当前位置:首页 >> 天文/地理 >>

Adaptation to climate change in Uganda Evidence from micro level data


Global Environmental Change 21 (2011) 1245–1261

Contents lists available at ScienceDirect

Global Environmental Change
journal homepage: www.elsevier.com/locate/gloenvcha

Adaptation to climate change in Uganda: Evidence from micro level data
Eria Hisali a,*, Patrick Birungi b, Faisal Buyinza a
a b

Faculty of Economics and Management, Makerere University, P.O. Box 7062, Kampala, Uganda National Planning Authority, Uganda

A R T I C L E I N F O

A B S T R A C T

Article history: Received 18 June 2010 Received in revised form 30 June 2011 Accepted 12 July 2011 Available online 15 August 2011 Keywords: Climate change Adaptation Multinomial logit

This study employed data from the 2005/06 Uganda national household survey to identify adaptation strategies and factors governing their choice in Uganda’s agricultural production. Factors that mediate or hinder adaptation across different shocks and strategies include age of the household head, access to credit and extension facilities and security of land tenure. There are also differences in choice of adaptation strategies by agro-climatic zone. The appropriate policy level responses should complement the autonomous adaptation strategies by facilitating technology adoption and availing information to farmers not only with regard to climate related forecasts but available weather and pest resistant varieties. ? 2011 Elsevier Ltd. All rights reserved.

1. Introduction and motivation The current unfavourable patterns in climatic outcomes in many parts of the world have brought the need to identify and understand possible adaptation strategies to the limelight. Changes in climatic conditions have been manifested through different channels including longer and more frequent drought spells, rising temperatures, as well as heavier and erratic rains with differing amplitudes on the spatial dimension (Below et al., 2010). The literature has identi?ed several channels through which these changes impact on ecosystems and human development. Both heavier rains and persistent droughts increase soil erosion and vegetation damage through run off with effects on agriculture and sustainable livelihoods. Higher temperatures also mediate faster loss of soil moisture, and prolonged droughts and increasing temperatures create favourable conditions for pests and diseases to multiply (Hoffmann, 2009). The 2007/2008 Human Development Report provides a detailed description of the processes through which current climatic changes might affect attainment of major human development goals as envisaged in the Millennium Development Goals (HDR, 2007). The primary impacts of climate change will invariably be more pronounced in the productivity of activities that depend more than proportionately on favourable climate outcomes—such as the rain fed agriculture that is practiced in most parts of sub Sahara Africa (McCarthy et al., 2001). In Uganda some climate change induced

outcomes have been identi?ed as integral components of the overall constraints to agricultural productivity. Hisali and Kasirye (2008) for example estimate that up to 34% of crop damage in the country is caused by climate induced stimuli such as rainfall shortage, crop diseases and insect damage. Their study also reports that 13%, 20%, and 23% of the community leaders interviewed for the 2005/06 Uganda national household survey (UNHS) indicate crop pests and diseases to be the ?rst, second, and third major constraints respectively to agricultural production. The secondary effects on economies and livelihoods derive from the importance of agriculture in sub Sahara Africa. The sector is the mainstay of economies of most countries in Africa providing primary employment to an average of 70% of the population (Thornton et al., 2006) and playing a major role in ensuring food security. Notwithstanding the signi?cant decline in the share of agriculture in GDP1 for example the sector remains of great importance in Uganda—at least with regard to employment. The interaction of limited resources, high incidences of poverty, ecosystem degradation, con?icts, and a weak institutional capacity, however, imply that the effects will be much stronger in the poorer countries of the world (Francisco, 2008). The adverse impacts of climate change combined with a weak adaptive capacity the micro level of households in Africa that brings the need to ascertain viable and sustainable adaptation strategies to the center of policy analysis and debate. In as much as

* Corresponding author. Tel.: +256 772418739; fax: +256 414532355. E-mail addresses: ehisali@fema.mak.ac.ug (E. Hisali), pbirungi2000@yahoo.com (P. Birungi), fbuyinza@fema.mak.ac.ug (F. Buyinza). 0959-3780/$ – see front matter ? 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.gloenvcha.2011.07.005

1 The contribution of agriculture to GDP reduced from 52% in 1992/93 to 15% in 2009/10 (Background to the Budget, 1994/95 and 2009/10). Nonetheless, the sector still employs about 77% of the rural adult population and accounts for more than about 60% of the merchandise exports.

1246

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

the literature (see for instance Patt and Schroter, 2008; Below et al., 2010; IPCC, 2007) has identi?ed some broad adaptation strategies, the diverse spatial difference with which climate change manifests itself implies that the relevant responses to the adverse impacts (as well as the opportunities) arising from climate change tend to be context speci?c and localised. Whereas analysis by the IPCC (2007) shows mean temperatures in Africa to have risen by one degree Celsius between 1900 and 2000, and with a projected increase of three degrees Celsius by 2050 the long term projections depict considerable variability. Boko et al. (2007) for example project a decrease in mean annual rainfall by up to 20 percent along the Mediterranean coast and the northern Sahara compared to a 7 percent increase in tropical and eastern Africa projected by Case, 2006. The IPCC on the other hand suggests that northern and southern Africa are likely to experience an increase in the number of people suffering from water stress whereas more parts of eastern and western Africa will likely experience a reduction in water stress (IPCC, 2007). Such variability with which climate change outcomes manifest themselves calls for appropriate adaptation responses in the different contexts. Presently however, there is very limited research on adaptation strategies and their determinants in Uganda’s agricultural production. It is in this context that this study sought to identify the major climatic changes that have taken place in Uganda and examine the various adaptation strategies as well as factors that in?uence choice of a particular strategy. This, we believe is imperative in efforts aimed at designing incentives to enhance private adaptation.2 In addition, the ?ndings should provide an informed basis for designing strategies that support existing adaptation measures of local farmers through appropriate public policy, investment and collective actions so as to reduce the negative consequences of predicted changes in future climate. The study employed the nationally representative Uganda national household survey (UNHS) 2005/06 data set collected by the Uganda Bureau of Statistics. In as much as the main aim of the survey was not to collect climate change information per se it probed households on the major climate change related shocks experienced as well as the response strategies they had adopted in the ?ve years preceding the survey. Of particular interest to climate change are questions related to ?oods/hailstorm, drought, livestock epidemic and pest attack. The Uganda Bureau of Statistics uses sampling weights to account for over- and under-sampling in various enumeration areas making the sample data nationally representative. After the introduction, the rest of this paper is organized as follows: an overview of the literature and adaptation benchmarking is presented in Section 2 followed by a description of the data in Section 3. The presentation of the climate change events and adaptation strategies of farmers in Uganda is undertaken in Section 4 followed by the methodology and empirical results in Section 5. Concluding remarks are contained in Section 6.

2.1. Perceptions and adaptation The ongoing changes in global climatic conditions are exposing communities to ever increasing risk and threatening the very sources of livelihood especially in the poorer parts of the world (Francisco, 2008). Contrary to what one would expect, the relationship between some of the climatic induced stimuli and autonomous responses are not automatic but rather, are cognitive processes. In particular, the decision to adapt to climate change risks is mediated by perceptual processes that underlie both the understanding and assessment of risk (Rogers, 1975, 1983). In estimating the likelihood of threats agents employ patterns to construct temporary internal models known as heuristics as opposed to using the more formal Bayesian approaches (Arthur, 1984). Among the most prominent of these heuristics are the availability and the representative heuristics. The availability heuristic posits that judgments will be based on what people can remember and that they search their memories for instances of a particular kind of event occurring which is theoretically expected to increase with freshness of memory of a given event or strength of emotional impact attached to particular memories (Tversky and Kahneman, 1973; Covello, 1990). The representativeness heuristic attaches a high likelihood to a particular event occurring if it is deemed to be representative (Tversky and Kahneman, 1974). It is the cognitive nature of autonomous adaptation decisions that explains the commonly observed con?icts between exposure to climate related risk and inaction, as well as con?ict between autonomous and policy adaptations. Theoretically these con?icts can be attributed to the endowment effect and the omission bias. The endowment effect explains the tendency of making decisions that guard against potential losses and detachments as opposed to potential gains.3 The omission bias attaches high probability values to inaction ‘‘. . . because [decision makers] assign more personal responsibility to the negative consequences of decisions than they do of omission, and want to avoid that personal responsibility for negative outcomes’’ (Patt and Schroter, 2008, p. 460). Grothmann and Patt (2005) used the sociocognitive Model of Private Proactive Adaptation to provide insights into the cognitive barriers to adaptation in rural Zimbabwe and showed that farmers were not partaking adaptation practices because they perceived both the associated risk and capacity to adapt to be rather low. In addition to insights inherent in perceptual processes, the required adjustments can also be mediated by economic, informational as well as social and ecological considerations such as threats to livelihood sources, awareness of severity of the problem and contextualization of external climate forecasts. The cognitive nature of autonomous adaptation decisions calls for partnership between analysts and decision makers in designing strategies that support existing adaptation measures of local farmers through appropriate public policy, investment and collective actions. 2.2. Choice of adaptation strategies

2. Overview of the literature This section presents a summary of the literature regarding perceptual processes that mediate response to actual or expected climatic stimuli or their effects. It also presents factors that in?uence the ability to use a particular strategy. Adaptation strategies are responses to actual or expected climatic stimuli (and their effects) which are intended to moderate harm or exploit associated bene?cial opportunities. The adjustments can be broadly categorized either as responses to current occurrences (climate variability) or planned adaptation to long term changes. In terms of characterization a range of practices have been identi?ed at different levels from household to more institutionalized settings. The review by Below et al. (2010)
3 This can be used to explain phenomena such as reluctance on the part of some people to cope to climate related risks by migrating for example, due to strong attachments they may have to their communities and ancestral roots.

2 Private adaptations are micro level responses to actual or expected climatic stimuli (and their effects) which are intended to moderate harm or exploit the associated bene?cial opportunities. Owing to limited resources to facilitate adaptation at a broader scale private adaptations are the most widely employed responses by farmers in Uganda to climate change.

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

1247

concludes that the most common ones involve some form of diversi?cation and portfolio shifts (including changes in land use and livelihood strategies, and variations in crops and improved crop varieties), changes in timing of cropping and cropping patterns, migration, water conservation techniques and irrigation. Deressa et al. (2009) identify tree planting, soil conservation, different crop varieties, variation in planting time, and irrigation to be the most common adaptation strategies by agriculturalists. With regard to the choice of a given adaptation strategy, the literature has identi?ed many considerations that in?uence decision making ranging from behavioral and socio-economic to gender and ecological. Behavioral factors such as the level of perceived risk determine the likelihood of partaking adaptation practices. Grothmann and Patt (2005) for instance show that the likelihood of partaking adaptation practices reduces when the perceived risk associated with climate change and the capacity to adapt are low. Patt and Schroter (2008) show that implementation of adaptation policies will fail if there is asymmetry of beliefs and risk assessments between bene?ciaries and policy makers and suggest a strong role for dialogue across groups in policy formulation. (Below et al., 2010) on the other hand quote studies which suggest that ‘‘. . . cognitive factors, such as experience, in?uence farmers’ processing of the probability of climate events, as well as their ability to apply climate forecasts. . .’’ (p. 9). Studies (see for instance Norris and Batie, 1987; Maddison, 2007; Deressa et al., 2009) also point to the important role of socioeconomic factors such as the individual decision maker’s position on the life cycle, experience, education, wealth status and availability of resources (such as land, credit and water storage facilities). Others include distance to markets, availability of extension advice (information), farm and household size, and tenure status. Social networks are a decisive factor in acceptance of forecast information (Valdivia et al., 2001, 2002). Societal construction of gender roles in much of the developing world has important implications for the medium of delivery of climate change information (Broad and Petty, 2004). In this respect, certain delivery vehicles such as time speci?c programmes through the electronic media which do not take into account the limited time ?exibility of women have are likely to delay adaptation. Gender discrimination may also make it dif?cult for some women to gain access to complementary inputs as well as relevant information (Tenge et al., 2004), though some studies argue that the dominant role of women in Africa’s agriculture gives them the relevant experience and information (Nhemachena and Hassan, 2008). The study by Deressa et al. (2009) concludes that agroecological zones also affect choice of adaptation strategies. Stringer et al. (2009) on their part focus on the interface between policy and autonomous adaptations and point to the role of national level adaptations in ensuring sustainability of autonomous adaptations. The empirical analysis in this paper is modeled as much as possible (subject to data limitations) along variables identi?ed in this section. 3. Data description The study employed the 2005/06 Uganda national household survey (UNHS) collected by the Uganda Bureau of Statistics. This was a multipurpose survey designed with three modules, namely, socioeconomic survey, community survey, and agriculture survey. Apart from the standard socio-economic information the socioeconomic survey probed households on the various shocks experienced, the timing of the shock as well as response and coping mechanisms over the ?ve-year period preceding the survey. Of particular interest to climate change are questions related to ?oods/hailstorm, drought, livestock epidemic and pest attack.

Information from this section was merged with other sections and modules to identify the socio-economic, community and other determinants of the choice of a particular adaptation strategy. The community survey was designed to collect data on the characteristics of local council one areas (LC 1),4 consumer markets, farm input markets and produce outlets, demographic information relating to communities residing in the sampled enumeration areas (EA) and various details on economic and social infrastructure in those areas. The agriculture survey was intended to collect information on the crop area, inputs, outputs and allied characteristics, covering farming households. In addition, the surveys also capture information relating to household access to key agricultural institutions such as ?nancial service providers, input and output markets. The information from the above three surveys were collected at the same time and from the same enumeration area. Adaptation strategies in the survey were captured by asking households to list the most important adaptation strategies they employed. The speci?c adaptation strategies and how they were reconstructed for purposes of analysis is presented in the next section. The survey information is collected and compiled on the basis of political and geographical units such as LC 1, districts and regions. To capture the spatial aspects of Uganda’s climate, agricultural households for purposes of this study are also classi?ed on the basis of agro-ecological zones. To facilitate meaningful transformation of the data, efforts were made to clean the data by checking for duplicates, inconsistencies and outliers. Sampling weights were used throughout the analysis to account for over- and under-sampling in various enumeration areas, making the sample data nationally representative. 4. Climate change events and adaptation strategies of farmers in Uganda Uganda’s climate is characterized by two rainy seasons. The ?rst rains, which also de?ne the main planting season are more intense and are realized between March and June. The second (lighter) rains take place between the months of October/November– December/January. These patterns are mediated by the La Nina and the El Nino phenomena in the Indian Ocean (LTS International, 2008). The country is demarcated into seven agro-climatic zones on the basis of spatial differences in soils, topography and to a certain extent climate (Table 1). Extreme weather events are not a new phenomenon in Uganda’s history. Reference is made here to the more than usual rainfall recorded in 1961/62, 1997/98 and 2007 and severe drought that hit the country in 1993/94. Unfortunately though, the available scienti?c projections on Uganda’s future climate outcomes are inconclusive. Some reports suggest that temperatures will increase by 4.3 degrees Celsius by 2080, but there are also indications that Uganda may become wetter and may experience changes in rainfall patterns with the second rains becoming more intense (LTS International, 2008; Ministry of Water and Environment, 2007). In any case the changing patterns bring the need to ascertain viable and sustainable adaptation strategies to the center of policy analysis and debate. Indeed, at the strategic level the country ?nalized a National adaptation plan of action (NAPA) to coordinate activities and to attempt to mainstream local adaptations. As part of the activities leading to production of Uganda’s NAPA a long list of adaptation related activities were identi?ed or suggested. These include: exploitation of aquatic resources, food preservation, herbal medicines, alternative livelihood systems, resorting to under-utilized and non-conventional foodstuffs and water harvesting. Others are changes in soil conservation and
4

The LC 1 is the smallest political administrative unit in Uganda.

1248 Table 1 Uganda’s agro-ecological zones. Farming system Banana/coffee system Districts

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table 2 Occurrence of climate related shocks by region. Drought Floods/hailstorm Pest attack Livestock epidemic Total Central Eastern Western Northern National 6.0 6.8 4.6 7.2 24.6 6.2 6.8 4.8 7.3 25.1 6.1 6.9 4.9 7.4 25.3 6.1 6.9 4.6 7.5 25.1 24.4 27.4 18.9 29.4 100

Banana/millet/cotton system Montane system

Respondents (Percent)

Teso systems Northern system Pastoral system West Nile system

Bundibugyo, parts of Hoima, Kabarole, Mbarara, Bushenyi, Mubende, Luweero, Mukono, Masaka, Iganga, Jinja, Kalangala, Mpigi and Kampala Kamuli, Pallisa, Tororo, parts of Masindi and Luweero Kabale, Kisoro, parts of Rukungiri, Bushenyi, Kasese, Kabarole, Bundibugyo, Mbarara, Mbale and Kapchorwa Soroti, Kumi and Kaberamaido Gulu, Lira, Apac and Kitgum Kotido, Moroto, parts of Mbarara, Ntungamo and Masaka, Ntungamo, Masaka and Rakai Moyo, Arua and Nebbi

Source: Authors’ calculations from the 2005/06 UNHS.

husbandry practices, greater involvement in self-help initiatives, traditional vector control approaches and indigenous approaches to rainmaking and thunderstorm prevention. Increased law enforcement, hygiene and sanitation strategies, renting land, shifting cultivation, sale of assets and use of starter stock and change in eating behavior were also identi?ed, among others. The socioeconomic module of the 2005/06 integrated household survey also asked respondents to describe major distress events that they had experienced, their frequency and severity, and duration in the ?ve years preceding the survey. They were also asked to list the adaptation strategies that they had employed. The distress events that were identi?ed and bear close relationship to climate change are drought, ?oods, pest attack, and livestock epidemics. Table 2 presents the percentage of respondents who reported that they had experienced some of climate change related event. The distress events do not show marked differences though pest attacks appear to be a more common threat. The regional totals show some variation in the geographical occurrence of shocks with 29.4 percent of respondents in the north reporting that they had experienced some type of climatic related shock compared to only 18.9 percent in the western part of the country. The climate related shocks as captured by the 2005/06 UNHS appear to be path dependent with their occurrence being reported by an increasing number of households over time (Fig. 1). The duration of shocks over the ?ve years preceding the survey shows variability but again pest attacks appear to be more persistent (Fig. 2).5 Drought is also a persistent problem with about 49 percent of the respondents reporting that it lasted for over ?ve months in each of the years preceding the survey. Whereas ?oods and livestock epidemics are becoming increasingly common, they last for shorter time periods. The increased episodes and persistence of climate distress events computed from the 2005/06 UNHS data appear to be indicative of long term climatic changes and are interestingly consistent with anecdotal and other evidence from reports on climate change in Uganda such as LTS International (2008) and Republic of Uganda (2007). In response to the observed changes in climate related outcomes households have been employing a range of adaptation practices. The 2005/06 survey identi?ed mortgaging household assets, selling assets, using past savings and withdrawing children from school as some of the adaptation mechanisms. Other strategies include sending children to live elsewhere, migration,
5 This is in line with ?ndings by Hisali and Kasirye (2008) which indicate that 13%, 20%, and 23% of the community leaders interviewed for the 2005/06 UNHS rank crop pests and diseases to be the ?rst, second, and third major constraints, respectively to agricultural production.

0 2000

10

20

30

40

2001

2002

2003

2004

2005

Year
Drought Pests floods Livestock epidemic

Fig. 1. Percentage of respondents reporting different shocks 2000–2005.

1 2 3 4 5 6 7 8 9 10 11 12 0 5 10 15 20 25

Duration of shock (months)

Percentage of respondents
drought pests floods livestock epidemic

Fig. 2. Duration of climate related shocks.

formal borrowing, informal borrowing, reducing consumption, and reliance on help from relatives, friends and local governments. More wage employment, changing crop choices to avoid bad weather, improving technology, working as self employed, and increasing agriculture labour supply were also commonly used. In line with the literature the identi?ed coping strategies were in turn collapsed into ?ve categories as follows: (i) borrowing, both from formal and informal sources; (ii) labour supply which was constructed to include more wage employment, working as self employed, increasing agriculture labour supply, migration to work elsewhere and withdrawing children from school and sending them to work; (iii) reducing consumption; (iv) running down assets and past savings including mortgaging assets, selling assets and utilising savings; and, (v) technology based adaptation strategies such as changes in crop choices to avoid bad weather and improving technology. The national and disaggregated description of the use of different strategies is presented in Tables A1–A4. Utilisation of household

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table 3 Credit status of households and adaptation strategy. Credit/strategy Drought Credit Borrowing Labour supply Technology Savings Reduce consumption Total 3.80 5.56 2.11 6.74 7.00 25.20 No credit 0.00 17.47 2.75 33.13 21.45 74.80 National 3.80 23.03 4.86 39.87 28.45 100.00 Floods Credit 5.30 5.97 2.94 4.62 5.80 24.63 No credit 0.00 17.88 4.66 35.38 17.44 75.37 National 5.30 23.85 7.60 40.00 23.24 100.00 Pests Credit 1.34 3.44 13.05 3.87 6.44 28.14 No credit 0.00 14.94 27.41 15.27 14.24 71.86 National 1.34 18.38 40.46 19.14 20.68 100.00 Livestock epidemic Credit 1.46 6.84 8.83 7.29 3.07 27.49 No credit 0.00 23.60 7.57 40.37 0.98 72.52

1249

National 1.46 30.44 16.40 47.66 4.05 100.00

Table 4 Extension services and coping strategies. Extension/strategy Drought Extension Borrowing Labour supply Technology Savings Reduce consumption Total 0.35 1.15 0.31 3.73 2.33 No extension 3.43 21.88 4.54 39.89 28.44 National 3.78 23.03 4.85 43.62 30.77 Floods Extension 0.31 1.43 0.82 2.68 1.03 No extension 4.98 22.41 6.79 37.24 22.31 National 5.29 23.84 7.61 39.92 23.34 Pests Extension 0.17 1.47 5.55 1.89 0.65 No extension 1.18 16.91 34.91 17.25 20.03 National 1.35 18.38 40.46 19.14 20.68 Livestock epidemic Extension 0.48 0.66 1.39 4.53 0.00 No extension 0.98 29.79 15.00 43.12 4.04 National 1.46 30.45 16.39 47.65 4.04

7.87

94.75

100.00

6.27

93.73

100.00

9.73

90.28

100.00

7.06

92.93

100.00

savings, scaling back on consumption and labour based strategies are the most frequently employed strategies in response to drought events. Technology based strategies and reducing consumption on the other hand are the most frequently employed strategies by households faced with a pest attack whereas utilisation of savings and increased labour supply are the most common response strategies to ?oods. Households faced with a livestock epidemic usually rely on savings and supply of additional labour. The bivariate relationship between adaptation strategies and some focal independent variables (such as sex and education of the household head, credit access and extension services as well as geographical location of the household) is generally weak (Tables A1–A4). In terms of sex of the household head both male and female headed households employ similar strategies in response to ?oods, pest attacks and livestock epidemics. The sex based difference in adaptation strategies is observed only in the event of a drought where female headed households accommodate drought related climate changes by reducing consumption while male headed use their savings (Table A1). Geographically, a higher percentage of households in the eastern and western parts of the country rely on savings in response to drought episodes whereas increased labour supply and reducing consumption are more commonly used in the north and central parts of the country, respectively. Use of household savings and technology based responses are robust across the different regions as responses to livestock epidemics and pest attacks, respectively. Savings are also the most common response to ?oods except in the north of the country where households respond by supplying more labour which is plausible given the higher incidence of poverty in the northern part of the country. Increased labour supply and reduced consumption as adaptation strategies are more prominent in the northern part of the country (Table A2). The disaggregated results fail to show differences in adaptation strategies on the basis of education level (Tables A3–A4). The limited variance of adaptation strategies on the basis of education and to a large extent sex of the household head is conjectured to re?ect either the local public good nature of indigenous knowledge or the role of externalities. In any case one important implication is that the policy adaptations that should be

implemented are those that complement some of the current practices that are sustainable. Borrowing does not only under perform all the other adaptation strategies but also contrary to standard thinking does not appear to enhance adoption of more sustainable adaptation strategies (Table 3). This is possibly mediated by the stringent repayment conditions coupled with the relatively high interest rates on the Ugandan ?nancial market. In relative terms the bivariate descriptive analysis results show that there is no difference in adaptation strategies on the basis of whether a household received extension advise or not which is conjectured to re?ect the possibility that the existing extension services being provided to farmers are yet to incorporate information on climate change (Table 4). Generally the most common adaptation strategies employed by households in Uganda are unsustainable and call for implementation of policies that seek to enhance the resilience of households to respond in a more sustainable manner to climate related climate shocks.6 The Kruskal–Wallis test was also employed to assess the statistical signi?cance of the adaptation strategies within and between regions as well as agro-ecological zones. The test results indicate a statistically signi?cant difference in the drought adaptation strategies between regions (Tables A5–A8) and agroecological zones (Tables A9–A12). At the national level the rank sum shows that utilisation of household savings is the most common drought adaptation strategy followed by reduced consumption and increased labour supply (Table A5). Geographically the use of household savings is ranked higher in the western and eastern parts of the country whereas households in the central region accommodate drought related climate changes by reducing consumption while those in the north mostly rely on increased supply of labour. The pest attack adaptation strategies are also signi?cantly different both within and between regions (Table A6) and agroecological zones (Table A10). Technology based strategies have a
6 Some of the viable policy options are presented in the concluding section of the paper.

1250

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

higher rank sum followed by reducing consumption and the ?nding is robust across the different regions (Table A6). In addition there is a statistically signi?cant difference in the ?oods adaptation strategies both within different regions and agro-ecological zones. At the national level savings rank higher than all other strategies followed by reduced consumption. Use of household savings also ranks higher in three of the four regions. Whereas the test fails to distinguish among the different ?oods adaptation strategies in the western region it distinguishes the strategies in all the other regions and nationally as well. Use of savings is ranked higher in the central and eastern regions and nationally whereas increased labour supply is more likely to be used in the north (Table A7). The livestock epidemic adaptation strategies are signi?cant at the national level but there is no signi?cant difference among the strategies in central and eastern regions. The strategies are statistically different in the western and northern regions with savings dominating in both and nationally (Table A8). The adaptation strategies employed by households in Uganda compare considerably well with those reported in studies from other developing countries (see for instance Boko et al., 2007; Below et al., 2010; Deressa et al., 2009). Reducing consumption as an adaptation strategy is in line with evidence from other studies which indicate that households accommodate exogenous changes that they may face by making adjustments to suit the new situation (Gregory et al., 2005). The use of technology based adaptation strategies is also in line with practices elsewhere (see for instance Thomas et al., 2007). In terms of policy, it will be helpful if these autonomous efforts by farmers were supported by research to facilitate technological developments that are suitable to changing environments. The prominence of increased labour supply as an adaptation strategy in Uganda probably re?ects agriculture extensi?cation which would conform to ?ndings from other studies (Gregory et al., 2005; Paavola, 2008). The bivariate analysis of the relationship between adaptation strategies and some focal independent variables undertaken in this section provides insights into factors that facilitate or hinder choice of a given adaptation strategy. It should, however, be pointed out that this kind of descriptive work cannot be relied on to provide conclusive analysis of the underlying relationships among the variables. It is for example not possible to explicitly control for other intervening variables in such settings. There is thus need to increase the statistical plausibility of the relationships by undertaking analysis in a multivariate setting. This is undertaken in the next section. 5. Determinants of choice of adaptation strategies 5.1. Empirical strategy and results This section employed the multinomial logistic method to establish factors that govern choice of a particular adaptation technique. The MNL technique compares any given outcome with a reference outcome. This technique is deemed suitable to study adaptation to climate change since households employ different adaptation strategies which are typically not mutually exclusive. In the MNL each pair of classes ?Y j ; Y 1 ? can be described by the ratio: log ?Y ? jjx? 0 ? a j ? b j x; ?Y ? 1jx?

strategy j on the basis of its features contained in the vector x. This may be calculated using the equation: P ?Y ? jjx? ? exp?a j ? b j x? : PJ 0 1 ? h?2 exp?ah ? b j x?
0

where x is the vector containing the predictor variables, aj the intercept parameter for the jth level and b the vector of regression coef?cients. On the basis of these J ? 1 regression equations, it is possible to compute the probability of a household employing a

Notice that for the baseline category (here, j = 1), a1 and b1 = 0. Thus, when looking for the probability of an adaptation strategy belonging to the baseline level, it is easy to compute the numerator, since exp(0) = 1. The value of the denominator is the same for each j The parameter estimates provide only the direction of the effect of the independent variables on the dependent variable, but the estimates do not represent either the actual magnitude of change nor probabilities. We subsequently utilise the odds ratios. The odds ratio assesses whether the odds of a certain event or outcome is the same for two groups. In line with the literature on autonomous adaptation strategies and subject to data availability we included as part of our explanatory variable vector, x, a range of variables that describe the probability of choosing a particular adaptation strategy. The multinomial logistic models crucially depend on the independence of irrelevant alternatives (IIA) assumption which posits that deleting (or adding) an outcome category should not affect the odds among the remaining categories. Appropriate tests were employed to assess the validity of the IIA assumption. The Wald and Likelihood ratio tests were employed to drop redundant explanatory variables whereas predicted probabilities were used to ascertain the relative importance of the different adaptation strategies. The variables used in the study are de?ned in Table A13. Factors that mediate a household’s choice of an adaptation strategy in response to drought are presented in Table 5. The Hausman test validates the IIA assumption (Table A14). The probability value associated with the ‘F’ test suggests that we can reject the null that all predictor variables in our models are jointly not different than zero. Whereas some predictors appear to be unable (in a statistical sense) to distinguish among outcomes, a number of them have suf?cient predictive power. The predictors in the latter category as well as their relationship to the various adaptation strategies are presented in what follows. Off farm employment. Lack of access to off farm employment opportunities reduces the relative risk of adapting by borrowing relative to using savings by 0.58 which underscores the negative effect that limited access to employment opportunities (formal or otherwise) has on credit worthiness hence diminishing chances of borrowing. Lack of access to off farm employment opportunities also reduces the relative risk of using labour supply and technology as coping strategies by about 42 percent and 30 percent respectively when compared to past savings. Extension services. Access to extension services reduces the relative risk of using labour supply as opposed to savings by about 40 percent. This result is plausible if extension advice is used to encourage households to diversify their crop portfolio to include some drought resistant crops. Other things being equal, this constrains the ability of household members to work off the farm as more labour may be required on the farm. Agro-ecological setting. The odds of borrowing relative to utilising savings reduce in the west Nile but increase by about 2.5 times and 1.65 times respectively in the pastoral and Montane zones. Utilisation of labour as opposed to savings is a more common practice in the west Nile and northern systems but less common in the Montane system. The same conclusion holds when use of technology and reduction in consumption are compared with utilisation of savings. Land tenure. A more secure land tenure arrangement is associated with an increase in the chances of adapting to drought through technology and through reduced consumption.

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table 5 Relative risk ratios (rrr) of the multinomial logistic model for coping with drought. Variable/coping strategy Borrowing rrr Age head Disttown1 Educhead1 Extension11 Offfarmh2 credith2 femalehhd11 tenure1 Banana/millet/cotton Montane Northern Pastoral West Nile Diagnostics Base category = Savings Number of observations = 2143 Number of strata = 4 Number of PSUs = 143 s.e. denotes the standard error. * Signi?cance at the 1% test level. ** Signi?cance at the 5% test level. *** Signi?cance at the 10% test level. 1.009 1.005 2.289 0.973 0.579*** 1.842*** 0.828 0.468 0.408 1.660*** 0.810 2.493** 0.000* s.e. 0.008 0.003 1.226 0.42 0.186 0.602 0.214 0.229 0.254 0.474 0.494 1.018 0.000 Labour supply rrr 0.984* 1.003 1.324 0.599** 0.578* 1.223 0.711** 0.931 1.071 0.540* 2.499* 1.515 5.405* s.e. 0.004 0.002 0.248 0.147 0.098 0.195 0.097 0.213 0.238 0.101 0.630 0.419 1.127 Technology rrr 1.007 1.007 1.925** 0.796 0.919 1.898** 0.886 2.142** 0.466 0.319** 7.755* 0.317 5.040* s.e. 0.008 0.005 0.530 0.362 0.295 0.475 0.234 0.702 0.228 0.148 2.967 0.332 1.964 Reduce consumption rrr 1.006 1.004*** 1.560* 0.918 0.704** 1.052 0.612* 1.767* 0.717 0.696** 1.975* 0.973 4.085*

1251

s.e. 0.004 0.002 0.250 0.168 0.111 0.169 0.090 0.252 0.156 0.102 0.508 0.307 0.809

Design d.f. = 139 F(52, 88) = 802.5 Prob > F = 0.000 Population size = 1,469,268

Sex of household head. The relative risk of increasing labour supply relative to using household savings in the face of a drought reduces by 0.71 in female headed households. This re?ects the limited ?exibility women have with respect to participating in wage employment. Societal construction of gender roles in Africa assigns household responsibilities which limits their time to competing for wage jobs. It might also represent discrimination and lack of requisite skills. Female headed households are also more likely to reduce consumption as opposed to use of past savings, re?ecting a strong endowment effect. Education of household head. Basic education of the father increases the chances of adopting technology and drought resistant varieties relative to use of savings. They would also rather reduce consumption than use their savings, again re?ecting a fairly strong endowment effect. Distance to the nearest town. Distance to the nearest town is used to proxy for availability of input and output markets. It increases the relative risk of reducing consumption as opposed to using savings by 0.42 percent in response to a drought. This might be attributed to the dif?culty in accessing markets for disposing off the ‘physical savings’. It might, however, also be possible to argue that the endowment effect is stronger in rural areas. Access to credit. Access to credit increases the chances of borrowing and adoption of technology as opposed to utilisation of savings in the adaptation to climate change. In particular, access to credit facilities increases the relative risk of borrowing relative to use of savings by 1.84 and increases the relative risk of technology adoption by 1.89. Age of household head. Age of the household head quite understandably reduces the odds of using labour supply as opposed to utilisation of savings which underscores the relatively higher vulnerability of the elderly. The predicted probabilities con?rm ?ndings from the descriptive analysis which indicated that agricultural households were more likely to use past savings in response to a drought episode. The probability that a household will rely on borrowing is the lowest when compared to all the other strategies (Fig. 3). The econometric results for factors that in?uence choice of an adaptation strategy in response to a pest attack are presented in Table 6.

Our model is signi?cantly better than the ‘know nothing’ speci?cation as judged by the ‘F’ probability value. The Small– Hsiao test supports the null of IIA since the probability values (P > chi2) range from 0.98 to 1.00 (Table A15). The predictor variables that in?uence choice of strategies to adapt to pests are explained in what follows. Off farm employment. Availability of off farm employment opportunities increases the relative risk of using labour as an adaptation strategy relative to technology by 1.75. Extension services. Access to extension services reduces the relative risk of reducing consumption as opposed to using technology by 0.14. This can be the case for instance if extension services are used as conduits for passing on advice and information about new crop varieties which, other things being equal, would encourage utilisation of technology as an adaptation strategy. Age of household head. Age of the household head increases the relative risk of reducing consumption relative to technology adoption by 1.02 which points to vulnerability of the elderly. Length of the last pest attack. Length of the last pest attack was included in the speci?cation to capture the availability heuristic.

Fig. 3. Predicted probabilities of different drought adaptation strategies.

1252

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

Table 6 Relative risk ratios (rrr) of the multinomial logistic model for coping with pests. Variable/coping strategy Labour supply rrr Lengthpests Agehead inputmktlc1 extension11 offfarmh1 credith1 Farm size bananacoffee bananamilletcotton montane pastoral Diagnostics Base category = Technology Number of observations = 496 Number of strata = 4 Number of PSUs = 126 s.e. denotes the standard error. * Signi?cance at the 1% test level. ** Signi?cance at the 5% test level. *** Signi?cance at the 10% test level. Design d.f. = 122 F(33, 90) = 3.7 Prob > F = 0.000 Population size = 346,219 0.959*** 0.987 0.787 0.613 1.756** 1.910*** 0.964 0.388** 0.818 0.818 0.450 se 0.024 0.011 0.234 0.267 0.480 0.664 0.036 0.169 0.362 0.352 0.232 Use savings rrr 0.932* 1.000 1.245 0.614 1.181 1.071 0.988 3.885* 2.865* 3.522* 0.668 se 0.021 0.009 0.349 0.266 0.362 0.337 0.008 1.388 1.080 1.392 0.780 Reduce consumption rrr 0.960*** 1.020** 1.449 0.143* 1.186 1.054 0.956 1.491 0.871 2.219*** 0.871 se 0.023 0.009 0.435 0.092 0.420 0.352 0.042 0.595 0.474 0.993 0.828

The results indicate that duration of the last pest attack reduces chances of using savings, labour supply and scaling back on household consumption relative to using technology based options. In particular, chances of utilising past savings, labour supply and scaling back on consumption reduce by 0.93, 0.96 and 0.96 respectively relative to technology adoption. Put another way, longer pest attack episodes other things being equal mediate technology usage as a coping strategy. Agro-ecological setting. The odds of increasing labour supply relative to utilising technology based adaptations are lower for those households living in the banana-coffee zone with the relative risk of relying on increased labour supply relative to technology reducing by about 60 percent. The odds of utilising savings as opposed to technology as a coping strategy to a pest attack is high in the banana-coffee, banana-millet-cotton and montane systems. The chances of scaling back on household consumption relative to technology usage are higher in the montane zone. The predicted probabilities indicate that households are more likely to utilise technology based strategies and increased labour supply in response to a pest attack (Fig. 4). The MNL regression results for factors that in?uence choice of an adaptation strategy in response to a livestock epidemic are presented in Table 7.

The Hausman test renders support the null of IIA (Table A16). The predictors of the different adaptation strategies to a live stock epidemic include: Extension services. Access to extension services reduces the relative risk of using labour supply as opposed to utilising savings. Agro-ecological setting. Households in the Teso and Montane systems are more likely to adapt to livestock epidemics using technology based practices relative to utilising savings. Land tenure. A more secure land tenure arrangement is associated with an increase in the chances of adapting to a livestock epidemic through technology relative to savings. A more secure land tenure arrangement increases the odds of using technology by about 4 times in relation to relying on savings. Access to credit. Quite surprisingly, households that did not get credit have higher chances of using technology adoption relative to utilisation of savings in the adaptation to a livestock epidemic. This might be the case where loan repayment reduces available resources for technology adoption or that credit can be used for other purposes other than climate change adaptation.

Table 7 Relative risk ratios (rrr) of the multinomial logistic model for coping with a livestock epidemic. Variable/coping strategy Labour supply rrr extension11 Credith2 femalehhd11 tenure1 montane Nothern teso Diagnostics Base category = Savings Number of observations = 190 Number of strata = 4 Number of PSUs = 87 s.e. denotes the standard error. ** Signi?cance at the 5% test level. *** Signi?cance at the 10% test level. Design d.f. = 83 F(14, 70) = 1.77 Prob > F = 0.0.06 Population size = 124,305 0.114*** 1.003 0.600 0.797 0.345*** 0.906 2.448 s.e. 0.142 0.437 0.291 0.379 0.216 0.410 1.460 Technology rrr 1.370 3.860** 1.833 3.748*** 5.725** 2.230 4.973*** s.e. 1.108 2.567 1.182 2.599 4.181 1.544 4.683

Fig. 4. Predicted probabilities of different pest attack adaptation strategies.

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table 8 Relative risk ratios (rrr) of the multinomial logistic model for coping with ?oods. Variable/coping strategy Labour supply rrr length?oods disttown1 offfarmh2 femalehhd12 tenure1 bananacoffee bananamilletcotton montane Nothern pastoral Diagnostics Base category = Savings Number of observations = 692 Number of strata = 4 Number of PSUs = 128 s.e. denotes the standard error. * Signi?cance at the 1% test level. ** Signi?cance at the 5% test level. *** Signi?cance at the 10% test level. 0.965 1.004 0.250* 1.129 1.231 0.159* 0.376** 0.117* 0.749 0.214** se 0.036 0.004 0.075 0.291 0.569 0.083 0.157 0.043 0.333 0.149 Technology rrr 0.999 0.998 0.406** 2.158** 12.860* 0.044* 0.521 0.013* 1.333 0.075* se 0.040 0.006 0.173 0.809 8.763 0.034 0.247 0.013 0.634 0.076 Reduce consumption rrr 0.863* 0.997 0.537** 0.831 1.759*** 0.698 0.480 0.647 1.239 1.126 se

1253

0.042 0.004 0.154 0.225 0.607 0.358 0.270 0.277 0.553 0.765

Design d.f. = 124 F(30, 95) = 4.64 Prob > F = 0.000 Population size = 483,606

The predicted probabilities indicate that technology based strategies, labour supply and use of savings are almost equally likely in response to a livestock epidemic (Fig. 5). The maximum likelihood estimation results for factors that in?uence choice of an adaptation strategy in response to ?oods are presented in Table 8. Whereas some variables included in the model fail to distinguish between the different adaptation outcomes a number of them have considerably good statistical predictive abilities (Table 8). These include: Off farm employment. Lack of access to off farm employment opportunities reduces the relative risk of adapting to ?oods by increasing labour supply relative to using savings. It also reduces chances of using technology and reducing consumption as opposed to reducing savings. This is plausible if decision makers at the household level perceive the ?ood occurrences as temporary phenomena. Agro-ecological setting. The relative risk of supplying more labour as opposed to using savings as a coping mechanism to ?oods reduces for households in the pastoral, montane, banana-coffee and banana/ millet/cotton systems. Households in the pastoral, montane and banana-coffee systems are less likely to use technology as opposed to using savings as a coping strategy to ?oods.

Land tenure. A more secure land tenure arrangement is associated with an increase in the chances of adapting to ?oods through technology and reduced consumption relative to savings. Length of the last ?ood. Length of the last ?ood reduces chances of scaling back on household consumption relative to using savings. Loosely speaking reduced consumption might only be a short term strategy which gives way to using savings as the shock duration increases. The predicted probabilities indicate that savings and labour supply are the most commonly used strategies in response to ?oods (Fig. 6). The Small–Hsiao test supports the null of IIA (Table A17). 5.2. Overview of results and comparison with ?ndings from other studies Whereas it was important to identify factors that explain the choice of adaptation strategies to different climate related outcomes the explanatory power of some factors is invariant across some climate shocks. The main variables in this category include age of the household head, access to credit and extension facilities and security of tenure. There are also notable differences in the choice of the adaptation strategies by agro-climatic zone possibly re?ecting the importance of indigenous knowledge and externalities in climate change adaptation.

Fig. 5. Predicted probabilities of different livestock epidemic adaptation strategies.

Fig. 6. Predicted probabilities of the different strategies for coping with ?oods.

1254

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

These ?ndings compare favourably with those from the existing literature. The results from this study show that older farmers are less likely to increase labour supply as an adaptation strategy and more likely to reduce consumption (see Tables 7 and 8) which points to their relatively higher vulnerability. This ?nding may be used to explain the con?icting ?ndings in part of the existing literature—Shiferaw and Holden (1998) for example ?nd a negative relationship between age and adoption of improved soil conservation practices whereas (Deressa et al., 2009) show that older farmers are more likely to employ adaptation strategies in the face of changes in climate related variables. The results also show that availability of extension services reduces the relative risk of using labour supply which can be possible if extension advice is used to encourage households to diversify their crop portfolio to include some drought resistant crops. Other things being equal, this constrains the ability of household members to work off the farm as more labour may be required on the farm. This result is in line with an interesting ?nding by (Maddison, 2007) who established that access to information and knowledge (through extension services for instance) about appropriate adaptation strategies promoted adaptation. The important role of other variables such as access to credit, security of tenure and agro-climatic zone is also borne out by other studies (see for instance Norris and Batie, 1987; Maddison, 2007; Deressa et al., 2009) who point to the important role of socioeconomic factors such as education, wealth status and availability of resources (such as land, credit and water storage facilities). The study by Deressa et al. (2009) also found that agroecological zones also affect choice of adaptation strategies. A summary of the study and possible policy options is presented in the next section. 6. Summary and policy implications The ongoing changes in global climatic conditions are exposing communities to ever increasing risk and threatening sources of livelihood especially in the poorer parts of the world. This calls for responses both to current occurrences and to more long term climate changes. This study employed data from the 2005/06 Uganda national household survey (UNHS) to shed light on some of the main climate change related shocks that have a bearing on the agricultural sector, how households were adapting as well as the perceptual, socioeconomic and institutional factors that mediate adaptation in Uganda. The most commonly observed climate related shocks include livestock epidemics, pest attacks, drought and ?oods. The study also identi?ed the major categories of autonomous adaptation strategies undertaken by households in response to climate related shocks as well as the behavioural and socioeconomic factors that mediate them. The main adaptation strategies can be broadly categorized to include reducing consumption, running down past savings, technology based options and to a lesser extent, borrowing. The sustainability of many of these strategies is, however, questionable. Whereas reducing consumption and utilising past savings for example are plausible short term strategies, they expose households to even more vulnerability in the face of persistent climatic variability. The exposure is likely to

be more pronounced in certain categories of households such as those headed by the elderly. In this regard our ?ndings call for policies that smooth the consumption patterns of households. Access to credit which is conditioned to climate change adaptation would not only enhance the ability of households to smooth short term consumption ?uctuations that result from climate related shocks but would also enable them to get access to resources that are required to implement certain adaptation practices. Policies such as reduction of information asymmetries so as to reduce the high cost of credit on the Uganda ?nancial market would go a long way in improving access to credit facilities and in strengthening the ability of households to adapt to climate change. The current prosperity for all programme is a right step in efforts aimed at improving access to credit. Af?rmative action targeting the disadvantaged and more vulnerable groups could be considered. Complementing credit access with extension advice would be of immense value. Increased labour supply (both on and off-farm) constitutes a major—and more sustainable category of responses to climate change stimuli. These adaptation strategies will, however, only be helpful in the long term if they are complemented by policies that enhance the productivity performance in sectors that employ larger sections of the population. Agricultural prosperity in Uganda is key to generating productive employment that strengthens the ability of households to adapt to climate change. Raising labour productivity in agriculture will also augment the low incomes of the work force and release workers to ?nd jobs elsewhere in the economy in order to solve the surplus labour problem in agriculture and hence reduce on the extensi?cation pressures. Increased productivity will require more public investment in research and advisory services, strategic types of infrastructure as well as access to credit and farm implements at favourable terms. These should be supplemented by value addition initiatives and increased access to markets. A number of these initiatives are already being pursued by government but they have not had signi?cant effect on the sector’s productivity performance. What needs to be done is to mobilize resources so as to scale up the investments. But most importantly, these initiatives need to be closely coordinated and where possible implemented simultaneously both spatially and temporally. It will also be necessary to compliment credit availability and job creation with information both on climate change and some examples of sustainable responses to support the autonomous technology related practices. This will require more public investment in research and advisory services to facilitate technological developments that are suitable to changing environments. Other policies that improve resilience such as investments in public health and education systems would compliment the autonomous responses by farmers. Programmes that control the pace at which the population is expanding will be helpful in curbing pressure to extensify agriculture.

Appendix A

Table A1 Climate change coping strategies and sex of household head. Strategy/gender of head Drought Male Borrowing Labour supply Technology Savings Reduce consumption 3.1 16.6 3.9 29.1 18.4 Female 1.5 6.3 1.3 9.0 10.7 Total 4.7 22.9 5.3 38.1 29.1 Floods Male 3.5 18.5 6.0 31.9 18.0 Female 1.4 5.6 1.9 9.1 4.3 Total 4.9 24.0 7.9 41.0 22.2 Pests Male 0.3 12.1 34.3 13.1 15.5 Female 0.2 4.0 7.7 5.9 7.1 Total 0.5 16.1 41.9 19.0 22.6 Livestock epidemic Male 0.5 18.3 15.8 43.1 1.5 Female 0.0 5.6 3.8 9.4 2.0 Total 0.5 23.9 19.6 52.5 3.5

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

Table A2 Climate change coping strategies and region. Strategy/region Drought Floods Pests Livestock epidemic

Central Western Eastern Northern National Central Western Eastern Northern National Central Western Eastern Northern National Central Western Eastern Northern National Borrowing Labour supply Technology Savings Reduce consumption 0.5 3.7 1.6 6.6 8.9 2.7 3.9 0.3 15.9 6.0 1.0 5.3 0.5 10.4 5.0 0.4 10.0 2.9 5.1 9.1 4.7 22.9 5.3 38.1 29.1 0.0 1.5 2.1 4.6 2.9 3.5 6.3 0.5 22.3 8.4 0.9 7.9 2.2 9.9 5.6 0.5 8.4 3.1 4.2 5.4 4.9 24.0 7.9 41.0 22.2 0.1 2.2 12.6 7.2 9.8 0.2 2.3 11.6 4.0 4.5 0.0 3.6 9.0 6.0 4.3 0.1 8.0 8.7 1.9 4.0 0.5 16.1 41.9 19.0 22.6 0.0 5.1 7.6 11.1 1.6 0.2 0.3 2.3 4.1 0.0 0.2 9.7 3.6 21.5 0.4 0.2 8.8 6.1 15.8 1.5 0.5 23.9 19.6 52.5 3.5

1255

1256

Table A3 Climate change coping strategies and education level of household head. Strategy/education level of head Drought Male Primary Borrowing Labour supply Technology Savings Reduce consumption 4.3 19.8 4.4 31.4 24.3 Secondary 0.3 3.0 0.8 5.8 4.5 Tertiary 0.1 0.2 0.1 0.9 0.2 Total 4.7 22.9 5.3 38.1 29.1 Female Primary 4.58 22.53 5.24 35.53 27.16 Secondary 0.1 0.4 0.0 2.4 1.8 Tertiary 0.0 0.0 0.0 0.3 0.0 Total 4.7 22.9 5.3 38.1 29.1 Floods Male Primary 4.07 20.05 6.47 35.25 20.18 Secondary 0.46 3.64 0.97 5.22 1.93 Tertiary 0.38 0.32 0.43 0.51 0.12 Total 4.91 24 7.87 41 22.2 Female Primary 4.65 23.22 6.93 37.86 21.75 Secondary 0.27 0.8 0.93 2.36 0.46 Tertiary 0 0 0 0.77 0 Total 4.9 24.0 7.9 41.0 22.2

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

Table A4 Climate change coping strategies and education level of household head. Strategy/education level of head Pests Male Primary Borrowing Labour supply Technology Savings Reduce consumption 0.32 11.67 33.3 15.07 17.5 Secondary 0.12 4.35 7.93 3.18 4.61 Tertiary 0 5.70E?02 0.7 0.75 0.43 Total 0.4 16.1 41.9 19.0 22.5 Female Primary 0.45 15.56 38.14 17.44 22.46 Secondary 0 0.54 3.24 1.51 0.11 Tertiary 0 0 0.49 6.90E?02 0 Total 0.5 16.1 41.9 19.0 22.6 Livestock Epidemic Male Primary 0.34 20.69 17.59 40.5 3.36 Secondary 0.16 3.2 2.01 11.86 0.13 Tertiary 0 0 0 0.16 0 Total 0.5 23.9 19.6 52.5 3.5 Female Primary 0.51 24.15 19.68 50.82 2.09 Secondary 0 5.50E?02 0.17 2.4 0.13 Tertiary 0 0 0 0 0 Total 0.5 24.2 19.9 53.2 2.2

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table A5 Kruskal–Wallis rank test within and between regions for drought coping strategies. Strategy/region Central Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 38 264 115 504 733 Rank sum 21,719.5 20,5954 120,450.5 428,681.5 591,879.5 Western Obs 177 238 17 1123 432 Rank sum 178,464 286,615 19,303 1,100,000 395,520.5 Eastern Obs 76 448 39 859 387 Rank sum 76,447 380,376 37,391.5 756,446.5 386,484 Northern Obs 41 995 262 474 832 Rank sum 53,248.5 1,250,000 376,417 651,520 1,060,000 National Obs 332 1945 433 2960 2380

1257

Rank sum 1,660,000 7,000,000 1,720,000 12,900,000 9,130,000

40.349 [0.0001] 40.356 [0.0001]

42.386 [0.0001] 42.392 [0.0001]

22.703 [0.0001] 22.705 [0.0001]

17.921 [0.0013] 17.923 [0.0013]

204.648 [0.0001] 204.649 [0.0001]

Table A6 Kruskal–Wallis rank test within and between regions for coping with pests. Strategy/region Central Obs Borrowing Labour supply Technology Savings Reduce Consumption chi-squared chi-squared with ties 2 45 243 146 182 Rank sum 643 10,709.5 78,625.5 39,890.5 61,402.5 Western Obs 3 34 175 85 65 Rank sum 697.5 7369 34,075.5 9087.5 14,473.5 Eastern Obs – 77 183 118 85 Rank sum – 16,148.5 40,801.5 30,283.5 20,182.5 Northern Obs 3 169 185 45 92 Rank sum 580 45,797 46,416 10,262.5 19,209.5 National Obs 8 325 786 394 424 Rank sum 8430 292,915 814,089 350,574.5 410,944.5

19.199 [0.0007] 19.209 [0.0007]

60.604 [0.0001] 60.681 [0.0001]

7.113 [0.0684] 7.117 [0.0683]

12.709 [0.0128] 12.719 [0.0127]

24.043 [0.0001] 24.044 [0.0001]

Table A7 Kruskal–Wallis rank test within and between regions for coping with ?oods. Strategy/region Central Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 1 39 51 123 77 Rank sum 35.5 5819 8685.5 15,941.5 12,004.5 Western Obs 76 143 10 515 181 Rank sum 39,527.5 68,055 4000 229,291.5 87,401 Eastern Obs 25 220 56 259 136 Rank sum 7491 66,143 21,315 93,761.5 53,845.5 Northern Obs 20 278 96 123 154 Rank sum 5480 88,228.5 29,792 44,144 57,811.5 National Obs 122 680 213 1020 548 Rank sum 178,149.5 747,847.5 242,938.5 1,410,000 755,657

11.349 [0.0191] 11.794 [0.0189]

7.642 [0.1056] 7.645 [0.1045]

24.102 [0.0001] 24.114 [0.0001]

14.375 [0.0062] 14.382 [0.0062]

83.454 [0.0001] 83.458 [0.0001]

Table A8 Kruskal–Wallis rank test within and between regions for coping with livestock epidemic. Strategy/region Central Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 35 49 86 11 Rank sum 3522 4353.5 7476.5 1119 2.243 [0.5235] 2.248 [0.5226] Western Obs 1 2 12 20 Rank sum 4 6 213 407 Eastern Obs 1 70 23 142 2 Rank sum 125 7668 2504.5 17,721.5 422 Northern Obs 1 73 41 121 11 Rank sum 214 8604 6372 13,877.5 1560.5 National Obs 3 180 125 369 24 Rank sum 1534 57,135.5 48,703.5 130,074 8604

7.211 [0.0655] 7.594 [0.0552]

6.390 [0.1718] 6.398 [0.1713]

12.794 [0.0123] 12.818 [0.0122]

11.432 [0.0221] 11.434 [0.0221]

1258

Table A9 Kruskal–Wallis rank test within and between agro-ecological zones for drought coping strategies. Strategy/agroecological Zone Banana/coffee Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 54 371 128 865 837 Rank 42,452 401,048 177,424 992,861 929,857 Banana/millet/ cotton Obs 12 219 13 364 151 Rank 2059.5 71,465 5491.5 143,348.5 66,055.5 Montane Obs 163 262 17 988 440 Rank 157,221.5 274,321.5 17,962 917,586 382,294 Teso Obs 30 80 13 168 43 Rank 7830 13,514 964 25,189 8448.5 Northern Obs 13 403 194 284 372 Rank 6856 238,258 113,983 178,635 264,280 Pastoral Obs 60 133 1 177 153 Rank 20,695 28,295 499 55,879 32,183 West Nile Obs – 477 67 114 388 Rank 257,721 45,767 63,040 181,054 National Obs 332 1945 433 2960 2384 Rank 1,660,000 7,000,000 1,720,000 12,900,000 9,130,000

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

38.312 [0.0001] 38.315 [0.0001]

36.266 [0.0001] 36.287 [0.0001]

19.363 [0.0007] 19.365 [0.0007]

49.727 [0.0001] 49.934 [0.0001]

26.006 [0.0001] 26.076 [0.0001]

74.585 [0.0001] 74.783 [0.0001]

35.006 [0.0001] 35.030 [0.0001]

204.648 [0.0001] 204.649 [0.0001]

Table A10 Kruskal–Wallis rank test within and between agro-ecological zones for pests coping strategies. Strategy/agroecological Zone Banana/coffee Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 2 53 335 220 196 Rank 808 17,330 146,443 85,317 75,324 Banana/millet/ cotton Obs 25 93 30 27 Rank 1560 8615 2951.5 2273.5 Montane Obs 2 34 107 78 98 Rank 399.5 6808 17,553 8575 17,706 Teso Obs 40 24 20 Rank 2053.5 803 713.5 Northern Obs 2 152 175 37 64 Rank 455 34,970.5 35842.5 8154 13243 Pastoral Obs 2 4 43 9 11 Rank 45 163.5 1857.5 79 270 West Nile Obs 17 9 28 Rank 550.5 335.5 599 National Obs 8 325 786 394 424 Rank 8430 292,915 814,089 350,575 410,945

15.054 [0.0046] 15.059 [0.0046]

8.573 [0.0355] 8.603 [0.0351]

34.954 [0.0001] 35.017 [0.0001]

10.114 [0.0064] 10.234 [0.0060]

3.763 [0.4390] 3.769 [0.4384]

26.661 [0.0001] 27.635 [0.0001]

9.333 [0.0094] 9.601 [0.0082]

24.043 [0.0001] 24.044 [0.0001]

Table A11 Kruskal–Wallis rank test within and between agro-ecological zones for ?oods coping strategies. Strategy/agroecological Zone Banana/coffee Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 13 75 60 230 83 Rank 1372.5 14,487 18,516.5 50,760 21,355 Banana/millet/ cotton Obs 10 93 30 127 69 Rank 1937 14,544.5 5150.5 20,536 12,117 Montane Obs 65 165 2 446 185 Rank 31,214.5 64,950.5 804.5 192,802.5 83,044 Teso Obs 3 51 16 42 12 Rank 276.5 2652.5 778 3086.5 956.5 Northern Obs 10 197 83 91 108 Rank 3185 43,428.5 17,783.5 25,502.5 29,905.5 Pastoral Obs 12 21 9 52 45 Rank 1206.5 1580 391.5 4298.5 2253.5 West Nile Obs 9 78 13 32 46 Rank 270 7879.5 996 2095 4690.5 National Obs 122 680 213 1020 548 Rank 178,149.5 747,847.5 242,938.5 1,410,000 755,657

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261

42.542 [0.0001] 42.566 [0.0001]

2.831 [0.5865] 2.840 [0.5850]

7.218 [0.1248] 7.221 [0.1247]

15.451 [0.0039] 15.636 [0.0035]

23.747 [0.0001] 23.769 [0.0001]

27.311 [0.0001] 27.945 [0.0001]

26.364 [0.0001] 26.474 [0.0001]

83.454 [0.0001] 83.458 [0.0001]

Table A12 Kruskal–Wallis rank test within and between agro-ecological zones for livestock epidemic coping strategies. Strategy/agroecological Zone Banana/coffee Obs Borrowing Labour supply Technology Savings Reduce consumption chi-squared chi-squared with ties 48 59 132 12 Rank 5904 7549.5 16463.5 1709 Banana/millet/ cotton Obs 30 10 40 Rank 1679.5 260.5 1300 Montane Obs 1 20 13 52 Rank 41 344 852 2504 Teso Obs 1 9 2 22 1 Rank 11.5 77.5 28 479 34 Northern Obs 1 51 41 86 2 Rank 148 4557.5 4360 7136 269.5 Pastoral Obs 16 24 Rank 335 485 West Nile Obs 6 13 9 Rank 72.5 159.5 174 National Obs 3 180 125 369 24 Rank 1534 57,135.5 48703.5 130,074 8604

0.779 [0.8544] 0.780 [0.8542]

21.926 [0.0001] 22.140 [0.0001]

34.130 [0.0001] 34.431 [0.0001]

13.683 [0.0084] 14.376 [0.0062]

8.160 [0.0859] 8.193 [0.0848]

0.037 [0.8468] 0.039 [0.8438]

4.581 [0.1012] 4.855 [0.0833]

11.432 [0.0221] 11.434 [0.0001]

1259

1260 Table A13 Variable de?nitions. Abbreviation tenure2 tenure1 credith2 credith1 offfarms2 offfarms1 offfarmh2 offfarmh1 edufather3 edufather2 edufather1 extension12 extension11 femalehhd malehhd westNile pastoral Nothern teso montane bananamill$n bananacoffee inputmktlc1 outputmktlc1 disttown1 farmsize extension coplepidemic coppests cop?oods copdrought lengthlepi$c lengthpests length?oods lengthdrou$t yrlepidemic yrpests yr?oods yrdrought tlepidemic tpests t?oods tdrought agrecology

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Table A17 Small–Hsiao test of IIA assumption for the ?oods coping model. Variable name Customary land tenure More formal tenure types No Access to credit by household head Access to credit by household head No access to off farm employment No access to off farm employment No access to off farm employment No access to off farm employment University education Secondary and other Primary education No access to extension services Access to extension services Female headed household Male headed household West Nile agro-climatic zone Pastoral agro-climatic zone Northern agro-climatic zone Teso agro-climatic zone Montane agro-climatic zone Banana/millet/cotton agro-climatic zone Banana/coffee agro-climatic zone Availability of input market in LC1 Availability of output market in LC1 Distance to nearest town Farm size Access to extension services Coping strategies for livestock epidemic Coping strategies for pests Coping strategies for ?oods Coping strategies for drought Length of last livestock epidemic Length of last pest attack Length of last ?oods Length of last drought Year last livestock epidemic occurred Year when the last pest attack occurred Year when the last ?oods occurred Year when the last drought occurred Frequency of livestock epidemics Frequency of pests Frequency of ?oods Frequency of drought Agro-ecological zone Omitted Labour supply Technology Savings chi2 33.185 33.691 30.586 d.f. 22 22 22 P > chi2 0.059 0.053 0.105 Evidence for Ho for Ho for Ho

References
Arthur, W.B., 1984. Inductive reasoning and bounded rationality. American Economic Review, Papers and Proceedings 406–411. Below, T., Artner, A., Siebert, R., Sieber, S., 2010. Micro-level practices to adapt to climate change for African small scale farmers: a review of selected literature. IFPRI Discussion Paper 00953. Boko, M., Niang, I., Nyong, A., Vogel, C., Githeko, A., Medany, M., Osman-Elasha, B., Tabo, R., Yanda, P., 2007. In: Parry, M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J., Hanson, C.E. (Eds.), Africa. 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 University Press, Cambridge, UK, pp. 433–467. Broad, O.B.K., Petty, A., 2004. Factors that in?uence the use of climate forecasts: evidence from the 1997/98 El Nino event in Peru. Bulletin of the American Meteorological Society 85, 1735–1743. Covello, V., 1990. Risk comparisons in communication: issues and problems in comparing health and environmental risks. In: Kasperson, R., Stallen, D. (Eds.), Communicating Risks to the Public: International Perspectives. Kluwer Academic Publishers, Dordrecht, pp. 79–124. Deressa, T.T., Hassan, R.M., Ringler, C., Alemu, T., Yesuf, M., 2009. Determinants of farmers’ choice of adaptation methods to climate change in the Nile Basin of Ethiopia. Global Environment Change 19, 248–255. Francisco, H.A., 2008. Adaptation to climate change: needs and opportunities in southeast Asia. ASEAN Economic Bulletin 25 (1), 7–19. Gregory, P., Ingram, J.S.I., Brklacich, M., 2005. Climate change and food security. Philosophical Transactions of the Royal Society B: Biological Sciences 360, 2139–2148. Grothmann, T., Patt, A., 2005. Adaptive capacity and human cognition: the process of individual adaptation to climate change. Global Environmental Change 15, 199–213. Hisali, E., Kasirye, I., 2008. Review of agricultural sector investments and institutional performance, ?nal report submitted to the Poverty Monitoring and Analysis Unit of the Ministry of Finance Planning and Economic Development (MFPED). Hoffmann, I., 2009. Livestock and climate change. <http://www.ifad.org/lrkm/ factsheet/cc.pdf>. IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Change 2007: Impacts, adaptation and vulnerability. Summary for policy makers. <http:// www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4_syr_spm.pdf> (accessed 01.07.09.). LTS International, 2008. Climate change in Uganda: understanding the implications and appraising the response. Scoping mission for DFID Uganda. <www.ltsi.co.uk>. McCarthy, J., Canziani, O.F., Leary, N.A., Dokken, D.J., White, C. (Eds.), 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK. Maddison, D., 2007. The perception of and adaptation to climate change in Africa. World Bank Policy Research Working Paper 4308. Nhemachena, C., Hassan, R., 2008. Micro-level analysis of farmers’ adaptation to climate change in southern Africa. IFPRI discussion paper 00714. Norris, E., Batie, S., 1987. Virginia farmers’ soil conservation decisions: and application of tobit analysis. Southern Journal of Agricultural Economics 19 (1), 89–97. Paavola, J., 2008. Livelihoods, vulnerability and adaptation to climate change in Morogoro, Tanzania. Environmental Science & Policy 11, 642–654. Patt, G.A., Schroter, D., 2008. Perceptions of climate risk in Mozambique: implications for the success of adaptation strategies. Global Environmental Change 18, 458–467. Republic of Uganda, 2007. Climate Change: Uganda National Programmes of Action. Ministry of Water and Environment. Rogers, R.W., 1975. A protection motivation theory of fear appeals and attitude change. Journal of Psychology 91, 93–114. Rogers, R.W., 1983. Cognitive and physiological processes in fear appeals and attitude change: a revised theory of protection motivation. In: Cacioppo, J., Petty, R. (Eds.), Social Psychophysiology. Guilford Press, New York. Shiferaw, B., Holden, S., 1998. Resource degradation and adoption of land conservation technologies in the Ethiopian highlands: a case study in Andit Tid, North Shewa. Agricultural Economics 18, 233–247. Stringer, L.C., Dyer, J.C., Reed, M.S., Dougill, A.J., Twyman, C., Mkwambisi, D., 2009. Adaptations to climate change, drought and deserti?cation: local insights to enhance policy in southern Africa. Environmental Science and Policy. doi:10.1016/j.envsci.2009.04.002.

Table A14 Hausman tests of IIA assumption for the drought model. Omitted Borrowing Labour supply Technology Reduce consumption chi2 ?1.476 0 0 0 d.f. 40 1 1 1 P > chi2 – 1 1 1 Evidence – for Ho for Ho for Ho

Table A15 Small–Hsiao test of IIA assumption for the pest attack model. Omitted Labour supply Savings Reduce consumption Technology chi2 11.509 7.194 12.410 7.743 d.f. 24 24 24 24 P > chi2 0.985 1.000 0.975 0.999 Evidence for for for for Ho Ho Ho Ho

Table A16 Small–Hsiao test of IIA assumption for the livestock epidemic model. Omitted Labour supply Technology Savings chi2 3.180 14.565 2.246 d.f. 8 8 8 P > chi2 0.923 0.068 0.973 Evidence for Ho for Ho for Ho

E. Hisali et al. / Global Environmental Change 21 (2011) 1245–1261 Tenge, De Graaff, J., Hella, J.P., 2004. Social and economic factors affecting the adoption of soil and water conservation in West Usambara highlands, Tanzania. Land Degradation and Development 15 (2), 99–114. Thomas, D.S.G., Twyman, C., Osbahr, H., Hewitson, B., 2007. Adaptation to climate change and variability: farmer responses to intra-seasonal precipitation trends in South Africa. Climatic Change 83 (3), 301–322. Thornton, P.K., Jones, P.G., Owiyo, T., Kruska, R.L., Herrero, M., Kristjanson, P., Notenbaert, A., Bekele, N., Omolo, A, with contributions from Orindi, V., Otiende, B., Ochieng, A., Bhadwal, S., Anantram, K., Nair, S., Kumar, V., Kulkar, U., 2006. Mapping climate vulnerability and poverty in Africa. Report to the Department for International development. ILRI, Nairobi, Kenya. Tversky, A., Kahneman, D., 1973. Availability: a heuristic for judging frequency and probability. Cognitive Psychology 5, 207–232.

1261

Tversky, A., Kahneman, D., 1974. Judgment under uncertainty: heuristics and biases. Science 211, 1124–1131. UNDP, 2007. Human Development Report 2007/2008. Fighting climate change: human solidarity in a divided world. <http://hdr.undp.org/en/media/ HDR_20072008_EN_Complete.pdf>. Valdivia, C.J., Gilles, J., Espejo, R., Carrillo, R., 2001. Current users and diffusion nodes of local climate forecasts in the Andes of Bolivia: lessons on potential users, timing, and content of climate forecast communications. In: IRI Communication of Climate Forecast Information Workshop. Palisades, New York. ? , C., Quiroz, R., Gilles, J.L., Materer, S., 2002. Peasant households’ Valdivia, C., Jette ?o strategies in the Andes and potential users of climate forecasts: El Nin of 1997–1998. Selected paper for the Annual Meeting of the American Agricultural Economics Association, July 30–August 2, Tampa, FL.


相关文章:
...LEVEL EVIDENCE FROM UGANDA, TANZANIA AND KENYA_论文.pdf
COMPETITION PRESSURE AND AGRO-PROCESSING: FIRM LEVEL EVIDENCE FROM UGANDA, TANZANIA AND KENYA_信息与通信_工程科技_专业资料。C0MPETIT10N PRESSURE AND AGR0....
(中学联盟)潍坊市2014届高三上学期期末考试 英语.doc
Fate, and Uganda. Jimmy’s life was in ...evidence C. guidance C. hoped C. compete C. ... a greenhouse gas linked to climate change , ...
IFPRI At a Glance_图文.pdf
IFPRI provides evidence-based policy solutions as ... Ghana, Malawi, Nigeria, Pakistan, and Uganda. ...to climate change, increasing crop yields, and ...
...for Political Development: Evidence From Uganda's Ge_论文_....pdf
Political Functionalism as a Prerequisite for Political Development: Evidence From Uganda's Ge_专业资料。Juoma f-iaPbi miirtnIS14-51lUSChn ulAdnsai,SN ...
How ICT can make a difference in_图文.pdf
effects of climate change and rapid decrease of... evidence of the contribution of ICT to ...by IICD in Bolivia, Uganda, Tanzania and Ghana...
10Food Security and Adaptation to Climate Change Wh....pdf
responses to past weather events are some of the only evidence we have to understand how 8 Food Security and Adaptation to Climate Change: What Do We ...
...in less-favored areas in developing countries evidence and....pdf
presents some preliminary evidence from Uganda, and... using district-level data from rural India for ... changes in rural wages, and changes in ...
...support for health services planning in Uganda_图文.pdf
There is evidence that poor access to health ... the second-level administrative unit in Uganda, ...Although some boundary changes were noted, ...
The Microeconomics of African Growth, 1950-2000”. Thematic ....pdf
Evidence for five African countries suggest that in fact most microenterprises... `Changes in Poverty in Uganda, 1992-1996’, Working Paper 98-15, ...
...market orientation and business performance evidence from_....doc
meta-analysis of the relationship between market orientation and business performance evidence from_文学研究_人文社科_专业资料。Collecting social contact data in...