environmental science & policy 12 (2009) 413–426

available at www.sciencedirect.com

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

Adaptation to climate change in Ethiopia and South Africa:

options and constraints

Elizabeth Bryan a,*, Temesgen T. Deressa b, Glwadys A. Gbetibouo b, Claudia Ringler a

a
International Food Policy Research Institute, 2033 K Street NW, Washington, DC 20006, USA
b
Center for Environmental Economics and Policy in Africa, Agricultural Annex, University of Pretoria, Pretoria, South Africa

article info abstract

Climate change is expected to adversely affect agricultural production in Africa. Because

Published on line 4 January 2009
agricultural production remains the main source of income for most rural communities in
the region, adaptation of the agricultural sector is imperative to protect the livelihoods of the
Keywords:
poor and to ensure food security. A better understanding of farmers’ perceptions of climate
Climate change
change, ongoing adaptation measures, and the decision-making process is important to
Adaptation
inform policies aimed at promoting successful adaptation strategies for the agricultural
Ethiopia
sector. Using data from a survey of 1800 farm households in South Africa and Ethiopia, this
South Africa
study presents the adaptation strategies used by farmers in both countries and analyzes the
Perceptions
factors influencing the decision to adapt. We find that the most common adaptation
strategies include: use of different crops or crop varieties, planting trees, soil conservation,
changing planting dates, and irrigation. However, despite having perceived changes in
temperature and rainfall, a large percentage of farmers did not make any adjustments to
their farming practices. The main barriers to adaptation cited by farmers were lack of access
to credit in South Africa and lack of access to land, information, and credit in Ethiopia. A
probit model is used to examine the factors influencing farmers’ decision to adapt to
perceived climate changes. Factors influencing farmers’ decision to adapt include wealth,
and access to extension, credit, and climate information in Ethiopia; and wealth, govern-
ment farm support, and access to fertile land and credit in South Africa. Using a pooled
dataset, an analysis of the factors affecting the decision to adapt to perceived climate
change across both countries reveals that farmers were more likely to adapt if they had
access to extension, credit, and land. Food aid, extension services, and information on
climate change were found to facilitate adaptation among the poorest farmers. We conclude
that policy-makers must create an enabling environment to support adaptation by increas-
ing access to information, credit and markets, and make a particular effort to reach small-
scale subsistence farmers, with limited resources to confront climate change.
# 2008 Elsevier Ltd. All rights reserved.

1. Introduction change is inevitable even if the international community

would exert its best effort to mitigate GHG emissions. Poor
Numerous studies have shown that atmospheric concentra- countries, which rely heavily on agricultural production, are
tions of greenhouse gases (GHG) are increasing as a result of expected to be most vulnerable to climate change and climate
human activity. Many scientists agree that some climate variability. The IPCC (2007) projects that in Sub-Saharan

* Corresponding author. Tel.: +1 202 862 5605.

E-mail address: [email protected] (E. Bryan).
1462-9011/$ – see front matter # 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envsci.2008.11.002
414 environmental science & policy 12 (2009) 413–426

Africa, warming is expected to be greater than the global as other drivers such as economic conditions, population
average and in parts of the region rainfall will decline. Even in growth, and government policies. One important question
East Africa, where most climate change models predict that arises in the literature is whether farmers adapt their
increased rainfall due to climate change, recent research behavior in response to short-term climate variability or long-
suggests that local circulation effects will result in decreased term climate change (Smit et al., 1996; Burton, 1997; Bryant
precipitation instead (Funk et al., 2008). These anticipated et al., 2000; Thomas et al., 2007). Some argue that adaptation to
climate changes pose great threats to food and water security, short-term climate variability may facilitate adaptation to
public health, natural resources, and biodiversity (McCarthy long-term climate changes (Burton, 1997). However, some
et al., 2001). adaptations taken in response to short-term climate varia-
Because agricultural production remains the main source bility, which could be classified as coping responses, may not
of income for most rural communities, adaptation of the be well suited for or could increase vulnerability to long-term
agricultural sector to the adverse effects of climate change will climate change (Smithers and Smit, 1997; Ziervogel et al.,
be imperative to protect the livelihoods of the poor and to 2008). Such ‘‘maladaptations’’ may serve short-term goals, but
ensure food security. Adaptation can greatly reduce vulner- come with future costs to society (Smithers and Smit, 1997).
ability to climate change by making rural communities better Thus, there is the need to anticipate long-term changes and
able to adjust to climate change and variability, moderating make the appropriate adjustments in addition to coping with
potential damages, and helping them cope with adverse current climate conditions. This strategic adaptation should
consequences (IPCC, 2001). A better understanding of how involve government intervention to promote and guide
farmers’ perceive climate change, ongoing adaptation mea- adaptation of the agricultural system (Smithers and Smit,
sures, and the factors influencing the decision to adapt 1997).
farming practices is needed to craft policies and programs The system characteristics component refers the condi-
aimed at promoting successful adaptation of the agricultural tions of the agricultural system including its sensitivity,
sector. Adaptation will require the involvement of multiple resilience, vulnerability, adaptive capacity, and other factors
stakeholders, including policymakers, extension agents, influencing its response to stressors (Smithers and Smit, 1997;
NGOs, researchers, communities, and farmers. Bryant et al., 2000). Other factors include the socioeconomic,
Using data from a survey of 1800 farm households in South cultural, political, and institutional characteristics, which can
Africa and Ethiopia, this study examines farmers’ perceptions either facilitate or hinder the adaptation process. The third
of climate change, the extent of adaptation, barriers to component refers to the multiple scales at which adaptation
adaptation, and the factors influencing adaptation and occurs. Climate impacts, adaptive capacity, and adaptation
adaptation choices. The paper is organized as follows. Section responses differ across multiple scales from the plot and farm
2 reviews the literature on adaptation to climate change and levels to the country and international levels. Therefore,
lays out the analytical framework for analysis. Section 3 analyzing the adaptive capacity of a system and appropriate
describes the dataset and analyzes farmers’ perceptions of adaptation responses should take into account the scale of
climate change, the adaptations taken by farmers, the analysis (Vincent, 2007).
constraints to adaptation, and the characteristics of the Numerous studies have identified, classified, and assessed
farmers in each country. This section explores questions adaptation responses, the fourth component in this frame-
regarding how farmers actually perceive climate change and work (Burton, 1997; Smithers and Smit, 1997; Bryant et al.,
what motivates adaptation. Section 4 presents the methodol- 2000; Dolan et al., 2001; Smit and Skinner, 2002; Svendsen,
ogy used to quantitatively analyze the factors influencing 2008). Responses to climate change can be either reactive or
farmers’ decisions to change their farming practices. The proactive (Smithers and Smit, 1997; Bryant et al., 2000).
factors shown to have an influence on farm-level adaptation Adaptive responses can also be classified according to other
based on the probit estimations are presented in Section 5. characteristics including their timing with respect to the
This section explores the decision to adapt across and within climate stress; duration (short- or long-term); form/type
both countries, and by the economic position of the house- (technological developments, government programs and
hold, defined by income terciles. Section 6 concludes with insurance, farm production practice, and farm financial
recommendations for policymakers. management); and effect (enhanced stability or resilience)
(Smithers and Smit, 1997; Smit and Skinner, 2002).
The nature of the response depends on the other
2. Literature review components of adaptation discussed above: the degree of
exposure to and nature of the stress, the properties of the
Agricultural adaptation to climate change is a complex, multi- system exposed to stress, and the scale and magnitude of the
dimensional, and multi-scale process that takes on a number event (Smithers and Smit, 1997). Burton (1997) argues that
of forms (Bryant et al., 2000). Bryant et al. (2000) identify four willingness to adapt to climate change and variability depends
main components of adaptation: (1) the characteristics of the on experience, time horizon, and the risk tolerance of
stress, (2) the characteristics of the system, including the individual decision-makers. Risk denial has been documented
cultural, economic, political, institutional and biophysical with respect to climate hazards—as the likelihood and
environment, (3) multiple scales, and (4) adaptive responses. frequency of extreme events increases due to climate change,
The first component, characteristics of the stress, refers to adaptation can be expected to increase as well.
the stimulus to which actors and systems respond. These A number of studies look particularly at the factors that
include climate signals (climate change and variability) as well influence farmers’ decision to adapt to climate change at the
 environmental science & policy 12 (2009) 413–426 415

farm level (Granjon, 1999; Roncoli et al., 2002; Hansen et al., more prominent role in determining farmers’ livelihood
2004; Vogel and O’Brien, 2006; Ziervogel et al., 2005). These strategies than climatic risk. Similarly, Adger (2000) found
studies examine farmers’ perceptions, use of information, and that Vietnam’s transition from state central planning
other factors influencing the decision-making process. The increased social vulnerability to climate change by decreasing
literature suggests that farmers’ perceptions of climate collective action for risk management by state institutions.
change and their behavioral responses may be more related This led to the emergence of civil institutions to mediate
to recent climate events or trends as opposed to long-term vulnerability to environmental change (Adger, 2000). Social
changes in average conditions (Smit et al., 1997; Granjon, 1999 networks and social capital make it possible for individuals
in Bryant et al., 2000; Thomas et al., 2007). Moreover, many and communities to organize collectively to manage climate
studies stress the importance of local knowledge in decision- risks (Adger, 2003; Pelling and High, 2005). Examples of how
making regarding climate risk (Roncoli et al., 2001, 2002; Vogel social networks can facilitate adaptation include providing
and O’Brien, 2006; Thomas et al., 2007). That is, farmers base informal sources of credit, and spreading information about
their decision to adapt their farming practices not only on climate change and appropriate adaptations.
changes in average conditions, but on a number of other A few studies also make use of household datasets to
climate factors observed through personal experience such as empirically examine the factors influencing adaptation. These
extreme events; rainfall frequency, timing, and intensity; and studies of farm-level adaptation confirm that farmers respond
early or late frosts. (Smithers and Smit, 1997; Roncoli et al., not only to climate stimuli but a number of other factors as
2002; Vogel and O’Brien, 2006; Thomas et al., 2007). well (Smit et al., 1996; Brklacich et al., 1997; Bryant et al., 2000;
Examining the role of forecast climate information in Bradshaw et al., 2004; Belliveau et al., 2006; Maddison, 2007;
decision-making, Hansen et al. (2004) suggest that information Nhemachena and Hassan, 2007). Therefore, farm-level
derived from personal experience and information from changes that might be expected given a certain climate signal
external description yield different choice results under may not actually occur due to other intervening factors, such
conditions of climate risk and uncertainty—decisions based as personal characteristics, economic conditions, and the
on personal experience are likely to give greater weight to recent policy environment (Bradshaw et al., 2004). Using data from
events. Ziervogel et al. (2005) find that the use of accurate over 15,000 Canadian prairie farms, the authors find that while
climate forecasts can improve household well-being while poor crop diversification would reduce risks from climate change
forecast information can actually be harmful to poor farmers. In and variability, farmers in the region are actually becoming
addition, the ability to respond to climate forecasts and the more specialized given economic considerations, such as the
benefits obtained from their use are determined by a number of high start-up costs and implications for achieving economies
factors including the policy and institutional environment and of scale (Bradshaw et al., 2004).
the socio-economic position of the household (Ziervogel et al., Smit et al. (1996) find that some farmers in southwestern
2005; Vogel and O’Brien, 2006). Given the potential for rural Ontario adopted short-term managerial adjustments or
climate information to support adaptation and manage climate more strategic adaptations in response to having experi-
risk, there is a need to make climate information more accurate, enced recent dry years, while most farmers reported no
accessible, and useful for farmers (Roncoli et al., 2002; Ziervogel purposeful response. The propensity to respond was related
et al., 2005; Hansen et al., 2007). Promoting the use of climate to farmers’ perceptions of dry-year frequencies, indicating
information for adaptation among the poorest farmers also that the strength of the climate signal influences adapta-
requires resources needed to implement adaptation options tion. The authors also indicate that the impact of climate
(Vogel and O’Brien, 2006). changes and variations cannot be understood without
Many studies underscore the importance of formal and considering the role of other economic policy and environ-
informal institutions and social relationships in facilitating or mental forces that not only influence how the climate
hindering adaptation to climate change (Agarwal, 2008; signals are felt, but also have a direct impact farm-level
Agarwal and Perrin, 2008; Isham, 2002; Eakin, 2005). Many decisions. Similarly, Brklacich et al. (1997) find that despite
studies describe the ways in which institutions influence and having perceived climate changes, Canadian farmers did
shape adaptation as well as climate vulnerability: they not adapt their farming practices. The lack of response was
structure the distribution of climate impacts, they shape the attributed to built in resilience of the agriculture system and
ability of individual actors to respond to climate change and declining relative importance of climate in relation to other
the options they choose, and they deliver and govern access to factors influencing farm-level decision-making (Brklacich
external resources to facilitate adaptation (Adger, 2000; et al., 1997).
Agarwal, 2008; Agarwal and Perrin, 2008). These studies also A study of Canadian farmers by Bryant et al. (2000) shows
highlight the potential for rural institutions to strengthen that farmers’ responses vary when faced with the same
adaptive capacity and facilitate local level adaptation to climate stimuli, even within the same geographic area, given
climate change (Adger, 2000; Agarwal, 2008; Agarwal and different agricultural systems and markets systems in which
Perrin, 2008). For instance, extension services have the farmers operate as well as different individual characteristics
potential to influence farmers’ decision to change their and contexts such as personal managerial style and entre-
farming practices in response to climate change (Maddison, preneurial capacity and family circumstances. Supporting the
2007; Nhemachena and Hassan, 2007). notion that personal characteristics and conditions influence
Based on case studies conducted in Mexico, Eakin (2003, adaptation, several studies find that farming experience,
2005) argues that institutional change, such as economic socioeconomic position, and access to resources, credit, and
restructuring, influences adaptive capacity and plays an even extension services increase the probability of uptake of
416 environmental science & policy 12 (2009) 413–426

adaptation measures to climate change (Maddison, 2007; Basin in collaboration with the Ethiopian Development
Nhemachena and Hassan, 2007). Furthermore, the nature of Research Institute (EDRI).
farmers’ response to climate change and variability also The South Africa dataset contains 800 observations from 19
depends on the socioeconomic position of the household— districts of 4 provinces of South Africa (Limpopo, North West,
poor farmers are likely to take measures to ensure their Mpumalanga, and Gauteng). The survey design aimed to
survival while wealthier farmers make decisions to maximize capture all the diverse agricultural patterns in the basin area
profits (Ziervogel et al., 2006). including farming strata, type of cultivation (dry land and
The literature on technology adoption also offers some irrigation), major and minor crops and livestock, and all the
important insights into the factors influencing farmers’ sub-catchment areas in each of the 5 water management areas
decision-making process. Although agricultural adaptation (WMAs) located in the Limpopo River Basin. Farmers were
involves far more than adopting a new technology, introdu- carefully selected with the assistance from producers’
cing new technology certainly plays a role in adaptation at the associations and the National Department of Agriculture.
farm-level. This literature reveals that a number of individual, The Ethiopia dataset contains 1000 observations from 20
household and farm characteristics, and institutional (market, districts (woredas) in 5 regions in Ethiopia (Tigray, Amhara,
policy, social) factors influence farmers’ decisions. Higher Oromiya, Benishangul Gumuz, and Southern Nations Nation-
levels of education are associated with access to information alities and Peoples (SNNP)). The sample districts were
on improved technologies and the adoption of improved purposely selected to include the different attributes of the
technologies (Norris and Batie, 1987; Igoden et al., 1990; Lin, basin including traditional typology of agro-ecological zones
1991). Some studies indicate that larger families may enable in the country, degree of irrigation activity (percent of
the household to accomplish a variety of agricultural tasks cultivated land), average annual rainfall, rainfall variability,
given fewer household labor shortages (Croppenstedt et al., and vulnerability (food aid dependent population). Peasant
2003; Nhemachena and Hassan, 2007). Studies also tend to associations (administrative units lower that districts) were
agree that households with higher income and greater assets also purposely selected to include households that irrigate
will be more likely to adopt new farming technologies, given their farms. One peasant association was selected from every
greater access to information and financial resources (Franzel, district making both the number of districts and peasant
1999; Knowler and Bradshaw, 2007). associations to be 20. Random sampling was used in selecting
Gender of the head of the household is also considered to 50 households from each peasant administration within the 20
influence the uptake of new technologies (Asfaw and woredas. After cleaning the data, the pooled dataset contains a
Admassie, 2004; Tenge et al., 2004; Nhemachena and Hassan, total of 1783 observations.
2007). However, depending on the context, studies differ on The survey collected data on the socioeconomic character-
whether male or female headed households are more likely to istics of the household; household expenditures; shocks
adopt new technologies. Farming experience and age also experienced by the household; land tenure; crop and livestock
appear to be significant determinants of technology adoption, production (inputs and costs); access to extension, markets
although the direction of the effect varies across studies. and credit; perceptions of climate change; adaptation
Studies in Ethiopia have shown a positive relationship responses; and constraints to adaptation. Farmers were asked
between number of years of experience in agriculture and open-ended questions about whether they had noticed any
the adoption of improved agricultural technologies (Kebede long-term changes in mean temperature and rainfall, and the
et al., 1990). However, a study by Shiferaw and Holden (1998) direction of the change. Questions about adjustments made in
indicated a negative relationship between age and adoption of response to perceived changes in temperature and rainfall and
improved soil conservation practices, which suggests that the constraints to adaptation were also posed. The survey
older farmers may be less willing to take the risks associated module on perceived climate change, adaptation responses
with new farming practices and technologies. and constraints, which contains the exact wording of these
questions, is included in the Supplementary Material section.

3. Data 3.2. Farm-level adaptation to climate change,

perceptions and barriers
3.1. Data sources and description of the survey
As the literature suggests, agricultural adaptation is a complex
The household-level data used in this study were collected as process involving action on a number of scales. This study is
part of the project ‘‘Food and Water Security under Global based on a dataset which looks at a specific sub-set of
Change: Developing Adaptive Capacity with a Focus on Rural adaptations at the farm level. That is, this study aims to
Africa’’ implemented by the International Food Policy describe actual adaptations taken at the farm level and to
Research Institute, funded by the German Federal Ministry analyze the factors that influence the decision to adapt.
for Economic Cooperation and Development (BMZ). The Farmer’s adaptation strategies were ascertained by asking
survey in the Limpopo Basin, South Africa, was carried out farmers open-ended questions about the actions they take to
in collaboration with the Center for Environmental Economics counteract perceived changes in temperature and rainfall. The
and Policy in Africa (CEEPA), University of Pretoria between way in which the question was phrased aimed to separate out
August and November 2005, covering the agricultural season a farmer’s response to perceived long-term climate changes
April/May 2004 to April/May 2005. In Ethiopia, the survey was from their response to other stimuli (such as short-term
conducted during the 2004/2005-production year in the Nile climate variability, economic factors, policies, etc.) Farmers
 environmental science & policy 12 (2009) 413–426 417

were asked ‘‘Have you noticed any long-term changes in the tures have in fact increased over the period from 1960 to 2003,
mean temperature/precipitation over the last 20 years?’’ If that with the increase mostly in summer. According to data from
question was too difficult farmers were asked ‘‘Has the South African Weather Service (SAWS), over this 43-year
number of hot days/rainfall days stayed the same, increased period, temperatures increased by around 1 8C. However, in
or declined over the last 20 years?’’ These questions were contrast to farmers’ perceptions, there is no clear statistically
designed to get a sense of farmers’ perceptions of long-term significant trend of declining rainfall over the 1960–2003 period
climate change. Farmers were then asked ‘‘What adjustments (SAWS). The high proportion of farmers noticing a decrease in
in your farming have you made to these long-term shifts in precipitation could be explained by the fact that during the last
temperature/rainfall?’’ This variable, which intends to capture few years (2001–2003), there was a substantial drop in the
farm-level adaptation to long-term changes in temperature volume of precipitation. Thus, farmers’ perceptions may be
and rainfall versus short-term climate variability, is used to based more on recent, short-term trends rather than long-
construct the dependent variable for the analysis. We then term changes (Gbetibouo, forthcoming). Moreover, Gbetibouo
analyze farmers’ responses to determine the factors which (forthcoming) finds that having access to water for irrigation
influence their decision to adapt or not. decreases the likelihood that the farmer perceives climate
change in South Africa. On the other hand, having access to
3.2.1. Farmers’ perceptions of climate change extension services increases the likelihood of perceiving
In order to adapt to climate change, farmers must first perceive climate change (Gbetibouo, forthcoming). This suggests that
that changes are taking place. The survey data show that, in perceptions are not entirely based on actual climate condi-
both countries, a large share of farmers perceive temperature tions and changes but are influenced by other factors as well.
has been increasing over time, and that rainfall has been In Ethiopia, farmers’ perceptions of climate change appear
decreasing. In South Africa, 86 percent of farmers responded to reflect actual temperature and rainfall data obtained from
that temperature has increased and 79 percent of farmers noted National Meteorological Services Agency (2001), which show
that rainfall has declined (see Fig. 1). A few farmers reported an increasing trend in temperature and decreasing trend in
seasonal changes in rainfall patterns (i.e. changes in the timing precipitation between 1952 and 2000. According to NMSA
of rainfall) (5 percent) and others reported both decreased (2001), the average annual minimum temperature over the
rainfall and seasonal changes (7 percent). country has been increasing by about 0.25 8C every 10 years
The results for Ethiopia show a significant number of while average annual maximum temperature has been
farmers also observed that temperature had increased (64 increasing by about 0.1 8C every decade. However, while there
percent) and that rainfall had decreased (65 percent) (see Fig. 2). is a decreasing trend in rainfall, which matches farmers’
Compared to South Africa, a greater number of farmers in perceptions, this downward trend is not statistically signifi-
Ethiopia reported that temperature (27 percent) and rainfall (17 cant. Moreover, a number of household and individual
percent) conditions had not changed over the 20-year period. characteristics were shown to influence farmers’ perceptions
For the most part, farmer’s perceptions of climate changes of climate change (Deressa, 2008). For instance, age, farm and
appear to be in line with actual climate data. However, there is non-farm income, and access to information and extension
some evidence that farmers’ perceptions of climate change services all influenced the likelihood of perceiving changes in
may not be based entirely on actual changes in climate climate (Deressa, 2008). Again, this suggests that farmers’
(Deressa, 2008; Gbetibouo, forthcoming). Statistical records of perceptions are not fully determined by actual long-term
actual temperature data in South Africa show that tempera- changes in climate.

Fig. 1 – Farmers’ perceptions of average temperature and rainfall changes in South Africa over the last 20 years (n = 783).
418 environmental science & policy 12 (2009) 413–426

Fig. 2 – Farmers’ perceptions of average temperature and rainfall changes in Ethiopia over the last 20 years (n = 1000).

Other studies confirm that while farmers report to be adaptation measures. Overall, 37 percent of farmers in
observing climate change, their perceptions of climate change Ethiopia did not adapt to either perceived changes in rainfall
are more related to recent experience (Smit et al., 1997; or temperature. Fig. 3 presents the range of adaptation options
Granjon, 1999 in Bryant et al., 2000). In addition, farmers are employed by farmers in Ethiopia. Only adaptation responses
often concerned with and adapt to climate characteristics mentioned by more than 1 percent of farmers are reported.
observed through personal experience (Smithers and Smit, Among those farmers who did adapt to climatic changes, the
1997; Roncoli et al., 2002; Vogel and O’Brien, 2006). These most common adaptation strategies include use of different
characteristics, such as rainfall frequency, timing, and crops or crop varieties, planting trees, soil conservation,
intensity; early or late frosts; and extreme temperatures; changing planting dates, and irrigation. Other responses
which influence farmers’ behavior are not captured in this reported less frequently, such as seeking off-farm activities,
study. Therefore, while this survey focused on measuring migrating to urban areas, changing farming type, using new
adaptation to changes in average climate conditions over the technologies, and water conservation, are included in the
long run, it is clear that other climate variables and factors are ‘‘other’’ category.
important determinants of farmers’ perceptions and behavior. In South Africa, 95 percent of farmers perceived changes in
Moreover, Hansen et al. (2004) find that ‘‘. . . farmers’ temperature and 97 percent felt that rainfall had changed, yet
memory of past climatic variability may be distorted in 66 percent of those who responded did not adapt to perceived
systematic ways, reflecting wishful thinking by distortions temperature changes and 65 percent did not adapt to
consistent with decision goals as well as being shaped by perceived rainfall changes. Overall, 62 percent of farmers
personality characteristics and preexisting beliefs . . .’’ Studies did not adapt to either changes in temperature or rainfall. In
by Deressa (2008) and Gbetibouo (forthcoming) confirm this South Africa, common measures used by farmers include
finding by pointing to a number of individual characteristics planting different crops or crop varieties, irrigation, changing
and factors which appear to influence farmers’ perceptions of planting dates, changing the amount of land under cultivation,
climate conditions. and supplementing animal feeds. Other responses, such as
More studies are needed to uncover the process by which crop diversification, which were reported by less than 1
farmers make decisions under uncertain and variable condi- percent of farmers, are included in the ‘‘other’’ category
tions and the relative importance of different risks to the (Fig. 4).
decision-making process. Given that few farmers adjusted their farming practices in
response to perceived changes in climate, it is clear that
3.2.2. Adaptation to climate change farmers’ decisions are not based purely on the long-term
A number of studies describe the disconnect between farmers’ changes in average climate conditions. Furthermore, a small
perceptions of climate change and actual adaptation (Smit number of farmers who responded that they did not perceive
et al., 1996; Brklacich et al., 1997; Granjon, 1999). Similarly, this long-term climate changes responded that they did change
study shows that despite having perceived changes in their farming practices in response to climate changes (37
temperature and rainfall, a large share of farmers in both farmers). These findings suggest that a number of other
countries did not take any adaptive measures. In Ethiopia, 73 factors influence the decision-making process as well,
percent and 83 percent of farmers perceived changes in including changes in other climate signals, such as short-
temperature or rainfall, respectively, but 56 percent and 41 term variability, extreme weather events, and rainfall inten-
percent of farmers, respectively, did not undertake any sity, timing, duration and frequency. Other factors including
 environmental science & policy 12 (2009) 413–426 419

Fig. 3 – Adaptation to perceived temperature and rainfall changes in Ethiopia (n = 995).

the economic or institutional environment and the character- of the various factors shaping farmers decision to adapt and
istics of the farm household also potentially influence farmers’ their adaptation response it is necessary to explore farmers
decision-making. The factors influencing farmers’ decision to own perceptions of the barriers they face.
adapt to perceived climate changes are further explored below The survey data on which this study is based contains
in Section 5. information on which factors farmers perceive to be the most
important barriers to changing their farming practices.
3.2.3. Barriers to adaptation Farmers that did not adjust their farming practices in response
The literature points to many factors which affect farmers’ to perceived climate change were asked ‘‘what were the main
ability to adapt to climate change. These factors include constraints/difficulties in changing your farming ways?’’
accessibility and usefulness of climate information (Roncoli While farmers referred to a number of barriers to adaptation,
et al., 2002), the policy and institutional environment (Eakin, the most important barriers cited by farmers were shortage of
2003, 2005; Agarwal, 2008), and the socio-economic position of land in Ethiopia (Fig. 5) and lack of access to credit/money in
the household (Ziervogel et al., 2006), among others. The South Africa (Fig. 6). In South Africa, 36 percent of farmers
influence of many of these factors on adaptation is examined pointed to the lack of access to credit/money as a serious
in the following sections using econometric techniques. constraint to adaptation. Other important barriers were: lack
However, in order to get a sense of the relative importance of access to water for irrigation (8 percent), lack of information

Fig. 4 – Adaptation to perceived temperature and rainfall changes in South Africa (n = 794).
420 environmental science & policy 12 (2009) 413–426

individual and household characteristics, socioeconomic

status, farm characteristics, and contextual factors (such as
access to extension and credit and distance from markets). For
data on the independent variables used in the analysis please
refer to Table 1 in the Supplementary Material section.
A comparison of the individual and household character-
istics across the two countries reveals that farmers in South
Africa tend to receive more education with an average of 7
years compared to 2 years in Ethiopia. Household size is
relatively large in both countries with an average of 6
household members. In South Africa, farmers tend to be
older on average (55 years) compared to Ethiopia (45 years).
However, in this analysis, age is not viewed as indicator of
experience because many farmers in South Africa began
Fig. 5 – Barriers to adaptation in Ethiopia. farming much later in life.
In terms of socioeconomic status, it is clear that farmers in
South Africa are better-off economically than farmers in
Ethiopia. While roughly the same percent of households have
about climate change and appropriate adaptation responses access to non-farm sources of income (26 percent in South
(5 percent), lack of market access (3 percent), and insecure Africa and 29 percent in Ethiopia), farmers in South Africa earn
property rights (1 percent). In Ethiopia shortage of land was much more farm income than farmers in Ethiopia (7646 USD
reported by 27 percent of respondents as a major constraint to compared to 266 USD). A larger percentage of farmers in South
adaptation, while lack of information (23 percent), lack of Africa also have a greater asset base than farmers in Ethiopia—
credit/money (21 percent), shortage of labor (8 percent), and a larger percentage of households in South Africa own radios;
lack of access to water for irrigation (3 percent) were also cars; refrigerators; stone, brick or concrete homes; and cellular
viewed as significant barriers. It is important to note that these phones; and have electricity. Farm households in Ethiopia
results are responses to the survey question ‘‘What are the tend to receive more food aid (24 percent compared to 7
main constraints/difficulties in changing your farming ways.’’ percent), while farmers in South Africa receive more farm
However, the main barriers reported by farmers apply more to support (21 percent compared to 5 percent). Farm support
certain adaptation measures than others. For instance, accounts for other inputs such as fodder for livestock, seeds,
shortage of land is more of a constraint to soil and water tools, machinery, and subsidies provided by the government.
conservation but not necessarily changing crop variety. The third set of variables, farm characteristics, shows that
Detailed questions were also posed to farmers regarding their farmers in Ethiopia tend to farm on larger plots of land (3 ha vs.
reasons for not adopting specific measures. 2 ha in South Africa) and have more fertile soils. However,
farmers in South Africa are more likely to have a public or
3.2.4. Characteristics of farmers in Ethiopia and South Africa private land title and more tools and farming equipment.
As suggested by the literature and the descriptive statistics Variables capturing the context in which the farmers
presented above, farmers’ decisions to adjust their farming operate show that farmers in Ethiopia tend to have greater
practices are influenced by a number of factors in addition to access to markets given their close proximity (6 km), while
the climate stimulus. Therefore, several variables reflecting farmers in South Africa have to travel farther to reach input
each of these categories were selected as independent (34 km) and output (49 km) markets. Farmers in Ethiopia also
variables for the analysis of farmers’ decision to adapt and appear to have greater access to both formal and informal
their choice of adaptation options. These variables include sources of credit—26 percent of farmers in Ethiopia have
access to formal sources of credit (from commercial banks,
farmer associations/cooperatives, private money lenders,
credit associations, microfinance institutions, NGOs, and the
government) and 30 percent have access to informal sources
of credit (from relatives, neighbors, and churches) compared
to 16 percent and 8 percent in South Africa, respectively.
However, the actual amount borrowed is much higher in
South Africa—the average amount borrowed from both
formal and informal sources in South Africa is 6322 USD
compared to just 39 USD in Ethiopia. A majority of farmers in
both countries have access to extension services, however,
the percent of farmers with access is higher in South Africa
where 78 percent receive training or advice from extension
officers on crop and livestock production compared to 56
percent in Ethiopia.
Farmers’ observations and actual climate data indicate that
Fig. 6 – Barriers to adaptation in South Africa. households in South Africa have been more negatively
 environmental science & policy 12 (2009) 413–426 421

affected by climate factors and extreme events. In South examine how farmers belonging to various income groups
Africa, 63 percent and 33 percent of households claim to have respond to perceived climate changes, the results are also
experienced droughts and floods, respectively, in the last 5 estimated by income terciles using the pooled dataset.
years compared to 37 percent and 14 percent, respectively in Marginal effects are reported for ease of interpretation. All
Ethiopia. The data also show that farmers in South Africa specifications contain dummy variables to control for district-
received less rainfall on average during the growing season level fixed effects. Pooled specifications also contain a dummy
between 1951 and 2000 with 564 mm of rainfall compared to variable to capture differences between the two countries.
972 mm in Ethiopia. However, the average amount of rainfall
variability over the 1951–2000 period is the same for both
countries. 5. Results

5.1. Pooled results

4. Empirical approach
The analysis of farmers’ decision to adapt to perceived climate
To analyze the factors which influence the decision to adapt to change across both countries shows that various socioeco-
perceived climate changes across both countries, a discrete nomic characteristics, contextual factors, and recent extreme
choice model is used. Data from both countries were pooled and climate events influence farmers’ decision to adapt. The results
the dependent variable for adaptation was created. The indicate that broad conclusions can be drawn about the factors
dependent variable is a dummy variable equal to 1 if the influencing adaptation across countries and suggest there may
farmer adopted any of the adaptation options presented above be common avenues for promoting adaptation.
in response to perceived changes in temperature or rainfall and Table 2 presents the marginal effects - the expected change
0 otherwise. Table 1 shows how the observations in the pooled in the probability of adaptation given a unit change in an
dataset are divided across country and adaptation response. independent variable from the mean value, ceteris paribus.
Analysis of this dependent variable requires a binary Only results that were statistically significant at the 10 percent
response model. Two options for this analysis are the logit and level or greater are reported. The results show that larger
probit models. Logit and probit models can be derived from an households are more likely to adapt to perceived climate
underlying latent variable model: change. Increasing household size by one member (from the
mean of 6 household members) increases the probability of
y ¼ b0 þ xb þ e; y ¼ 1½y > 0
adaptation by 1 percent. These results suggest that having
where y* is the unobserved, or latent, variable; x denotes the additional household labor (i.e. husbands, extended family
set of explanatory variables, e is the error term and 1[y* > 0] members, and older children) facilitates adaptation.
defines the binary outcome. The main difference between the
probit and logit models lies in the assumption of the distribu-
tion of the error term, e. The error term is assumed to have the Table 2 – Pooled results, marginal effects reported.
standard logistic distribution in the case of the logit, and the Variables Coefficient P value
standard normal distribution in the case of the probit model.
***
Economists tend to prefer the normality assumption of the HH size 0.012 0.010
Radio 0.056* 0.068
probit model, given that several specification problems are
Land area (ha) 0.016* 0.087
more easily analyzed because of the properties of the normal Land title 0.061* 0.061
distribution (Wooldridge, 2006). Therefore, a probit model is Distance to output markets (km) 0.001** 0.030
used to determine the factors which influence farmers’ deci- Access to formal credit 0.127*** 0.000
sions to adapt to perceived climate change across South Africa Access to extension 0.081** 0.012
and Ethiopia. A set of independent variables were used (dis- Flood in last 5 years 0.138*** 0.000
Country 0.585** 0.035
cussed above), which were comparable across both countries.
The results are also estimated separately for each country District fixed effects Yes
to determine if there are differences in the factors that Obs. 1756
influence farmers’ decision to adapt between the two Pseudo R 2 0.316

countries. This analysis enables one to draw particular Note: Only statistically significant coefficients are shown in the
conclusions about the ways in which adaptation could be table. Variables omitted from the table include: age and education
promoted within the given context of each country. To of the household head, whether the household receives food aid
and farm support, whether the household owns a radio, toilet, and
has electricity, whether the house is made of concrete, brick, or
stone, whether the farm has fertile soil, current value of tools and
equipment, distance to input markets, whether the household has
Table 1 – Distribution of observations by country and
access to non-farm income, informal sources of credit, and
adaptation response.
information on climate change, and whether the household has
South Africa Ethiopia Total experienced a drought in the last 5 years. The full table is available
in Supplementary Material.
Adapted 296 623 919 *
Significant at 10 percent probability level.
Did not adapt 487 374 861 **
Significant at 5 percent probability level.
***
Total 783 997 1780 Significant at 1 percent probability level.
422 environmental science & policy 12 (2009) 413–426

The results also suggest that wealthier households are Ethiopian farmers are poorer and more risk averse than South
more likely to adapt to perceived climate changes. Households African farmers (Yesuf and Bluffstone, 2007). Moreover, the
that own a radio and households with a larger area of land climate signal for Ethiopia is considered to be weaker (IPCC,
were more likely to adapt. The coefficients on many other 2007). These results indicate a greater awareness among
household assets were positive, yet the results were not Ethiopian farmers that the risk of not adapting is greater than
statistically significant. These results support other studies the risk of adapting to climate change. As noted above, far
which find that wealthier households are better able to act fewer households in Ethiopia receive farm support from the
quickly to offset climate risk than poorer households (Adger government (equipment, inputs, subsidies, etc.) compared to
et al., 2001; Downing et al., 2005; Ziervogel et al., 2006). South Africa. Given a more limited social safety net, Ethiopian
Similarly, the coefficients on food aid and farm support were farmers must take measures to reduce their vulnerability to
also positive, but statistically insignificant. Farmers that have climate change. The fact that households that experienced a
a public or private title to their land (as opposed to a communal shock within the past 5 years were more likely to adapt, also
or informal land rights) are 6 percent more likely to adapt. This supports the notion that farmers act to reduce the risk of
suggests that creating stronger individual property rights future climate shocks. Farmers are more willing to adapt when
would promote farm-level adaptation. they are aware of the risks posed by extreme climate events.
Access to rural services, such as extension and credit, also We find that households that wish to reduce the risks
has a positive influence on adaptation. Farmers with access to associated with climate change and have the resources or
extension services are 8 percent more likely to adapt to climate access to resources needed to make the appropriate changes
change and those with access to formal sources credit (from are more likely to take up adaptation. However, while we do
banks, NGOs, microfinance institutions, etc.) are 13 percent not examine particular adaptation options in this analysis, it is
more likely to adapt. These results are not only highly important to point out that some adaptations may be risk
statistically significant, but the magnitude of the effect is also reducing while others may be risk increasing. Poorer farmers
rather large. These results indicate that improving market may be less able to take strategic adaptations to reduce the
access for small-scale subsistence farmers would increase risks imposed by climate change over the long run or to
their ability to adapt to climate change. However, contrary to decrease their vulnerability to climate change by making
what one would expect, the results also show that adaptation decisions that would increase their economic position. Rather
increases with distance to output markets. This suggests that these farmers are forced to employ adaptations to cope with
there is less of an opportunity cost for households in remote short-term climate variability in order to survive (Ziervogel
areas to adopt adaptation practices that may be labor et al., 2006), which may increase long-term risk.
intensive. That is, where fewer income earning opportunities The scope of the present analysis is limited in that
are available, households in remote areas may be more willing adaptations are grouped into one dummy variable so that
to take up adaptation in order to reduce climate-related risk. the decision to adapt could be compared across both study
The coefficient on floods indicates that farmers’ decision to countries where adaptation strategies are different. As
adapt is greatly influenced by perceived extreme weather Bradshaw et al. (2004) point out, this type of analysis is
events. Those farmers that reported experiencing a flood essential to understand the likely uptake of adaptation among
within the last 5 years were 14 percent more likely to change farmers. However, while this analysis reveals what factors
their farming practices. Thus, while farmers are likely to induce farmers to change their behavior, it does not reveal the
employ coping strategies in response to shocks-coping effectiveness of particular adaptations over the long run.
responses reported by farmers in this survey included selling Separate analyses of farmers’ choice of adaptation options in
livestock, borrowing from relatives or the bank, receiving food Ethiopia and South Africa show that the factors influencing
aid, participating in food for work programs, migration, adaptation vary across the different adaptation measures
seeking off-farm employment, and reducing consumption- (Deressa, 2008; Gbetibouo, forthcoming). Therefore, additional
farmers also appear likely to take proactive measures to research is needed to distinguish between adaptations that are
reduce the risk of shocks in the future. However, the risk reducing and risk enhancing over the long run. This
coefficient of variation of rainfall between 1951 and 2000 is information would help policymakers and other stakeholders
statistically insignificant, suggesting that farmers are not craft strategies to promote the kind of adaptation measures
reacting to measured differences in the average amount of that will reduce risk to long-term climate change.
rainfall variability over the long-run. Rather, farmers appear to
adapt to reduce the risk of future extreme weather shocks. 5.2. Results by country1
These results also suggest that while farmers were asked
about adaptations to long-term changes in average tempera- The pooled results reported above may mask the factors which
ture and rainfall conditions, farmers’ behavior depends more influence adaptation in a specific country context. There are
on short-term climate variations and extremes. likely to be differences across countries not only in terms of
Finally, the country dummy shows that farmers in Ethiopia
are more likely to adapt to perceived long-term changes in
1
The factors influencing farmers’ choice of adaptation method
climate than farmers in South Africa. This result reflects the
are examined by Deressa (2008) and Gbetibouo (forthcoming)
descriptive statistics presented above, which show that 62
using multinomial logit models and this same dataset. For more
percent of farmers in Ethiopia adapted to perceived climate detailed results on the determinants of adaptation in Ethiopia
changes while only 38 percent of farmers in South Africa please refer to Deressa (2008) and for more detailed results for
adapted. These results are somewhat surprising given that South Africa please refer to Gbetibouo (forthcoming).
 environmental science & policy 12 (2009) 413–426 423

the extent of adaptation but also in terms of the particular climate change. In Ethiopia, access to extension services and
factors that influence adaptation. Therefore, the results were information on climate change and appropriate adaptation
estimated individually for each of the study countries, using responses are important, whereas these factors are not
roughly the same set of explanatory variables. Given that significant in South Africa.
households in Ethiopia tend to be poorer and have fewer large Similarly, other studies have found that a number of factors
assets compared to South African farm households, the set of discourage the uptake and use of climate information in the
assets included in the analysis was limited to those owned by a southern African region, including the type of information
significant percent of households in Ethiopia. Because produced, the way in which it is presented, unequal access to
information was available on the number of relatives in the information, and lack of attention to delivery processes (Vogel
local area in the case of Ethiopia, this variable was added as a and O’Brien, 2006). This suggests that, in South Africa,
measure of social capital. In the case of South Africa, because government agencies, NGOs, extension agents, and other
age is not viewed as an indicator of experience given that actors should work harder to ensure that information on
many farmers in South Africa began farming much later in life, climate change and appropriate adaptation responses is
the number of years farming was included as an explanatory presented in a useful and acceptable way, enabling farmers
variable. This variable was not collected in the case of to reduce the risks posed by climate change. Better commu-
Ethiopia. These results provide more insight into how policy- nication and coordination between relevant actors and end
makers in each of the countries can better promote adaptation users would also strengthen delivery mechanisms and
given the particular context in which the farmers operate. increase the likelihood of use of climate information.
Table 3 in the Supplementary Material section presents the
country-specific results. These results show that there are 5.3. Results by income tercile
both similarities and differences regarding the factors that
influence adaptation in South Africa and Ethiopia. Wealth, Another way to examine the differential influence of the
measured by the households’ ownership of key assets, is determinants of adaptation is to divide the survey households
shown to be an important determinant of adaptation in both into groups according to their income level. Therefore, the
countries. Households in South Africa that own cellular adaptation decisions of households in Ethiopia and South
telephones and cars were more likely to adapt to perceived Africa were examined at various levels of farm income. This
climate changes. In Ethiopia, where farmers do not tend to analysis provides policymakers with information on how to
own cell phones or cars, radio ownership was shown to be an best support adaptation among different groups of farmers. In
important factor influencing adaptation. Moreover, the cur- particular, the results show how policymakers can best
rent value of all farm tools and equipment is also an important support poor farmers, who are most vulnerable to climate
determinant of adaptation in Ethiopia. impacts given limited resources to make changes in their
Government aid is also shown to have a positive effect on farming practices. See Table 4 in the Supplementary Material
adaptation in both countries. In Ethiopia, where 24 percent of section for the results by income tercile.
farmers surveyed receive assistance in the form of food aid, The results show that the factors influencing adaptation
food aid is shown to increase the likelihood of adaptation by 5 vary by economic position. Among the poorest farmers, food
percent. This finding shows that food aid relieves the aid and the value of farm tools and equipment are shown to
consumption pressure many farm households face, enabling increase the probability adaptation. This suggests that a slight
them to take up adaptation options. This result is supported by increase in wealth and well being among the poorest farmers
other studies which show that high consumption risk could significantly increase adaptation by enabling them to
decreases the likelihood of technology adoption (Dercon and undertake measures that require the investment of resources.
Christiaensen, 2007). In South Africa, where farm support is Wealth also appears to enable farmers in the middle income
more common with 21 percent of farmers receiving this form tercile to adapt. Farmers in this income group that own radios
of assistance, farm support increases the likelihood of and have more land area were more likely to take up
adaptation by 12 percent. adaptation. This suggests that farmers in this income group
Access to formal sources of credit is also an important are somewhat constrained by lack of financial resources in
determinant of adaptation in both countries. Farmers with their decision to adapt.
access to credit were 11 percent more likely to adapt in South Farm support (inputs and subsidies) is shown to have a
Africa and 6 percent more likely to adapt in Ethiopia. Another negative influence on adaptation among the poorest while
similarity is that farmers in both countries appear to be those that receive farm support in the highest income tercile are
adapting in part in response to past experience with floods. more likely to adapt. This suggests that government farm
Farmers that reported experiencing a flood in the last 5 years support is tailored to meet the needs of wealthier farmers and
were 19 percent and 4 percent more likely to adapt their while the poor are less able to use inputs and subsidies provided
farming practices in South Africa and Ethiopia, respectively. by the government to make the necessary changes in their
Despite many similarities between the two countries, there farming practices. Policymakers, therefore, should carefully
are also important distinctions. Having access to larger land consider the needs of poor farmers and provide support that
area and fertile soils are shown to be important determinants enables them to reduce the risks of climate change.
of adaptation in South Africa. In addition, number of years Access to land and property rights are important determi-
farming is also significant in the case of South Africa, nants of adaptation for farmers in the middle and highest
suggesting that farmers with more experience are more likely income terciles. From the mean, an additional hectare of land
to change their farming practices in response to perceived increases the likelihood of adaptation by 6 percent among
424 environmental science & policy 12 (2009) 413–426

farmers in the middle income tercile and 2 percent among nants of adaptation decisions based on household surveys
farmers in the highest income tercile. In addition, having a conducted in Ethiopia and South Africa. This analysis aimed to
public or private land title increases the probability of strengthen understanding about farmers’ decision-making
adaptation by 14 percent among farmers in the middle income process to enable policymakers and other stakeholders to
tercile. support adaptation to climate change at the farm-level. While
Rural services increase the likelihood of adaptation across agricultural adaptation to climate change involves more than
income terciles. Access to extension services appears more farm-level changes in farming practices, farm-level adaptations
important for poorer farmers—farmers with access to exten- are an essential component of adaptation of agricultural
sion services in the lowest and middle income terciles were 10 systems. Therefore, it is crucial to understand how the social,
percent and 14 percent more likely to adapt their farming economic, institutional, and ecological context mediates the
practices, respectively. Credit appears to be more important for climate impacts and influences the adaptation response.
wealthier farmers. Farmers in the middle and highest income The results show that farm-level adaptation involves more
terciles with access to formal sources of credit were 15 percent than adopting new agricultural technologies such as improved
more likely to adapt. Receiving information on climate change water storage facilities, additional irrigation, and new crop
also promotes adaptation among farmers in the lowest and varieties. Given the importance influence of having access to
highest income terciles by 7 percent and 9 percent, respectively. extension services and formal sources of credit on farmers’
Thus, improving and expanding the provision of rural services, decision to adapt, policy-makers should extend and improve
particularly extension and the dissemination of climate upon such services, ensuring that they reach small-scale
information, appear to have positive effects on all farmers. subsistence farmers. Providing support to the poorest farmers
However, the information and material support provided to is critically important, given that this group is the most
farmers would have to be tailored to the particular needs of the vulnerable to long-term climate change, and least-equipped to
community based on their economic position, environmental make the changes needed to sustain their livelihoods in the
features of the area, and the expressed need of farmers. face of such a threat. Addressing these market imperfections,
In support of the pooled results, the results by income of lack of access to information and credit, and ensuring
terciles show that farmers respond to recent climate shocks effective targeting requires strong leadership and involvement
(droughts and floods) rather than differences in the amount of of the government in planning for adaptation and implement-
rainfall over time. Farmers in the lowest income tercile were 7 ing measures to facilitate adaptation at the farm level.
percent more likely to adapt if they had experienced a drought The results by country and income terciles suggest that
in the last 5 years while farmers in the highest income tercile strategies should also be tailored to meet the particular needs
were 23 percent more likely to adapt if they had experienced a and constraints of different countries and groups of farmers.
flood. However, recent drought appears to have a negative There are no one-size-fits-all strategies for promoting adapta-
effect on adaptation among farmers in the middle income tion of the agriculture sector. In Ethiopia, the results show that
tercile. For these farmers, drought had a debilitating effect, raising awareness about climate change and the available
reducing the likelihood of adaptation by 11 percent. This adaptation options is important to encourage farmers to
suggests that the poorest farmers are forced to adjust their adapt. Other policy tools for promoting adaptation in Ethiopia
farming practices to cope with climate shocks and minimize include providing farm support (including tools and equip-
risks from future shocks in order to survive while wealthier ment) and supporting the poorest of the poor with food aid and
farmers are better equipped to deal with shocks given the other forms of social assistance.
financial resources at their disposal. Middle income house- In South Africa, policymakers should make it easier for
holds, on the other hand, appear to be less willing to put farmers to receive land titles and have access to more fertile
safeguards in place to protect their livelihoods. This may be lands. Given the fact that very few farmers adapted to
the case, particularly if adaptation options, such as crop perceived climate change in South Africa and the fact that
diversification, have a yield reducing effect—middle income the survey found a much wider array of adaptations being
farmers may be less willing to take up such measures as they taken up by farmers in South Africa, policymakers should also
are not as desperate as the poorest farmers. More research is focus on which adaptations are most needed and make an
needed to uncover the types of adaptation options taken by effort to promote uptake of those options. Because receiving
rich, poor, and middle income farmers. climate information did not increase the likelihood of
The coefficient on the country dummy variable reveals that adaptation in South Africa, greater efforts should be made
Ethiopian farmers in the poorest and middle income terciles to improve the accessibility and usefulness of information
were much more likely to adapt their farming practices. Again provided to farmers to facilitate their adaptation. This would
this suggests that the absence of a strong social safety net involve a large public relations effort aimed at making farmers
increases the vulnerability of these farmers to climate change aware of the need to adapt and the options available to them. It
and variability, causing them to take steps to reduce the threat also entails presenting climate information in a manner that is
to their livelihoods. understandable and useful for farmers.
The results by income terciles show that different factors
influence farmers’ decision to adapt to climate change
6. Conclusions depending on their particular needs and constraints. Govern-
ment aid, extension services, and information on climate
This study examined farmers’ perceptions of long-term climate change appear to facilitate adaptation among the poorest
change, adaptation measures undertaken, and the determi- farmers while wealthier farmers are more likely to adapt given
 environmental science & policy 12 (2009) 413–426 425

access to land, credit and information about climate change. W081-6. International Forestry Resources and Institutions
This analysis masks important differences in the types of Program, University of Michigan.
Asfaw, A., Admassie, A., 2004. The role of education on the
adaptation options chosen by different income groups. More
adoption of chemical fertilizer under different
research is needed on the types of adaptation options that are
socioeconomic environments in Ethiopia. Agricultural
selected by various farmers, and the long-term implications of Economics 30 (3), 215–228.
these choices. Belliveau, S., Bradshaw, B., Smit, B., Reid, S., Ramsey, D.,
While this study focuses on farm-level adaptations, actions Tarleton, M., Sawyer, B., 2006. Farm-level adaptation to
on multiple scales are needed to promote adaptation. multiple risks: climate change and other concerns.
Additional actors, such as the private sector, NGO, local Occasional Paper, 27. University of Guelph, Canada.
Bradshaw, B., Dolan, H., Smit, B., 2004. Farm-level adaptation to
institutions, such as farmers associations, and the media,
climatic variability and change: crop diversification in the
should become more involved in promoting adaptation. Canadian Prairies. Climatic Change 67, 119–141.
Government investments in enhanced and expanded water Brklacich, M., McNabb, D., Bryant, C., Dumanski, I., 1997.
control, development of better crop varieties, and improved Adaptability of agriculture systems to global climatic
crop management practices, such as agro-forestry also require change: a Renfrew County, Ontario, Canada Pilot Study. In:
government support to be taken up by a larger number of Iibery, B., Chiotti, Q., Richard, T. (Eds.), Agricultural
Restructuring and Sustainability: A Geographical
farmers. Ultimately, given the constraints to adaptation
Perspective. CAB International, Wallingford, CT.
highlighted in this study, many farmers may turn to
Bryant, R.C., Smit, B., Brklacich, M., Johnston, R.T., Smithers, J.,
adaptation options outside the agriculture sector, including Chiotti, Q., Singh, B., 2000. Adaptation in Canadian
migration, or finding wage employment. agriculture to climatic variability and change. Climatic
While the purpose of this study is to examine the ways in Change 45, 181–201.
which policymakers can facilitate farm-level adaptations, it is Burton, I., 1997. Vulnerability and adaptive response in the
equally important to determine which adaptations are most context of climate and climate change. Climatic Change 36,
185–196.
effective at increasing farmers’ resilience in a particular
Croppenstedt, A., Demeke, M., Meschi, M.M., 2003. Technology
context. This will require more spatially disaggregated projec-
adoption in the presence of constraints: the case of fertilizer
tions of climate change impacts and more information on the demand in Ethiopia. Review of Development Economics 7
long-term benefits and risks of various adaptation options. (1), 58–70.
Dercon, S., Christiaensen, L., 2007. Consumption risk,
technology adoption, and poverty traps: evidence from
Acknowledgements Ethiopia. World Bank Policy Research Working Paper, 4257.
Available at: SSRN: http://ssrn.com/abstract=995078.
Deressa, T.T., 2008. Determinants of farmers’ choice of
This work is supported by the Federal Ministry for Economic adaptation methods to climate change in the Nile Basin of
Cooperation and Development, Germany under the project Ethiopia. IFPRI Discussion Paper, 798. International Food
‘‘Food and Water Security under Global Change: Developing Policy Research Institute (IFPRI), Washington, DC, 36 pp.
Adaptive Capacity with a Focus on Rural Africa,’’ and forms Dolan, A.H., Smit, B., Skinner, M.W., Bradshaw, B., Bryant, C.R.,
part of the CGIAR Challenge Program on Water and Food. 2001. Adaptation to Climate Change in Agriculture:
Evaluation of Options. Department of Geography
Occasional Paper, 26.
Downing, T.E., Patwardhan, A., Klein, R.J.T., Mukhala, E.,
Stephan, L., Winograd, M., Ziervogel, G., 2005. Assessing
Appendix A. Supplementary data vulnerability for climate adaptation. In: Lim, B., Spanger-
Siegfried, E., Burton, I., Malone, E., Huq, S. (Eds.),
Supplementary data associated with this article can be Adaptation Policy Frameworks for Climate Change:
found, in the online version, at doi:10.1016/j.envsci.2008.11.002. Developing Strategies, Policies and Measures. Cambridge
University Press, Cambridge.
Eakin, H., 2003. Rural Responses to Climatic Variability and
references Institutional Change in Central Mexico. University of
California at San Diego, Center for U.S.-Mexican
Studies 1004, Center for U.S.-Mexican Studies, UC San
Diego.
Adger, W.N., 2000. Institutional adaptation to environmental Eakin, H., 2005. Institutional change, climate risk, and rural
risk under the Transition in Vietnam. Annals of the vulnerability: cases from Central Mexico. World
Association of American Geographers 90 (4), 738–758. Development 33 (11), 1923–1938.
Adger, W.N., Kelly, P.M., Ninh, N.H., 2001. Living with Funk, C., Dettinger, M.D., Michaelsen, J.C., Verdin, J.P., Brown,
Environmental Change: Social Resilience, Adaptation and M.E., Barlow, M., Hoell, A., 2008. Warming of the Indian
Vulnerability in Vietnam. Routledge, London, UK. Ocean threatens Eastern and Southern African food
Adger, W.N., 2003. Social capital, collective action, and adaptation security but could be mitigated by agricultural
to climate change. Economic Geography 79, 387–404. development. PNAS 105 (32), 11081–11086.
Agarwal, A., 2008.In: The role of local institutions in adaptation Franzel, S., 1999. Socioeconomic factors affecting the adoption
to climate change, Paper Prepared for a workshop on potential of improved tree fallows in Africa. Agroforestry
‘‘Social Dimensions of Climate Change’’, organized by the Systems 47 (1–3), 305–321.
Social Development Department, The World Bank, Gbetibouo, G.A. Understanding farmers’ perceptions and
Washington, DC, March 5–6, 2008. adaptations to climate change and variability: the case of
Agarwal, A., Perrin, N., 2008. Climate adaptation, local the Limpopo Basin, South Africa. IFPRI Discussion Paper,
institutions, and rural livelihoods. IFRI Working Paper, forthcoming.
426 environmental science & policy 12 (2009) 413–426

Granjon, D., 1999. Ennquêtes et resultants sur l’adaptation de Shiferaw, B., Holden, S., 1998. Resource degradation and
l’agriculture aux different types de stress: le cas de la zone adoption of land conservation technologies in the Ethiopian
de Napierville, Research Report Submitted to Singh, B. and highlands: case study in Andit Tid, North Shewa.
Bryant, C.R. as Part of a Research Contract with Agricultural Economics 27 (4), 739–752.
Atmospheric Environment Services, Environment Canada, Smit, B., Skinner, M.W., 2002. Adaptations options in
Downsview. Agriculture to climate change: a typology. Mitigation and
Hansen, J., Baethgen, W., Osgood, D., Ceccato, P., Ngugi, R.K., 2007. Adaptation Strategies for Global Change 7, 85–114.
Innovations in climate risk management: Protecting and Smit, B., Blain, R., Keddie, P., 1997. Corn hybrid selection and
building rural livelihoods in a variable and changing climate. climatic variability: gambling with nature? Canadian
Journal of Semi-Arid Tropical Agricultural Research 4 (1). Geographer 42 (4), 429–438.
Hansen, J., Marx, S., Weber, E., 2004. The role of climate Smit, B., McNabb, D., Smithers, J., 1996. Agricultural adaptation
perceptions, expectations, and forecasts in farmer decision to climatic variation. Climatic Change 33, 7–29.
making: the Argentine Pampas and South Florida. Final Smithers, J., Smit, B., 1997. Human adaptation to climatic
Report of an IRI Seed Grant Project. International Research variability and change. Global Environmental Change 7 (3),
Institute for Climate Prediction (IRI), The Earth Institute at 129–146.
Columbia University. South African Weather Service (SAWS). Available at: http://
Igoden, C., Ohoji, P., Ekpare, J., 1990. Factors associated with the www.weathersa.co.za/.
adoption of recommended practices for maize production Svendsen, M., 2008. Adaptation to Hydrological impacts of
in the Lake Basin of Nigeria. Agricultural Administration global climate change in developing countries. Study
and Extension 29 (2), 149–156. Prepared for the Climate Protection Programme for
Intergovernmental Panel on Climate Change (IPCC), 2001. Developing Countries. Environment and Infrastructure
Climate change 2001: impacts, adaptation and vulnerability. Division, GTZ.
Contribution of Working Group II to the Third Assessment Tenge, A.J., de Graaff, J., Hella, J.P., 2004. Social and economic
Report of the IPCC. factors affecting the adoption of soil and water
Intergovernmental Panel on Climate Change (IPCC), 2007. The conservation in West Usambara highlands, Tanzania. Land
physical science basis. Contribution of Working Group I to Degradation and Development 15 (2), 99–114.
the Fourth Assessment Report of the IPCC. Thomas, D.S.G., Twyman, C., Osbahr, H., Hewitson, B., 2007.
Isham, J., 2002. The effect of social capital on fertiliser adoption: Adaptation to climate change and variability: farmer
evidence from Rural Tanzania. Journal of African responses to intra-seasonal precipitation trends in South
Economies 11, 39–60. Africa. Climatic Change 83, 301–322.
Kebede, Y., Kunjal, K., Coffin, G., 1990. Adoption of new Vincent, K., 2007. Uncertainty in adaptive capacity and the
technologies in Ethiopian agriculture: the case of Tegulet- importance of scale. Global Environmental Change 17 (1),
Bulga District, Shewa Province. Agricultural Economics 4, 12–24.
27–43. Vogel, C., O’Brien, K., 2006. Who can eat information?
Knowler, D., Bradshaw, B., 2007. Farmers’ adoption of Examining the effectiveness of seasonal climate forecasts
conservation agriculture: a review and synthesis of recent and regional climate-risk management strategies. Climate
research. Food Policy 32 (1), 25–48. Research 33, 111–122.
Lin, J., 1991. Education and innovation adoption in agriculture: Wooldridge, J.M., 2006. Introductory econometrics: a modern
evidence from hybrid rice in China. American Journal of approach, Third Edition. Thomson South-Western.
Agricultural Economics 73 (3), 713–723. Yesuf, M., Bluffstone, R., 2007. Risk aversion in low income
Maddison, D., 2007. The perception of and adaptation to climate countries: experimental evidence from Ethiopia, IFPRI
change in Africa. World Bank Policy Research Working Discussion Paper, 715.
Paper, 4308. The World Bank, Washington, DC. Ziervogel, G., Bithell, M., Washington, R., Downing, T., 2005.
McCarthy, J., Canziani, O.F., Leary, N.A., Dokken, D.J., White, C., Agent-based social simulation: a method for assessing the
2001. Climate change 2001: impacts, adaptation, and impact of seasonal climate forecasts among smallholder
vulnerability. Contribution of Working Group II to the Third farmers. Agricultural Systems 83 (1), 1–26.
Assessment Report of the Intergovernmental Panel on Ziervogel, G., Bharwani, S., Downing, T.E., 2006. Adapting to
Climate Change. Cambridge University Press, Cambridge. climate variability: pumpkins, people and pumps. Natural
National Meteorological Services Agency, 2001. Available at: Resource Forum 30, 294–305.
http://www.ethiomet.gov.et/ Ziervogel, G., Cartwright, A., Tas, A., Adejuwon, J., Zermoglio, F.,
index.php?Page_No=1.1&item=5&PHPSESSID= Shale, M., Smith, B., 2008. Climate change and adaptation in
c5db0a38883fe500388c8b3a20c469ee. African agriculture. Report Prepared for the Rockefeller
Nhemachena, C., Hassan, R.M., 2007. Micro-level analysis of Foundation by the Stockholm Environment Institute.
farmers’ adaptation to climate change in Southern Africa.
IFPRI Discussion Paper, 714. International Food Policy Elizabeth Bryan is a senior research assistant in the Environment
Research Institute (IFPRI), Washington, DC, 30 pp. and Production Technology Division of the International Food
Norris, E., Batie, S., 1987. Virginia farmers’ soil conservation Policy Research Institute.
decisions: an application of Tobit analysis. Southern Journal
of Agricultural Economics 19 (1), 89–97.
Temesgen T. Deressa is a PhD student at the Center for Environ-
Pelling, M., High, C., 2005. Understanding adaptation: what can
mental Economics and Policy in Africa, University of Pretoria.
social capital offer assessments of adaptive capacity?
Global Environmental Change 15A (4), 308–319.
Glwadys A. Gbetibouo is a PhD student at the Center for Environ-
Roncoli, C., Ingram, K., Kirshen, P., 2001. The costs and risks of
mental Economics and Policy in Africa, University of Pretoria.
coping with drought: livelihood impacts and farmers’
responses in Burkina Faso. Climate Research 19 (2), 119–132.
Roncoli, C., Ingram, K., Kirshen, P., 2002. Reading the rains: local Claudia Ringler is a senior research fellow in the Environment and
knowledge and rainfall forecasting among farmers of Production Technology Division of the International Food Policy
Burkina Faso. Society and Natural Resources 15, 411–430. Research Institute.