and Patt 2005). The objective
determinants of adaptive capacity include factors such as
financial capital and accesses to technology, which have
been well established and have received much attention in
the literature (Burnham and Ma 2015; Harmer and Rahman
2014). The subjective determinants of adaptive capacity
are mostly related to how individuals and communities
perceive the process of adaptation, which has been shown
to influence adaptation decision-making (Wolf et al. 2013),
and is receiving increasing attention within the scholarly
literature (e.g., Blennow and Persson 2009; Frank et al.
2011; Grothmann and Patt 2005; Kuruppu and Liverman
2011; Patt and Schroter 2008). In particular, the work of
Grothmann and Patt (2005) highlighted that individuals
with a low estimation of their adaptive capacity may be
more vulnerable to climate change than those with a high
estimation of their adaptive capacity because the low
estimation decreases the likelihood that they will engage in
adaptive behavior, thereby increasing the likelihood that
climate change will negatively impact their livelihood.
In this study, we continue the work of examining indi-
vidual climate change adaptation decision-making with a
focus on the role that identified determinants of adaptive
capacity play in shaping Chinese smallholder farmers’
perceptions of their ability to adapt to climate change, as
well as their stated intent to do so. We begin by briefly
reviewing the literature on the determinants of adaptive
capacity, paying particular attention to the role they play in
influencing the subjective dimensions of adaptive capacity.
We then provide a contextual overview of our study site,
the Loess Plateau region of China. This is followed by a
description of the methods we used to collect and analyze
data, the results of our study, and a discussion of their
significance.
Theoretical background and methodology
There is broad agreement within the literature on factors
that determine adaptive capacity, though no definitive
typology exists (Barnett et al. 2013). Smit and Pilifosova
(2001) argued that for communities, regions, or countries,
the ‘‘determinants of adaptive capacity relate to the eco-
nomic, social, institutional, and technological conditions
that facilitate or constrain the development and deployment
of adaptive measures.’’ Adger et al. (2007) listed five
general categories of barriers to adaptation: financial,
technological, cognitive, cultural, and institutional. Com-
munication and information (Moser and Ekstrom 2010);
values, beliefs, and norms (Jones and Boyd 2011; Moser
and Ekstrom 2010); and physical and ecological factors
(Jones and Boyd 2011) have also been identified as factors
influencing adaptive capacity. Further, researchers have
shown that social capital can contribute to and constrain
adaptive capacity (Adger 2003; Pelling and High 2005).
Specifically, Adger (2003) argued that adaptive capacity is
the product of access to resources, how those resources are
distributed within and between groups, and the institutions
that govern the resources. Identified social barriers to
adaptation include knowledge, emotions, and cultural fac-
tors such as place attachment and identity (Adger et al.
2013; Barnett et al. 2013). At the smallholder household
level, recent representative research has empirically
demonstrated that the ability of smallholder households to
adapt is determined by factors such as access to crop
insurance (Panda et al. 2013), the availability of credit
(Bryan et al. 2013; Hisali et al. 2011; Mertz et al. 2010;
Tambo and Abdoulaye 2013), local government and mar-
ket-based institutions (Wang et al. 2013), property own-
ership (Below et al. 2012), and access to technical
information about agricultural management and climate
change through agricultural extension services (Bryan et al.
2013; Deressa et al. 2010; Young et al. 2009). Commonly
reported constraints to adaptive capacity include lack of
land (Barbier et al. 2009; Piya et al. 2012; Tucker et al.
2010) and lack of human capital (Young et al. 2009).
M. Burnham, Z. Ma
123
Grothmann and Patt (2005) have argued that previous
research has not fully taken into account the cognitive
factors that impede individual adaptive capacity. Specifi-
cally, while it is generally acknowledged that perceived
adaptive capacity is an important component of an indi-
vidual’s decision to adapt to climate change, little research
has been done to examine the factors that play a role in
shaping it (Grothmann and Patt 2005; Kuruppu and
Liverman 2011). To address this gap, Grothmann and Patt
(2005) developed a model of private proactive adaptation
to climate change (MPPACC), and posited that subjective
determinants of adaptive capacity related to an individual’s
cognitive processes are at least as important as objective
determinants of adaptive capacity in determining a per-
son’s ability to adapt.
The model identified two ‘‘bottlenecks’’ in an individual’s
adaptation decision-making process. The first, ‘‘risk apprai-
sal,’’ consists of two components: (1) a person’s determination
of the probability that they will be ‘‘exposed to the threat’’ and
(2) a person’s determination of how much harm the threat will
do to the things they value. The second bottleneck, termed
‘‘adaptation appraisal,’’ is an individual’s assessment of the
positive and negative consequences that would result from
taking an action, as well as their ability to perform the action.
Adaptation appraisal only occurs if the individual’s appraisal
of the risk posed by climate change exceeds a minimum
threshold. The adaptation appraisal process has three com-
ponents through which a person determines: (1) whether an
adaptive action will succeed in protecting them from the threat
(i.e., ‘‘perceived adaptive efficacy’’); (2) whether they have
the ability to carry out the adaptation (i.e., ‘‘perceived self-
efficacy’’); and, (3) the costs of taking the action (i.e., ‘‘per-
ceived adaptation costs’’). According to Grothmann and Patt
(2005), perceived self-efficacy, in part, determines a person’s
perceived adaptive capacity.
Drawing on the MPPACC, Blennow and Persson (2009)
found a significant association between Swedish forest
landowners who had not adapted to climate change and
those who lacked an understanding of how to adapt or did
not believe in the efficacy of particular adaptive strategies.
Frank et al. (2011) investigated the role social identity
plays in shaping perceived self-efficacy among coffee
farmers in Chiapas, Mexico. They found that how an
individual viewed herself in terms of social group mem-
bership influenced her perceptions of the risk posed by
climate change and her adaptive capacity. Kuruppu and
Liverman (2011) examined the role of perceived self-effi-
cacy in determining the formation of intention to adapt to
climate change-induced water stress in the central Pacific
islands of Kiribati. They found that high levels of per-
ceived self-efficacy were an important driver of adaptation
intent and that a person’s belief in their own self-efficacy
may depend more on past experience with water stress than
a detailed understanding of climate impacts. Jones and
Boyd (2011) found that discrimination against Dalit and
Humli populations in Nepal and India restricted the avail-
ability and type of job opportunities available to them,
resulting in low perceived self-efficacy and limiting their
ability to cope with climate stress. Finally, Lo (2013)
demonstrated that perceptions of social norms played a
mediating role between risk perception and the adaptive
action of purchasing flood insurance in Australia, and
better explained flood insurance purchasing behavior than
did risk perception alone.
In this paper, we seek to better understand the specific
factors that constitute an individual’s perceived self-efficacy
in the MPPACC, and how this relates to the adaptation intent
of smallholders in the Loess Plateau region of China. In the
MPPACC, both objective and subjective determinants of
adaptive capacity shape an individual’s perceived self-effi-
cacy and their intention to adapt. It is important to identify
which of these determinants have the most influence on
perceived self-efficacy and adaptation intent as it may allow
for them to be systematically redressed, potentially
increasing the likelihood of adapting localized conditions to
enhance smallholders’ ability to undertake adaptive actions
on their own or participate in planned adaptation projects. To
do so, we constructed an empirical model (described in detail
in the methods section) by drawing on a set of factors relating
to both the physical elements (e.g., technology, wealth) and
social/institutional elements (e.g., human capital, institu-
tions) noted to be crucial to determining a system’s adaptive
capacity in Eakin and Lemos (2006). Their typology was
adapted from earlier work by Smit and Pilifosova (2001) and
Yohe and Tol (2002), and broadly captures the determinants
of adaptive capacity established in the literature. Their
typology has seven categories of determinants of adaptive
capacity: (1) human capital; (2) information and technology;
(3) material resources and infrastructure; (4) organization
and social capital; (5) political capital; (6) wealth and
financial capital; and (7) institutions and entitlements. In this
paper, we adapt this typology based on the data we were able
to collect from our field work. We were not able to include
the organization and social capital category nor the political
capital category. We also combined the information and
technology category and the material resources and infras-
tructure category to form a new category that consists of
material and non-material resources. Thus, using our adapted
typology we were able to measure the effects of four cate-
gories of determinants on smallholder farmers’ perceived
self-efficacy to adapt to climate change (Table 1).
We next apply our typology to determine which factors
influence the adaptation intent of smallholder farmers. Pre-
vious research on the relationship between attitudes and
behavior positions behavioral intention as playing an
important role in predicting actual behavior (Ajzen 1991;
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
Fishbein and Ajzen 1975; Triandis 1977). However, as
Grothmann and Patt (2005) noted, intention does not always
lead to realized behavior. Recent research on attitude–be-
havior relations has shown that intentions can only be
expressed as behavior when a person is able to control the
behavior (Webb and Sheeran 2006). Thus, not all stated
adaptation intentions result in actual adaptive actions, and
one of the barriers that prevents realization of a linear rela-
tionship between the two is a lack of adaptive capacity,
including factors such as access to resources, technical skill,
and social support as these affect a person’s ability to control
their behavior (Grothmann and Patt 2005; Liska 1984).
While a gap between intended and actual behavior exists,
intentions are considered a good predictor of future behavior
in unstable or changing conditions such as those brought
about by climate change (Kuruppu and Liverman 2011;
Webb and Sheeran 2006). As Kuruppu and Liverman (2011)
noted, including adaptation intention in the MPPACC pro-
vides insight into people’s commitment to adapt to climate
change. A stated intent demonstrates how much effort a
person is willing to exert in order to achieve their desired
outcome (Ajzen 1991; Webb and Sheeran 2005), thus indi-
cating how motivated they are to adapt (Ajzen 1991). And
motivation has been shown to be a critical determinant of
adaptive action (Frank et al. 2011). Thus, we examine the
role of determinants of adaptive capacity in shaping small-
holders’ adaptation intent in order to provide insight into the
cognitive processes that motivate or impede adaptive action
and how those processes are shaped by the external world.
This task is important as planned adaptations will have to
take into account the cognitive barriers to adaptive action in
order to succeed (Patt and Schroter 2008).
Study site and methods
Climate and adaptation on the Loess Plateau
The Loess Plateau region of China sits at about 37� north
latitude in the middle and upper reaches of the Yellow
River, and covers an area of approximately 250,000 square
miles across parts of five provinces. It is situated just
northwest of the East Asian monsoon (EAM) zone. The
climate is a typical continental monsoonal climate, and the
advance of the EAM each year causes the region to have
distinctive wet and dry seasons (Ding and Chan 2005). The
climate in the region ranges from arid/semiarid to sub-
humid, and averages about 140 mm of rainfall in the
northwest to 800 mm in the southeast. Approximately
70 % of annual precipitation falls between June and
September in the form of heavy storms, leaving many crops
vulnerable to drought in the early growing season (Li et al.
2012). Between 1970 and 2010, mean annual temperature
increased by 0.06 �C per year and annual precipitation
decreased by 0.51 mm per year (Zhang et al. 2012).
Drought frequency and intensity also increased across
much of the region (Zhang et al. 2013).
The Loess Plateau has been identified as one of the most
agriculturally vulnerable regions to climate change in China
(Wu et al. 2013). Researchers have predicted increases in
average annual temperature, drought frequency, and soil
erosion, as well as higher levels of precipitation in the winter
when it is less useful for agriculture and lower levels of
precipitation in the summer when it is crucial. Future climate
scenarios project that by the 2080s, annual mean temperature
in the Loess Plateau region will increase by as much as 5 �C.
Although average annual precipitation is expected to
increase by 54 mm to 150 mm (Liu et al. 2011), the evapo-
transpiration rate is expected to increase by 12 % over its
1961–2009 average (Li et al. 2012). Additionally, the
intensity of precipitation events are likely to increase (Li
et al. 2010), while runoff from the Yellow River is expected
to decrease, leading to water shortages that will be exacer-
bated by a growing population (Wang and Zhang 2011; Li
et al. 2010; Piao et al. 2010). In certain areas, the duration of
seasonal snow cover will likely be shortened, with snow
packs thawing in advance of spring onset, and runoff pos-
sibly being reduced by 20–40 % (Wang and Zhang 2011).
Despite these dire predictions, climatic and ecological
changes are not new to the Loess Plateau region (Zhao
Table 1 Determinants of adaptive capacity, adapted from Eakin and Lemos (2006)
Categories of determinants Specific factors/variables within each category
Human capital Knowledge (scientific, ‘‘local,’’ technical, political), education level, health,
individual risk perception, labor
Material and non-material resources (i.e., information,
technology, material resources, and infrastructure)
Communication networks, technology transfer and data exchange, innovation
capacity, early warning systems, technological relevance, transport, water
infrastructure, buildings, sanitation, environmental quality
Wealth and financial capital Income and wealth distribution, marginalization, accessibility and availability of
financial instruments (insurance, credit), fiscal incentives for risk management
Institutions and entitlements Informal and formal rules for resource conservation, risk management, regional
planning, participation, property rights, and risk sharing mechanisms
M. Burnham, Z. Ma
123
et al. 2013), and the farming community has long been
involved in the process of perceiving and adapting to
aridity and rainfall variability to reduce their vulnerability
(Li et al. 2013), similar to agricultural producers in many
other parts of the world (Thomas et al. 2007). Specifically,
farmers on the Loess Plateau have reported that drought
conditions and temperatures have increased since the early
1980s and precipitation has decreased in the last decade,
and these perceptions correspond with local weather station
data (Li et al. 2013; Otswald and Chen 2006). In addition,
studies have shown that smallholders attribute different
levels of risk to different types of climate change. For
instance, farmers on the Loess Plateau have perceived
changes in rainfall timing to be more important in deter-
mining their livelihood success than changes in the amount
of rainfall (Hageback et al. 2005). They have also been
shown to perceive changes in spring temperature and
growing season precipitation, but not other aspects of the
changing climate that are less important to their agricul-
tural livelihoods (Burnham et al. 2016). Compared to the
literature on smallholder climate perception, little has been
done to investigate smallholder adaptation to perceived
climate variability and change in the Loess Plateau region.
An exception is Hageback et al. (2005), which provides a
preliminary understanding of how and why smallholders
adapt to climate change and other stressors. This study
shows that smallholders in Ansai County, Shaanxi, have
become less vulnerable to climate change over the pre-
ceding 20 years because they have been able to diversify
their livelihoods and have become less dependent on
agriculture. Further, this study shows that the adaptations
that farmers have made are more likely to be driven by new
economic opportunities, such as new markets for crops, and
policy changes than they are to be by climate variability
and change. Likewise, Li et al. (2013) demonstrated that
smallholders in Ningxia have applied a range of practices
to retain or increase soil moisture to maintain agricultural
production in the face of climate variability and drought
conditions. However, their work also showed that changing
socioeconomic conditions were more important determi-
nants of livelihood trajectories than climate variability or
change.
Data collection
This study used a mixed-methods approach to determine
the suite of determinants of adaptive capacity that shape
smallholders’ perceived self-efficacy and their intent to
adapt to climate change. We combined a household survey
with qualitative interviews to allow for a sufficient level of
generalizability, while at the same time taking into account
how local social processes produce differential outcomes in
specific places (Birkenholtz 2012). Data for this study were
collected in two steps. In 2011, 28 semistructured and 38
unstructured interviews were completed in nine villages
across six townships in Shaanxi and Ningxia Provinces
(Fig. 1).
To identify interviewees, we consulted with the local
agricultural bureau in each county to identify households
for whom agriculture was an important component of their
livelihoods. We then used a snowball sampling method to
recruit additional interviewees (Noy 2008). The interviews
were designed to elicit information about general village
and farming life; changes smallholders had made to their
farm management and livelihood practices over the last
30 years and why; their perceptions of past and future
climate change and attendant risk; the major challenges and
risks they face; and their social and professional networks
and socioeconomic status. A qualitative approach is useful
for obtaining data about a little understood topic or area
(Didier and Brunson 2004), which can then be used to
inform the development of an appropriate set of questions
for a broader survey (Tremblay 1957).
In 2012, a survey of 483 smallholder households was
completed. The survey questionnaire was designed itera-
tively, drawing on theoretical insights established in a lit-
erature review and findings from the interviews conducted
in 2011. The survey was designed to collect information
about household socioeconomic characteristics, percep-
tions of past and future climate change, adaptations made
in response to climate change and other livelihood stres-
sors, livelihood challenges and risks from climatic and non-
climatic sources, the perceived impacts of future climate
change on farming and livelihoods, and factors affecting
farm management decision-making.
The survey was conducted in three counties in the Loess
Plateau region: Mizhi and Yangling, in Shaanxi, and
Hongsipu (county-level municipality) in Ningxia. Graduate
students from Northwest Agriculture and Forestry
University (NWAFU) in Yangling, Shaanxi, were trained
and employed as survey enumerators. The survey was
pretested in two villages near NWAFU, and subsequent
refinements were made. Within each county, stratified
random sampling was used to select eight villages in which
household surveys were conducted. The eight villages were
stratified according to their distance from the county pop-
ulation center to account for socioeconomic differences,
with seven villages categorized as far ([20 km), eight
villages categorized as middle ([10 km and B20 km), and
nine villages categorized as near (B10 km). Within each
village, simple random sampling was used to select 20
households and the primary agricultural decision-maker
was surveyed face-to-face in each household by an
enumerator.
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
Empirical model
To assess the role of determinants of adaptive capacity in
shaping smallholders’ perceived self-efficacy and adapta-
tion intent, we constructed two empirical models, one for
each dependent variable. One dependent variable is
EFFICACY, which represents smallholder perceived self-
efficacy. The other is INTENT, which represents small-
holder adaptation intent (Supplementary Table 1). To
measure EFFICACY, we asked respondents to indicate
their ability to make changes to their farming practices or
livelihoods to prevent damage caused by climate change
without assistance from agricultural professionals or the
government. The respondents were able to choose from
five options: (1) not possible, (2) very difficult, (3) some-
what difficult, (4) somewhat easy, and (5) very easy.
Because only five respondents indicated that adapting to
climate change would be very easy, those responses were
combined with option four in our models. To measure
INTENT, we asked respondents to indicate the likelihood
that they would make changes to their farming practices or
livelihoods to prevent damage caused by climate change
without assistance from agricultural professionals or the
government based on their current understanding of the
situation. The respondents were able to choose from four
options: (1) very unlikely, (2) somewhat unlikely, (3)
somewhat likely, and (4) very likely. The independent
variables in our empirical models were drawn from the
determinants of adaptive capacity identified by Eakin and
Lemos (2006). As discussed above, we were able to collect
data on variables in four adapted categories of objective
determinants (Table 1). A detailed description of each
independent variable is provided in Supplementary
Table 1. In the INTENT model, we included EFFICACY
as an independent variable to examine the role of small-
holders’ perception of their ability to adapt in determining
their adaptation intent.
Because both EFFICACY and INTENT are ordinal
variables, we adopted a proportional odds ordered logistic
regression model to estimate the empirical models. A
nested regression approach was also used to assess each of
the four adapted categories of objective determinants of
adaptive capacity and to identify specific determinants
within each category that affect the model outcomes (i.e.,
each category of determinants constitutes a separate block
of the nested model). Using a proportional odds ordered
logistic regression model, the probability of a smallholder
perceiving their ability or stating their intent to adapt at
level j can be written as follows (Long 1997; Rifaat et al.
2012):
P yi [ jð Þ ¼ g Xib0ð Þ ¼
exp Xib0 � sj
� �
1 þ exp Xib0 � sj
� � ;
j ¼ 1; 2. . .;M � 1
where M is the total number of levels of the dependent
variable, Xi is a vector of independent variables, b is a
Fig. 1 Map of study area
M. Burnham, Z. Ma
123
vector of logit coefficients, and sj is a cut point for being at
level j or lower versus a higher level.
A key of assumption of the proportional odds model is
that logit coefficients are equal across logit equations for
the different cut points, implying that the effect of each
independent variable on the log odds of the dependent
variable is the same regardless of which levels of the
dependent variable are being compared. This enables the
reporting of a single coefficient for each independent
variable, similar to a binary logit model (Fullerton 2009).
We used a Wald test to determine whether our data violate
this assumption (Brant 1990; Long and Freese 2006). The
overall equality of the coefficients (i.e., omnibus test) and
each individual variable were tested. No statistically sig-
nificant test statistics were returned at the 0.05 level,
indicating that the assumption was not violated. Pairwise
correlations were calculated for all independent variables
in the models to check for multicollinearity. All pairwise
correlations fell below 0.6; thus, all independent variables
were retained. The variance inflation factor (VIF) was also
calculated for both empirical models. The VIF for the full
EFFICACY model was 1.70, while the VIF for the full
INTENT model was 1.71. Both are below the commonly
used heuristic that a VIF of 10 may indicate a problem of
multicollinearity.
Principal component analysis
Three sets of survey items in the smallholder household
survey questionnaire measured various aspects of small-
holder perceptions of climatic and non-climatic sources of
risk. The first set included 15 items measuring perceived
risks and challenges from non-climatic sources to liveli-
hoods. The second set included 16 items measuring per-
ceived risks and challenges from climatic sources to
livelihoods. The third set included 10 items measuring
perceived impacts of climate change on farming and
livelihoods. High correlations among several of the survey
items within each set indicated that the data are not one-
dimensional, warranting a data reduction procedure. Prin-
cipal component analysis (PCA) is a statistical technique
that can be used to reduce a large number of correlated
variables into a smaller number of uncorrelated, composite
variables called principal components with a minimal loss
of information. The results of a PCA are usually discussed
in terms of PC loadings, which represent the correlations
between the survey items and the PCs, and are used to
define and name the PCs. An absolute PC loading of 0.50
or greater indicates a strong association among survey
items used to generate that PC, and in this study we used a
cutoff point of 0.55. We also calculated Cronbach’s alpha
for each PC to measure the internal consistency of survey
items to determine the reliability of the PC (Cronbach
1951). Generally speaking, values of 0.70 or higher indi-
cate sufficient scale reliability (Nunnally 1978), and PCs
with a Cronbach’s alpha value lower than 0.70 should be
interpreted with caution. To determine which PCs to retain,
Kaiser (1958) stated that all PCs with an eigenvalue of one
or greater should be retained.
We applied PCA to the aforementioned three sets of
survey items. The first set of survey items measuring per-
ceived risks and challenges from non-climatic sources were
reduced to four PCs (Table 2). Based on the associated item
themes, the first PC, MRKT_RISK, was defined as the
amount of perceived risk posed to livelihood from market
sources; the second PC, FARM_RISK, was defined as the
amount of perceived risk posed to livelihood from on-farm
sources; the third PC, ILL_RISK, was defined as the amount
of perceived risk posed to livelihood from illness or injury;
and the final PC, EX_RISK, was defined as the amount of
perceived risk posed to livelihood from sources outside of
household’s control. The second set of survey items mea-
suring perceived risks and challenges from climatic sources
were reduced to three PCs (Table 3). Based on the associated
item themes, the first PC, PRECIP_RISK, was defined as the
amount of perceived risk posed to livelihood from changes in
precipitation-related events; the second PC, PRE-
DICT_RISK, was defined as the amount of perceived risk
posed to livelihood from increased unpredictability of cli-
mate events; and the third PC, SPRING_RISK, was defined
as the amount of perceived risk posed to livelihood from
changes in spring climate events. Finally, the third set of
survey items measuring perceived impacts of climate change
were reduced to three PCs (Table 4). Based on the associated
item themes, the first PC, WATER_IMPACT, was defined as
the level of perceived impact climate change will have on
water availability and growing season; the second PC,
PROFIT_IMPACT, was defined as the level of perceived
impact climate change will have on the profitability of
agriculture; and the third PC, YIELD_IMPACT, was defined
as the level of perceived impact climate change will have on
crop yields. All but two PCs met the suggested Cronbach’s
Alpha minimum of 0.70, indicating sufficient levels of
internal consistency and scale reliability.
Results
Profile of smallholder respondents and their farms
The mean age of our respondents was 51, and the average
respondent completed five years of school. Respondents in
Yangling earned an average household income of 23,717
Yuan,1 with 37 % of it derived from agricultural activities
1 6.47 Chinese Yuan is equal to 1 U.S. dollar.
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
(Table 5). In Mizhi, the average household income was
11,261 Yuan, with 66 % coming from agriculture. In
Hongsipu, the average household income was 22,233
Yuan, with 68 % coming from agricultural activities.
Across respondent households, working off-farm, whether
through labor migration or local wage work, played an
important role in household livelihood strategies. Eighty-
six percent of smallholding households in Yangling
engaged in some form of off-farm work, while 54 % did in
Mizhi and 61 % in Hongsipu. Farm sizes across the three
counties were relatively small. On average, participants
had usufruct rights to 8.3 mu (5666 m2) of land, divided
across three non-contiguous plots. Farm labor was pro-
vided by an average of 2.3 household members.
Generally speaking, respondents in each of the three
counties were dependent on agriculture to both provide
income and supply family food needs, though to varying
degrees. In Mizhi, 58 % of respondents identified them-
selves as employing a diversified subsistence/market
strategy, and only 8 % farmed solely for market purposes.
In Yangling, 51 % of respondents employed a mixed
strategy while 31 % farmed solely for market purposes. In
Hongsipu, 25 % employed a mixed strategy while 63 %
farmed solely for market purposes. Maize was a major crop
across the three counties, with 83 % of respondents
growing it. In response to locally led initiatives to alleviate
poverty, respondents in Yangling and Hongsipu have
started to slowly move into kiwi (6 %) and goji (11 %)
production. Aggregated across the three counties, 33 % of
respondents grew vegetables or fruit as their primary cash
crops, while 64 % of respondents raised livestock, pri-
marily chickens for personal use and pigs for market pur-
poses. In Hongsipu, respondents also raised sheep for sale
in times of financial emergency.
Access to irrigation water was differentially distributed
across the three counties. In Hongsipu, 100 % of respon-
dent households had irrigated land, while 79 % in Yangling
and 41 % in Mizhi did. Of the respondents with irrigated
land, 48 % reported using an irrigation schedule. Nearly all
households had access to and used some basic technolo-
gies, such as pesticides (94 %), fertilizer (91 %), and store-
bought seeds (90 %). More expensive technologies such as
plastic mulching (37 %), greenhouses (7 %), and various
types of water saving irrigation (e.g., drip irrigation) (7 %)
were less prevalent.
Perceived self-efficacy
Respondents were asked to describe their ability to make
changes to their farming practices or livelihood to prevent
damage caused by climate change without information or
assistance from agricultural professionals or the
Table 2 Description of survey items measuring smallholder perceptions of risk and challenge to livelihood from non-climatic sources
Survey items: perceived risk and challenge from non-climatic
sourcesaRotated principal component loadingb Cronbach’s
alphaMRKT_RISKc FARM_RISKd ILL_RISKe EX_RISKf
High cost of farming inputs 0.66 0.90
Low income from farming 0.73
Unpredictable crop markets 0.85
Lack of available markets to sell crops 0.75
Low or unfair crop prices from crop buyers 0.87
Low market prices 0.88
Pests and diseases 0.62 0.73
Not enough water to irrigate sufficiently 0.70
Lack of mechanized tools 0.67
Lack of access to technical help 0.63
Not enough labor to farm properly 0.82 0.75
Personal illness or injury 0.82
Land being taken away (i.e., lack of tenure security) 0.80 0.61
Low-quality seeds 0.65
Polluted irrigation water 0.71
a Item scale: 1, not a difficulty; 2, slight difficulty; 3, moderate difficulty; 4, significant difficultyb Blanks represent rotated PC loadings\0.55c MRKT_RISK was defined as the amount of perceived risk posed to livelihood from market sourcesd FARM_RISK was defined as the amount of perceived risk posed to livelihood from on-farm sourcese ILL_RISK was defined as the amount of perceived risk posed to livelihood from illness or injuryf EX_RISK was defined as the amount of perceived risk posed to livelihood from sources outside of household’s control
M. Burnham, Z. Ma
123
Table 3 Description of survey items measuring smallholder perceptions of risk and challenge to livelihood from climatic sources
Survey items: perceived risk and challenge from climatic sourcesa Rotated principal component loadingb Cronbach’s
alphaPRECIP_RISKc PREDICT_RISKd SPRING_RISKe
Decreased length of rainy season 0.83 0.93
Less rain during rainy season 0.82
Increased rainfall intensity during rainy season 0.67
Increased spring temperature 0.78
Increased summer temperature 0.76
Increased winter temperature 0.79
Increased morning/afternoon temperature difference during growing
season
0.68
Increased drought 0.56
Change in the timing of 24 solar terms 0.60 0.88
Unpredictable rainfall 0.85
Unpredictable drought 0.86
Unpredictable temperature 0.85
Spring arriving late 0.63
Longer time for frozen ground to thaw in the spring 0.60 0.83
Increased spring frosts 0.90
Increased spring cold spells 0.90
a Item scale: 1, not a difficulty; 2, slight difficulty; 3, moderate difficulty; 4, significant difficultyb Blanks represent rotated PC loadings\0.55c PRECIP_RISK was defined as the amount of perceived risk posed to livelihood from changes in precipitation-related eventsd PREDICT_RISK was defined as the amount of perceived risk posed to livelihood from increased unpredictability of climate eventse SPRING_RISK was defined as the amount of perceived risk posed to livelihood from changes in spring time climate events
Table 4 Description of survey items measuring smallholder perceptions of impacts of climate change on farming and livelihood
Survey items: perceived impact of climate change on the
respondent’s farmaRotated principal component loadingb Cronbach’s
alphaWATER_IMPACTc PROFIT_IMPACTd YIELD_IMPACTe
My crops will need more water 0.59 0.79
The timing of when I need to irrigate my crops will change 0.62
Less water will be available for irrigation 0.62
The amount of time during which I can harvest my crops will
become shorter
0.81
The growing season will become shorter 0.80
The amount of land that can be farmed in my village will be
reduced
0.61 0.73
Climate change will negatively affect the profitability of my
farm in the future
0.80
Farming will no longer be profitable in my village 0.79
My crop yields will decrease 0.72 0.55
Pest invasions will increase 0.65
a Item scale: 1, highly disagree; 2, somewhat disagree; 3, somewhat agree; 4, highly agreeb Blanks represent PC loadings\0.55c WATER_IMPACT was defined as the level of perceived impact climate change will have on water availability and growing seasond PROFIT_IMPACT was defined as the level of perceived impact climate change will have on the profitability of agriculturee YIELD_IMPACT was defined as the level of perceived impact climate change will have on crop yields
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
government (i.e., perceived self-efficacy). Of the 483
respondents, 45 % described adaptation as not possible,
33 % very difficult, 16 % somewhat difficult, and 6 %
somewhat or very easy. Respondents were then asked to
describe their ability to make changes with information or
assistance from agricultural professionals or the govern-
ment. In response, 2 % of respondents described it as not
possible, 2 % very difficult, 15 % somewhat difficult, and
81 % somewhat or very easy.
The nested proportional odds ordered logistic regres-
sion model assessing factors that influence smallholders’
perceived self-efficacy to adapt to climate change was
statistically significant (p\ 0.01) across each of the four
blocks of the full model (Supplementary Table 2). Several
variables were consistently statistically significant at the
5 % level across the nested models. Of the variables
measuring the effect of human capital on perceived self-
efficacy, having completed middle school or higher, the
presence of more household members who provided
agricultural labor, income from a local off-farm source,
and previous adaptation to rainfall-related climatic chan-
ges all had positive significant relationships with per-
ceived self-efficacy.
The three PCs (PRCP_RISK, PRDCT_RISK, FROS-
T_RISK) measuring respondents’ perceived risk posed to
livelihood from climatic sources had little to no statistically
significant relationship with respondents’ perceived self-
efficacy. The same was true for each of the three PCs
(WTRSSN_IMPACT, PRFT_IMPACT, YLD_IMPACT)
measuring perceived impacts of climate change on farming
and livelihood. In contrast, each of the four PCs measuring
respondents’ perceived risk posed to livelihood from non-
climatic sources had a statistically significant negative
relationship with respondents’ perceived self-efficacy in at
least one of the four nested models. The largest negative
effect was that of perceived market risk on perceived self-
efficacy. Specifically, in the full model (i.e., model 4), for a
one unit of change in MRKT_RISK, the odds of a
respondent indicating the highest level of perceived self-
efficacy versus the three lower levels was 0.54 when
holding other variables constant.
Building upon model 1, we added four variables to
measure the effect of material and non-material resources
(i.e., information and technology, material resources and
infrastructure) on perceived self-efficacy. The resulting
model 2 did not have a statistically better fit than model 1.
Table 5 Household and farm characteristics of smallholder survey respondents
Household characteristics by county Yangling Mizhi Hongsipu
(n = 160) (n = 162) (n = 161)
Household size 5.1 4.5 5.1
Gender (% men) 52.5 59.9 60.2
Age 53.0 55.1 43.6
Years of schooling 6.5 4.5 4.2
No. of household members providing farm labor 2.7 2.2 2.2
% employed solely a market-oriented strategy 31.0 8.0 63.0
% employed solely a subsistence strategy 18.4 34.0 13.0
% employed a mixed subsistence and market-oriented strategy 51.0 58.4 25.0
Landholding size (mua) 5.2 9.1 10.5
No. of parcels 3.5 4.8 2.1
Years of agricultural experience 32.3 34.3 20.6
% with access to irrigation water 78.0 49.1 100
% as member of a farmer association 19.0 8.0 10.0
No. of extension visits last year 0.8 0.9 0.3
No. of times that land was reallocated in the last 10 years \1.0 \1.0 \1.0
% perceiving land reallocation will occur in the next 10 years 41.0 7.0 20.0
% of households with at least one member engaged in off-farm work 86 54 61
Annual household income (Yuanb) 23,717 11,261 22,233
% of income from agriculture 36.9 65.9 68.3
% of income from general off-farm work 49.5 29.5 31.1
% of income from local labor or business 25.0 14.5 21.1
% of income from migration 24.0 15.0 10.0
a 1 Chinese mu = 668 m2
b 1 Chinese Yuan = 0.16 U.S. Dollars
M. Burnham, Z. Ma
123
We added five additional variables to measure the effect of
wealth and financial capital on perceived self-efficacy. The
resulting model 3 had a statistically better fit than models 1
and 2 (likelihood-ratio test, p\ 0.01), and all the added
variables were statistically significantly associated with
perceived self-efficacy. Specifically, having a higher
annual household income was positively associated with
reporting higher self-efficacy. For example, for respondents
with an annual household income of more than 30,000
Yuan, the odds of them reporting higher self-efficacy was
3.17 times higher than the odds for respondents with an
annual household income of \10,000 Yuan. Those who
indicated market access as an important factor for making
farm management decisions were also more likely to per-
ceive a higher level of self-efficacy. In the full model (i.e.,
model 4), we added seven variables to measure the effect
of institutions and entitlements on perceived self-efficacy.
While the model was a statistically better fit than the pre-
vious three models (likelihood-ratio test, p\ 0.01), the
only variable that had a statistically significant association
with self-efficacy was HONGSIPU. This result indicates
that for respondents from Hongsipu, the odds of them
perceiving higher self-efficacy was 0.39 times lower than
the odds for respondents from Yangling. In addition, two
previously added variables became statistically significant
in the full model. The use of an irrigation schedule was
positively associated with perceived self-efficacy, while
using plastic mulching (usually for maize production) was
negatively associated with perceived self-efficacy.
Adaptation intent
Respondents were asked to indicate the likelihood that
they would make changes to their farming practices or
livelihood to prevent damage caused by climate change
based on their current understanding and situation (i.e.,
adaptation intent). Of the 483 survey respondents, 32 %
indicated it was very unlikely they would adapt, 42 %
somewhat unlikely, 24 % somewhat likely, and 3 % very
likely. Respondents were also asked about their likelihood
to participate in a program run by the government or local
university that would help them make changes to their
farming practices or livelihood to prevent damage caused
by climate change. The majority of respondents (97 %)
indicated they were somewhat or very likely to participate
in such a program.
The nested proportional odds ordered logistic regression
model assessing factors that influence smallholders’ adap-
tation intent was statistically significant (p\ 0.01) across
each of the four blocks of the full model (Supplementary
Table 3). Several variables were consistently statistically
significant at the 5 % level across the nested models. Of the
variables measuring the effect of human capital on
adaptation intent, having completed middle school or higher,
having previously adapted to rainfall-related climatic chan-
ges, and the amount of perceived risk to livelihood from
market sources were all positively associated with adapta-
tion intent. As opposed to the EFFICACY models, several
PCs that measured respondents’ perceived risk from climatic
sources and perceived impacts of climate change had sta-
tistically significant associations with adaptation intent.
Specifically, respondents who perceived a higher amount of
risk to their livelihood from changes in precipitation-related
events, a higher amount of risk to their livelihood from
increased unpredictability of climate events, and a higher
level of impact of climate change on the profitability of
agriculture reported a lower likelihood of make changes to
their farming practices or livelihood to adapt to climate
change. Related to this, respondents who stated that the
climate had been changing over the last 30 years reported a
lower adaptation intent than those who stated that the cli-
mate had not been changing. Across the nested models, the
largest positive effect was that of perceived self-efficacy on
adaptation intent. Specifically, in the full model (i.e., model
8), for respondents with the highest level of perceived self-
efficacy (i.e., somewhat or very easy), the odds of them
indicating an higher level of adaptation intent was 21.29
times higher than the odds for respondents with the lowest
level of perceived self-efficacy (i.e., not possible).
Model 6, which included four variables measuring the
effect of material and non-material resources, had a sta-
tistically better fit than model 5, which only included
variables measuring human capital (likelihood-ratio test,
p\ 0.01). Specifically, having a greenhouse increased the
odds of a respondent indicating a higher level of adaptation
intent. Building upon model 6, we added variables to
measure the effect of wealth and financial capital on
adaptation intent. The resulting model 7 did not have a
statistically better fit than models 5 and 6. Finally, we
included variables measuring institutions and entitlements
in the full model (i.e., model 8). While the full model had a
significantly better fit than the previous three models
(likelihood-ratio test, p\ 0.01), the only variable that had
a statistically significant association with adaptation intent
was DIST_FAR. This result indicates that for respondents
who lived more than 20 km away from a local population
center, the odds of them indicating a higher adaptation
intent was 0.42 times lower than the odds for respondents
who lived within 10 km of a local population center.
Discussion
The profile of smallholder farmers who participated in our
survey broadly agrees with previous findings from studies
conducted in the Loess Plateau region of China (Burnham
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
et al. 2016; Hageback et al. 2005). Our results generally
confirm the findings of earlier research examining the
cognitive dimensions of climate change adaptation. In
particular, our results provide strong empirical support for
the hypothesis that an individual’s perception of their self-
efficacy plays an important role in determining intention to
adapt (Grothmann and Patt 2005).
In terms of factors influencing perceived self-efficacy
and adaptation intent, our results suggest an association
between higher incomes and higher levels of perceived
self-efficacy. Previous research has offered some possible
explanations. One is that households with more assets may
be more likely to undertake adaptation activities because
they have capital reserves that they can draw on to pay for
the activities (Lemos et al. 2013). Another is that higher
incomes may enable households to take on risks associated
with adaptation activities without jeopardizing the short- or
long-term welfare of household members (Panda et al.
2013). It is worth noting that our results suggest no asso-
ciation between income level and adaptation intent while
holding other variables, including perceived self-efficacy,
constant. One possible explanation is that households with
higher incomes in our study were dependent on off-farm
jobs more than they were dependent on agriculture, thus
making them less likely to adapt their agricultural liveli-
hood to climate change.
Our study also suggests that smallholders with experi-
ence adapting to precipitation-related changes in the last
30 years were more likely to report a higher level of per-
ceived self-efficacy and adaptation intent. This result is
consistent with what Kuruppu and Liverman (2011) found
in Kiribati, suggesting that individual perceived self-effi-
cacy may be more influenced by past experiences dealing
with climatic stressors than an understanding of likely
climate change impacts.
In addition to past experiences, our results suggest that
the use of technology may contribute to higher levels of
perceived self-efficacy and adaptation intent by mediating
the impacts of climate change on smallholder production.
Specifically, we found that smallholders who used an irri-
gation schedule were more likely to report higher levels of
perceived self-efficacy and those who used greenhouses
were more likely to report higher levels of adaptation
intent. Greenhouses protect crops from climate variability
because they allow temperature and water availability to be
manipulated in a climate-controlled setting. Similar to
greenhouses, irrigation schedules allow smallholders to
reduce the threat of water scarcity, a major climatic con-
straint to crop production in the Loess Plateau region,
particular in the early growing season. Interestingly, the
use of plastic mulching had a negative relationship with
perceived self-efficacy. In the Loess Plateau region, plastic
mulching is mostly used in maize production. It enables
smallholders to mitigate the negative impacts of early
growing season water shortages. In our study, plastic
mulching was primarily used in rainfed systems. Thus,
while it does enable a certain degree of control over the
climate, smallholders may also view it as being all that can
be done to deal with aridity short of irrigation. As the
MPPACC predicts, smallholders may view plastic mulch-
ing as having a low adaptation efficacy, and, as a result,
they would have lower perceived self-efficacy levels.
Our study provides further evidence of the role of local
institutions in mediating perceived self-efficacy, as shown
in previous research (Agrawal and Perrin 2009; Wang et al.
2013). Specifically, respondents from Hongsipu tended to
have a lower level of perceived self-efficacy. This may be
partly related to the history of Hongsipu and the role of
government in shaping smallholder livelihoods in the area.
Hongsipu was created in 1998 as a dam construction pro-
ject relocation site for households from nearby marginal
hillsides where government services and irrigation infras-
tructure were lacking. Many respondents raised livestock in
their former villages and did not grow crops. To promote
agricultural production, the government sent extension
agents to live in Hongsipu for a minimum of 1 year to
teach villagers how to farm. Many villagers also partici-
pated in government-sponsored workshops four to six
times a year to learn agricultural skills, such as pruning.
The government set up a canal system to deliver irrigation
water from the Yellow River to all the villages included in
our study. Furthermore, local and regional governments
have assisted in the development of goji berry as a cash
crop for the area. Thus, the government as a whole has had
a heavy hand in developing agriculture as a viable liveli-
hood pursuit in Hongsipu. As a result, smallholders have
become dependent on state involvement. It is not surprising
that such dependency would dampen how smallholders
view their own ability to make changes to their farming
practices and livelihoods to adapt to climate change.
In our study, availability of farm labor and performing
local wage work both positively affect perceived self-effi-
cacy, while households with members migrating for work
on a permanent or semi-permanent basis did not have
higher perceived self-efficacy. We suggest that this is
because the relationship between off-farm work and per-
ceived self-efficacy is mediated through the impact of off-
farm work on farm labor availability. Even though local
wage work and labor migration may provide the same level
of income, they have different impacts on the availability
of farm labor. Local wage workers are able to provide farm
labor during peak times such as harvest season, while
migrant workers may not be able to. Thus, holding income
and availability of farm labor constant, local wage work
contributes to smallholders’ perceived self-efficacy, but
labor migration does not. Similarly, our results indicate a
M. Burnham, Z. Ma
123
negative relationship between perceived risk from personal
illness or injury and perceived self-efficacy, which we
believe is also mediated through the impact of personal
illness or injury on the availability of farm labor. Other
studies have also found that farmers perceived the risk of
illness and injury to be similar to or greater than the risk
posed by climate change (Thomas et al. 2007).
In much of the literature that examines individual
response to natural hazards or climate change, researchers
assume that higher levels of perceived risk should be
positively correlated with higher levels of adaptation intent
(Lo 2013). In the MPPACC, the influence of perceived risk
is mediated by perceived adaptive capacity, which is partly
determined by perceived self-efficacy. Our study provides
empirical evidence of this contention. We found that per-
ceiving the climate has changed over the last 30 years, as
well as perceiving high levels of risk to agriculture from
changes in precipitation-related events and from increased
unpredictability of climate events, did not affect small-
holders’ perceived self-efficacy, but lowered their adapta-
tion intent. One possible explanation for this relationship
between higher perceived climate-related risks and lower
adaptation intent is that smallholders may view the risk of
adapting as greater than the risk of not adapting because
they are concerned about other non-climatic risks (Patt and
Schroter 2008). Specifically, our study shows that small-
holders generally perceive risk to livelihoods from market
sources. Market-related factors are frequently reported by
smallholders to pose greater risk to their livelihoods than
climate-related factors (Burnham et al. 2015; Frank et al.
2011; Gandure et al. 2013; Tucker et al. 2010). In the Loess
Plateau region of China, agricultural strategies for helping
smallholders adapt to climate change generally require
changes to current cropping or irrigation practices, which
will likely increase their engagement with markets, such as
finding markets for new crops and selling more crops from
increased production. Thus, smallholders may view these
adaptation strategies as sources of greater or new risk to
their livelihoods, decreasing their adaptation intent.
In the MPPACC, reliance on public adaptation is one
factor that decreases an individual’s appraisal of the risk
that climate change poses to them, as well as the likelihood
they would engage in adaptive behavior on their own.
While we did not directly measure the effect of public
adaptation on smallholder risk perception, our results add a
new dimension to thinking about the role public adaptation
plays in the MPPACC. Throughout the Loess Plateau
region, local and regional governments, as well as agri-
cultural research centers and universities, have been
heavily involved in building irrigation infrastructure and
disseminating irrigation technologies, developing and dis-
tributing drought-resistant hybrid maize seeds, and offering
subsidies that enable smallholders to purchase pesticides,
fertilizers, and plastic mulching, among other things. Thus,
it is not surprising that in our study smallholders reported
low levels of perceived self-efficacy and adaptation intent
without information or assistance from agricultural pro-
fessionals or the government. Conversely, they reported a
much greater ability to adapt with information or assistance
from agricultural professionals or the government, as well
as an overwhelming likelihood to participate in a program
run by the government or local university to help them
adapt. Indeed, during our interviews we asked smallholders
what could be done to adapt to climate change and the most
common response was that it is impossible for them to
adapt on their own and that it is the government’s
responsibility to lead adaptation efforts.
Further evidence of the relationship between public
adaptation and perceived self-efficacy is demonstrated
by the result that smallholders from Hongsipu tended to
have a lower level of perceived self-efficacy. As previously
discussed, the government has been heavily involved in the
establishment and development of Hongsipu and agricul-
tural livelihoods in the area. Thus, smallholders seem to
have developed a reliance on state intervention and are less
confident in their own ability to adapt without government
help. Similar results have been reported previously (Crate
2011; Eakin and Bojorquez-Tapia 2008; Lemos et al. 2013;
Saldana-Zorrilla 2008), noting that households’ long-term
capacities may be eroded when state-sponsored welfare
programs create ‘‘dependencies between state and society.’’
Such dependencies can create rigidities that may decrease
the ability of smallholders to adapt to climatic or other
stressors on their own (Lemos et al. 2013). Our study
further demonstrates the role that state-society dependen-
cies play in determining smallholders’ perceived self-effi-
cacy, which will shape their adaptation intent.
Conclusion
Our study contributes to understanding the cognitive
dimensions of adaptation decision-making as conceptual-
ized in the MPPACC (Grothmann and Patt 2005). We
systematically examined how smallholders’ perceptions of
various factors that may enhance or hinder adaptation
shape their perceived self-efficacy and adaptation intent.
Our study provides further evidence that self-efficacy
beliefs are a strong, positive predictor of adaptation intent,
as suggested by recent research (Grothmann and Patt 2005;
Kuruppu and Liverman 2011). Further, our study indicates
that higher income and past adaptation experiences con-
tribute to higher perceived self-efficacy. Farm labor
availability also contributes to perceived self-efficacy and
mediates the effect of off-farm work as a livelihood
diversification strategy on smallholders’ ability to adapt.
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
Our study challenges the notion that perceiving risks to
agriculture and livelihoods from climate change leads to
intention to adapt. Our results show that perceiving climate
change risks and impacts on agriculture were either nega-
tively associated with or had little effect on adaptation
intent. Finally, our study adds a new dimension to under-
standing the role that local institutions and public adapta-
tion play in determining perceived self-efficacy and
adaptation intent. Specifically, our study indicates that
state-society dependencies, developed through govern-
ment-led livelihood interventions and agricultural devel-
opment projects, may reduce smallholders’ perceived self-
efficacy. Overall, our findings highlight the importance of
incorporating both objective and subjective determinants of
smallholder adaptive capacity into future climate change
adaptation programs and policies in order to facilitate
adaptation actions.
Acknowledgments This research was partially funded by Northwest
Agriculture and Forestry University (NWAFU) through the 111
project of Chinese Ministry of Education (No. B12007). The authors
are grateful to Drs. Pute Wu, Delan Zhu, Youke Wang, Xining Zhao,
Xiping Liu, Yubao Wang from NWAFU for their support for this
research. The authors also thank Chunyan Qi, Mengying Sun, and
several other NWAFU undergraduate and graduate students for their
assistance during fieldwork. Finally, we would like to thank our
anonymous reviewers for their helpful comments, which significantly
strengthened the paper.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
References
Adger WN (2003) Social capital, collective action, and adaptation to
climate change. Econ Geogr 79(4):387–404. doi:10.1111/j.1944-
8287.2003.tb00220.x
Adger WN, Vincent K (2005) Uncertainty in adaptive capacity. CR
Geosci 337(4):399–410. doi:10.1016/j.crte.2004.11.004
Adger WN, Agrawala S, Mirza MMQ et al (2007) Assessment of
adaptation practices, options, constraints and capacity. Climate
change 2007: impacts, adaptation and vulnerability. Contribution
of working group II to the fourth assessment report of the
intergovernmental panel on climate change. Cambridge Univer-
sity Press, Cambridge. http://www.ipcc.ch/pdf/assessment-
report/ar4/wg2/ar4-wg2-chapter17.pdf. Accessed 29 May 2015
Adger WN, Barnett J, Brown K, Marshall N, O’Brien K (2013)
Cultural dimensions of climate change impacts and adaptation.
Nat Clim Change 3:112–117. doi:10.1038/nclimate1666
Agrawal A, Perrin N (2009) Climate adaptation, local institutions and
rural livelihoods. In: Adger WN, Lorenzone I, O’Brien KL (eds)
Adapting to climate change: thresholds, values, governance.
Cambridge University Press, Cambridge, pp 350–367
Ajzen I (1991) The theory of planned behavior. Organ Behav Hum Decis
Process 50(2):179–211. doi:10.1016/0749-5978(91)90020-T
Barbier B, Yacouba H, Karambiri H, Zorome M, Some B (2009)
Human vulnerability to climate variability in the Sahel: farmers’
adaptation strategies in northern Burkina Faso. Environ Manag
43(5):790–803. doi:10.1007/s00267-008-9237-9
Barnett J, Waters E, Pendergast S, Puleston A (2013) Barriers to
adaptation to sea-level rise: the legal, institutional and cultural
barriers to adaptation to sea-level rise in Australia. National
Climate Change Adaptation Research Facility, Gold Coast
Below TB, Mutabazi KD, Kirschke D, Franke C, Sieber S, Siebert R,
Tscherning K (2012) Can farmers’ adaptation to climate change
be explained by socio-economic household-level variables? Glob
Environ Change 22(1):223–235. doi:10.1016/j.gloenvcha.2011.
11.012
Birkenholtz T (2012) Network political ecology: method and theory
in climate change vulnerability and adaptation research. Prog
Hum Geogr 36(3):295–315. doi:10.1177/0309132511421532
Blennow K, Persson J (2009) Climate change: motivation for taking
measure to adapt. Glob Environ Change 19(1):100–104. doi:10.
1016/j.gloenvcha.2008.10.003
Brant R (1990) Assessing proportionality in the proportional odds
model for ordinal logistic regression. Biometrics
46(4):1171–1178. doi:10.2307/2532457
Brooks N, Adger WN (2005) Assessing and enhancing adaptive
capacity. In: Lim B, Spanger-Siegfried E (eds) Adaptation policy
framework for climate change: developing strategies, policies,
and measures. Cambridge University Press, Cambridge,
pp 165–182
Bryan E, Ringler C, Okoba B, Roncoli C, Silvestri S, Herrero M
(2013) Adapting agriculture to climate change in Kenya:
household strategies and determinants. J Environ Manag
114:26–35. doi:10.1016/j.jenvman.2012.10.036
Burnham M, Ma Z (2015) Linking smallholder farmer climate change
adaptation decisions to development. Clim Dev. doi:10.1080/
17565529.2015.1067180
Burnham M, Ma Z, Zhu D (2015) The human dimensions of water
saving irrigation: lessons learned from Chinese smallholder
farmers. Agric Hum Values 32:347–360. doi:10.1007/s10460-
014-9565-8
Burnham M, Ma Z, Zhang B (2016) Making sense of climate change:
hybrid epistemologies, socio-natural assemblages, and small-
holder knowledge. Area 48:18–26. doi:10.1111/area.12150
Crate SA (2011) A political ecology of ‘‘water in mind’’: attributing
perceptions in the era of global climate change. Weather Clim
Soc 3(3):148–164. doi:10.1175/WCAS-D-10-05006.1
Cronbach LJ (1951) Coefficient alpha and the internal structure of
tests. Psychometrika 16:297–334. doi:10.1007/BF02310555
Deressa TT, Hassan RM, Ringler C (2010) Perception of and
adaptation to climate change by farmers in the Nile basin of
Ethiopia. J Agric Sci 149(01):23–31. doi:10.1017/
S0021859610000687
Didier EA, Brunson MW (2004) Adoption of range management
innovations by Utah ranchers. J Range Manag 57(4):330–336.
doi:10.2111/1551-5028(2004)057[0330:AORMIB]2.0.CO;2
Ding Y, Chan JCL (2005) The East Asian summer monsoon: an
overview. Meteorol Atmos Phys 89(1–4):117–142. doi:10.1007/
s00703-005-0125-z
Eakin H, Bojorquez-Tapia LA (2008) Insights into the composition of
household vulnerability from multicriteria decision analysis.
Glob Environ Change 18(1):112–127. doi:10.1016/j.gloenvcha.
2007.09.001
Eakin H, Lemos MC (2006) Adaptation and the state: Latin America
and the challenge of capacity-building under globalization. Glob
Environ Change 16(1):7–18. doi:10.1016/j.gloenvcha.2005.10.
004
Engle NL (2011) Adaptive capacity and its assessment. Glob Environ
Change 21(2):647–656. doi:10.1016/j.gloenvcha.2011.01.019
Fishbein M, Ajzen I (1975) Belief, attitude, intention and behavior: an
introduction to theory and research. Addison Wesley, Reading
M. Burnham, Z. Ma
123
Frank E, Eakin H, Lopez-Carr D (2011) Social identity, perception
and motivation in adaptation to climate risk in the coffee sector
of Chiapas, Mexico. Glob Environ Change 21(1):66–76. doi:10.
1016/j.gloenvcha.2010.11.001
Fullerton AS (2009) A conceptual framework for ordered logistic
regression models. Sociol Methods Res 38(2):306–347. doi:10.
1177/0049124109346162
Gandure S, Walker S, Botha JJ (2013) Farmer’s perceptions of
adaptation to climate change and water stress in a South African
community. Environ Dev 5:39–53. doi:10.1016/j.envdev.2012.
11.004
Grothmann T, Patt A (2005) Adaptive capacity and human cognition:
the process of individual adaptation to climate change. Glob
Environ Change 15(3):199–213. doi:10.1016/j.gloenvcha.2005.
01.002
Hageback J, Sundberg J, Ostwald M, Chen D, Yun X, Knutsson P
(2005) Climate variability and land-use change in Danangou
Watershed, China—examples of small-scale farmers’ adaptation.
Clim Change 72(1–2):189–212. doi:10.1007/s10584-005-5384-7
Harmer N, Rahman S (2014) Climate change response at the farm
level: a review of farmers’ awareness and adaptation strategies in
developing countries. Geogr Compass 8(11):808–822. doi:10.
1111/gec3.12180
Hisali E, Birungi P, Buyinza F (2011) Adaptation to climate change in
Uganda: evidence from micro level data. Glob Environ Change
21(4):1245–1261. doi:10.1016/j.gloenvcha.2011.07.005
Jones L, Boyd E (2011) Exploring social barriers to adaptation:
insights from Western Nepal. Glob Environ Change
21(4):1262–1274. doi:10.1016/j.gloenvcha.2011.06.002
Kaiser HF (1958) The varimax criterion for analytic rotation in factor
analysis. Psychometrika 23(3):187–200. doi:10.1007/BF02289233
Klein RJT, Midgley GF, Preston BL, Alam M, Berkhout FGH, Dow
K, Shaw MR (2014) Adaptation opportunities, constraints, and
limits. In: Field CB, Barros VR, Dokken DJ, Mach KJ,
Mastrandrea MD, Bilir TE, Chatterjee M, Ebi KL, Estrada
YO, Genova RC, Girma B, Kissel ES, Levy AN, MacCracken S,
Mastrandrea PR, White LL (eds) Climate change 2014: impacts,
adaptation, and vulnerability. part a: global and sectoral aspects.
Contribution of working group II to the fifth assessment report of
the intergovernmental panel on climate change. Cambridge
University Press, Cambridge, pp 899–943
Kuruppu N, Liverman D (2011) Mental preparation for climate
adaptation: the role of cognition and culture in enhancing
adaptive capacity of water management in Kiribati. Glob
Environ Change 21(2):657–669. doi:10.1016/j.gloenvcha.2010.
12.002
Leichenko R, O’Brien K (2008) Environmental change and global-
ization: double exposures. Oxford University Press, Oxford
Lemos MC, Agrawal A, Eakin H, Nelson DR, Engle NL, Johns O
(2013) Building adaptive capacity to climate change in less
developed countries. In: Asrar G, Hurrell J (eds) Climate science
for serving society. Springer, Dordrecht, pp 437–457
Li Z, Zheng FL, Liu WZ, Flanagan DC (2010) Spatial distribution and
temporal trends of extreme temperature and precipitationevents on the
Loess Plateau of China during 1961–2007. Quat Int 226(1):92–100.
doi:10.1016/j.quaint.2010.03.003
Li Z, Zheng FL, Liu WZ (2012) Spatiotemporal characteristics of
reference evapotranspiration during 1961–2009 and its projected
changes during 2011–2099 on the Loess Plateau of China. Agric
For Meteorol 154–155:147–155. doi:10.1016/j.agrformet.2011.
10.019
Li Y, Conway D, Wu Y, Gao Q, Rothausen S, Xiong W, Ju H, Lin E
(2013) Rural livelihoods and climate variability in Ningxia,
Northwest China. Clim Change 119(3–4):891–904. doi:10.1007/
s10584-013-0765-9
Liska AE (1984) A critical examination of the causal structure of the
Fishbein/Ajzen attitude-behavior model. Soc Psychol Q
47(1):61–74. doi:10.2307/3033889
Liu L, Liu Z, Ren X, Fischer T, Xu Y (2011) Hydrological impacts of
climate change in the Yellow River Basin for the 21st century
using hydrological model and statistical downscaling model.
Quat Int 244(2):211–220. doi:10.1016/j.quaint.2010.12.001
Lo AY (2013) The role of social norms in climate adaptation:
mediating risk perception and flood insurance purchase. Glob
Environ Change 23(5):1249–1257. doi:10.1016/j.gloenvcha.
2013.07.019
Long JS (1997) Regression models for categorical and limited
dependent variables. Sage Publications, Thousand Oaks
Long JS, Freese J (2006) Regression models for categorical depen-
dent variables using Stata. Stata Press Books, College Station
Mertz O, Mbow C, Maiga A, Diallo D, Reenberg A, Diouf A, Barbier
B et al (2010) Climate factors play a limited role for past
adaptation strategies in West Africa. Ecol Soc 15(4):1–15. http://
www.ecologyandsociety.org/vol15/iss4/art25/
Morton JF (2007) The impact of climate change on smallholder and
subsistence agriculture. Proc Natl Acad Sci
104(50):19680–19685. doi:10.1073/pnas.0701855104
Moser S, Ekstrom J (2010) A framework to diagnose barriers to
climate change adaptation. Proc Natl Acad Sci USA
107(51):1–6. doi:10.1073/pnas.1007887107
Noy C (2008) Sampling knowledge: the hermeneutics of snowball
sampling in qualitative research. Int J Soc Res Methodol
11:327–344. doi:10.1080/13645570701401305
Nunnally JC (1978) Psychometric theory, 2nd edn. McGraw-Hill,
New York
Ostwald M, Chen D (2006) Land-use change: impacts of climate
variations and policies among small-scale farmers in the Loess
Plateau, China. Land Use Policy 23(4):361–371. doi:10.1016/j.
landusepol.2005.04.004
Panda A, Sharma U, Ninan KN, Patt A (2013) Adaptive capacity
contributing to improved agricultural productivity at the house-
hold level: empirical findings highlighting the importance of
crop insurance. Glob Environ Change 23(4):782–790. doi:10.
1016/j.gloenvcha.2013.03.002
Patt AG, Schroter D (2008) Perceptions of climate risk in Mozam-
bique: implications for the success of adaptation strategies. Glob
Environ Change 18(3):458–467. doi:10.1016/j.gloenvcha.2008.
04.002
Pelling M, High C (2005) Understanding adaptation: what can social
capital offer assessments of adaptive capacity? Glob Environ
Change 15(4):308–319. doi:10.1016/j.gloenvcha.2005.02.001
Piao S, Ciais P, Huang Y, Shen Z, Peng S, Li J, Zhou L et al (2010)
The impacts of climate change on water resources and agricul-
ture in China. Nature 467:43–51. doi:10.1038/nature09364
Piya L, Maharjan KL, Joshi NP (2012) Determinants of adaptation
practices to climate change by Chepang households in the rural
Mid-Hills of Nepal. Reg Environ Change 13(2):437–447. doi:10.
1007/s10113-012-0359-5
Ribot J (2009) Vulnerability does not fall from the sky: towards
multiscale, pro-poor climate policy. In: Mearns R, Norton A
(eds) The social dimensions of climate change: equity and
vulnerability in a warming world. World Bank Publications,
Washington, pp 47–74
Rifaat SM, Tay R, de Barros A (2012) Severity of motorcycle crashes
in Calgary. Accid Anal Prev 49:44–49. doi:10.1016/j.aap.2011.
02.025
Saldana-Zorrilla SO (2008) Stakeholders’ views in reducing rural
vulnerability to natural disasters in southern Mexico: hazard
exposure and coping and adaptive capacity. Glob Environ
Change 18(4):583–597. doi:10.1016/j.gloenvcha.2008.09.004
Climate change adaptation: factors influencing Chinese smallholder farmers’ perceived…
123
Smit B, Pilifosova O (2001) Adaptation to climate change in the
context of sustainable development and equity. In: IPCC
Working Group II (eds) Climate change 2001: impacts, adap-
tation and vulnerability. Earth Scan, Geneva, pp 879–906
Tambo JA, Abdoulaye T (2013) Smallholder perceptions of and
adaptations to climate change in the Nigerian Savanna. Reg
Environ Change 13(2):375–388. doi:10.1007/s10113-012-0351-
0
Thomas DSG, Twyman C, Osbahr H, Hewitson B (2007) Adaptation
to climate change and variability: farmer responses to intra-
seasonal precipitation trends in South Africa. Clim Change
83(3):301–322. doi:10.1007/s10584-006-9205-4
Tremblay MA (1957) The key informant technique: a non-ethno-
graphic application. Am Anthropol 59(4):688–701. doi:10.1525/
aa.1957.59.4.02a00100
Triandis HC (1977) Interpersonal behavior. Brooks/Cole Publishing
Company, Monterey
Tucker CM, Eakin H, Castellanos EJ (2010) Perceptions of risk and
adaptation: coffee producers, market shocks, and extreme
weather in Central America and Mexico. Glob Environ Change
20:23–32. doi:10.1016/j.gloenvcha.2009.07.006
Turner BL, Kasperson RE, Matson PA, McCarthy JJ, Corell RW,
Christensen L, Eckley N et al (2003) A framework for
vulnerability analysis in sustainability science. Proc Natl Acad
Sci USA 100(14):8074–8079. doi:10.1073/pnas.1231335100
Wang S, Zhang Z (2011) Effects of climate change on water resources
in China. Clim Res 47:77–82. doi:10.3354/cr00965
Wang J, Brown DG, Agrawal A (2013) Climate adaptation, local
institutions, and rural livelihoods: a comparative study of herder
communities in Mongolia and Inner Mongolia, China. Glob
Environ Change 23(6):1673–1683. doi:10.1016/j.gloenvcha.
2013.08.014
Webb TL, Sheeran P (2005) Integrating concepts from goal theories
to understand the achievement of personal goals. Eur J Soc
Psychol 35(1):69–96. doi:10.1002/ejsp.233
Webb TL, Sheeran P (2006) Does changing behavioral intentions
engender behavior change? A meta-analysis of the experimental
evidence. Psychol Bull 132(2):249–268. doi:10.1037/0033-2909.
132.2.249
Wolf J, Allice I, Bell T (2013) Values, climate change, and
implications for adaptation: evidence from two communities in
Labrador, Canada. Glob Environ Change 23(2):548–562. doi:10.
1016/j.gloenvcha.2012.11.007
Wu W, Verburg PH, Tang H (2013) Climate change and the food
production system: impacts and adaptation in China. Reg
Environ Change 14(1):1–5. doi:10.1007/s10113-013-0528-1
Yohe G, Tol RS (2002) Indicators for social and economic coping
capacity-moving toward a working definition of adaptive
capacity. Glob Environ Change 12(1):25–40. doi:10.1016/
S0959-3780(01)00026-7
Young G, Zavala H, Wandel J, Smit B, Salas S, Jimenez E, Fiebig M
et al (2009) Vulnerability and adaptation in a dryland community
of the Elqui Valley, Chile. Clim Change 98(1–2):245–276.
doi:10.1007/s10584-009-9665-4
Zhang B, Wu P, Zhao X, Wang Y, Wang J, Shi Y (2012) Drought
variation trends in different subregions of the Chinese Loess
Plateau over the past four decades. Agric Water Manag
115:167–177. doi:10.1016/j.agwat.2012.09.004
Zhang B, Wu P, Zhao X, Gao X (2013) Spatiotemporal analysis of
climate variability (1971–2010) in spring and summer on the
Loess Plateau, China. Hydrol Process 28(4):1689–1702. doi:10.
1002/hyp.9724
Zhao G, Mu X, Wen Z, Wang F, Gao P (2013) Soil erosion,
conservation, and eco-environment changes in the Loess Plateau
of China. Land Degrad Dev 24(5):499–510. doi:10.1002/ldr.
2246
M. Burnham, Z. Ma
123
Table 1 Independent variables used in the empirical models for estimating smallholders’ perceived self-efficacy to adapt to climate
change and their adaptation intent.
Variable name Description Type/scale
Human capital
AGE Respondent’s age Continuous
GENDER Respondent’s gender Binary/1=male
EDU Respondent’s education level Binary/1=completed
middle school or higher
FARMEXP Number of years of farming experience Continuous
FARMLABOR Number of people who provide labor on farm Continuous
LOC_WORK Having at least one household member who works off-farm locally Binary/1=yes
MIG_WORK Having at least one household member who migrates for work on a (semi)-
permanent basis
Binary/1=yes
ADAPT_RAIN Respondent has adapted to precipitation-related changes in the last 30 years Binary/1=yes
REALLOCATE_RISK Respondent perceives a risk that their land will be reallocated in the next 10 years Binary/1=yes
MRKT_RISKa The amount of perceived risk posed to livelihood from market sources Continuous
ILL_RISKa The amount of perceived risk posed to livelihood from illness or injury Continuous
FARM_RISKa The amount of perceived risk posed to livelihood from on-farm sources Continuous
EX_RISKa The amount of perceived risk posed to livelihood from sources beyond household’s
control
Continuous
PRECIP_RISKa The amount of perceived risk posed to livelihood from changes in precipitation-
related events
Continuous
PREDICT_RISKa The amount of perceived risk posed to livelihood from increased unpredictability of
climate events
Continuous
SPRING_RISKa The amount of perceived risk posed to livelihoods from changes in spring time
climate events
Continuous
WATER_IMPACTa The level of perceived impact climate change will have on water availability and
growing season
Continuous
PROFIT_IMPACTa The level of perceived impact climate change will have on the profitability of
agriculture
Continuous
YIELD_IMPACTa The level of perceived impact climate change will have on crop yields Continuous
CLIMATE Respondent’s answer to question “In general, do you think the climate in your
county has been changing over the last 30 years?”
Binary/1=yes
Variable name Description Type/scale
SELFEFFICACY_2b Respondent perceives adapting to climate change on their own as very difficult Binary/1=yes
SELFEFFICACY_3b
Respondent perceives adapting to climate change on their own as somewhat
difficult
Binary/1=yes
SELFEFFICACY_4b
Respondent perceives adapting to climate change on their own as somewhat or very
easy
Binary/1=yes
Material and non-material resources (i.e., information and Technology, material resources and infrastructure)
MULCH Respondent uses plastic mulching Binary/1=yes
GREENHOUSE Respondent has a greenhouse Binary/1=yes
IRR_SCHD Respondent uses an irrigation schedule Binary/1=yes
IRRIGATION Respondent’s land is irrigated Binary/1=yes
Wealth and financial capital
INC_2 Respondent’s annual household income is between 10,000-19,999 Yuanc
Binary/1=yes
INC_3 Respondent’s annual household income is between 20,000-29,999 Yuanc
Binary/1=yes
INC_4 Respondent’s annual household income is over 30,000 Yuanc
Binary/1=yes
CREDIT Respondent perceives access to credit as an important factor when making decisions
about how to manage farm
Binary/1=yes
MRKTACCESS Respondent perceives market access as an important factor when making decisions
about how to manage farm
Binary/1=yes
Institutions and entitlements
EXTENSION Respondent was visited by an agricultural extension agent in the last year Binary/1=yes
REALLOCATE Respondent’s land was reallocated in the last 10 years Binary/1=yes
COOP Respondent is a member of a local farmer cooperative Binary/1=yes
MIZHI Respondent lives in Mizhi Binary/1=yes
HONGSIPU Respondent lives in Hongsipu Binary/1=yes
DIST_MIDDLE Respondent lives > 10 km and < 20 km from a local population center Binary/1=yes
DIST_FAR Respondent lives > 20 km from a local population center Binary/1=yes a This variable is the composite score from the principal component analysis (see tables 2, 3, and 4).
b This variable was only used to in the empirical model for predicting smallholder farmer adaptation intent.
c 1 Chinese yuan = 0.16 U.S. dollar
Table 2 Logistic estimates of the empirical model for estimating smallholders’ perceived self-efficacy to adapt to climate change
(n=420).
Explanatory variable
Human capital
(model 1)
Material and non-material
resources (model 2)
Wealth and financial
capital (model 3)
Institutions and
entitlements (model 4)
Coefficienta,b
Odds
ratio Coefficient
a,b
Odds
ratio Coefficient
a,b
Odds
ratio Coefficient
a,b
Odds
ratio
AGE -0.01 (0.01) 0.99 -0.01 (0.01) 0.99 -0.01 (0.01) 0.99 -0.01 (0.01) 0.99
GENDER 0.29 (0.21) 1.34 0.27 (0.21) 1.31 0.34 (0.21) 1.40 0.36 (0.22)* 1.44
EDU 0.75 (0.22)*** 2.11 0.77 (0.23)*** 2.16 0.63 (0.24)*** 1.89 0.61 (0.25)** 1.84
FARMEXP 0.00 (0.01) 1.00 0.00 (0.01) 1.00 0.00 (0.01) 1.00 0.00 (0.01) 1.00
FARMLABOR 0.21 (0.09)** 1.23 0.19 (0.09)** 1.21 0.24 (0.09)** 1.27 0.22 (0.10)** 1.25
LOC_WORK 0.49 (0.21)** 1.64 0.49 (0.21)** 1.63 0.51 (0.21)** 1.67 0.49 (0.22)** 1.63
MIG_WORK 0.20 (0.21) 1.22 0.25 (0.21) 1.28 0.14 (0.22) 1.15 0.07 (0.22) 1.08
ADAPT_RAIN 1.10 (0.21)*** 3.00 1.00 (0.22)*** 2.71 0.87 (0.23)*** 2.38 0.84 (0.24)*** 2.31
REALLOCATE_RISK -0.01 (0.26) 0.99 -0.12 (0.26) 0.88 -0.03 (0.27) 0.97 -0.26 (0.28) 0.77
MRKT_RISK -0.56 (0.13)*** 0.57 -0.58 (0.13)*** 0.56 -0.62 (0.13)*** 0.54 -0.62 (0.14)*** 0.54
ILL_RISK -0.29 (0.11)*** 0.75 -0.31 (0.11)*** 0.73 -0.25 (0.12)** 0.78 -0.29 (0.13)** 0.75
FARM_RISK -0.24 (0.12)** 0.79 -0.23 (0.13)* 0.80 -0.23 (0.13)* 0.80 -0.18 (0.14) 0.84
EX_RISK -0.04 (0.11) 0.96 -0.07 (0.12) 0.93 -0.17 (0.12) 0.85 -0.35 (0.14)** 0.71
PRECIP_RISK -0.05 (0.13) 0.96 0.00 (0.14) 1.00 0.04 (0.15) 1.04 -0.13 (0.16) 0.88
PREDICT_RISK 0.06 (0.12) 1.06 0.07 (0.12) 1.07 0.13 (0.13) 1.14 0.22 (0.13)* 1.25
SPRING_RISK 0.06 (0.11) 1.06 0.11 (0.11) 1.12 0.10 (0.11) 1.10 0.07 (0.12) 1.07
WATER_IMPACT 0.03 (0.11) 1.02 0.01 (0.11) 1.01 0.00 (0.12) 1.00 -0.02 (0.13) 0.98
PROFIT_IMPACT -0.04 (0.10) 0.96 -0.01 (0.10) 0.99 -0.07 (0.11) 0.93 -0.14 (0.12) 0.87
YIELD_IMPACT 0.03 (0.11) 1.03 0.02 (0.11) 1.02 0.03 (0.12) 1.03 0.08 (0.12) 1.09
CLIMATE -0.41 (0.28) 0.66 -0.37 (0.28) 0.69 -0.38 (0.29) 0.68 -0.43 (0.31) 0.65
MULCH
-0.35 (0.22) 0.71 -0.35 (0.22) 0.71 -0.61 (0.24)** 0.55
GREENHOUSE
0.30 (0.41) 1.34 0.14 (0.43) 1.15 0.04 (0.44) 1.04
IRR_SCHD 0.36 (0.25) 1.44 0.53 (0.26)** 1.69 0.51 (0.27)** 1.67
IRRIGATION 0.05 (0.31) 1.05 -0.31 (0.32)*** 0.73 -0.04 (0.34) 0.96
INC_2 1.05 (0.29)*** 2.86 1.20 (0.31)*** 3.32
INC_3 0.94 (0.33)*** 2.55 1.04 (0.35)*** 2.82
Explanatory variable
Human capital
(model 1)
Material and non-material
resources (model 2)
Wealth and financial
capital (model 3)
Institutions and
entitlements (model 4)
Coefficienta,b
Odds
ratio Coefficient
a,b
Odds
ratio Coefficient
a,b
Odds
ratio Coefficient
a,b
Odds
ratio
INC_4 1.02 (0.32)*** 2.76 1.15 (0.34)*** 3.17
CREDIT -0.65 (0.23)*** 0.52 -0.46 (0.24)** 0.63
MRKTACCESS 1.10 (0.37)*** 3.00 1.01 (0.38)*** 2.75
EXTENSION 0.46 (0.24) 1.59
REALLOCATE 0.51 (0.32) 1.67
COOP 0.41 (0.31) 1.51
MIZHI 0.06 (0.39) 1.07
HONGSIPU -0.94 (0.36)*** 0.39
DIST_MIDDLE 0.08 (0.25) 1.08
DIST_FAR -0.02 (0.28) 0.98
LR chi-square 124.09*** 129.51*** 162.50*** 185.63***
Psuedo R2 0.13 0.131 0.16 0.18
Log likelihood -432.358*** -429.64 -413.151*** -401.586** a Unstandardized regression coefficients (standard errors)
b *p<0.1, **p<0.05, ***p<0.01
Table 3 Logistic estimates of the empirical model for estimating smallholders’ stated intent to adapt to climate change (n=420).
Explanatory Variable
Human capital
(model 5)
Material and non-material
resources (model 6)
Wealth and financial
capital (model 7)
Institutions and
entitlements (model 8)
Coefficienta,b
Odds
Ratio Coefficient
a,b
Odds
Ratio Coefficient
a,b
Odds
Ratio Coefficient
a,b
Odds
Ratio
AGE 0.00( 0.01) 1.00 0.00 (0.01) 1.00 0.00 (0.01) 1.00 0.00 (0.01) 1.00
GENDER 0.05 (0.21) 1.05 0.05 (0.21) 1.05 0.05 (0.21) 1.05 0.11 (0.22) 1.12
EDU 0.56 (0.24)** 1.76 0.49 (0.24)** 1.63 0.49 (0.24)** 1.64 0.41 (0.25)* 1.51
FARMEXP 0.01 (0.01) 1.01 0.01 (0.01) 1.01 0.01 (0.01) 1.01 0.01 (0.01) 1.01
FARMLABOR -0.10 (0.09) 0.91 -0.12 (0.09) 0.89 -0.11 (0.09) 0.89 -0.10 (0.09) 0.91
LOC_WORK -0.31 (0.21) 0.73 -0.35 (0.21)* 0.71 -0.34 (0.21) 0.71 -0.43 (0.22)* 0.65
MIG_WORK 0.01 (0.21) 1.01 0.05 (0.21) 1.05 0.05 (0.21) 1.06 0.03 (0.22) 1.04
ADAPT_RAIN 0.73 (0.21)*** 2.08 0.60 (0.22)*** 1.83 0.57 (0.23)** 1.77 0.75 (0.24)*** 2.13
REALLOCATE_RISK -0.37 (0.27) 0.69 -0.48 (0.27)* 0.62 -0.49 (0.28)* 0.61 -0.52 (0.29)* 0.60
MRKT_RISK 0.41 (0.13)*** 1.50 0.40 (0.13)*** 1.50 0.38 (0.13)*** 1.46 0.32 (0.14)** 1.38
ILL_RISK 0.08 (0.11) 1.08 0.09 (0.11) 1.09 0.07 (0.12) 1.07 0.04 (0.12) 1.04
FARM_RISK 0.22 (0.13)* 1.25 0.28 (0.13)** 1.32 0.29 (0.13)** 1.34 0.25 (0.14)* 1.29
EX_RISK 0.07 (0.11) 1.07 -0.02 (0.11) 0.98 -0.02 (0.12) 0.98 -0.05 (0.13) 0.95
PRECIP_RISK -0.30 (0.13)** 0.74 -0.31 (0.14)** 0.73 -0.30 (0.14)** 0.74 -0.47 (0.16)*** 0.62
PREDICT_RISK -0.47 (0.12)*** 0.63 -0.45 (0.12)*** 0.63 -0.44 (0.13)*** 0.64 -0.48 (0.13)*** 0.62
SPRING_RISK 0.01 (0.11) 1.01 0.03 (0.11) 1.03 0.03 (0.11) 1.03 0.07 (0.12) 1.08
WATER_IMPACT -0.02 (0.11) 0.98 -0.05 (0.11) 0.95 -0.06 (0.11) 0.94 0.02 (0.12) 1.02
PROFIT_IMPACT -0.26 (0.10)** 0.77 -0.23 (0.11)** 0.79 -0.26 (0.11)** 0.77 -0.21 (0.11)* 0.81
YIELD_IMPACT 0.01 (0.12) 1.01 0.03 (0.12) 1.03 0.04 (0.12) 1.04 0.00 (0.12) 1.00
CLIMATE -0.50 (0.28)* 0.61 -0.52 (0.28)* 0.60 -0.51 (0.28)* 0.60 -0.64 (0.29)** 0.53
SELFEFFICACY _2 1.62 (0.25)*** 5.04 1.60 (0.26)*** 4.97 1.57 (0.26)*** 4.81 1.57 (0.27)*** 4.80
SELFEFFICACY _3 1.92 (0.32)*** 6.85 1.90 (0.32)*** 6.66 1.83 (0.34)*** 6.26 1.87 (0.34)*** 6.49
SELFEFFICACY _4 3.02 (0.51)*** 20.50 3.07 (0.51)*** 21.63 3.06 (0.53)*** 21.42 3.06 (0.54)*** 21.29
MULCH
0.29 (0.21) 1.33 0.27 (0.22) 1.31 0.11 (0.23) 1.11
GREENHOUSE
0.86 (0.42)** 2.37 0.83 (0.42)** 2.30 0.85 (0.43)** 2.33
IRR_SCHD 0.39 (0.26) 1.48 0.42 (0.26)* 1.52 0.39 (0.26) 1.47
IRRIGATION -0.11 (0.30) 0.90 -0.15 (0.31) 0.86 0.13 (0.33) 1.14
Explanatory Variable
Human capital
(model 5)
Material and non-material
resources (model 6)
Wealth and financial
capital (model 7)
Institutions and
entitlements (model 8)
Coefficienta,b
Odds
Ratio Coefficient
a,b
Odds
Ratio Coefficient
a,b
Odds
Ratio Coefficient
a,b
Odds
Ratio
INC_2 0.08 (0.28) 1.09 0.20 (0.29) 1.22
INC_3 0.00 (0.32) 1.00 -0.05 (0.34) 0.95
INC_4 0.09 (0.31) 1.09 0.14 (0.32) 1.15
CREDIT 0.07 (0.22) 1.07 0.17 (0.23) 1.18
MRKTACCESS 0.30 (0.35) 1.35 0.09 (0.37) 1.10
EXTENSION 0.28 (0.23) 1.32
REALLOCATE -0.08 (0.33) 0.93
COOP -0.35 (0.31) 0.71
MIZHI 0.72 (0.39)* 2.05
HONGSIPU -0.31 (0.35) 0.73
DIST_MIDDLE -0.28 (0.24) 0.75
DIST_FAR -0.87 (0.28)*** 0.42
LR chi-square 150.32*** 161.84*** 161.12*** 179.22***
Psuedo R2 0.15 0.16 0.16 0.18
Log likelihood -418.06*** -413.22** -412.66 -403.61** a Unstandardized regression coefficients (standard errors)
b *p<0.1, **p<0.05, ***p<0.01