IWMI Research Determinants of Adoption Report of Rainwater Management Technologies among Farm 154 Households in the Nile River Basin Gebrehaweria Gebregziabher, Lisa-Maria Rebelo, An Notenbaert, Kebebe Ergano and Yenenesh Abebe Research Reports The publications in this series cover a wide range of subjects—from computer modeling to experience with water user associations—and vary in content from directly applicable research to more basic studies, on which applied work ultimately depends. Some research reports are narrowly focused, analytical and detailed empirical studies; others are wide-ranging and synthetic overviews of generic problems. Although most of the reports are published by IWMI staff and their collaborators, we welcome contributions from others. Each report is reviewed internally by IWMI staff, and by external reviewers. The reports are published and distributed both in hard copy and electronically (www.iwmi.org) and where possible all data and analyses will be available as separate downloadable files. Reports may be copied freely and cited with due acknowledgment. About IWMI IWMI’s mission is to improve the management of land and water resources for food, livelihoods and the environment. In serving this mission, IWMI concentrates on the integration of policies, technologies and management systems to achieve workable solutions to real problems—practical, relevant results in the field of irrigation and water and land resources. IWMI Research Report 154 Determinants of Adoption of Rainwater Management Technologies among Farm Households in the Nile River Basin Gebrehaweria Gebregziabher, Lisa-Maria Rebelo, An Notenbaert, Kebebe Ergano and Yenenesh Abebe International Water Management Institute (IWMI) P O Box 2075, Colombo, Sri Lanka i The authors: Gebrehaweria Gebregziabher is a Research Economist at the East Africa and Nile Basin Office of the International Water Management Institute (IWMI) in Addis Ababa, Ethiopia; Lisa-Maria Rebelo is a Senior Researcher – Remote Sensing and GIS at the Southeast Asia Office of IWMI in Vientiane, Lao PDR; An Notenbaert is a Spatial Analyst at the headquarters of the International Livestock Research Institute (ILRI) in Nairobi, Kenya; Kebebe Ergano is a Research Officer at ILRI in Addis Ababa, Ethiopia; and Yenenesh Abebe is a Database – GIS Specialist at the East Africa and Nile Basin Office of IWMI in Addis Ababa, Ethiopia. Gebregziabher, G.; Rebelo, L-M.; Notenbaert, A.; Ergano, K.; Abebe, Y. 2013. Determinants of adoption of rainwater management technologies among farm households in the Nile River Basin. Colombo, Sri Lanka: International Water Management Institute (IWMI). 34p. (IWMI Research Report 154). doi: 10.5337/2013.218 / water management / water conservation / rainwater / farmers / gender / households / living standards / economic aspects / river basins / watersheds / landscape / soil conservation / Ethiopia / ISSN 1026-0862 ISBN 978-92-9090-781-7 Copyright © 2013, by IWMI. All rights reserved. IWMI encourages the use of its material provided that the organization is acknowledged and kept informed in all such instances. Front cover photograph shows smallholder agriculture in the Ethiopian Highlands (photo: Petterik Wiggers). Please send inquiries and comments to: IWMI-Publications@cgiar.org A free copy of this publication can be downloaded at www.iwmi.org/Publications/IWMI_Research_Reports/index.aspx Acknowledgements The authors gratefully acknowledge the Amhara Regional Agricultural Research Institute (ARARI), Ethiopia; the Oromia Agricultural Research Institute (OARI), Ethiopia; and the Ethiopian Rainwater Harvesting Association (ERHA), for their advice and assistance with data collection. In particular, we would like to thank Gerba Leta, Research Assistant, International Livestock Research Institute (ILRI), Addis Ababa, Ethiopia; Catherine Pfeifer, Postdoctoral Fellow, ILRI, Addis Ababa, Ethiopia; Zerihun Abebe and Bayisa Gedefa, Researchers at Bako Agricultural Research Center, OARI, Ethiopia; Gizaw Desta, Senior Researcher, Soil and Water Management, ARARI, Ethiopia; Ashenafi Teklay, Researcher, Gonder Agricultural Research Center, ARARI, Ethiopia; and Tewodros Teshome, Head, ERHA, for their assistance with field data collection. This research study was supported by the CGIAR Challenge Program on Water and Food (CPWF), and the CGIAR Research Program on Water, Land and Ecosystems (WLE), which is led by IWMI. Project This research study was initiated as part of the CGIAR Challenge Program on Water and Food (CPWF) Nile Basin Development Challenge (NBDC), ‘To improve rural livelihoods and their resilience through a landscape approach to rainwater management’. The NL3 project focused specifically ‘On targeting and scaling out’. Collaborators International Water Management Institute (IWMI) International Livestock Research Institute (ILRI) Ethiopian Rainwater Harvesting Association (ERHA) Amhara Regional Agricultural Research Institute (ARARI), Ethiopia Oromia Agricultural Research Institute (OARI), Ethiopia Donors Funding for the research carried out in this report was provided by the following organizations, among others, through the CGIAR Challenge Program on Water and Food (CPWF). UK Department for International Development (DFID) European Commission International Fund for Agricultural Development (IFAD) Swiss Agency for Development and Cooperation (SDC) The views expressed herein do not necessarily reflect the official opinion of DFID, European Commission, IFAD or SDC. This work has been undertaken as part of the CGIAR Research Program on Water, Land and Ecosystems. IWMI is a member of the CGIAR Consortium and leads this program. Contents Summary vii Introduction 1 Study Area and Data 3 Methodology 5 Independent Variables and Hypotheses 8 Results and Discussion 10 Conclusions 20 References 21 Appendix 24 v Summary Agriculture is the main sector of the Ethiopian landscape approach to rainwater management in economy, as is the case in many sub-Saharan the Ethiopian part of the Blue Nile River Basin. African countries. In this region, rainfal l The conceptual framework of this study is distribution is extremely uneven both spatially based on the premise that farmers are more and temporally. Drought frequently results in likely to adopt a combination of promising crop failure, while high rainfall intensities result rainwater management technologies as a coping in low infiltration and high runoff, causing soil mechanism against climate variabil ity and erosion and land degradation, which contribute agricultural production constraints. For example, to low agricultural productivity and high levels multi-purpose trees are likely to complement of food insecurity. High population growth and bunds/terraces, while bunds/terraces are likely to cultivation of steep and marginal lands, together result in increased infiltration and groundwater/ with poor land management practices and lack surface water recharge, thereby leading to of effective rainwater management strategies, the adoption of shallow/hand-dug wells and, aggravate the situation. subsequently, orchards and irrigated crops. Over the past two decades, the Government On the other hand, when farmers adopt area of Ethiopia has attempted to address these enclosures and/or gully rehabilitation, they usually issues through the large-scale implementation of supplement these with different trees and grasses a range of soil and water conservation measures, for animal feed and food production. Similarly, including stone terraces, soil bunds and area farmers usually invest in river diversion structures enclosures. Despite these efforts, adoption of for irrigation to produce high-value cash crops. the interventions remains low. Studies from the In general, farmers are faced with alternative, Ethiopian Highlands show that the adoption of but correlated, technologies in their adoption rainwater management technologies is influenced decisions, implying the interdependence of by a variety of factors, including biophysical technologies. This is in contrast to much of characteristics such as topography, slope, soil the previous work carried out on the adoption fertility, rainfall amount and variability. However, of rainwater management technologies, which even when technologies are appropriate to typically examined a single technology without a particular biophysical setting, they may not considering the interdependence between be implemented, because farmers usually technologies. This study contributes to filling this consider a variety of factors when making their research gap based on evidence from the Blue decisions to adopt technologies. Thus, gaining Nile River Basin. The study assesses the patterns an understanding of the factors that influence the of adoption and the factors that influence farm adoption of rainwater management technologies household adoption of rainwater management is crucial for improved management of land technologies, and draws recommendations and and water resources. In this context, this study policy implications from the results. has been carried out within the framework of The data for this study were obtained from a the Nile Basin Development Challenge (NBDC) household survey conducted in seven watersheds project of the CGIAR Challenge Program on in the Ethiopian part of the Blue Nile River Basin Water and Food (CPWF), which aims to improve in 2011. The sample farm households were rural livelihoods and their resilience through a selected using a multi-stage stratified random vii sampling design. The total sample size was 671, and biophysical performance of conservation with information from 654 farm households used measures, but also on the socioeconomic and in the analysis. In addition to descriptive statistics, livelihood benefits; (ii) adoption of rainwater a multivariate probit model was used to analyze management technologies are interdependent, the data. hence any act iv i ty to promote ra inwater The regression results showed significant management technologies need to consider such correlat ion and interdependence between interdependence. Failure to do so may mask the rainwater management technologies. Differences reality that farmers are typically faced with a set in the estimated coefficients across equations of choices, and will result in poor performance of also support the appropriateness of differentiating the technologies; (iii) targeting women groups to between technology opt ions. Household address their constraints to actively participate demographic characteristics (such as age, in rural economic activities can have a positive gender, marital status and family size in adult impact on the adoption of rainwater management equivalent), participation in off-farm activities, technologies; (iv) farmers with better experience migration, ownership of livestock, ownership of and information are more likely to adopt and try land, landholding size per adult equivalent, access out new technologies. Identifying these farmers to credit centers and markets captured by walking and working with them can help to promote distance, social capital captured by household successful and proven technologies; (v) in membership in formal/informal networks, and addition to the socioeconomic and demographic farm household location captured by the fixed characteristics, it is important to understand the effects (dummies) of the woredas (districts) were biophysical suitability of technologies instead of the main determinants (positive or negative) of promoting blanket recommendations; and (vi) adoption of rainwater management technologies. externally driven technical solutions are rarely The main recommendations of the study are sustained by farmers unless consideration is given as follows: (i) rainwater management interventions to socioeconomic, cultural and institutional, as well should focus not only on the engineering as biophysical and technical factors. viii Determinants of Adoption of Rainwater Management Technologies among Farm Households in the Nile River Basin Gebrehaweria Gebregziabher, Lisa-Maria Rebelo, An Notenbaert, Kebebe Ergano and Yenenesh Abebe Introduction Agriculture is the main sector of the Ethiopian The use of rainwater management (RWM) economy. It contributes to approximately 42% of interventions, including soil and water conservation the gross domestic product (GDP), generates more (SWC) techniques, is widely accepted as a than 85% of the foreign exchange earnings and key strategy to improve agricultural productivity employs over 80% of the population (CSA 2004; by alleviating growing water shortages, the MoFED 2010). The Government of Ethiopia is effects of droughts and worsening soil conditions committed to rapid agricultural growth as a means (Kurukulasuriya and Rosenthal 2003). In the of accelerating economic growth and reducing rainfed agroecological landscapes of Ethiopia, poverty. Despite impressive achievements over the the low yield, which is, on average, about 35% last three decades, Ethiopia remains one of the of the potential, is typically not due to the lack poorest countries in the world with over 12 million of water but rather a result of the inefficient people being food-insecure. Agricultural productivity management of water, soils and crops (Amede is low, dominated by low input-low output rainfed 2012). The gap between actual and attainable mixed crop-livestock production in the Ethiopian yields suggests that there is a large untapped Highlands (Merrey and Gebreselassie 2011). In potential for yield increases (Rockström et al. this region, agricultural productivity is constrained 2010). It has been demonstrated that access to by high climate variability rather than low water RWM interventions can reduce poverty levels availability; rainfall distribution is extremely uneven by approximately 22% (Awulachew et al. 2012). both spatially and temporally, which has negative These interventions can also provide a buffer implications for the livelihoods of the population against production risks associated with increasing (FAO 2005). Drought frequently results in crop rainfall variability due to climate change (Kato et failure, while high rainfall intensities result in low al. 2009). While various studies have highlighted infiltration and high runoff, causing soil erosion the potential of RWM interventions to increase and land degradation, which contribute to low agricultural productivity and improve livelihoods in agricultural productivity and high levels of food Ethiopia (Pender and Gebremedhin 2007; Kassie insecurity (Lautze et al. 2003; Deressa 2007). High et al. 2008; Awulachew et al. 2010), in practice population growth rates and the cultivation of steep adoption rates of these interventions remain low and marginal land exacerbates the problem; soil (Santini et al. 2011). erosion in the cultivated highlands of Ethiopia is The Government of Ethiopia along with estimated to be 42 tonnes/ha, on average (Hurni development agencies has invested substantial 1990; Tamene and Vlek 2008). resources to promote SWC pract ices in 1 part icular , and a range of intervent ions income, wealth of the household, access to including, but not limited to, stone terraces, extension and credit, information on climate, soil bunds and area enclosures. These have farmer-to-farmer extension and number of been introduced at a large scale, but with relatives (as a proxy of social capital) influence limited success (Zemadim et al. 2011). Since farmers’ adoption of RWM technologies. Hence, the early 1970s, for example, food for work one way to improve productivity and build (FFW) programs have been widely implemented climate-resilient livelihoods in the Ethiopian as a means of providing much-needed food Highlands is to target promising technologies aid to rural communities, which are earned to a particular (biophysical and socioeconomic) by undertaking rural public works. A major environment. Santini et al. (2011), for example, component of this has been the construction highlighted the need for new models of planning of soil and water conservation structures, with for RWM investments by recognizing the diversity the intention of preventing or reversing erosion and complexity of the country contexts, and by processes (Merrey and Gebreselassie 2011). tailoring the interventions to suit the priorities However, the outcome of these conservation and livelihood strategies of the rural population. measures was not as expected, and it has This helps to overcome the limited success and been emphasized that interventions should not impact of practices that are often adopted using only focus on the engineering and biophysical ‘blanket’ approaches (ILRI and IWMI 2010). In performance of conservation measures but addition, a package of technologies should be also on the socioeconomic and livelihood considered rather than individual interventions. benefits (Zemadim et al. 2011). Studies from the However, there is no agreement on the Ethiopian Highlands show that the adoption of factors that encourage or discourage adoption RWM technologies are influenced by a variety of specific technologies and practices (which of factors, including biophysical characteristics can be farm- and/or technology-specific), and such as topography, slope, soil fertility, rainfall a better understanding of these factors is amount and rainfall variability (Deressa et al. needed. Furthermore, relatively little empirical 2009). Experience also shows that even when work has been undertaken to examine the technologies are appropriate for the biophysical factors that affect the adoption of RWM setting, they are not always adopted (Guerin technologies at the watershed level, or as 1999; Amsalu and Graaff 2007) because farmers a ‘package’ or combination of technologies. consider a variety of factors when making Most of the previous research that has been a decision to adopt a particular intervention conducted has focused on the adoption of (McDonald and Brown 2000; Soule et al. 2000). individual technologies over a large (country In addition, studies have found that farmers’ or regional) scale, while farmers typically recognition of the problem (e.g., soil erosion, adopt multiple RWM technologies that deal low agricultural productivity) and awareness with their overlapping constraints and are of the potential solutions are necessary, but suitable to specific landscapes and the position not sufficient conditions for the adoption and within the landscape. Kato et al. (2009), for continued use of SWC technologies (Merrey and example, highlighted that the effectiveness of Gebreselassie 2011). Externally driven technical various SWC technologies in Ethiopia depend solutions are rarely sustained by farmers unless on whether they are used independently or consideration is given to socioeconomic, cultural as a package. As the suitabil ity of RWM and institutional as well as biophysical and technologies may depend on the position within technical factors (McDonald and Brown 2000; a landscape, interventions implemented in the Merrey and Gebreselassie 2011). An empirical upper slopes are likely to be different from those study from the Nile Basin (Deressa et al. 2009) implemented in the lower slopes. Under such demonstrated that, among others, the level of a scenario, analyses which do not take into education, gender, age, farm and non-farm account landscape or watershed variability may 2 underestimate or over-estimate the influence of adoption or dis-adoption of a particular RWM factors affecting the adoption of technologies. technology and combinations of technologies in As limited rigorous empirical work has the Ethiopian Highlands. The outcome of this been carried out on the economic factors that study contributes to the growing evidence base influence the adoption of particular technologies for the adoption of RWM technologies as a (Kassie et al. 2008), the objective of this study strategy for sustainable agriculture and climate- is to understand the factors that influence resilient livelihoods. Study Area and Data This study has been carried out within the new watersheds should be relatively small and framework of the Nile Basin Development managed by one or two communities. The Challenge (NBDC) project, which aims to biophysical description of the study sites is improve livelihoods of the rural population and presented in Table 1. build their resilience to climate change through Cross-sectional data have been collected from a landscape approach to rainwater management 671 randomly selected sample households (see in the Ethiopian part of the Blue Nile River Table 2) for the purpose of this study. Basin. Within this region, three landscapes A multi-stage stratified random sampling were identified and selected for this study – procedure was used to select the sample Jeldu, Diga and Fogera. These landscapes households. In the f i rst stage, a l ist of were different in their state of development, households from each kebele (community) agroecology and livelihood systems, and there within the watershed was used to stratify were opportunities for the implementation them by location of landscape. Following of rainwater management strategies. Action this, households in the selected watersheds research sites have been selected within these were stratified into female- and male-headed study landscapes, providing a nested set of households, and according to adopt ion sites for learning and research. The sample status (i.e., adopter and non-adopter), in watersheds for the current study include these order to generate a reasonable proportion three NBDC research action sites (Meja in Jeldu, of both these household categories for a Dapo in Diga and Mizewa in Fogera) along with counterfactual analysis. Finally, proportional four new sites (Boke, Laku, Zefe and Maksegnit) random sampling was employed to identify which are also located within the Blue Nile River 671 sample households. The adoption rate of Basin (Figure 1). RWM technologies disaggregated by gender The four new sites were selected by the is presented in Appendix, Table A1. Following national partners, Oromia Agricultural Research data collection, it became apparent that 17 Institute (OARI) and Amhara Regional Agricultural households from the Meja watershed had Research Institute (ARARI), for the purpose of been wrongly surveyed. Thus, by excluding this study and to fit in with the NBDC definition these households, only 654 of the sample of a landscape. The criteria used to select households were considered in the analysis. For the new watersheds include the presence of data collection, a structured questionnaire that RWM interventions, and size and slope of the comprised a set of household and community watershed. Similar to the NBDC sites, the four level information was used. 3 4 TABLE 1. Biophysical characteristics of the sample watersheds. Woreda Sample Nearest rainfall and Mean Aridity Erosion Average travel time Rainfall Temperature Soil type* watershed temperature station Index (rate/tonnes/ha/year) to nearest town variability (AI) in hours coefficient of variance (CV) Gondar Zuria Maksegnit Gondar Zuria 0.68* Humid 20.82* High 3* 0.298 27 Chromic Luvisols, Eutric Leptosols and Eutric Vertisols Fogera Mizewa Worota/Addis Zemen 0.66* Humid 1.78* Low 4* 0.348 29 Haplic Luvisols and Eutric Fluvisols Farta Zefe Debre Tabor 0.81** Humid 28.66** High 4** 0.316 22 Chromic Luvisols Diga Dapo Nekemte 0.96* Humid 4.45* Low 4* 0.305 24 Haplic Alisols Horo Laku Shambu 1.03** Humid 13.14** Moderate 5** 0.328 22 Haplic Alisols Jeldu Meja Ambo 0.89* Humid 1.94* Low 2* 0.854 23 Haplic Alisols Ambo Boke/Gorosole Tikur Inchini 0.72** Humid 29.36** High 5** 0.228 23 Chromic Luvisols Source Zomer et al. 2007, Haileslassie Global Agriculture NMA 2007; MoWE 1998a 2008; Spinoni et al. 2005; and Food Security National et al. 2013 Shiferaw 2011 Program (GAFSP) Meteorology Consortium for Ethiopia – Average Agency (NMA), Spatial Information travel time to nearest Ethiopia, (CGIAR-CSI) town over 20K dataset and Global Aridity and (hours) (2000) information PET Database (Available at http: resources (Available at http: //maps.worldbank. (http://www. //www.cgiar-csi.org) org/overlays/7406) ethiomet.gov.et/ data_access/ information) Notes: Soil erosion severity classes: 0-10, 10-20, 20-30, 30-45, 45-60, 60-80 and > 80 classified as Low, Moderate, High, Very high, Severe, Very severe and Extremely severe, respectively. Aridity Index classes: < 0.05, 0.05-0.2, 0.2-0.5, 0.5-0.65 and > 0.65 classified as Hyper-arid, Arid, Semi-arid, Dry Sub-humid and Humid, respectively. * Watershed scale ** Woreda scale FIGURE 1. Location of selected sample watersheds. TABLE 2. Watersheds surveyed and the number of randomly selected sample households. Region Woreda/District Watershed Site Number of Data sample collected by households Oromia Jeldu Meja NBDC 120 ERHA Guder Boke New 90 OARI Shambu Laku New 90 OARI Diga Dapo NBDC 90 OARI Amhara Farta Zefe New 90 ARARI Fogera Mizewa NBDC 101 ARARI Gondar Zuria Gumera/Maksegnit New 90 ARARI Total 671 Methodology The methodological framework is based on t e c h n o l o g i e s , w h i c h m a y b e a d o p t e d the premise that farmers are more likely to s i m u l t a n e o u s l y a n d / o r s e q u e n t i a l l y a s adopt a combination of rainwater management a complement or supplement to each other. 5 For example, multi-purpose trees are likely to In this context, we employ a multivariate complement bunds/terraces, while bunds/terraces probit model (MVP) (Kassie et al . 2012; are likely to result in increased infiltration and Cappellari and Jenkins 2003) as shown in groundwater/surface water recharge, thereby Equations (1) and (2). leading to the adoption of shallow/hand-dug wells and, subsequently, orchards and irrigated crops. Y * ht = b ' t Xht + eht, t = 1,....m and (1) On the other hand, when farmers adopt area enclosures and/or gully rehabilitation, they usually supplement these with different trees and grasses Y * ht = 1 if Y * ht > 0 and 0 otherwise (2) for animal feed and food production. Similarly, farmers usually invest in river diversion structures where: T = 1,...m represents the choices for irrigation to produce high-value cash crops. In of rainwater management technologies. The general, farmers are faced with alternative, but assumption is that hth farm household has a correlated, technologies in their adoption decisions, latent variable Y * ht that captures the choices which implies the interdependence of technologies. associated with the Tth rainwater management Furthermore, the choice of technologies selected technology. may be partly dependent on earlier experiences The estimation is based on the observed (Kassie et al. 2012). Various empirical studies binary discrete variables Y * ht that indicate (Moyo and Veeman 2004; Marenya and Barrett whether or not hth farm household has adopted 2007; Nhemachena and Hassan 2007; Yu et al. a particular rainwater management technology 2008; Kassie et al. 2009) argued that farmers (denoted by 1 for adoption and zero for non- usually consider a set of possible technologies and adoption). The status of adoption is assumed select the single one that they assume will have to be influenced by observed characteristics the best results; hence, the adoption decision is (Xht ), including household characteristics, access inherently multivariate. However, most previous to services, markets, social capital (captured by studies of technology adoption (such as rainwater household’s membership in formal and/or informal management and conservation technologies) social groups), and biophysical characteristics assume a single technology without considering captured by woreda/district dummies. The the possible correlation/interdependence between unobserved characteristics are captured by the different technologies (Yu et al. 2008), thereby error term denoted by eht, while bt is a parameter masking the reality that decision makers are to be estimated. often faced by a set of choices. In general, In line with this, we assume that rainwater when technologies are correlated, univariate management technologies considered in this modeling excludes useful information contained study are interdependent, implying that the in the interdependence and adoption decision adoption of one technology is likely to influence analysis. A single technology approach may, (posit ively or negatively) the adoption of therefore, underestimate or over-estimate the another technology, hence the error terms influence of factors on the adoption decision. In (eht , t = 1,...,m) in Equation (1) are distributed general, univariate models ignore the potential as multivariate with zero mean and variance correlation among unobserved disturbances in the 1, where eht » MVN(0,V). The value of variance adoption equations as well as the relationships (V) is, therefore, normalized to unity on the between the adoption of different rainwater diagonal and correlations as off-diagonal management technologies, because farmers elements in Equat ion (3). The non-zero may consider some combination of technologies value of the off-diagonal elements allow for as complementary and/or competing. Failure to correlation across the error terms of several capture such interdependence will lead to biased latent equations, which represent unobserved and inaccurate estimates. characteristics that affect the choice of alternative 6 rainwater management technologies (Kassie et Al though nine rainwater management al. 2012). The covariance matrix V is given in technologies were initially considered, it was Equation (3): clear from the survey data that adoption rates of shallow wells and ponds were low and insufficient 1 r12 r13 ... r to undertake further statistical analysis. Moreover, 1m as the estimation of the multivariate probit model r21 1 r23 ... r2m was cumbersome to converge, bunds/terraces V = r (3) 31 r33 1 ... r with vegetation and without vegetation were 3m merged as one technology. Finally, we reviewed 1 whether the technologies were adopted on rm1 rm2 rm3 1 private or public (communal) land and the result showed that river diversions and area enclosures were adopted on public lands (Figure 2), as In general, the multivariate probit model is a collective action. Since the analysis is based generalization of the probit model that is used to on household-level data, both river diversions estimate numerous correlated binary outcomes and area enclosures were excluded from the jointly, where the source of correlation can analysis. Hence, only four rainwater management be complementarity (positive correlation) and technologies (multi-purpose trees, orchards, substitutability (negative correlation) between bunds/terraces and gully rehabilitation) were different technologies (Belderbos et al. 2004). considered in the analysis. FIGURE 2. Adoption of rainwater management technologies and ownership of land. 7 Independent Variables and Hypotheses The explanatory variables considered in the production assets and livestock ownership. analysis and their expected effects on the These constraints will clearly have a direct effect adoption of rainwater management technologies on technology adoption (including rainwater are discussed below. management technologies), where women are Household characteristics: In this regard, we usually less likely to adopt these technologies as considered different household characteristics they are resource-demanding and labor-intensive and family member composition as a proxy (Ndiritu et al. 2011). for the human capital of the households. For Capital ownership: This variable is captured example, the level of education, age and gender by the number of livestock (Total Livestock of the family members, and family size are Units [TLU]), farm size per adult equivalent important indicators of the available human (a dummy variable that captures whether or capital, which has an influence on the adoption not a farm household owns the land) and of technologies. Households with more educated the value of durable household assets. The members are likely to have better access to assumption is that households that own more information, and are more aware about the capital are wealthier and more likely to take merits and demerits of the technologies. They risks associated with the adoption of new are also able to interpret new information to technologies. Moreover, such households are make knowledge-based decisions in favor of less constrained financially and are able to appropriate/suitable technologies. On the other purchase inputs. Household expenditure is hand, households with more educated members also considered as a proxy for income level. may be less likely to invest in labor-intensive Hence, the expected effect of capital on the technologies and practices, because they are adoption of rainwater management technologies more likely to earn higher returns from their labor is positive. However, since households with and capital investment through other activities relatively large landholdings may be able to (Kassie et al. 2012; Pender and Gebremedhin diversify their crops and income sources, they 2007). The age of the members of the household may be less susceptible to risks and shocks; as may imply farming experience and the ability such, they may be less interested in investing to respond to unforeseen events/shocks. Older in rainwater management technologies as a household heads may have an accumulation of coping mechanism. capital and respect in their community, implying Off-farm activity: Economic incentives play greater social capital. On the other hand, age an important role in the adoption of rainwater can be associated with loss of energy and short management technologies. Households’ access planning horizons, and the reluctance towards to off-farm employment and alternative sources new technologies due to risk aversion behavior. of income are likely to influence the adoption of Gender is an important factor in terms of rainwater management technologies in different access to resources. The general argument ways. For example, those who have alternative is that women have less access to important sources of income are better able to adopt resources and services, such as land, labor, and invest in these technologies. On the other credi t and educat ion, and are general ly hand, participation in off-farm income-generating discriminated against in terms of access to activities is likely to divert labor from on-farm external inputs and information (De Groote and activities and working on rainwater management Coulibaly 1998; Quisumbing et al. 1995). In technologies, both as a private investment and sub-Saharan Africa, there are gender-specific as collective action. The findings of Deressa constraints that women face, such as less et al. (2009) supported this hypothesis. Off- education, inadequate access to land, and farm activity is captured by the participation of 8 household members in the FFW program and/ products, which in turn can affect technology or whether any member of a household had adoption (Lee 2005; Pender and Gebremedhin migrated. Both these variables are defined as 2007; Wollni et al. 2010). Moreover, farmers who dummy variables (a value of 1 for participation have limited contacts with extension agents can and zero otherwise). be informed about the methods and benefits of Access to markets, extension, credit and new technologies from their networks, as they inputs: The walking distance (in minutes) was share information and learn from each other. used as a proxy of access to markets, extension On the other hand, having more relatives may and input supply centers. Access to credit was reduce incentives for hard work and induce captured by the household response when asked inefficiency, such that farmers may exert less whether they had requested for credit and the effort to invest in technologies (Kassie et al. actual amount of the loan they received in the 2012). The expected effect of the social capital previous year. Access to markets can influence coefficient is, therefore, ambiguous prior to the use of various inputs as well as access to empirical testing. information and support services. For example, Biophysical characteristics: Various rainwater Deressa et al. (2009) revealed that access to management technologies can be used as a credit has a significant positive impact on the coping mechanism in areas with low rainfall and likelihood of using soil conservation techniques, moisture stress, while others are more suited changing planting dates and using irrigation in the to areas with high rainfall. Unfavorable rainfall Blue Nile River Basin. Therefore, the hypothesis amounts, such as too little rainfall, may encourage is that the longer the walking distance to markets farmers to adopt soil and water conservation and other service centers, the less likely it is practices. On the other hand, the high rainfall that households will adopt a particular rainwater intensities that result in high runoff can augment harvesting technology. soil erosion leading to nutrient depletion. It can Social capital: This is represented by also increase waterlogging (Kassie et al. 2010), variables such as the household membership in which may negatively influence the likelihood of informal institutions (such as Equib and Edir)1. adoption of soil and water conservation practices. In Ethiopia, it is common for rural communities Hence, farm households may adopt certain to form informal groups for labor sharing, and rainwater management technologies (e.g., bunds/ saving and risk-sharing mechanisms. This can terraces) to reduce exposure to rainfall hazards take place in the form of friendship or kinship by increasing soil moisture, reducing soil loss networks, implying that households with a from erosion and flooding, and diversifying large number of relatives and wider networks cropping patterns. are likely to be more resilient to risk and have In the Blue Nile River Basin, the topography fewer credit constraints; they are more likely to follows a gradient from the flat lowlands in the adopt technologies because they are in a better West to mountainous areas in the East. While position to take risks (Fafchamps and Gubert it is acknowledged that topographical and soil 2007). With limited information and imperfect characteristics will affect the suitability of rainwater markets, social networks can facilitate the management technologies (due to the lack of exchange of information, enabling farmers to site-specific biophysical data), we considered access inputs and overcome credit constraints. district/woreda dummies, for example, equal to Social networks also reduce transaction costs 1 if the woreda is Guder and zero otherwise, and increase farmers’ bargaining power, helping assuming that such dummy variables can capture them to earn higher returns when marketing their unobserved biophysical properties of the sites. 1 Equib is an informal saving group. Edir is an informal group formed by members of the community, mainly for self-support. 9 Results and Discussion Descriptive Results leads to farmers’ lack of confidence to invest in the technology. The number of households that have adopted In many parts of the Ethiopian Highlands, rainwater management technologies varies across farmers have been practicing rainwater management the watershed (Table 3). Bunds/terraces followed technologies, such as bunds/terraces, to preserve by gully rehabilitation, multi-purpose trees, area the topsoil and ensure sustainable cultivation of enclosures, river diversions, orchards, shallow/ crops for their sustenance. Slope is a major factor hand-dug wells and ponds are the most adopted in determining whether bunds or terraces should technologies across the study sites. Furthermore, be constructed for soil conservation in a given Zefe, Gumera/Maksegnit and Mizewa watersheds place. Terraces are usually found on medium to (all in the Amhara region) were found to have steep slopes and can be created by moving soil the highest rates of adoption of rainwater from one place to another on the slope, which management technologies. involves a lot of work. Data in Table 3 show that Zemadim et al. (2011) indicated that there are bunds/terraces were practiced by about 70% of the successful situations of RWM programs as part of sample households. sustainable land management (SLM) to increase in- Multi-purpose trees are part of the RWM situ water availability and increase aquifer recharge strategies. Farmers adopt this technology for in the Blue Nile River Basin. On the other hand, soil and water conservation and for obtaining despite massive investments in ponds, the adoption fuelwood. Despite the promotion of multi- rate is minimal and possibly due to its low rate of purpose fodder trees for livestock feed and soil success. This is consistent with findings of a study improvement by many organizations, the number carried out by Arba Minch University (AMU 2009), of farmers practicing this technology remains low which stated that most of the 40,000 rainwater (Mekoya 2008). For instance, about 32% of the harvesting ponds were individually owned and total sample households adopted multi-purpose mainly used for supplementary irrigation. Most of trees, while about 54%, 47% and 46% of sample the ponds constructed between 2003 and 2008 households in Boke, Laku and Dapo watersheds, in the Amhara and Tigray regions of Ethiopia respectively, adopted this technology. Area have, however, failed, which was mainly due to enclosures and gully rehabilitation were adopted faulty design, wrong location of ponds and lack by about 18% of the total sample households, of monitoring after their construction. This all but these were mostly adopted by watersheds TABLE 3. Number of households that adopted rainwater management technologies in watersheds. RWM technology Watersheds Total Meja Zefe Maksegnit Boke Dapo Laku Mizewa Multi-purpose trees 22 18 16 53 52 45 15 221 Orchards 1 25 7 2 31 13 10 89 Bunds/terraces 13 78 78 67 76 70 86 468 Shallow/hand-dug wells 0 9 1 2 1 3 1 17 Ponds 0 3 2 1 0 2 5 13 River diversions 9 13 12 16 7 12 28 97 Area enclosures 1 41 43 7 1 2 19 114 Gully rehabilitation 54 74 77 36 18 36 57 352 Total 100 261 236 184 186 183 221 1,371 10 in Amhara rather than Oromia. On the other a rainwater management technology. Based on hand, about 34% of sample households in Dapo the responses from our sample households, we watershed have practiced orchards, while the observed that, except for bunds/terraces, most least number of adopters of this technology were of the other rainwater management technologies from Meja watershed (see Table 3). Rainwater were adopted on degraded lands, which was management technologies are interdependent probably because these technologies were and correlated, and hence farmers are more likely used as ex-post land rehabilitation and resource to adopt a combination of these technologies as conservation mechanisms (Figure 4). complements or substitutes. Among the sample In addition to land degradation, land-use households, for example, 55, 173, 109, 82, 48, type is also likely to influence the suitability and and 277 adopted a combination of multi-purpose decision of farm households to adopt a RWM trees and orchards, multi-purpose trees and technology. Multi-purpose trees have a higher bunds/terraces, multi-purpose trees and gully rate of adoption on croplands and grasslands rehabilitation, orchards and bunds/terraces, (Figure 5). Similarly, orchards and bunds/terraces orchards and gully rehabilitation, and bunds/ were adopted on both land-use types, although terraces and gully rehabilitation, respectively, the rate of adoption seems to favor croplands. where most of them were positively correlated Gully rehabilitation has been adopted more on (Appendix, Table A4). grasslands. Finally, the survey data indicate that The suitability of rainwater management river diversions and area enclosures are more technologies is l ikely to be inf luenced by suited to croplands and grasslands, respectively. landscape. Figure 3 shows that most households’ This result is not unexpected, because river adoption of multi-purpose trees, orchards, bunds/ diversions are used for irrigation and area terraces and area enclosures were on lands with enclosures are used for land conservation and a gentle slope, while river diversions and gully natural resource regeneration. rehabilitation were suited to lands with a flat and Table 4 presents the def ini t ion and steep slope, respectively. summary statistics of both dependent and The level of land degradation is also more independent variables used in this analysis. likely to affect a household’s decision to adopt Accordingly, about 35% of sample households FIGURE 3. Adoption of rainwater management technologies according to landscape. 11 adopted multi-purpose trees, while about 15%, adopted orchards, bunds/terraces and gully 72% and 56% of the sample households rehabilitation, respectively. FIGURE 4. The effects of land degradation on the adoption of rainwater management technologies. FIGURE 5. Adoption of rainwater management technologies according to land-use type. 12 TABLE 4. Definition and descriptive statistics of dependent and independent variables. Variable description Dependent variables Frequency Yes No Multi-purpose trees (1 = yes, 0 = no) 221 433 Orchards (1 = yes, 0 = no) 89 565 Bunds/terraces (1 = yes, 0 = no) 468 186 Gully rehabilitation (1 = yes, 0 = no) 352 302 Independent variables Mean Std. Dev. Age of household head (years) 46.996 15.342 Gender of household head (1 = male, 0 = female) 0.846 0.361 Farming experience (years) 27.121 15.705 Marital status of household head (1 = married, 0 = single/separated/divorced) 0.838 0.369 Family size in adult equivalent (number) 4.684 2.073 Household head is educated or at least can read and write (1 = yes, 0 = no) 0.200 0.400 Number of household members with elementary (1-8) education level (number) 1.979 1.618 Number of household members with high school and above (>= 9) education level (number) 0.787 1.216 At least one household member participates in off-farm activities (1 = yes, 0 = no) 0.276 0.447 At least one household member has migrated (1 = yes, 0 = no) 0.133 0.339 Total household expenditure during the previous year (ETB) 2,939 14,539 Household’s livestock holding in TLU (number) 5.234 4.612 Household’s own land (1 = yes, 0 = no) 1.002 0.723 Landholding per adult equivalent (ha) 0.428 0.399 One-way walking distance to all-weather road (minutes) 29.241 29.596 One-way walking distance to woreda center (minutes) 47.076 36.354 One-way walking distance to farmer training center (minutes) 35.408 27.626 One-way walking distance to credit center (minutes) 47.375 39.422 Household participates in Debo (1 = yes, 0 = no) 0.890 0.313 Household participates in Equib (1 = yes, 0 = no) 0.125 0.331 Household participates in Edir (1 = yes, 0 = no) 0.925 0.285 Household member participates in women’s association (1 = yes, 0 = no) 0.201 0.401 Jeldu District (woreda) (1 = yes, 0 = no) control woreda 0.180 0.385 Guder District (woreda) (1 = yes, 0 = no) 0.320 0.467 Horo (Shambu) District (woreda) (1 = yes, 0 = no) 0.314 0.465 Diga District (woreda) (1 = yes, 0 = no) 0.314 0.465 Farta District (woreda) (1 = yes, 0 = no) 0.314 0.465 Gondar Zuria District (woreda) (1 = yes, 0 = no) 0.310 0.463 Fogera District (woreda) (1 = yes, 0 = no) 0.328 0.470 Notes: Debo – a traditional labor sharing system; Std. Dev. = Standard deviation. 13 These results are based on the responses of invest in and adopt the technologies. In line with this, farm households that have already adopted some Figure 6 presents the perceptions of non-adopters of the technologies, and hence do not capture the and highlight the constraints that impede farmers’ perceptions of non-adopters and their limitations to investment in rainwater management technologies. FIGURE 6. Number of non-adopters who consider factors as constraints for not adopting rainwater management technologies. (a) Shortage of land. (b) Lack of labor. (c) Lack of cooperation with neighborhood. (d) Technology is not suitable on farmer's land. (e) Lack of capital/credit. (f) Lack of proper technical advise. (Continued) 14 FIGURE 6. Number of non-adopters who consider factors as constraints for not adopting rainwater management technologies (Continued). (g) It is labor- and time-intensive (tiresome). (h) Lack of awareness about the technology. (i) Limited access to markets. Regression Results the rainwater management technology adoption equations are not independent of each other, and Parameter estimates from the multivariate probit hence a multivariate probit approach is appropriate model are presented in Table 5. The regression in this case. Similarly, the likelihood ratio test results revealed that the determinants of [c 2 (6) = 36.324 and probability > c 2 = 0.000] adoption of rainwater management technologies indicates a significant joint correlation between the can be broadly classified into household and technologies, and supports the estimation of the socioeconomic characteristics, access to multivariate probit model as opposed to a separate markets and services, social capital and district- univariate probit model. specific characteristics. Furthermore, the positive and significant Coefficients that capture correlation between correlation coefficients of the error terms indicate the technologies are presented in Table 6. These that there is complementarity (positive correlation) essentially indicate pair-wise correlations between between di f ferent rainwater management the error terms in the system of equations of the technologies being used by farmers, and supports multivariate probit model. Results show that, with the assumption of interdependence between the exception of orchards and gully rehabilitation, the different rainwater management technology all the other rainwater management technologies options. Differences in the estimated coefficients are positively and significantly correlated, which across equations also support the appropriateness supports the hypothesis. The error terms in of differentiating between technology options. 15 16 TABLE 5. Results of the multivariate probit model. Independent variables Technologies (dependent variables) Multi-purpose trees Orchards Bunds/terraces Gully rehabilitation Coefficient Coefficient Coefficient Coefficient Human capital Age of household head (years) -0.011*** (-0.004) -0.018*** (-0.006) -0.013*** (-0.005) -0.0122*** (-0.004) Gender of household head (1 = male) 0.760*** (-0.221) 0.379 (-0.312) 0.159 (-0.233) 0.056 (-0.214) Marital status of household head (1 = married) -0.555*** (-0.202) -0.242 (-0.291) 0.202 (-0.214) -0.0725 (-0.211) Family size in adult equivalent 0.147*** (-0.043) 0.213*** (-0.054) 0.024 (-0.052) 0.040 (-0.044) Household head is educated (1 = yes) 0.175 (-0.154) -0.016 (-0.185) 0.122 (-0.169) 0.0306 (-0.150) Number of household members with elementary (1-8) education level -0.032 (-0.046) -0.001 (-0.059) 0.035 (-0.054) 0.036 (-0.045) Number of household members with high school and above (>= 9) education level -0.018 (-0.058) -0.014 (-0.075) -0.080 (-0.066) -0.012 (-0.058) Total household expenditure during the previous year 0.001 (0.001) 0.001 (0.001) 0.001 (-0.001) 0.001 (0.001) Physical capital Livestock holding in TLU 0.015 (-0.015) 0.031 (-0.018) 0.050*** (-0.018) 0.003 (-0.015) Landholding per adult equivalent 0.363** (-0.167) 0.559*** (-0.196) -0.395** (-0.174) -0.145 (-0.169) Household’s own land (1 = yes, 0 = no) 0.179 (-0.144) 0.120** (-0.054) 0.516 (-0.348) -0.313 (-0.409) Access to markets and services One-way walking distance to all-weather road (minutes) 0.002 (-0.002) -0.006** (-0.003) -0.001 (-0.002) -0.004* (-0.002) One-way walking distance to woreda center (minutes) 0.001 (-0.002) -0.006*** (-0.002) 0.003 (-0.002) -0.012 (-0.002) One-way walking distance to farmer training center (minutes) -0.001 (-0.002) 0.001 (-0.003) 0.001 (-0.002) -0.001 (-0.002) One-way walking distance to credit center (minutes) -0.003 (-0.002) 0.003* (-0.002) 0.001 (-0.002) 0.002 (-0.002) Social capital Participation in off-farm activities (1 = yes) 0.172 (-0.135) -0.232 (-0.181) -0.070 (-0.153) -0.281** (-0.136) At least one household member migrates (1 = yes) -0.267 (-0.188) -0.541** (-0.272) -0.436** (-0.209) 0.015 (-0.187) Household participates in Debo (1 = yes) -0.112 (-0.197) 0.411 (-0.304) 0.521*** (-0.194) 0.371* (-0.204) Household participates in Equib (1 = yes) -0.193 (-0.169) 0.065 (-0.230) -0.262 (-0.188) 0.170 (-0.171) Household participates in Edir (1 = yes) 0.253 (-0.275) 0.987*** (-0.310) 0.491** (-0.246) 0.057 (-0.223) Household participates in women’s associations (1 = yes) 0.531*** (-0.140) 0.408** (-0.186) -0.030 (-0.165) 0.038 (-0.141) (Continued) 17 TABLE 5. Results of the multivariate probit model (Continued). Independent variables Technologies (dependent variables) Multi-purpose trees Orchards Bunds/terraces Gully rehabilitation Coefficient Coefficient Coefficient Coefficient District (woreda) dummies Woreda is Guder (1 = yes, 0 = no) 0.243 (-0.153) -2.003*** (-0.378) -0.928*** (-0.168) -0.608*** (-0.150) Woreda is Horo (Shambu) (1 = yes, 0 = no) 0.282* (-0.161) -0.404* (-0.224) -0.892*** (-0.185) -0.369** (-0.158) Woreda is Diga (1 = yes, 0 = no) 0.411*** (-0.145) 0.505*** (-0.196) -0.624*** (-0.185) -1.044*** (-0.156) Woreda is Farta (1 = yes, 0 = no) -0.211 (-0.179) 0.808*** (-0.248) -0.133 (-0.215) 0.795*** (-0.172) Woreda is Gondar Zuria (1 = yes, 0 = no) -0.454*** (-0.173) -0.177 (-0.236) -0.037 (-0.209) 1.053*** (-0.188) Woreda is Fogera (1 = yes, 0 = no) -0.690*** (-0.161) -0.093 (-0.230) -0.327** (-0.163) 0.063 (-0.140) Omitted (control) woreda is Jeldu - - - - Constant -1.226*** (-0.441) -2.952*** (-0.566) 0.207 (-0.479) 0.632 (-0.526) Regression diagnostics Number of observations 654 LR test of rho=0: c 2 (6) 36.324*** Wald (c 2 ) 773.730 Log pseudolikelihood -1108.127 Prob > c 2 0.000*** Notes: *, ** and *** indicates levels of significance at 10%, 5% and 1%, respectively. Figures within parenthesis are robust standard errors. Social capital captured in the form of While this is in agreement with the findings household membership and participation in of Adesina et al. (2000), Kassie et al. (2009) informal and formal community networks is reported that female-headed households are more unobservable when used as a proxy of social likely to adopt sustainable agricultural technologies capital. The age of the household head was in Tanzania. Although the impact of gender on found to be negatively and significantly correlated technology adoption is likely to be technology- with adoption, and indicates that older farmers specific and generalization is not possible (Kassie are less likely to adopt rainwater management et al. 2009), our results indicate that male-headed technologies than younger farmers. This may be households have a comparative advantage in the because young farmers are more able to provide adoption of rainwater management technologies in the labor required to implement the technologies, the Blue Nile River Basin (Figure 7). and/or older farmers may have shorter planning Most of the agricultural work is typically horizons and are more risk-averse. undertaken by men, while women are usually The results also disclose that male-headed restricted to household and backyard activities. households are more likely to adopt multi-purpose This suggests that men are more likely to have trees compared to female-headed households. better farming experience. Farming experience TABLE 6. Relationships between RWM technologies from the multivariate probit model regression equations (robust standard error is shown within parentheses). Rainwater management Multi-purpose trees Orchards Bunds/terraces technology Orchards r21 0.448***(0.080) Bunds/terraces r31 r32 0.154*(0.084) 0.232***(0.085) Gully rehabilitation r41 r42 r43 0.127*(0.069) - 0.006(0.076) 0.156**(0.071) Note: *, ** and *** indicates level of significance at 10%, 5% and 1%, respectively. FIGURE 7. Gender-disaggregated data on the adoption of rainwater management technologies. 18 usually increases the probability of technology bunds/terraces and gully rehabilitation, which adoption, because experienced farmers are are commonly collective action activities usually more likely to have better access to information carried out on a FFW basis, those who own more and knowledge of climatic conditions and coping land are more likely to defect collective action mechanisms. In this context, the policy implication as they may not expect to benefit from FFW is that targeting women groups to address their payments. In this respect, the policy implication constraints to actively participate in rural economic is that tenure arrangement and security is likely activities can have a significant impact on the to facilitate long-term investments in rainwater adoption of rainwater management technologies. management technologies. Furthermore, farmers with better experience and Although some of the results are statistically information are most likely to take initiatives in insignificant, a longer distance to farmer training adopting and testing new technologies. Targeting and credit centers (captured by the walking time of such progressive and model farmers during the to the nearest center) were found to negatively promotion of technologies can, therefore, have a affect the adoption of rainwater management significant positive effect. technologies. This implies that farmers who As expected, family size in adult equivalent have better access to these services are has a positive and significant effect on the better informed about the role of rainwater adoption of multi-purpose trees and orchards. management technologies. Also, improved This may imply that these technologies are access to markets, extension services and credit labor-intensive and, therefore, households centers have the potential to increase farmers’ who have more labor are more likely to adopt adoption of rainwater management technologies. them. Participation in off-farm activities has a Furthermore, access to credit centers and significant negative effect on the adoption of markets improves options to address liquidity gully rehabilitation, while migration is negatively constraints associated with investments in related with the adoption of orchards and bunds/ rainwater management technologies. terraces. The implication is that both off-farm Socia l cap i ta l captured by household activities and migration are likely to compete membership in social networks (group membership) for labor, which in turn could have been used was defined as binary (equal to 1 if the household to finance investment in rainwater management is a member, and zero otherwise). The regression technologies. Since labor is a serious constraint, results suggested that social capital positively RWM technologies that require less labor or affects a household’s decision to adopt rainwater increasing labor efficiency may help to foster the management technologies. For example, a adoption of these technologies. household’s membership in Debo (a traditional Ownership of livestock has a significant labor sharing system) has a significant positive positive impact on the adoption of orchards effect on the probability of adoption of bunds/ and bunds/terraces, implying that household terraces and gully rehabilitation. Similarly, a wealth positively affects their decision to adopt household’s participation in Edir (a traditional a technology. On the other hand, ownership of peer support system) has a positive relationship land is positively correlated with the adoption of with the adoption of orchards and bunds/terraces. multi-purpose trees and orchards, but negatively Membership in a women’s association also has correlated with the adoption of bunds/terraces. a positive and significant effect on the adoption This is likely because a household that owns a of multi-purpose trees and orchards. Women’s large farm is less resource-constrained and has associations commonly play the role of facilitating better options to diversify its income, which in access to affordable (low interest rate) credit and turn may negatively affect willingness/incentive to technologies to their members. In general, the invest in bunds/terraces. results suggest that social networks (both formal Since multi-purpose trees and orchards and informal) help members to use their peer are typically private investments as opposed to support to overcome labor and/or credit constraints. 19 The fixed effects of woredas were included to and Gondar Zuria woredas, is also higher than capture unobserved site-specific factors. Results in Jeldu. Finally, the results indicate that the show that farm households in the Guder woreda adoption of multi-purpose trees in Gondar Zuria are less likely to adopt orchards, bunds/terraces and Fogera, and bunds/terraces in Fogera, is and gully rehabilitation than in Jeldu. Also, farm less likely when compared to Jeldu. In general, households in Horo (Shambu) and Diga are more the results suggest that it might be important to likely to adopt multi-purpose trees, but less likely examine the socioeconomic and demographic to adopt bunds/terraces and gully rehabilitation characteristics of households, and biophysical than those in Jeldu (the control woreda). The suitability of watersheds, instead of promoting probability of adoption of orchards in Diga and blanket recommendations for the adoption of Farta woredas, and gully rehabilitation in Farta rainwater management technologies. Conclusions The factors that influence the adoption of In general, the results suggest that it might RWM technologies in the Blue Nile River Basin be important to examine the socioeconomic and have been presented, in order to improve the demographic characteristics of households, and understanding of why farmers do not adopt biophysical suitability of watersheds, instead these technologies despite their suitability and of promoting blanket recommendations for the potential benefits. The results indicate a joint adoption of rainwater management technologies. correlation (interdependent) between RWM The regression results, together with insights technologies, implying that the adoption decision gained from qualitative analysis, suggest that the of a specific technology is correlated with the most appropriate target groups for adoption of adoption of another technology. This supports rainwater management technologies are those farm the assertion that it is important to consider households with (a) limited landholdings; (b) limited packages of technologies. access to markets, information and extension The main variables influencing the adoption services; (c) bigger family size in adult equivalent, of rainwater management technologies in the as an indication of labor endowment and the ability Blue Nile River Basin include (i) demographic and to engage in labor-intensive activities; (d) capital family size of farm household (i.e., age, gender constraints and limited access to credit; (e) limited and marital status); (ii) education status; (iii) livestock and asset ownership; and (f) constraints participation in off-farm activities; (iv) ownership faced by women to actively participate in rural of livestock; (v) ownership of land; (vi) access economic activities, and by addressing these to markets, extension services and credit constraints. Correlation coefficients indicate that centers; (vii) social capital captured in the form the adoption of RWM technologies are correlated, of household membership, and participation in implying interdependence between different informal and formal community networks; and (viii) technologies. Thus, the adoption and promotion of site-specific factors captured by the fixed effects rainwater management technologies should follow of each location in the form of woreda dummies. a holistic approach. 20 References Adesina, A.A.; Mbila, D.; Nkamleu, G.B.; Endamana, D. 2000. Econometric analysis of the determinants of adoption of alley farming by farmers in the forest zone of Southwest Cameroon. Agriculture, Ecosystems & Environment 80: 255-265. Amede, T. 2012. Rainwater management for Sustainable Agricultural Intensification in the Ethiopian Highlands. Presentation from the 2012 World Water Week in Stockholm. Available at http://www.worldwaterweek.org/ documents/WWW_PDF/2012/Tue/Rainfed-production-under-growing/Tilahun-Amede.pdf (accessed on August 13, 2013). Amsalu, A.; de Graaff, J. 2007. Determinants of adoption and continued use of stone terraces for soil and water conservation in an Ethiopian highland watershed. Ecological Economics 61(2-3): 294-302. AMU (Arba Minch University). 2009. Inventory of water storage types, their distribution and characteristics in the Abbay River basin. Unpublished report prepared for the International Water Management Institute (IWMI) by Arba Minch University, Ethiopia. Awulachew, S.B.; Demissie, S.S.; Hagos, F.; Erkossa, T.; Peden, D. 2012. Water management intervention analysis in the Nile Basin. In: The Nile River Basin: Water, agriculture, governance and livelihoods, eds., Awulachew, S.B.; Smakhtin, V.; Molden, D.; Peden, D. Oxon, UK: Routledge. Pp. 292-311. Awulachew, S.B.; Rebelo, L-M.; Molden, D. 2010. The Nile Basin: Tapping the unmet agricultural potential of Nile waters. Water International 35(5): 623-654. Belderbos, R.; Carree, M.; Diederen, B.; Lokshin, B.; Veugelers, R. 2004. Heterogeneity in R&D cooperation strategies. International Journal of Industrial Organization (22): 1237-1263. Cappellari, L.; Jenkins, S.P. 2003. Multivariate probit regression using simulated maximum likelihood. The Stata Journal 3(3): 278-294. CSA (Central Statistical Agency). 2004. The Federal Democratic Republic of Ethiopia statistical abstract for 2003. Addis Ababa, Ethiopia: CSA. De Groote, H.; Coulibaly, N. 1998. Gender and generation: An intra-household analysis on access to resources in southern Mali. African Crop Science Journal 6(1): 79-95. Deressa, T. 2007. Measuring the economic impact of climate change on Ethiopian agriculture: Ricardian approach. World Bank Policy Research Paper No. 4342. Washington, DC: World Bank. Deressa, 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 Environmental Change 19: 248-255. Fafchamps, M.; Gubert, F. 2007. The formation of risk sharing networks. Journal of Development Economics (83): 326-350. FAO (Food and Agriculture Organization of the United Nations). 2005. Irrigation in Africa in Figures: AQUASTAT survey - 2005. FAO Water Report No. 29. Rome: FAO. ftp://ftp.fao.org/agl/aglw/docs/wr29_eng.pdf (accessed on August 13, 2012). Guerin, T. 1999. An Australian perspective on the constraints to the transfer and adoption of innovations in land management. Environmental Conservation 24(4): 289-304. Haileslassie, A.; Priess, J.; Veldkamp, E.; Teketay, D.; Lesschen, J.P. 2005. Assessment of soil nutrient depletion and its spatial variability on smallholders’ mixed farming systems in Ethiopia using partial versus full nutrient balances. Agriculture, Ecosystems and Environment 108: 1-16. Hurni, H. 1990. Degradation and conservation of soil resources in the Ethiopian Highlands. In: African mountains and highlands: Problems and perspectives, eds., Messerli, B.; Hurni, H. Produced for the African Mountains Association (AMA). Marceline, Mo., USA: Walsworth Press. Pp. 51-63. 21 ILRI (International Livestock Research Institute); IWMI (International Water Management Institute). 2010. Rainwater management in the Ethiopian highlands: Mapping, targeting and scaling out interventions. Nairobi, Kenya: International Livestock Research Institute (ILRI). Available at http://mahider.ilri.org/handle/10568/2428 (accessed on August 18, 2012). Kassie, M.; Jaleta, M.; Shiferaw, B.; Mmbando, F.; Mekuria, M. 2012. Interdependence in farmer technology adoption decisions in smallholder systems: Joint estimation of investments in sustainable agricultural practices in rural Tanzania. Paper prepared and selected for presentation at the International Association of Agricultural Economists (IAAE) Triennial Conference, Foz do Iguaçu, Brazil, August 18-24, 2012. Kassie, M.; Pender, J.; Yesuf, M.; Kohlin, G.; Bluffstone, R.; Mulugeta, E. 2008. Estimating returns to soil conservation adoption in the northern Ethiopian highlands. Agricultural Economics 38(2): 213-232. Kassie, M.; Zikhali, P.; Manjur, K.; Edwards, S. 2009. Adoption of organic farming technologies: Evidence from semi-arid regions of Ethiopia. Natural Resources Forum 33: 189-198. Kassie, M.; Zikhali, P.; Pender, J.; Kohlin, G. 2010. The economics of sustainable land management practices in the Ethiopian Highlands. Journal of Agricultural Economics 61(3): 605-627. Kato, E.; Ringler, C.; Yesuf, M.; Bryan, E. 2009. Soil and water conservation technologies: A buffer against production risk in the face of climate change. Insights from the Nile Basin in Ethiopia. IFPRI Discussion Paper 00871. Washington, DC: International Food Policy Research Institute (IFPRI). Kurukulasuriya, P.; Rosenthal, S. 2003. Climate change and agriculture: A review of impacts and adaptations. Climate change series Paper no. 91. Paper prepared and published for the Rural Development Group and Environment Department of the World Bank. 106p. Lautze, S.; Aklilu, Y.; Raven-Roberts, A.; Young, H.; Kebede, G.; Learning, J. 2003. Risk and vulnerability in Ethiopia: Learning from the past, responding to the present, preparing for the future. Report for the United States Agency for International Development (USAID), Addis Ababa, Ethiopia. Lee, D.R. 2005. Agricultural sustainability and technology adoption: Issues and policies for developing countries. American Journal of Agricultural Economics 87(5): 1325-1334. Marenya, P.P.; Barrett, C.B. 2007. Household-level determinants of adoption of improved natural resources management practices among smallholder farmers in western Kenya. Food Policy 32: 515-536. McDonald, K.; Brown, K. 2000. Soil and water conservation projects and rural livelihoods: Options for design and research to enhance adoption and adaptation. Land Degradation and Development 11: 343-361. Mekoya, A.K. 2008. Multi-purpose fodder trees in Ethiopia: Farmers’ perception, constraints to adoption and effects of long-term supplementation on sheep performance. PhD thesis. The Netherlands: Wageningen University. Merrey, D.J.; Gebreselassie, T. 2011. Promoting improved rainwater and land management in the Blue Nile (Abay) basin of Ethiopia. NBDC Technical Report 1. Nairobi, Kenya: International Livestock Research Institute (ILRI). MoFED (Ministry of Finance and Economic Development). 2010. Annual report on macroeconomic developments. Available at http://www.mofed.gov.et/English/Resources/Documents/2002MEDRAnnual.pdf (accessed on April 8, 2013). MoWE (Ministry of Water and Energy). 1998a. Abbay River Basin integrated development master plan project (1998), phase 2 report, section II, volume VIII, Addis Ababa, Ethiopia. MoWE. 1998b. Abbay River Basin integrated development master plan project (1998), phase 2 report, section II, volume X, Addis Ababa, Ethiopia. Moyo, S.; Veeman, M. 2004. Analysis of joint and endogenous technology choice for protein supplementation by smallholder dairy farmers in Zimbabwe. Agroforestry Systems 60: 199-209. Ndiritu, S.W.; Kassie, M.; Shiferaw, B.; Ouma, J.; Odendo, M. 2011. Adoption of agricultural technologies in Kenya: How does gender matter? Nairobi, Kenya: International Maize and Wheat Improvement Center (CIMMYT). 22 Nhemachena, C.; Hassan, R. 2007. Micro-level analysis of farmers’ adaptation to climate change in Southern Africa. IFPRI Discussion Paper No. 00714. Washington, DC: International Food Policy Research Institute (IFPRI). NMA (National Meteorology Agency). 2007. National Adaptation Program of Action (NAPA) on climate change of Ethiopia. Bonn, Germany: United Nations Framework Convention on Climate Change (UNFCCC). Available at http://unfccc.int/resource/docs/napa/cpv01.pdf (accessed on April 17, 2013). Pender, J.; Gebremedhin, B. 2007. Determinants of agricultural and land management practices and impacts on crop production and household income in the highlands of Tigray, Ethiopia. Journal of African Economics 17: 395-450. Quisumbing, A.R.; Brown, L.; Hillary, R.; Feldsten, S.; Haddad, L.; Pena, C. 1995. Women: The key to food security. Food Policy Statement No. 21. Washington, DC: International Food Policy Research Institute (IFPRI). Rockström, J.; Karlberg, L.; Wani, S.P.; Barron, J.; Hatibu, N.; Oweis, T.; Bruggeman, A.; Farahani, J.; Qiang, Z. 2010. Managing water in rainfed agriculture—The need for a paradigm shift. Agricultural Water Management 97: 543-550. Santini, G.; Peiser, L.; Faurès, J.M.; Neves, B.; Vallée, D. 2011. Planning smart investments in agricultural water management through a livelihood mapping approach: The case of Ethiopia (Draft). Shiferaw, A. 2011. Estimating soil loss rates for soil conservation planning in the Borena Woreda of South Wollo Highlands, Ethiopia. Journal of Sustainable Development in Africa 13(3): 87-106. Clarion, Pennsylvania: Clarion University of Pennsylvania. Soule, J.M.; Tegene, A.; Wiebe, D.K. 2000. Land tenure and the adoption of soil conservation practices. American Journal of Agricultural Economics 82(4): 993-1005. Spinoni, J.; Micale, F.; Carrao, H.; Naumann, G.; Barbosa, P.; Vogt, J. 2013. Global and continental changes of arid areas using the FAO Aridity Index over the periods 1951-1980 and 1981-2010. Geophysical Research Abstracts Vol. 15. EGU General Assembly, April 7-12, 2013, Vienna, Austria, EGU2013-9262-1. Tamene, L.; Vlek, P.L.G. 2008. Soil erosion studies in Northern Ethiopia. Land Use and Soil Resources 73-100. Wollni, M.; Lee, D.R.; Thies, J.E. 2010. Conservation agriculture, organic marketing, and collective action in the Honduran hillsides. Agricultural Economics 41: 373-384. Yu, L.; Hurley, T.; Kliebenstein, J.; Orazen, P. 2008. Testing for complementarity and substitutability among multiple technologies: The case of U.S. hog farms. Working Paper No. 08026. Ames, IA, USA: Iowa State University, Department of Economics. Zemadim, B.; McCartney, M.; Sharma, B.; Wale, A. 2011. Integrated rainwater management strategies in the Blue Nile Basin of the Ethiopian highlands. International Journal of Water Resources and Environmental Engineering 3(10): 220-232. Zomer, R.J.; Bossio, D.A.; Trabucco, A.; Yuanjie, L.; Gupta, D.C.; Singh, V.P. 2007. Trees and water: Smallholder agroforestry on irrigated lands in Northern India. Colombo, Sri Lanka: International Water Management Institute (IWMI). 45p. (IWMI Research Report 122). Zomer, R.J.; Trabucco, A.; Bossio, D.A.; van Straaten, O.; Verchot, L.V. 2008. Climate change mitigation: A spatial analysis of global land suitability for clean development mechanism afforestation and reforestation. Agriculture, Ecosystems and Environment 126: 67-80. 23 Appendix TABLE A1. Adoption rates of rainwater management technologies disaggregated by gender. Rainwater Male Female Total T-test management (significance of technology Number Mean Number Mean differe nce) Multi-purpose trees 202 0.380 (0.486) 19 0.204 (0.405) 221 0.000*** Orchards 82 0.157 (0.364) 7 0.087 (0.284) 89 0.033** Bunds/terraces 408 0.745 (0.436) 60 0.602 (0.492) 468 0.003*** Gully rehabilitation 298 0.551 (0.498) 54 0.544 (0.501) 352 0.938 Total 990 140 1,130 Notes: *, ** and *** indicates level of significance at 10%, 5% and 1%, respectively. Figures in parenthesis are standard deviations. TABLE A2. Adoption of rainwater management technologies and ownership of land, slope, degradation and land use. Rainwater Ownership Slope Degradation Land use management technology Public Private Flat Gentle Steep Degraded Less Grassland Cropland land land slop slop degraded Multi-purpose trees 36 201 92 102 41 143 92 104 131 Orchards 1 98 41 55 52 44 31 65 Bunds/terraces 45 322 31 281 165 185 292 104 373 (without vegetation) Bunds/terraces 21 44 8 53 29 42 48 35 55 (with vegetation) Shallow/hand-dug wells 33 Ponds 1 22 River diversions 85 27 69 41 69 41 110 Area enclosures 101 29 24 62 35 78 43 121 Gully rehabilitation 18 278 19 103 175 205 92 204 93 24 25 TABLE A3. Reasons for not adopting rainwater management technologies: Frequency of farmers’ response. Multi-purpose Orchards Bunds/terraces Bunds/terraces Shallow Ponds River Area Gully trees (without vegetation) (with vegetation) wells diversions enclosures rehabilitation No problem of land degradation 19 11 23 9 10 9 13 13 18 Availability of sufficient rainfall 12 10 15 5 17 13 15 8 10 Have enough land 17 18 18 7 15 14 13 17 14 Shortage of land 56 44 51 20 34 34 36 49 32 Lack of labor 37 31 46 17 44 42 31 25 43 It is labor- and time-intensive 17 16 31 12 22 36 24 19 39 (tiresome) Availability of sufficient surface 14 13 11 5 28 18 16 8 9 water/groundwater Lack of awareness about the 54 40 36 12 35 34 19 31 24 technology Technology is not suitable on 32 47 28 10 42 41 50 32 32 my land Lack of capital/credit 43 38 44 16 13 37 27 25 27 Not profitable to invest 12 15 17 5 23 15 9 14 14 Lack of proper technical advise 40 33 32 11 10 24 17 24 23 I have better options 6 7 8 3 13 8 7 7 7 Limited access to markets 19 14 15 5 24 11 8 9 11 Lack of cooperation with 29 24 33 12 NA 25 25 28 31 neighborhood TABLE A4. Number of sample households that adopted a combination of RWM technologies and level of correlation between these technologies. Combination of RWM technologies Number of adopting Correlation coefficient and households significance level Multi-purpose trees and orchards 55 0.235*** Multi-purpose trees and bunds/terraces 173 0.106** Multi-purpose trees and gully rehabilitation 109 0.065 Orchards and bunds/terraces 82 0.181*** Orchards and gully rehabilitation 48 0.001 Bunds/terraces and gully rehabilitation 277 0.171*** Multi-purpose trees, orchards and bunds/terraces 49 NA Multi-purpose trees, orchards and gully rehabilitation 22 NA Multi-purpose trees/gully rehabilitation and bunds/terraces 84 NA Orchards, gully rehabilitation and bunds/terraces 44 NA Multi-purpose trees, orchards, bunds/terraces and gully rehabilitation 18 NA Notes: ** and *** indicates level of significance at 5% and 1%, respectively. NA = not applicable. 26 IWMI Research Reports 154 Determinants of Adoption of Rainwater Management Technologies among Farm Households in the Nile River Basin. Gebrehaweria Gebregziabher, Lisa-Maria Rebelo, An Notenbaert, Kebebe Ergano and Yenenesh Abebe. 2013. 153 Facilitating Outcomes: Multi-stakeholder Processes for Influencing Policy Change on Urban Agriculture in Selected West African and South Asian Cities. Priyanie Amerasinghe, Olufunke O. Cofie, Theophilus O. Larbi and Pay Drechsel. 2013. 152 Agricultural Water Storage in an Era of Climate Change: Assessing Need and Effectiveness in Africa. Matthew McCartney, Lisa-Maria Rebelo, Stefanos Xenarios and Vladimir Smakhtin. 2013. 151 Managed Aquifer Recharge: The Solution for Water Shortages in the Fergana Valley. Akmal Karimov, Vladimir Smakhtin, Aslon Mavlonov, Vecheslav Borisov, Inna Gracheva, Fazleddin Miryusupov, Jamol Djumanov, Tatyana Khamzina, Rustam Ibragimov and Botir Abdurahmanov. 2013. (Also available in Russian). 150 Glacier Systems and Seasonal Snow Cover in Six Major Asian River Basins: Hydrological Role under Changing Climate. Oxana S. Savoskul and Vladimir Smakhtin. 2013. 149 Glacier Systems and Seasonal Snow Cover in Six Major Asian River Basins: Water Storage Properties under Changing Climate. Oxana S. Savoskul and Vladimir Smakhtin. 2013. 148 Evaluating the Flow Regulating Functions of Natural Ecosystems in the Zambezi River Basin. Matthew McCartney, Xueliang Cai and Vladimir Smakhtin. 2013. 147 Urban Wastewater and Agricultural Reuse Challenges in India. Priyanie Amerasinghe, Rajendra Mohan Bhardwaj, Christopher Scott, Kiran Jella and Fiona Marshall. 2013. 146 The Water Resource Implications of Changing Climate in the Volta River Basin. Matthew McCartney, Gerald Forkuor, Aditya Sood, Barnabas Amisigo, Fred Hattermann and Lal Muthuwatta. 2012. 145 Water Productivity in Context: The Experiences of Taiwan and the Philippines over the Past Half-century. Randolph Barker and Gilbert Levine. 2012. 144 Revisiting Dominant Notions: A Review of Costs, Performance and Institutions of Small Reservoirs in Sub-Saharan Africa. Jean-Philippe Venot, Charlotte de Fraiture and Ernest Nti Acheampong. 2012. Electronic copies of IWMI's publications are available for free. Visit www.iwmi.org/publications/index.aspx Postal Address P O Box 2075 Colombo Sri Lanka Location 127 Sunil Mawatha Pelawatta Battaramulla Sri Lanka Telephone +94-11-2880000 Fax +94-11-2786854 E-mail iwmi@cgiar.org Website www.iwmi.org ISSN: 1026-0862 ISBN: 978-92-9090-781-7