You are on page 1of 18

This article was downloaded by: [Anandaraj K] On: 10 February 2014, At: 05:18 Publisher: Taylor & Francis

Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Agricultural Sustainability


Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tags20

Roles of extension and ethno-religious networks in acceptance of resourceconserving agriculture among Ethiopian farmers
Petr Matou , Yasuyuki Todo & Dagne Mojo
a a b c

School of Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 110-0001, Japan
b

Department of International Studies, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8563, Japan
c

Ethiopian Institute for Agricultural Research and Graduate Program in Sustainability Science, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8563, Japan Published online: 10 Dec 2012.

To cite this article: Petr Matou, Yasuyuki Todo & Dagne Mojo (2013) Roles of extension and ethno-religious networks in acceptance of resource-conserving agriculture among Ethiopian farmers, International Journal of Agricultural Sustainability, 11:4, 301-316, DOI: 10.1080/14735903.2012.751701 To link to this article: http://dx.doi.org/10.1080/14735903.2012.751701

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or

distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Downloaded by [Anandaraj K] at 05:18 10 February 2014

International Journal of Agricultural Sustainability, 2013 Vol. 11, No. 4, 301 316, http://dx.doi.org/10.1080/14735903.2012.751701

Roles of extension and ethno-religious networks in acceptance of resourceconserving agriculture among Ethiopian farmers
a , Yasuyuki Todob and Dagne Mojoc Petr Matous
School of Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 110-0001, Japan; Department of International Studies, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8563, Japan; cEthiopian Institute for Agricultural Research and Graduate Program in Sustainability Science, University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8563, Japan
b a

Downloaded by [Anandaraj K] at 05:18 10 February 2014

This article analyses roles of social and extension networks in adoption of resource-conserving practices among Ethiopian farmers. We gathered data from 297 randomly sampled households on their agricultural practices, social networks, access to the extension, and geographical location. After examining general determinants of practising resource-conserving agriculture, we employ a two-stage regression with full-maximum likelihood correction for selection bias to establish the roles of general social networks and external professionals in acceptance of conservation techniques. In accordance with previous research, probit regression in the rst stage shows that the access to extension increases with farmers wealth and the size of their personal networks, and decreases with the distance of their households from village centres. However, after accounting for this unequal access to extension, the second-stage linear regression shows that regardless of education, wealth or geographical location, those whose religion and ethnicity match with their agent, report learning more about conservation from extension sources. Furthermore, farmers who are socially well connected within the community tend to be less receptive to agents recommendations regarding resource conservation. Dissemination policy of conservation agriculture should consider the ethnic and religious afnity between farmers and their extension agents. It also needs to pay more attention to socially and geographically isolated individuals. Keywords: resource-conserving agriculture; extension systems; social networks; technology adoption; Ethiopia

1.

Introduction

In view of natural resource depletion, application of resource-conserving practices is becoming increasingly important for sustainability of agricultural systems, particularly in most environmentally and economically constrained agriculture-dependent regions such as sub-Saharan Africa (Yila and Thapa 2008, Isaac and Kimaro 2011, Pretty et al. 2011, Beddington et al. 2012). Conservation agriculture, for example, promotes soil health and contributes to affordable and sustainable intensication of production but in most countries it is still a little-known concept (Kassam et al. 2009). This is partly because it is a knowledge-intensive system with counterintuitive elements that cannot be easily discovered by farmers without institutional support (Kassam et al. 2009).

Corresponding author. Email: petr@civil.t.u-tokyo.ac.jp

# 2013 Taylor & Francis

302

P. Matous et al.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Traditionally, expectations were upon extension agents to spread effective practices by informing farmers about their benets but researchers of extension have eventually recognized the limits of this top-down linear model and their attention has shifted more towards the farmers learning and decision-making processes (Douthwaite et al. 2001, Leeuwis 2004, Warner 2007, Spielman et al. 2009). It is understood now that farmers do not simply accept facts from experts, nor do they rationally choose technologies in social isolation. They are inuenced by other farmers, the nature of their relationship to the intervening intermediaries (Rogers 2003, Solano et al. 2003, Schneider et al. 2009). Informal networks, social capital, social learning, and social memory can play even a larger role in protection of natural resources than formal institutions (Ostrom 1990, Folke 2006, Janssen et al. 2006, Bodin and Crona 2009, Pretty et al. 2011). Consequently, scholars have started to call for application of social network analytical concepts in studies of agricultural innovation systems and social learning (Leeuwis 2004, Hoang et al. 2006, Isaac et al. 2007, Warner 2007, Spielman et al. 2008, Conley and Udry 2010, Bartholomay et al. 2011, Spielman et al. 2011). However, despite the rise of interest in networks among farmers, recent studies reconrm that ties to external extension experts, and training by public institutions, do play an important role in smallholders access to information and uptake of resource-conserving practices (Kursat 2010, Tefera and Sterk 2010, Pretty et al. 2011, Yorobe Jr et al. 2011, Isaac 2012). Warner (2007) argues that while the technology transfer model has become obsolete, an emerging model of extension which promotes sustainable practices based on ecological science is indispensable. Undoubtedly, it is meaningful to consider the inuence of both peers and experts (Cramb and Culasero 2003, Millar and Connell 2009, Klerkx et al. 2010, Prell et al. 2010). A recommendable step forward is to start employing network methods which enable rigorous verication of the interplay between these two categories of relationships. Using original data of Ethiopian agricultural farmers social networks, personal characteristics, and geographical locations of their households, the aim of this article is rst to examine which individuals in which network positions tend to practice resource-conserving agriculture. Next, we analyse the relative perceived importance of connections with extension agents versus other social contacts. We nd signicant effects of intra-religious and intra-ethnic afnity on formation of relationships between farmers and agents as well as on the outcomes that the farmers attribute to extension in terms of knowledge and adoption of resource-conserving practices. 2. Extension and networks: analytical framework Opinions about the role of extension are diverse and keep evolving (see, for example, Leeuwis 2004, Ison and Russell 2007, Warner 2007, Klerkx and Leeuwis 2008, Klerkx and Leeuwis 2009, Rivera and Sulaiman 2009). Our informants from the Ethiopian extension understood their role to be mainly to communicate ndings from agricultural research centres to farmers and also, to a lesser degree, to report farmers feedback to the research centres. Intermediaries in such structural positions in between otherwise disconnected heterogeneous entities are crucial for facilitation of information ows across the whole system (Burt 1995, Howells 2006). However, not all farmers have ties with extension agents. Studies point out tendencies in extension agents selection of farmers with whom to work. Less is known about which farmers are actually receptive to the extension workers agenda. Rendering from existing research, the following factors may inuence farmers access to and a relationship with their extension agent, which in turn might potentially inuence the farmers agricultural practices. Farmers with larger personal networks are a popular target for demonstration of new practices because their behaviour is expected to inuence many others (Rogers 2003, Hoang et al. 2006,

International Journal of Agricultural Sustainability

303

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Prell et al. 2009). In addition to the sheer size of farmers networks, network composition, i.e. what type of people are predominantly known by the respondent, may also matter (Klerkx et al. 2010, Prell et al. 2010). The social capital literature also suggests that social ties to people within communities and outside have different relevance and hence it is useful to distinguish these (Woolcock 1998, Matous et al. 2011, Matous and Ozawa 2010). Moreover, for sustainable resource management, structural attributes of social networks, such as density (i.e. the proportion of existing ties out of all theoretically possible ties), are considered highly relevant (Isaac et al. 2007, Newman and Dale 2007, Bodin and Crona 2009, Bodin and Prell 2011, pp. 29 43). It is necessary to account also for potentially important non-network factors such as wealth (Stommes and Sisaye 1979, Chambers 1997, Haug 1999, Belay and Abebaw 2004, Zekaria and Beyene 2005, Hoang et al. 2006), education (Haug 1999), ethnicity (Isham 2002, Romani 2003), religion (Bandiera and Rasul 2006), geographical accessibility (Stommes and Sisaye 1979, Belay 2003), sex (Saito and Weidermann 1990, Haug 1999), and age (Rogers 2003). To nd the relative roles of networks and the other above-listed factors in farmers receptiveness to extension, we rst examine general determinants of resource-conserving agriculture and also tease out the effects of farmers unequal access to extension. Although resource-saving techniques should ideally be applied in combination, extension agents do not necessarily promote them as a set and some farmers adopt only some of them. We investigate to which degree farmers adopted the set of the selected technologies. Using measures aggregated from diverse practices minimizes the risk that some techniques may be suitable only for everyone. In the main analysis, we specically investigate the determinants of (1) the absolute number and the proportion of known resource-conserving techniques that the farmers had learnt from extension agents; and (2) the absolute number and the proportion of resource-conserving techniques the farmers had ever used for which an extension agent was identied as the most inuential person in their decision to adopt.

3.

Resource-conserving agriculture and the extension system in Ethiopia

3.1. Resource-conserving agriculture in Ethiopia Ethiopia, with about 80% of its population directly employed in agriculture, is one of the most agrarian countries in the world (Central Statistical Agency 2004). The sector is dominated by small-scale-subsistence farmers with rain-fed plots typically of less than 1 ha, characterized by low inputs and low outputs (Deressa 2007). Such farming families cultivate 95% of Ethiopian crop land and account for 90% of the national production (Deressa 2007). Despite this predominance of agriculture across the society, the country is still dependent on food-aid because of inadequate farming practices and ongoing soil erosion on its steep mountainous lands (Bewket 2007, Deressa 2007, Mojo et al. 2010). Since 1970, there have been extension efforts to increase agricultural sustainability but with little success (Gebremedhin and Swinton 2003). For thousands of years Ethiopian farmers have practised excessively heavy cross-plowing with traditional Maresha ploughs, leading to long-term soil deterioration and water loss (Biazin et al. 2011). They have been resistant to change this practice because on structurally weak and semi-arid Ethiopian soils, reduced tillage tend to have negative short-term effects on yield and farmers are unaware of the long-term benets of appropriate tillage (Temesgen et al. 2008, Giller et al. 2009, Temesgen et al. 2012). Such lack of information is generally the main barrier to Ethiopian farmers adoption of new practices (Deressa et al. 2009). Typically, they react to perceived degradation of environmental conditions and decreased yields just by adopting improved seed varieties as opposed to fundamentally changing their traditional routines (Deressa et al. 2009).

304

P. Matous et al.

Conservation agriculture is a system of production that enhances productivity while conserving the environment (Kassam et al. 2009). It has three main principles: (1) minimum soil disturbance, (2) organic soil cover, and (3) crop diversication (Kassam et al. 2009). This study focuses on a battery of six complementary agricultural technologies, which have potential to save constrained natural resources. Three of them correspond directly to the principles conservation agriculture outlined above; two are land management strategies that preserve soil fertility, and one focuses on the usage of efcient varieties. Specically, these six technologies are conservation tillage, fallowing, crop residue management, application of manure or compost, crop rotation, and usage of improved varieties of crops. These practices range widely in terms of their popularity and level of ofcial endorsement from the Ethiopian extension system. The selected techniques do not require a large amount of land nor labour, and they tend to decrease farmers input expenditures. Brief descriptions follow: (1) Conservation tillage is a general name given to any kind of reduced-, ridge-, minimum- or no-tillage techniques, which conserves or reduces soil, water, and nutrient loss, or which reduces draft power and labour requirements for crop production (Fowler and Rockstrom 2001). The level of endorsement of this technique by extension varies across the country. (2) Crop rotation is a farming practice in which a sequence of different crops are grown on a specic eld to preserve the organic matter in the soil for sustainable production (Bullock 1992). Ethiopian researchers and extension workers have highly recommended the use of this practice. (3) Fallowing is a land management system, which involves leaving the farm uncultivated for one or more seasons to conserve soil and to restore its fertility (Carswell 2002). Ethiopian extension workers do not strongly promote fallowing, stating shortage of land as the reason. (4) Crop residue management is an agronomic practice, which includes retention or burning of stubble or straw on the soil surface to reduce erosion and increase water retention. This is particularly important in semi-arid Ethiopia which lacks off-season rainfall (Wiebe and Gollehon 2006, Temesgen et al. 2012). Extension workers generally recommend this practice. (5) The use of organic resources such as farmyard manure and compost instead of chemical fertilizers is highly encouraged by Ethiopian researchers and extension workers. (6) The use of improved less input-demanding varieties of wheat, barley, faba bean, maize, or potato, although not an element of conservation agriculture, is most strongly promoted by Ethiopian extension workers. We focused on varieties that Ethiopian Institute for Agricultural Research introduced in the area within 3 years prior to the study for their potential in sustainable productivity intensication. 3.2. Challenges for Ethiopian extension

Downloaded by [Anandaraj K] at 05:18 10 February 2014

The number of extension workers in Ethiopia has been estimated at 14,000 which is considered insufcient to serve the entire farming population across the country (Belay and Abebaw 2004). Spielman et al. (2008) found that Ethiopian public extension exerts a strong inuence over smallholders access to knowledge. In another study, farmers found information provided by extension agents to be very useful since this was the only source of external information about new technologies, but at the same time, they were critical of the low availability of extension agents (Lemma and Hoffman 2005). The Ethiopian extension system has also been criticized for its top-down structure and lack of participation or feedback from the farmers (Haug 1999, Belay 2003, Belay and Abebaw 2004,

International Journal of Agricultural Sustainability

305

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Anderson and Feder 2007). Under a pressure to meet prescribed technology adoption rates but unable to collaboratively engage their entire assigned population, extension agents reportedly choose most accessible farmers and try to persuade them into adoption (Botha and Stevens 1999, Belay and Abebaw 2004, Lemma and Hoffman 2005). According to the Ethiopian extension policy, three extension agents are supposed to be located in every village but their assignment periods are short and turnover rates are high. Their work is complicated by being typically outsiders to the communities to which they are allocated. They need to work with farmers with diverse mother tongues. The language of Amharic minority used to be the only ofcial language across all nations of Ethiopia. In 1991, a multi-ethnic government was formed and the right of Ethiopian nations to self-determine their language of instruction and administration was ofcially recognized (for detailed elaboration see Bulcha 1997, Hamesa 1997, Anteneh and Ado 2006, Dugassa 2006). At present, Afan Oromo is the ofcial language in the Oromo region where this study was conducted and prociency in this language is the necessary condition for employment within the regional extension system (except for researchers working for the federal research centres).

4. Data collection 4.1. Description of the household survey The data in this article come from a survey by the authors between December 2009 and February 2010 in Tiyo District, Oromia Region, Ethiopia. This district is composed of 18 villages in three agro-ecologies: high altitude (altitude over 2,700 m), mid-altitude (around 2,300 m), and low altitude (below 1,800 m). One village from each agro-ecological zone was randomly chosen for the survey. Each selected village had around 600 households; 100 households were randomly selected from each. Local research assistants administered questionnaires face-to-face to self-identied heads of selected household. Among the total of 300 households, 8 were not available for an interview, 5 were randomly added, so the total number of surveyed households is 297. The ethnic and religious composition of the district and the three studied villages is displayed in Table 1. In addition to the survey of farmers, the authors carried out semi-structured interviews with seven extension agents who were based in the surveyed village. The interviews covered organization of the local extension, general working style of the agents and the specic ways they endorse resource-saving technologies. 4.2. Measuring social networks and their consequences1 We employed the personal network concept (Wellman 2007) to operationalize social networks. The focus is on comparisons at the individual level, i.e. how individuals direct social contacts, or
Table 1. Demography. Ethnicity (%) All Ethiopia Oromia Tiyo District Surveyed sample Known agents

Religion (%) Gurage 2.5 0.93 3.41 6.06 0.57 Christian 53.1 48.6 72.67 87.12 40.23 Muslim 33.9 47.5 27.02 12.88 55.17

Oromo 34.5 87.8 53.92 53.54 83.52

Amhara 26.9 7.22 37.63 40.4 15.34

73,918,505 27,158,471 182,193 297

Source: Ethiopian Central Statistical Authority, 2007 data.

306
Table 2. Variable

P. Matous et al.
Consequences of extension. Number of valid observations 294 218 218 218 218 Standard Mean deviation 0.74 4.87 2.66 3.60 1.87 0.55 0.59 0.44 0.93 1.39 1.31 1.39 0.28 0.36 Minimum Maximum 0 2 0 0 0 0 0 1 6 6 6 6 1 1

Tie with an agent dummy (1 yes) For those with a tie: Number of the selected techniques that the respondent knows how to apply Number of techniques learnt from extension agents Number of applied techniques Number of techniques whose application is attributed due to agents inuence

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Proportion of techniques learnt from agents 218 Proportion of techniques applied because 205 of agents

their personal networks, relate to their outcomes of interest? To specically quantify the extent and composition of the personal network of each household head, we adapted the rst-namecue method (McCarty et al. 1997). We have randomly drawn rst names from the list of household heads in the three villages and asked the respondents whether they knew anyone with the particular rst name. Knowing was dened such that the respondent could directly contact the person when needed and vice versa. We repeated this process until the total number of rst names that the respondent knew reached 14. Owing to the randomness of the cues, the generated 14 people are a representative subsection of everyone the respondent knew by name. The enumerators further asked about each of the elicited 14 partners to get a full picture of the overall composition of the respondents network. To approximate the density of networks that the informants were embedded in, they also elicited the proportion of the respondents associates who are connected to each other. The interviews also included questions about knowledge and usage of the selected resourceconserving agricultural techniques. For each technique that the farmers knew, they were asked how they learned about it and for each technique they had ever used, they were asked who they considered to be the most inuential person in their decision to try it (Table 2).

5.

Empirical methodology

We start the analysis with an ordinary least squares regression to see which farmers in which network positions tend to practice resource-conserving agriculture. Next, we explore the relative ` -vis other social contacts. In order to perceived importance of connection to extension agents vis-a do that, we need to account for the selection bias caused by the fact that not all farmers have ties to the specialists. Therefore, we employed the following two-stage analytical framework. In the rst stage, the determinants of a probit model of knowing any extension agent (either a present or a previous one) are analysed. In the second stage, among the farmers who do have a tie with any agent, the relative informational value of these ties and their perceived consequences for their choice of practices are analysed. As results from the rst stage show, the farmers who know an extension agent are indeed not a random subsample of the local population but rather a group with specic characteristics (and not all of these can be observed). Therefore, using simple regression models could produce biased estimates. Heckman (1979) showed in his seminal

International Journal of Agricultural Sustainability

307

paper that this can be thought of as an omitted-variable bias and can be treated by including the inverse of the Mills ratio (i.e. the non-selection hazard value) computed in the rst-stage probit model as an additional regressor in the second-stage models. The second-stage calculations in this article have been calculated by maximum likelihood estimation and also conrmed by the Heckman two-step model, which produced substantively similar results. A detailed description of the estimation methods and their implementation can found elsewhere (Cameron and Trivedi 2009, pp. 541 547). The independent variables describe household, personal network and geography characteristics. 5.1. Household heads characteristics

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Wealth is operationalized as the area of cultivated land. This measure is closely correlated with possessed assets, but it is less likely to be affected by farmers networks. To allow for nonlinear effects of age, dummy variables for 5-year intervals representing all age groups are used in the analyses. Religion and ethnicity are not completely independent. Therefore, we use a composite categorical variable for the four present combinations of ethnic and religious afliations: Oromo-Christians (the base category), Oromo-Muslims, Amhara-Christians, and Gurage-Christians (Table 3). 5.2. Network characteristics When asking people whether they know someone with a particular rst name, those who know more people need to be given less rst name cues before they provide some predetermined number of positive answers. We use the inverse of the number of cues needed to reach 14 positive answers as an index of the respondents personal network size. For easy interpretation, this index is standardized and centred in the presented tables. In addition to this, we also used indices of personal network density and variables indicating the proportion of specic types of ties in respondents networks. 5.3. Geographical characteristics The non-straight distance in kilometres from the administration ofce in the centre of each village to the household by actual recorded access paths is used as the measure of households remoteness. Household altitude was also measured. Finally, village dummies are included in the models to account for unobserved variations of extension systems across the villages. Summary statistics are in Table 4. 6. Estimation results

More educated, wealthier farmers, with larger and denser networks use more resource-conserving practices (Table 5). Overall, the prevalence of these practices is similar among those who have ties
Table 3. Distribution of respondents and named agents across ethno-religious groups. A Respondents (%) Amhara-Christian Oromo-Christian Oromo-Muslim Gurage-Christian 40.3 40.7 12.9 6.1 B Known agents (%) 17.0 19.6 62.8 0.7 B/A 0.42 0.48 4.87 0.11

308 P. Matous et al.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Table 4.

Summary statistics. Number of observations 295 282 295 295 296 297 297 292 Standard deviation 0.34 4.14 16.33 1 0.40 1.22 0.28 0.20 Mean for those connected to an agent 0.17 4.63 45.06 0.14 1.28 1.66 0.44 0.79 Mean for the others 0.01 4.24 47.38 2 0.36 1.14 2.52 0.43 0.81

Variable Oromo-Muslim dummy (1 yes) Years of completed education Age Network size (standardized) Wealth (square root of cultivated land in hectares) Distance from the administration ofce in kilometres Proportion of communal ties Network density

Mean 0.13 4.54 45.64 0.00 1.25 1.88 0.44 0.79

Minimum 0 0 19 2 1.94 0.00 0.01 0.00 0.00

Maximum 1 14 95 2.67 2.89 5.07 1.00 1.00

P value 0.000 0.485 0.284 0.000 0.008 0.000 0.737 0.429

t tests of the null hypothesis that the means are equal for respondents who have a tie with an agent and the rest.

International Journal of Agricultural Sustainability


Table 5. Determinants of resource-conserving agriculture. Coefcient Household and personal characteristics Wealth (square root of cultivated land in hectares) Ethno-religious group dummies (reference group: Oromo-Christian) Amhara-Christian Gurage-Christian Oromo-Muslim Years of completed education Age dummy (4550 years 1) Network characteristics Network size index (standardized) Network density index Proportion of communal ties Presence of tie with an extension agent (yes 1) Geographical characteristics Village dummies (base category: mid-altitude village) Low-altitude village High-altitude village Distance from the village administration ofce in kilometres Intercept Number of observations R2 0.649 0.129 0.371 0.214 0.042 0.707

309

Standard error 0.194 0.191 0.274 0.342 0.023 0.413 0.076 0.391 0.285 0.190

Downloaded by [Anandaraj K] at 05:18 10 February 2014

0.149 0.873 0.219 2 0.083

0.346 0.925 0.015 1.351 271 0.2077

0.213 0.221 0.073 0.527

Note: Dummies for all age categories are included in the model but the insignicant ones are not displayed in this table. Base category is age above 70 years. Signicance levels: p , 0.01, p , 0.05, p , 0.1. Ordinary least square regression; dependent variable: number of applied resource-conserving techniques.

to extension agents and those who do not, which casts doubt about the effectiveness of extension. There is also no statistically signicant difference in the knowledge and prevalence of resourceconserving practices among different ethnic and religious groups or among men and women. A two-stage statistical analysis of determinants of contact with an agent and resource-conservation follows.

6.1. Stage 1: Determinants of access to extension Extension agents focus on geographically accessible wealthier farmers has been noted in Ethiopia already in the 1960s (Stommes and Sisaye 1979), and it still persists in our sample at present (Table 6). Keeping other covariates at their means, moving away from the village centre by 1 km decreases the chance of knowing an agent by 9 percentage points. Age matters too. Farmers between 40 and 45 years of age have 36 percentage points higher probability of being tied to an agent compared with farmers in the oldest category. Farmers with larger social networks are more likely to know extension agents which may be an outcome of the agents stated strategy of targeting locally inuential individuals. Education of the household head is not a signicant predictor of knowing an extension agent and substituting alternative measures of human capital did not change this result. Ethno-religious afliation matters signicantly. When other covariates are kept at their means, Oromo-Muslims have 37 percentage points higher probability of knowing an extension agent compared with Oromo-Christians. Note that Oromo-Muslims form the majority among known

310
Table 6.

P. Matous et al.
Probit regression of presence of a tie with an agent. Coefcient Standard error 0.271 0.222 0.410 0.523 0.503 0.100 0.091 0.439 dy/dx (at means) 0.258 not signicant not signicant 0.365 0.362 0.098 2 0.092

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Wealth (square root of cultivated land in hectares) Ethno-religious group dummies (reference group:Oromo-Christian) Amhara-Christian Gurage-Christian Oromo-Muslim Age dummy (40 45 years 1) Network size index (standardized) Distance from the administration ofce in kilometres Intercept Number of observations Pseudo R2 Log likelihood

0.970 2 0.175 0.657 1.376 1.364 0.367 2 0.347 2 0.036 288 0.242 2 124.45

Note: Dummies for all age categories are included in the model but the insignicant ones are not displayed in this table. Base category is age above 70 years. Village dummies are also included. dy/dx signies marginal effects on probability of knowing an agent for independent variables kept at their means. Signicance levels: p , 0.01.

extension agents in the studied area but are only a minority among the inhabitants of the surveyed villages (Table 3). The next stage of the analysis examines which characteristics of farmers and their social networks affect the consequences of ties to agents.

6.2. Stage 2: Relevance of ties to extension agents The second stage of analysis is carried out only on respondents who know an agent. Thus, we can replace here ethno-religious afliation of the respondent with dummy variables directly indicating match of religion and ethnicity between household heads and the agent they know (Table 7). Results from the maximum likelihood procedure are presented in Table 8. Correlation of errors caused by unobserved variables (r) is signicantly different from 0 for all models, conrming that using simple OLS without a correction for the sample selection would have produced biased results. The distributions of residual errors are close to normal which suggests that
Table 7. Number of learnt and applied techniques by match of ethnicity and religion between farmers and their respective agents. Number of techniques Learnt from agents Farmers whose: Religion and ethnicity differ from their agent Only religion matches Only ethnicity matches Both match Number of observations 72 26 76 44 Mean 2.21 2.54 2.83 3.18 Standard deviation 1.48 1.21 1.41 1.04 Applied due to agents Mean 1.44 1.73 2.07 2.30 Standard deviation 1.32 1.22 1.51 1.21

Table 8.

Relevance of networks, ethnicity and religion.

Heckman selection model (a regression model with a correction for sample selection)

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Number of techniques learnt from agents

Proportion of techniques learnt from agents

Number of techniques applied because of agents

Proportion of techniques applied because of agents

Coefcient Standard error Coefcient Standard error Coefcient Standard error Coefcient Standard error Dummies indicating match of ethnicity and religion with Only religion matches 0.358 Only ethnicity matches 0.590 Both match 0.886 Proportion of communal ties 2 0.461 Network density index 0.704 Intercept 3.015 Number of observations Censored observations ( does not know an agent) Uncensored observations 286 74 212 the agent (reference group: neither matches) 0.294 0.092 0.060 0.218 0.087 0.044 0.257 0.178 0.052 0.324 2 0.161 0.066 0.470 0.130 0.095 0.431 0.481 0.086 286 74 212 18.76 0.002 2 142.06 2 0.392 2 0.106 0.036 0.277 0.616 0.717 2 0.645 1.329 1.130 286 74 212 24.22 0.000 2 472.31 2 0.798 2 1.143 0.000 0.278 0.208 0.247 0.317 0.454 0.416 0.054 0.072 0.204 2 0.313 0.092 0.643 275 74 201 21.08 0.001 2 184.53 2 0.423 2 0.144 0.038 0.077 0.057 0.067 0.084 0.126 0.115

International Journal of Agricultural Sustainability

16.98 Wald x2(10) 0.005 Prob . x2 Log likelihood 2 480.08 r 2 0.573 lambda 2 0.079 Test of independency of equations (r 0): 0.002 Prob . x2

Notes: The variables included in the selection equation are wealth (cultivated land), ethno-religious dummies, age category dummies, network size, and distance from the administration centre. Signicance levels: p , 0.01, p , 0.05.

311

312

P. Matous et al.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

aggregating the data for all practices and using them in the presented models as continuous dependent variables is legitimate from a statistical viewpoint. We nd no effects of wealth (the area of cultivated land), education, age, network size, nor distance from administration centres. Inclusion of these variables do not signicantly change the results from those presented in Table 8. What determines the reported consequence of ties with extension agents is (1) matching on ethnicity and religion between farmers and agents, and (2) the farmers network composition. Matching on religion alone does not increase the potential for learning but it has to be noted that in the case of the Muslim agents majority, it is not possible to match only on religion (all Muslims are Oromo). That is possible only for Christian agents, who are relatively rare. Knowing an agent of the same ethnicity and religion increases the proportion of techniques learnt from the agent on average by 16 percentage points. The proportion of techniques whose adoption was attributed to agents inuence increases by 20 percentage points. Interviewed agents stated that they do not prioritize any religious or ethnic group but that some farmers apparently become closer with agents of similar lifestyle and mother tongue. Literature on social learning shows that less-knowledgeable farmers tend to be more responsive to information obtained from their peers (Bandiera and Rasul 2006, Conley and Udry 2010). Our ndings further suggest that farmers who are less connected within their community are more receptive to ofcial information sources in their farming decisions. Ten percentage points increase of the proportion of communal ties in ones personal network leads to a decrease of 3 percentage points in the proportion of techniques adopted due to extension workers. Regarding network structure, respondents embedded in denser networks attribute information and inuence about more practices to ties with extension agents. However, people who are embedded in denser networks use more of the resource-conserving practices overall (Table 5) and therefore the effect of density disappears in the relative inuence model.

7.

Conclusions and policy implications

7.1. Conclusions To tackle challenges to agricultural sustainability and for better design of agricultural interventions, it is useful to understand how farmers characteristics and social networks inuence their ` -vis information from their peers. reception of information from extension workers vis-a We nd that farmers who live closer to village centres, who have larger farms, who have larger overall personal networks and who belong to the same ethnic and religious groups as the majority of the front-line extension workers are more likely to be included within the extension system. Among those who are included, there is a difference in their reported levels of reliance on the extension workers. Our sequential analysis shows which factors are associated with high farmers receptiveness to the ofcial information on resource-saving practices. After accounting for the inequality in access to extension, there are no observable effects of education or wealth on the farmers perceived outcomes of extension agents interventions. On the other hand, matching on religion and ethnicity with an extension agent is highly consequential as well as the degree to which the composition of the farmers social networks provides them with opportunities to learn from other community members. While individuals with larger networks are more likely to know an extension agent, those who ultimately report being inuenced in terms of their conservation practices tend to be less socially connected members of the community. Better-educated and wealthier farmers know more and use more of the promoted resource-conserving technologies but they do not attribute this to links with extension agents.

International Journal of Agricultural Sustainability

313

Overall, outcomes of the extension are elusive from the viewpoint of resource conservation. Although a signicant proportion of informants reported receiving useful information from extension agents, those who have no contact with extension have learnt at least as much about conservation agriculture from alternative sources as those with links to extension agents.

7.2.

Implications

However, conrming the importance of informal farmers social learning networks and shortcomings of the current extension should not lead to decrease in attention paid to ofcial information channels and educational systems. We argue that it is often not feasible and ethical for outsiders to try to manipulate local social networks. The extension, on the other hand, is the element of the overall agricultural system that policy makers can directly inuence and improve. Like in other regions (Prell et al. 2009), change agents in the studied area tend to work with individuals who are central in local social networks. These so-called opinion leaders are expected to inuence many others through their social connections (Rogers 2003). However, our data suggest that such well-connected individuals with abundant information sources pay less attention to the extensionists agenda. Therefore, the practice of targeting those who need it least is not only inequitable (Hoang et al. 2006), it may actually be also ineffective. The importance to accommodate the needs of poor small-holders and women has been recognized (Prell et al. 2010). For example, Hoang et al. (2006) recommends working with separate socio-economic groups to ensure true participation in the extension for everyone. Saito and Weidermann (1990) propose that female farmers are approached by female agents who are supposedly more successful in delivering the extension message to them. From our data we can make an analogical point about religion and ethnicity. The messenger is a part of the message. Those in charge of dissemination of sustainable agricultural practices need to understand well the culture, lifestyle, and languages of local populations. This is difcult if the agents of change get frequently transferred between regions and communities, particularly in a multi-cultural country like Ethiopia. Both formal and informal information networks need to be synergistically utilized to spread knowledge-intensive conservation practices. Geographical and social distance presents an obstacle for social learning just as it does for extension operations. Naturally, it is not easy for agents to reach distant households in a context without transportation infrastructure. Nevertheless, for the same reason, little spontaneous behavioural diffusion can be expected across distant locations (Matous et al. forthcoming) and therefore extension services need to operate particularly well in the most remote areas. Similarly, if learning and imitation are less likely to occur across social divides, extension agents must assure to engage interested members of all ethnic and religious groups as well as socially disconnected individuals in search for sustainable practices.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

Acknowledgements
, Kazu te The authors thank Marney Isaac, Ryo Takahashi, Susan Johnson, Steve McDonald, Rochelle R. Co masa Ozawa, Christopher McCarty, H. Russell Bernard, Linton C. Freeman, and Riki Honda, for their comments. They are also grateful to all research assistants and informants for their cooperation.

Note
1. Detailed description of the data gathering and analytical methods as well as estimation results and proofs of models robustness that are not fully presented in this article because of space limitations are available from authors upon request.

314

P. Matous et al.

References
Anderson, J.R. and Feder, G., 2007. Agricultural extension. In: R. Evanson and P. Pingali, eds. Handbook of agricultural economics. New Haven: Elsevier, 23432378. Anteneh, G. and Ado, D., 2006. Language policy in Ethiopia: history and current trends. Ethiopian journal of education and sciences, 2 (1), 3762. Bandiera, O. and Rasul, I., 2006. Social networks and technology adoption in northern Mozambique. Economic journal, 116 (514), 869902. Bartholomay, T., et al., 2011. Mapping extensions networks: using social network analysis to explore extensions outreach. Journal of extension, 49 (6), 114. Beddington, J.R., et al., 2012. What next for agriculture after Durban? Science, 335 (6066), 289290. Belay, K., 2003. Agricultural extension in Ethiopia: the case if participatory demonstration and training extension system. Journal of social development in Africa, 18 (1), 4983. Belay, K. and Abebaw, D., 2004. Challenges facing agricultural extension agents: a case study from southwestern Ethiopia. African development review, 16 (1), 139168. Bewket, W., 2007. Soil and water conservation intervention with conventional technologies in northwestern highlands of Ethiopia: acceptance and adoption by farmers. Land use policy, 24 (2), 404416. Biazin, B., et al., 2011. The effect of long-term Maresha ploughing on soil physical properties in the Central Rift Valley of Ethiopia. Soil and tillage research, 111 (2), 115122. Bodin, O. and Crona, B.I., 2009. The role of social networks in natural resource governance: what relational patterns make a difference? Global environmental change, 19 (3), 366374. Bodin, O. and Prell, C., eds., 2011. Social networks in natural resource management. Cambridge: Cambridge University Press. Botha, C.a.J. and Stevens, J.B., 1999. The need for changes in veld management technology generation and dissemination. South African journal of agricultural extension, 28 (1), 108126. Bulcha, M., 1997. The politics of linguistic homogenization in Ethiopia and the conict over the status of Afaan Oromo. African affairs, 96 (384), 325352. Bullock, D.G., 1992. Crop rotation. Critical reviews in plant sciences, 11 (4), 309326. Burt, R.S., 1995. Structural holes: the social structure of competition. Cambridge, MA: Harvard University Press. Cameron, A.C. and Trivedi, P.K., 2009. Microeconometrics using stata. College Station, TX: Stata Press. Carswell, G., 2002. Farmers and fallowing: agricultural change in Kigezi District, Uganda. Geographical journal, 168 (2), 130140. Central Statistical Agency, 2004. The federal democratic Republic of Ethiopia statistical abstract for 2003. Addis Ababa: Central Statistical Agency. Chambers, R., 1997. Whose reality counts? Putting the last rst. Bourton on Dunsmore: Practical Action Publishing. Conley, T.G. and Udry, C., 2010. Learning about a new technology: pineapple in Ghana. American economic review, 100 (1), 3569. Cramb, R.A. and Culasero, Z., 2003. Landcare and livelihoods: the promotion and adoption of conservation farming systems in the Philippine Uplands. International journal of agricultural sustainability, 1 (2), 141154. Deressa, T.T., 2007. Measuring the economic impact of the climate change on Ethiopian agriculture: Ricardian approach. Washington, DC: The World Bank. Deressa, T.T., et al., 2009. Determinants of farmers choice of adaptation methods to climate change in the Nile Basin of Ethiopia. Global environmental change, 19 (2), 248255. Douthwaite, B., Keatinge, J.D.H., and Park, J.R., 2001. Why promising technologies fail: the neglected role of user innovation during adoption. Research policy, 30, 819836. Dugassa, B.F., 2006. Ethiopian language policy and health promotion in Oromia. Journal of sociology & social welfare, 33 (4), 6986. Folke, C., 2006. Resilience: the emergence of a perspective for social-ecological systems analyses. Global environmental change, 16 (3), 253267. Fowler, R. and Rockstrom, J., 2001. Conservation tillage for sustainable agriculture: an agrarian revolution gathers momentum in Africa. Soil and tillage research, 61 (1 2), 93108. Gebremedhin, B. and Swinton, S.M., 2003. Investment in soil conservation in northern Ethiopia: the role of land tenure security and public programs. Agricultural economics, 29 (1), 6984. Giller, K.E., et al., 2009. Conservation agriculture and smallholder farming in Africa: the heretics view. Field crops research, 114 (1), 2334.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

International Journal of Agricultural Sustainability

315

Hamesa, S.Y., 1997. The language of education in Africa: the key issues. Language, culture and curriculum, 10 (1), 113. Haug, R., 1999. Some leading issues in international agricultural extension, a literature review. Journal of agricultural education and extension, 5 (4), 263274. Heckman, J.J., 1979. Sample selection bias as a specication error. Econometrica, 47 (1), 153161. Hoang, L.A., Castella, J.C., and Novosad, P., 2006. Social networks and information access: implications for agricultural extension in a rice farming community in northern Vietnam. Agriculture and human values, 23 (4), 513527. Howells, J., 2006. Intermediation and the role of intermediaries in innovation. Research policy, 35 (5), 715728. Isaac, M.E., 2012. Agricultural information exchange and organizational ties: the effect of network topology on managing agrodiversity. Agricultural systems, 109 (0), 915. Isaac, M.E. and Kimaro, A.A., 2011. Diagnosis of nutrient imbalances with vector analysis in agroforestry systems. Journal of environmental quality, 40 (3), 860866. Isaac, M.E., et al., 2007. Transfer of knowledge on agroforestry management practices: structure of informal social networks. Ecology and society, 12 (2), 32. Isham, J., 2002. The effect of social capital on fertilizer adoption: evidence from rural Tanzania. Journal of African economies, 11 (1), 3960. Ison, R. and Russell, D., 2007. Agricultural extension and rural development: breaking out of knowledge transfer traditions. Cambridge: Cambridge University Press. Janssen, M.A., et al., 2006. Toward a network perspective of the study of resilience in social-ecological systems. Ecology and society, 11 (1), 15. Kassam, A., et al., 2009. The spread of conservation agriculture: justication, sustainability and uptake. International journal of agricultural sustainability, 7 (4), 292320. Klerkx, L. and Leeuwis, C., 2008. Balancing multiple interests: embedding innovation intermediation in the agricultural knowledge infrastructure. Technovation, 28 (6), 364378. Klerkx, L. and Leeuwis, C., 2009. Establishment and embedding of innovation brokers at different innovation system levels: insights from the Dutch agricultural sector. Technological forecasting and social change, 76 (6), 849860. Klerkx, L., Aarts, N., and Leeuwis, C., 2010. Adaptive management in agricultural innovation systems: the interactions between innovation networks and their environment. Agricultural systems, 103 (6), 390400. Kursat, D., 2010. Analysis of information systems and communication networks for organic and conventional hazelnut producers in the Samsun province of Turkey. Agricultural systems, 103 (7), 444452. Leeuwis, C., 2004. Communication for rural innovation: rethinking agricultural extension. Ede: Blackwell Publishing Ltd. Lemma, M. and Hoffman, V., 2005. The agricultural knowledge system in Tigray, Ethiopia: empirical study about its recent history and actual effectivenessed. Conference on International Agricultural Research for Development, Stutgart: Hohenheim, 1 5. Matous, P. and Ozawa, K., 2010. Measuring social capital in a Philippine slum. Field methods, 22 (2), 133153. Matous, P., Todo, Y., and Mojo, D. forthcoming. Boots are made for walking: interactions across physical and social space in infrastructure-poor regions. Transport Geography. Matous, P., Tsuchiya, T., and Ozawa, K., 2011. Farmers access to resources via networks in remote rural areas with mobile phone reception: creating a resource battery for a mountain tribe in south India. Rural society journal, 20 (2), 174 186. McCarty, C., et al., 1997. Eliciting representative samples of personal networks. Social networks, 19 (4), 303323. Millar, J. and Connell, J., 2009. Strategies for scaling out impacts from agricultural systems change: the case of forages and livestock production in Laos. Agriculture and human values, 27 (2), 213225. Mojo, D., Todo, Y., and Matous, P., 2010. Perception of farmers and agricultural professionals on changes in productivity and water resources in Ethiopia. World academy of science, engineering and technology, 1 (66), 860868. Newman, L. and Dale, A., 2007. Homophily and agency: creating effective sustainable development networks. Environment, development and sustainability, 9 (1), 7990. Ostrom, E., 1990. Governing the commons: the evolution of institutions for collective action. Cambridge: Cambridge University Press.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

316

P. Matous et al.

Prell, C., Hubacek, K., and Reed, M., 2009. Stakeholder analysis and social network analysis in natural resource management. Society & natural resources, 22 (6), 501518. Prell, C., et al., 2010. Competing structure, competing views: the role of formal and informal networks social structures in shaping stakeholder perceptions. Ecology and society, 15 (4), 34. Pretty, J., Toulmin, C., and Williams, S., 2011. Sustainable intensication in African agriculture. International journal of agricultural sustainability, 9 (1), 524. Rivera, W. and Sulaiman, R.V., 2009. Extension: object of reform, engine for innovation. Outlook on agriculture, 38 (3), 267273. Rogers, E.M., 2003. Diffusion of innovations. 5th ed. New York: Free Press. Romani, M., 2003. Love thy neighbour? Evidence from ethnic discrimination in information sharing within villages in Cote dIvore. Journal of African economies, 12 (4), 533563. Saito, K.A. and Weidermann, C.J., 1990. Agricultural extension for women farmers in Africa. Washington DC: The World Bank. Schneider, F., et al., 2009. Social learning processes in Swiss soil protectionthe from farmer - to farmer project. Human ecology, 37 (4), 475489. Solano, C., et al., 2003. The role of personal information sources on the decision-making process of Costa Rican dairy farmers. Agricultural systems, 76 (1), 318. Spielman, D., et al., 2011. Rural innovation systems and networks: ndings from a study of Ethiopian smallholders. Agriculture and human values, 28 (2), 195212. Spielman, D.J., Ekboir, J., and Davis, K., 2009. The art and science of innovation systems inquiry: applications to Sub-Saharan African agriculture. Technology in society, 31 (4), 399405. Spielman, D.J., et al., 2008. An innovation systems perspective on strengthening agricultural education and training in Sub-Saharan Africa. Agricultural systems, 98 (1), 19. Stommes, E. and Sisaye, S., 1979. The administration of agricultural development programmes: a look at the Ethiopian approach Part 2. Agricultural administration, 6 (4), 253267. Tefera, B. and Sterk, G., 2010. Land management, erosion problems and soil and water conservation in Finchaa watershed, western Ethiopia. Land use policy, 27 (4), 10271037. Temesgen, M., et al., 2008. Determinants of tillage frequency among smallholder farmers in two semi-arid areas in Ethiopia. Physics and chemistry of the earth, parts A/B/C, 33 (1), 183191. Temesgen, M., et al., 2012. Assessment of strip tillage systems for maize production in semi-arid Ethiopia: effects on grain yield, water balance and water productivity. Physics and chemistry of the earth, parts A/B/C, 4748, 156 165. Warner, K.D., 2007. Agroecology in action. Cambridge, MA: The MIT Press. Wellman, B., 2007. The network is personal: introduction to a special issue of Social Networks. Social networks, 29 (3), 349356. Wiebe, K. and Gollehon, N., 2006. Agricultural resources and environmental indicators. Washington, DC: United States Department of Agriculture. Woolcock, M., 1998. Social capital and economic development: toward a theoretical synthesis and policy framework. Theory and Society, 27 (2), 151208. Yila, O.M. and Thapa, G.B., 2008. Adoption of agricultural land management technologies by smallholder farmers in the Jos Plateau, Nigeria. International journal of agricultural sustainability, 6 (4), 277288. Yorobe Jr, J.M., Rejesus, R.M., and Hammig, M.D., 2011. Insecticide use impacts of Integrated Pest Management (IPM) Farmer Field Schools: evidence from onion farmers in the Philippines. Agricultural systems, 104 (7), 580587. Zekaria, S. and Beyene, H., 2005. Multivariate discriminant analysis: participation in the new extension package program and its effectiveness on crop productivity in Oromia Region. In: A. Mottram, D.P. James, W.I. Robinson, S. Zekaria and G. Taddesse, eds. The Ethiopian agricultural sample enumeration in-depth analysis. Addis Ababa, Ethiopia: Center for Arid Zone Studies, 3343.

Downloaded by [Anandaraj K] at 05:18 10 February 2014

You might also like