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

moburnha@esf.edu

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