An Article by Andrei Parvan
Most of the world’s poor are engaged in agriculture in rural areas. Governments and development agencies promote income-generating projects as a way of encouraging growth through increased agricultural production and the protection of the natural resource base. Not all targeted communities participate in the agricultural development projects at the ideal rates and intensity, or for the proscribed length of time. This paper presents a review of agricultural technology adoption literature, specifically community characteristics that have the most significant associations with technology adoption, disadoption and non-adoption. The characteristics this paper investigates come from Feder, Just, and Zilberman’s seminal World Bank study published in 1982: farm size, land tenure system, credit access, labor availability, biophysical characteristics, risk preferences, human capital, and access to commodity markets. The paper recommends policy options for governments and aid agencies to increase the likelihood that a targeted community will adopt an introduced agricultural technology. Finally, the paper focuses on how the UN World Food Programme can apply these recommendations in scaling up a degraded-land-reclamation project in Ethiopia called Managing Environmental Resources to Enable Transitions to More Sustainable Livelihoods (MERET).
About the Author
Andrei Parvan graduated in 2010 with a Master of Public Administration from Cornell Institute for Pubic Affairs. His academic and career interests focus on the overlap between food policy, humanitarian policy and international development. He graduated from the University of Arizona in 2008 with a B.A. in International Studies. While at the University of Arizona he spent the 2005-2006 academic year studying English and American Literature at the Université de Fribourg in Switzerland. He has interned with the Office of Congresswoman Gabrielle Giffords in Tucson, AZ and with the UN World Food Programme in Rome, Italy, and in Addis Ababa, Ethiopia.
Introduction to Agricultural Technology Adoption Literature
The vast majority of the world’s poor lives in rural areas and is engaged in agriculture, and therefore activities designed to address the vulnerability of these rural poor are often geared toward improving agricultural practices as a means of increasing productivity, efficiency and, ultimately, income. Governments, NGOs, aid agencies and extension workers have long known that the success of any project depends, in part, on whether farmers adopt the offered technologies and, if they do, whether those farmers adopt the technologies in an ideal combination and for the proscribed length of time needed to produce designed results. Researchers have conducted decades’ worth of surveys and analyses around the world in an attempt to understand the adoption decisions of individual farmers and the diffusion patterns among communities of farmers and rural poor. By understanding how farmers and communities decide whether to adopt a technology, aid professionals can refine their agricultural technology outreach projects to address the conscious and subconscious concerns of targeted communities, and increase the probability that farmers will be willing and able to participate in project activities.
List of Acronyms
|FAO Food and Agriculture Organization
FFW Food for Work
GoE Government of Ethiopia
GR Green Revolution
HDI Human Development Index
HYV High Yield Variety
IPM Integrated Pest Management
LEI Low External Input
LEISA Low External Input Sustainable Agriculture
|MERET Managing Environmental Resources to Enable Transitions to More Sustainable Livelihoods
NGO Non-Governmental Organization
NRM Natural Resource Management
SA Sustainable Agriculture
SLM Sustainable Land Management
SRI System of Rice Intensification
UNDP United Nations Development Programme
WFP World Food Programme
In trying to measure the process of agricultural technology adoption and diffusion, researchers most commonly use three methods to understand the factors that determine the adoption of technology across space and time: time series analysis, cross-sectional analysis, and panel data analysis.1 Each approach involves collecting and analyzing different types of data and methods, and explains a different aspect of the adoption process. Researchers use time-series data extensively to explain how the rate of technology adoption varies with time, but time-series data does not address the fundamental reasons for adoption.2, 34 Cross-sectional data analyses come in two forms: “snapshot” and “recall.”5 The former associates farmer characteristics with likelihoods of adoption and the latter links characteristics with the time at which adoption occurred.6 The shortfalls of these data are the unrealistic assumptions required to make the data applicable, mainly that characteristics are consistent over time.7 Panel data bring together cross-sectional and time-series data and can be used to explain both adoption process and the characteristics associated with adoption.8 They are rarely used because they are difficult to collect and hard to manipulate. These three empirical methodologies describe the parts of agricultural technology adoption which must be understood if governments and NGOs are to craft their activities for optimum effect: what characteristics, and across what time intervals are associated with which probabilities of participation.
Technology is assumed to mean a new, scientifically derived, often complex input supplied to farmers by organizations with deep technical expertise. Neill and Lee point out that the majority of existing literature on agricultural technology adoption is focused on Green Revolution (GR) technologies such as irrigation, fertilizer use, and the adoption patterns of high-yield variety (HYV) seeds.9 Due to the development process of HYV and the inputs required to make them productive, studies examining HYV adoption look at very advanced forms of technology; HYV seeds are often the product of intensive laboratory research, and when they are targeted to farmers they are bundled with other technology inputs such as chemical fertilizers, pesticides and extensive irrigation because these are necessary for the HYV seeds to perform as designed. Because so many studies of agricultural technology adoption and diffusion focus on HYV and other GR inputs, their findings are concentrated on a “high-tech” definition of agricultural technology.
However, the association between most agricultural technology adoption literature and “high technology” inputs is incidental; it just so happens that at this point in time, most agricultural technologies being measured are scientifically advanced. This coincidence should not obstruct the point that a technology is simply the application of scientific knowledge for a certain end. A project or a technique can still be considered a technology even if the science is many steps removed from the eventual implementer. For example, a project where extension workers encourage farmers to rotate legumes into their planting cycles is quite “low-tech,” but the chemistry behind the process of nitrogen fixation is extensive and elaborate. There are many lessons and best practices that can be gleaned from existing studies if technology is looked at in broader terms. Gershon and Umali define technology as “… a factor that changes the production function and regarding which there exists some uncertainty, whether perceived or objective (or both). The uncertainty diminishes over time through the acquisition of experience and information, and the production function itself may change as adopters become more efficient in the application of the technology.”10
The UN World Food Programme (WFP) developed a project called Managing Environmental Resources to Enable Transitions to More Sustainable Livelihoods (MERET)11 comprised of different bundles of agricultural technologies. MERET, like other forms of sustainable agriculture (SA) and natural resource management (NRM) activities, does not have the same immediately obvious “technical” implication as GR activities, but it is an agricultural technology nonetheless. In this paper, technology is any discrete input — either as a good or as a method — with the purpose of controlling and managing animal, vegetative growth, or both. This more inclusive concept allows us to look at the adoption dynamics and diffusion patterns of an expanded MERET project using criteria established by a wide body of scholarly research and publications. The existing research is ultimately concerned with understanding the farming choices of rural communities, and there is no evidence to suggest that the decision-making process is dependent on the scientific sophistication of the input. The characteristics associated with higher rates of HYV adoption are the same as the ones associated higher participation rates in terrace construction, save for context-specific exemptions.
Just as there are different types of technologies, there are different kinds of adoption. Feder, Just and Zilberman make three distinctions in types of adoption: 1) individual vs. aggregate adoption, 2) singular vs. packets of technologies available for adoption, and 3) divisible vs. non-divisible technologies. The first option is between final adoption at the individual level, which involves an internal deliberative process but is ultimately manifested as a dichotomous decision, and the aggregate adoption behavior observed as the diffusion of a technology, and its corresponding adoption, throughout a discrete space.12 Individual adoption can measure the degree of use in the long run, but it is ultimately a binary observation. Aggregate adoption, on the other hand, is measured as the aggregate level of use of a particular technology among one specific group of farmers or within one particular area.13 These farmers, whether observed individually or collectively, can choose to adopt in different ways. In some instances, farmers are presented with a single choice: the adoption of one discrete technology such as a new HYV seed, or some other single input. But in most cases, as with MERET, agricultural technologies are introduced in bundles, and these bundles are often complementary.14 A HYV seed is introduced along with the fertilizer and corresponding land preparation practices needed to make the HYV work as designed.
Similarly with MERET, a community site may be recruited to construct dams, bunds, gully controls and terracing. They may also be taught new forms of organic and green manure application, and trained on different income-generating agriculture, including high-value fruits and vegetables and targeted animal-fattening programs. This gives farmers several distinct technological options, and it gives those trying to measure and model that adoption more to consider because farmers may adopt the complete package of innovation, they may adopt nothing, or they may pick subsets of bundles. Doing so produces several simultaneously occurring adoption and diffusion processes, although these processes have been shown to follow specific and predictable patterns.15
These descriptions of adoption focus on the degree of use, but some technology options are non-divisible, so their adoption either happens or does not. Variable inputs such as HYVs can be adopted in part and planted on a percentage of farmland, and fertilizer can be applied selectively, so modeling their adoption and diffusion involves first measuring if it has been adopted at all, and second assessing the extent to which farmers have adopted it. Technologies such as wells, tractors and other mechanized inputs are not divisible, thus farmers have only a discrete choice: either adopt the technology entirely or not at all. Modeling this adoption behavior at the individual level produces dichotomous outcomes, but an aggregate analysis turns these discrete choices into continuous measures of the percentage of farmers using the non-divisible inputs.16
Understanding the different kinds of possible adoption is important in understanding how traditional indicators relate to that adoption process. A study looking at MERET adoption as a binary choice would offer vastly different results than if MERET were looked at in terms of technology bundles, analyzing which bundles are adopted in what combination by which types of farmers and communities.
Ethiopia and MERET
Ethiopia is a country so beset by poverty and vulnerability to natural and man-made shocks that it has become synonymous with famine and starvation. It is both one of the poorest countries in Africa17 and one of the most populous.18 That it consistently generates some of the lowest human development indicators in the world19 due in part to the reinforcing cycles of poverty and high population growth.20 The United Nations Development Programme’s (UNDP) annual ranking of countries based on development indicators places Ethiopia among the bottom 15 countries in 2007; Human Development Index (HDI) trends since 1990 have been well below those of the Sub-Saharan average.21 Even after decades of large humanitarian and development operations, millions of people in Ethiopia are chronically food insecure, requiring regular and repeated food transfers to meet basic caloric requirements. The United Nations Food and Agriculture Organization (FAO) and WFP report that over 7.5 million people were chronically food insecure in 2009 and another 4.9 million people were in need of emergency food assistance from January to July 2009.22 As high as these numbers are, they likely underestimate the true food insecurity problem by several millions of people.23
Development agencies are often influenced by the misconception that chronic food insecurity results from inadequate food supply. Agencies influenced by this line of thinking have tended to favor direct transfers of food to meet immediate needs. And when the targeted communities remain food insecure in subsequent years, these aid agencies are left searching for reasons why a one-time in-kind transfer was insufficient for addressing the underlying causes of the original food insecurity. In the past few decades donors and aid professionals have come to learn that food insecurity is the result of insufficient access to food, not insufficient availability. Thus, development projects addressing long-term food insecurity have been refined to promote income generation, reduce vulnerability to shocks, increase sustainable local production and strengthen local capacity and infrastructure.
In Ethiopia this paradigm shift has manifested itself in a project called Managing Environmental Resources to Enable Transitions to More Sustainable Livelihoods (MERET), which has been implemented by WFP, along with assistance from the Government of Ethiopia (GoE) since 1980. Roughly 80 percent of Ethiopian households live in rural areas and are highly dependent on small-scale local agriculture to meet their food needs.24 WFP finds that much of the need for direct and consistent food transfers results from low agricultural productivity and heavily degraded agricultural lands, population growth, and extremely low household incomes.25 Compounding the degradation problem and confounding attempts to address chronic food insecurity is the understanding that those factors are interrelated and inter-propagating. The natural resource base is degraded from unsustainable farming practices and forest removal, these unsustainable practices being the byproducts of growing population pressures. But these population pressures were in part caused by widespread poverty, political and military conflict, and high incidences of natural shocks, especially drought-flood cycles. These forces locked the growing population out of school, out of the cities, out of non-agricultural work, and consequently forced them to stay on increasingly smaller and more heavily exploited land parcels. Responding to this income insecurity and relative unavailability of larger plots, more productive lands or non-farm opportunities, many desperate Ethiopians were forced to over-exploit their meager holdings to ensure some short-term food availability.
To break this cycle, MERET attempts to promote long-term food security by providing targeted communities with income-generating opportunities and access to more productive lands through Sustainable Land Management (SLM) practices, degraded land reclamation activities and participatory, community-based watershed development.26 Due to the variation in biophysical characteristics among different MERET sites, techniques are context-appropriate and not necessarily uniform. There are, nevertheless, some shared features common to most MERET sites. They include soil and stone bunds, gully-control constructions, trenches, bench terracing, water-pond construction, organic fertilizer application, and the planting of strategically chosen tree, shrub and grass varieties.27 In the past 30 years MERET has covered more than 600 sub-watershed systems, each with 300 to 2,000 participating households. The program has directly benefited over 1.3 million people.28 Biophysical outcomes of MERET include:
• increased vegetative covers and increasing or rebounding biodiversity;
• reduced rates of soil loss from both cultivated and uncultivated lands, and reductions in the on- and off-site impacts of soil erosion;
• improved availability of surface and sub-surface water; and
• regulated microclimates.29
The impact that MERET has had on human lives has been just as dramatic. With the rehabilitation of once unproductive lands, MERET helped poor communities create income-generating opportunities via activities such as the managed collection of foodstuffs, wood and other biomass from the newly verdant project sites. Indicators showing positive impacts on human wellbeing include:
• improved availability of, and access to, food;
• modernized housing conditions and home amenities;
• increased financial assets and investments in savings accounts, livestock and other assets;
• increased school enrollment and participation of children;
• abandonment of out-migration as a coping mechanism;
• improved quality and quantity of water;
• increased access to fuel, construction wood and grasses; and
• increased community confidence, sense of self-reliance and control over own destiny.30
Communities are expected to provide the labor required for area enclosures, for bund, pond and terracing construction, and are paid for this labor with food assistance. Participants receive assistance for immediate food needs in the form of Food for Work (FFW) payments, and they also receive investment to improve future food availability through their MERET work with WFP and the GoE. Even with these dual incentives for participation, not all targeted communities choose to take part in MERET, and of those that do opt in, not all of them stay in for the entire five-to-seven years of the project cycle.
Although MERET has been active for over three decades and receives tens of millions of dollars each year, it is still a relatively small project. As the benefits of MERET become publicized WFP may decide to expand the project to other marginalized and food-insecure communities, and WFP’s understanding of the technology adoption decision-process can ensure the successful implementation of MERET among these new communities. The scholarly literature is replete with studies that discuss agricultural technology adoption patterns to see which determinants have significant impacts on the adoption decision, a review of which can be used to understand MERET participation rates. Governments, aid agencies, and development NGOs can then tailor their agriculture outreach projects to be attractive to their targeted communities.
Factors Influencing Adoption
The most often cited factors that have been used to explain the variability seen in agricultural technology adoption and its patterns of diffusion, are those described by Feder, Just and Zilberman.31 A range of literature measuring technology adoption, including Besley and Case,32 Zeller, Diagne and Mataya,33 Neill and Lee,34 Arellanes and Lee,35 Fuglie and Kascak,36 Adesina and Baidu-Forson,37 and Moser and Barrett38 start with the factors spelled out by Feder, Just and Zilberman. These explanatory indicators vary from study to study based on their contextual applicability, but traditionally include: 1) farm size, 2) risk exposure and capacity to bear risk, 3) human capital, 4) labor availability, 5) credit constraints, 6) tenure, and 7) access to commodity markets. In delineating these particular factors, they point out that the categories are not discrete or exclusive and that boundaries may blur and overlap due to the interdependent relationship between indicators.39 For example, inadequate rural financial systems decrease the availability of affordable credit; a lack of credit increases aversion to risky undertakings such as new technology adoption; higher levels of risk aversion — or decreased ability to mitigate and bear risk — are correlated with higher levels of poverty and vulnerability to shocks; higher poverty levels are themselves associated with smaller farm sizes, lower levels of education and less allocative ability to manage change. Many studies have shown that each of these indicators significantly influences the agricultural technology adoption process; trying to separate each characteristic from the others is difficult and may even be unnecessary. The objective of adoption surveys — and of this paper — is to show how each variable affects adoption, allowing implementing actors to refine their strategies based on a wide body of empirical and qualitative results.
Farm size is often one of the first factors measured when modeling adoption processes. Farm size does not always have the same effect on adoption; rather the literature finds that the effects of farm size vary depending on the type of technology being introduced, and the institutional setting of the local community.40 Fixed costs are often a primary barrier to adoption; therefore, spreading fixed costs over a larger farm may be one explanation for the observed positive association between farm size and propensity to adopt. That is not to suggest uniform causation; farm size may act as a proxy for other socio-economic indicators such as access to credit because the larger farm has more collateral value. It very well may be the case that these correlated indicators also influence the adoption decision, and therefore a failure to account for them in the regression models may tend to inflate the reported relationship between farm size and adoption likelihoods. Looking at soil conservation techniques in the Philippines, Shively finds that the decision to adopt depends on farm size, partially because soil conservation on small farms is especially costly due to increases in the short-run risk of consumption shortfall with certainty.41 The adoption of System of Rice Intensification (SRI) — a rice-growing technique for increased yields through decreased non-labor inputs — in Madagascar follows a similar pattern, with adopters allocating larger amounts of land to the practice than those farmers who adopted and later disadopted.42
In analyzing the diffusion of conservation tillage practices, integrated pest management (IPM) activities and soil fertilizer testing among American farmers, Fuglie and Kascak begin with the traditional explanatory factors, including farm size.43 They report that larger farms were more likely to adopt the technology bundles sooner than small farms, and that the adoption lags steadily increase for smaller farm sizes.44 Using a cross-sectional approach with recall, the researchers were able to account for underlying dynamic influences in adoption, finding that differing rates of technology diffusion among regions persist over time.45 And as predicted by Besley and Case, the use of recall data forced them to assume exogenous farm characteristics were constant over the period of technology diffusion, noting that “while this assumption is probably valid for natural resource characteristics, it is possible that other farm characteristics may change over time.”46
Looking at maize-mucuna adoption in Honduras, Neill and Lee report that above a minimum required threshold, farm size has the expected positive association with propensity to adopt the agricultural technology.47 Feder and Umali find that while larger farms adopt lumpy (non-divisible) and divisible technologies faster than smaller farms, the latter adopt the divisible technology more intensively, and may eventually adopt the lumpy technology.48 This positive relationship between farm size and likelihood to adopt represents a significant problem for MERET. Because the primary targets of MERET activities are the poorest and most food-insecure communities, WFP is specifically targeting those farmers whose poverty, and consequently smaller farm sizes, indicate they are the least able to adopt the types of agriculture technologies which aim to address the root cause of that poverty.
Risk and Uncertainty
All technology adoption decisions carry with them some mixture of subjective risk — such as human tendencies to assume more uncertainty in outcomes from unfamiliar techniques — and objective risks resulting from variations in rainfall, pests, diseases and other blights, and the timely access to critical inputs.49 The observed patterns of technology adoption are typically influenced by the farmers’ individual risk preferences and their ability to bear the risk of a new and uncertain endeavor.
Contributions to the understanding of the role played by farmer uncertainty in SA and NRM practices made by Lee are of particular significance to a survey of MERET adoption. Lee notes that unlike GR technologies, benefits from techniques employed in SA and NRM activities are more heavily skewed towards the future, while the costs are immediate.50 This extreme delay in benefits gives a more prominent role to risk-preferences and uncertainty in the technology-adoption decision making process because for farmers to opt into a SA/NRM project like MERET, they require certain guarantees that future access to land, inputs and outputs will not be a point of uncertainty. Without some level of assurance that access to future benefits is not at risk, farmers have little incentive to invest their time, labor and capital into technology adoption. The study of SRI adoption in Madagascar shows how institutional deficiencies can exacerbate risk aversion: for the poorest and most food-insecure households, weak rural financial systems drive up the implicit interest rate on credit, making the net present value of even high future returns seem less valuable than income earned today.51 Holden and Shiferaw find that Ethiopian farm households’ planning horizons are short, discount rates are high, and their willingness to invest in productivity-increasing activities is “so low as not to even partially internalize long-term land degradation externalities.”52 Although the correlative relationship between risk-aversion and the economic measure of poverty is complicated, there is strong evidence to suggest a strong and significant relationship between low return on assets, low asset levels, the ability to diversify and manage risk, and income poverty.53 This relationship between poverty and extreme risk aversion (or extreme inability to bear risk) may serve to caution against the implicit logic, which assumes that poor farmers will accept any technology that is expected to produce increased future yields. Poverty leaves farmers vulnerable to food and income shocks, against which they have little capacity to insure; therefore even large future returns may not seem attractive if the immediate costs, and immediate risks, are sufficiently distorted. And if the farmers are from the lowest socio-economic cross-section, their lack of access to agricultural and financial resources will prevent them from being able to bear risks, even if they would otherwise prefer the riskier option. In other words, they are kept from experimenting with new techniques and technologies by the amount of risk they are able to take on, not by the amount of risk they prefer to accept.54
Studies looking at technology adoption behaviors following shocks find that the consequences of covariate shocks affect welfare for many years after the initial impact, and in anticipation of such outcomes, poor households opt for less risky technologies to avoid permanent damage.55 This suggests the existence of risk-induced poverty traps, where the most vulnerable are kept in low-yielding agriculture by their high sensitivity to the possible negative outcomes of risky investments. Zimmerman and Carter find that subsistence constraints and imperfect credit led to bifurcated optimal consumption strategies, with poor farmers adopting lower risk (and lower yield) crop varieties and absorbing shocks by reducing their consumption to maintain asset levels.56 In Ethiopia, as in many developing nations, input levels have to be decided before the uncertainty over yields, climate, tenure and other indicators, has been resolved; this early commitment compounds the original aversion to adopting the high-yielding technologies which may break them out of their vulnerability-induced preference for traditional, yet insufficient, farming techniques.
These variables are comprised of individual or community characteristics such as education, human health indicators, age and gender demographics, and their relationship to technology adoption is one of potential. Welch breaks down human capital into worker ability and allocative ability, with the latter defined as the ability to adjust to change.57 It is suggested that farmers with higher education possess higher allocative abilities and are able to adjust faster to farm and market conditions.58 Looking at U.S. farmers, Fuglie and Kascak find that human capital is positively correlated with innovators or early adopters; farmers with higher levels of education adopt new technology more rapidly than farmers with only a high school diploma; and laggards are associated both with lower education and also with poor soil quality where technology does not perform well on marginal lands.59 In examining technology adoption among poor households in Bangladesh, Mendola finds that human capital features, such as improvements in health and education, foster the adoption of new technologies.60
While human capital traditionally focuses on education and health indicators, Adesina and Baidu-Forson expand the category to examine farmer rationality in the technology adoption process. Surveying sorghum farmers in Burkina Faso, they note that economists investigating the adoption of new technologies often overlook how farmers’ subjective perceptions of the applicability of technological outputs influence adoption decisions. In Burkina Faso the main use of sorghum is the making of tô, a paste derived from the grinding of sorghum, and this tô is the cornerstone meal for most caloric intake in Burkina Faso. Adesina and Baidu-Forson find that adoption of different sorghum varieties was based more on the applicability to grinding each variety than on output increases.61 These findings corroborate similar ones by Zinnah et al., which show that farmers’ assessments of the relevance of technology is more important than contact with the technology or with extension workers in the adoption process.62 Thus the consideration of non-agronomic implications by farmers is another manifestation of human capital, which influences the adoption decisions and which a scale-up of MERET should take into account.
Just as risk-aversion can create a poverty cycle due to the poor’s increased vulnerability to even minor shocks, negative human capital indicators, highly correlated to income, can also reinforce unsustainable agricultural practices and aversion to technology adoption. Yamauchi, Yohannes and Quisumbing find that investment in human capital development, specifically education, decreases aftershocks.63 They also find that human capital accumulation prior to disasters increases resilience to the adverse effects of those shocks.64 The prevalence of negative human capital indicators already makes the most vulnerable Ethiopians less likely to adopt capital-improving technologies, but add to that the projected effects of climate change and the situation looks even more dire. The World Bank estimates that climate change will have marked negative impacts on Ethiopia’s agriculture. A one-unit increase in temperature during summer and winter would reduce the net revenue per hectare by US$177.72 and US$464.71 respectively, whereas the marginal impact of increased precipitation during the spring would result in net revenue increases of US$224.09 per hectare.65 If nothing is done to mitigate these projected effects, increased poverty and decreasing human capital indicators will push more vulnerable Ethiopians into a poverty-shock cycle and make it more unlikely that they will adopt the types of risky agricultural technologies that can break them out of that cycle. Mendola asserts that better targeting of resource-poor producers might be the main vehicle for maximizing direct poverty-alleviating effects.66 And it is in this capacity that MERET is designed to operate. Income generating activities and land rehabilitation schemes hold the promise of creating assets and reducing risk-aversion to the point where more farmers will adopt MERET activities, creating a cycle of growth and adoption. But to overcome the initial resistance to adoption, MERET should address the risk-preferences of the rural poor and break the current cycle of vulnerability, poverty and risk-aversion.
The average household size in MERET-targeted areas is 5.9 persons — 50 percent of households have an average size of five persons and 30 percent have between seven and nine persons. The populations in these areas skew young, with 59 percent of the population under the age of 19 and almost one-third below the age of nine. Women-headed households account for a very small percentage, roughly 4.1 percent of total households. However, due to the goals of MERET and WFP, women make up a full 35 percent of project participants. Almost 90 percent of people participating in MERET activities are illiterate,67 compared with a national average of 64 percent.68 Because MERET is developed for areas with marginal lands and marginalized people, a study of the association between human capital indicators and MERET participation could show a negative relationship, contrary to the findings of the literature cited above.
The labor market affects technology adoption differently depending on whether the area targeted with the technology has a net labor shortage or net labor surplus; seasonal availability adds another dimension. Another consideration is whether the proposed technology is labor-saving or labor-intensive. Higher labor supply is associated with higher rates of adoption of labor-intensive technologies;69 the inverse is also true. Lee70 sums up findings showing that household size and labor availability have been shown to influence adoption of soil conservation investments in the Philippines71 and Ethiopia.72 He also points out the dual nature of off-farm labor possibilities, noting that increased liquidity can allow farmers to invest in SA and NRM, but can also reduce the availability of labor and thereby decrease the likelihood of adopting high-labor technologies.73 Polson and Spencer, looking at HYV adoption of cassava among subsistence farmers in Nigeria, found that family size (and therefore labor availability) was not a significant influencer of adoption.74 They explain this discrepancy by suggesting that subsistence farming does not experience the same types of labor shortages as income-generating agriculture.75
Examining the diffusion of SRI among food-insecure Malagasy rice farmers, Moser and Barret report slightly more nuanced findings. Although SRI is a low external-input (LEI) technology, it does in fact require 38-54 percent more labor than traditional rice-growing methods.76 While the returns to labor seem to significantly surpass those of traditional methods, many farmers either never participate in SRI, or opt out of the project after only a few years. A major reason for this is the timing of increased labor demand, and labor’s intertwined relationship to credit constraints. The need for added labor inputs means farmers have less time to sell their labor to other farmers. While the added income from output increases would serve to offset the opportunity cost of lost labor wages in the present, farmers can only conceivably do this if they have access to credit to meet their financial obligations until the SRI crop is harvested. But weak rural financial institutions and non-existent credit means that attractive returns to SRI are less impressive to farmers because they are unable to meet immediate needs; therefore they choose not to participate in SRI, but rather to sell their excess labor to other farmers in order to earn immediate wages.77
Access to credit is an indicator which manifests itself in other factors, such as farm size (since a farmer can borrow more money against a larger farm than a smaller farm, all other things being equal), human capital (because farmers with more education are better informed about credit practices and can even shop around for competitive interest rates), and tenure (since a sharecropper does not own land, and cannot borrow against its value). Lee notes that increased access to credit sources can help farmers surmount short-run liquidity constraints and increase technology adoption.78
Credit considerations are of indirect concern as well because explicit and implicit interest rates determine the future value of money, and when interest rates are high, they can make modest immediate income seem more attractive than even large future returns. Rational farmers, comparing present opportunities against future income streams, can therefore be expected to exhibit sensitivity to interest rates and other credit considerations. This makes farmers from areas with high interest rates less likely to participate in any activity in which they forgo immediate cash for any future returns. In areas where this is the case, aid agencies should include cash transfers, or payment for project participation, in order to overcome the distorted discounting caused by high interest rates. Another option is to provide financing to the communities at more reasonable interest rates, although both options risk angering local moneylenders. MERET typically uses food transfers as payment, rather than cash. This approach works for WFP because the targeted communities are significantly food-insecure; food transfers in areas with sufficient food production are not advisable as they can distort food prices, leading to lost revenue for farmers and lower production in the future.
Tenure incorporates issues addressed in the sections on credit constraints and risk and uncertainty. As mentioned above, the uncertainty associated with a change of course is an impediment to technology adoption. It is the most vulnerable communities, those that are least able to afford a decrease in output, that are the most risk-averse. The most vulnerable communities are also more likely to have insecure tenure rights. The self-reinforcing nature of vulnerability means that those who can least afford to take on risk are the ones who are trapped in a cycle of poverty due to that risk-aversion. Poverty status is also related to land insecurity, further reducing these communities’ incentives to adopt risky technology, and further promoting the risk-poverty-tenure cycle.
The history of land certification in Ethiopia is yet another explanation for the country’s chronically insufficient agricultural production. Land rights in the 20th century fall into three main episodes, characterized by the form of government ruling Ethiopia at that time. Before 1975 Ethiopia was an absolute monarchy and followed a traditionally imperial landowning system. The vast majority of land was owned by a few noble and absentee landowners,79 but worked on by peasants. Large tracts were underutilized.80 In 1975 the imperial government was overthrown by a military coup (the Derg), which instituted a command-economy modeled after the USSR and Eastern Bloc countries. Under this order, land was confiscated from the large landowners and transferred to the State. The State then allocated parcels to families and gave them user-rights, although the State remained the ultimate owner of the land. Land could not be sold, transferred or mortgaged.81 The Derg were overturned in 1991 by the current government in Ethiopia. This government maintained the land-tenure structure instituted by the Derg with a few alterations. It was written into the constitution that anyone wishing to farm had the right to land, and people were, for the first time, given the opportunity to rent out their land to sharecroppers.82
Deininger, Ali and Alemu measured the impacts of land certification on investment and tenure security. They find that tenure security reduces the fear of land redistribution, thereby addressing uncertainty over land position.83 Their findings also note that land tenure security is strongly correlated with increased likelihood to invest in soil and water conservation activities, and that it more than doubles the predicted number of hours spent on each activity.84 Increased land security also increases the propensity to rent out land, which may lead to more efficient allocation of resources if the landowner is unable or unwilling to cultivate her plot.85 In a different paper these same researchers find the inverse holds as well. Because sharecropper farmers will, in any given season, receive only a part of their marginal product, they have limited incentive to invest more time, labor or capital than is minimally required.86 The results suggest that input and output intensities are significantly lower on sharecropped lands compared to owned plots.87 Ali, Dercon and Gautam find that the share of land allocated to coffee (which is both a high input and high output crop) increases if transfer rights are present, while the expectation of land loss results in higher amounts of land allocated to low-input and low-returning crops such as q’at and eucalyptus.88
While WFP can do little to influence property rights of the land tenure system in Ethiopia, it will do well to take note of the relationship between tenure and agricultural technology adoption. The response of farmers to technology adoption based on their tenure situation is yet another example of people responding to rational incentives. If farmers are not somehow ensured access to the land and its outputs, they have no incentive to invest their time, money, or both, into what they perceive as risky technology, regardless of the output increases that may occur.
Commodity Market Access
New technologies often require repeated and consistent use of new inputs such as fertilizers and pesticides. Even low external-input sustainable agriculture (LEISA) activities usually demand significant amounts of construction materials for land preparation activities. If farmers are not secure in their access to these resources and the markets that provide them, adopting the technologies that require such inputs would place them at the mercy of supply streams. Having seen that the ability to bear risk decreases with poverty, the poorest farmers may need the greatest assurances that they will not be left without the inputs needed to sustain their families, and also earn extra income. But access to markets is needed also as an outlet for production, and not just as a means of securing inputs. Farmers need something to do with their increased output. If there are no markets that can bear the extra supply without creating a reactionary price decline, their investment in new agricultural technologies will be for naught.
Poor infrastructure in many developing nations results in inefficiencies and expensive cycles in the prices of commodities. Due to the lack of good transportation infrastructure and storage capacity, local markets are often flooded with agricultural commodities immediately following the harvest, and this drives down the unit price for each commodity. Poor storage means a large amount of output rots before it can be sold, leaving very little available for purchase in the months before the next harvest. Access to wider markets offers the possibility of increased food availability due to less spoilage and loss, higher profits for farmers because prices are not deflated due to the post-harvest flooding of local markets, and the minimization of commodity-price fluctuations. Studies often use a farmer’s distance to a major road as a simple proxy for commodity market access, and they show that the likelihood of a farmer to adopt an agricultural technology decreases with distance from a road.89, 90 Roads also imply the level of access farmers have to information. Studies suggest the likelihood that a farmer will continue using an agricultural technology is related to the frequency of contact with trained extension workers, especially for technically complex technologies,91 contact with neighboring farmers who possess knowledge of the proposed technology also increases the likelihood of adoption.92
The first step in creating a more successful agriculture development project is to collect comprehensive data on the characteristics associated with the highest rates of adoption. Good baseline data will let the project designers know what each farmer is able to bear: how much risk they can take, how many resources they can commit, and what their constraints are. Once the implementing agency understands the positions of their targeted communities, they can build the appropriate incentives into their projects to overcome the farmers’ limitations. Governments and aid professionals are next faced with a significant dilemma: it is often the farmers with the best indicators — the largest farms, the highest education, and the closest to major roads — who are most likely to adopt a technology and benefit from increased yields. If they focus on these “poorest of the poorest” farmers, development agencies would abandon the most vulnerable communities. But with enough data, agencies will know who are the “richest of the poorest,” “average of the poorest,” and “poorest of the poorest,” and take extra steps to bridge the gaps.
Farm Size and Tenure
Farm size may be the most important characteristic to measure because it can act as a proxy for many other wealth-related variables. Some agricultural technologies need a certain amount of land in order to be successful. Watershed rehabilitation and topsoil conservation projects work best on medium- to large-scale tracts of land because of the need to stabilize the surrounding environment from wind, animal and water degradation. Farmers with the largest parcels can afford to be more experimental because for them even a relatively small percentage of their total land may be large enough to support land-intensive agricultural technologies. That makes these farmers the most likely to not only adopt these large-scale projects, but to stay in for the life of the project because their extra land, and associated wealth, means they can weather small or medium shocks that may dissuade smaller farmers.
Large farmers are also good candidates for risky or experimental agricultural technologies. Their land resources means they can devote a relatively small percentage of arable land to a new technology, while still having enough buffer land to plant their regular crops and still be assured of those economic returns. As these farmers become more familiar with the new technology, it becomes less risky, and they can assign more land to its use. Their increased familiarity will convince neighboring farmers, whose smaller plots prevented them from participating at the beginning, to take up the new technology. Development project designers should seek out large farmers for these risky and experimental technologies, give them the technical assistance to be successful and to share their results with interested neighbors. In this way, farmers become the extension workers for the very projects that can increase incomes and provide food security.
Farmers with larger farms are more likely to adopt an agricultural technology, and also more likely to remain adopters. Yet because MERET targets the poorest communities, and is often applied first to communal lands, the effects of farm size on MERET adoption and participation are more nuanced. WFP should expect more resistance to full MERET adoption from communities allocating smaller shares of land for project activities, and it may be helpful to offer MERET activities to individual farms as well. An early focus on private land will address the negative relationship between tenure security and project participation, because offering services to private farms can convince a community to also accept operations on communal land, in which not all households feel an equal sense of ownership, or from which they may not draw as much income as other households.
Communities without secure tenure systems are less likely to participate in labor-intensive or capital-intensive projects if they are not guaranteed future access to the returns from those investments. But these communities are also often the most marginalized and vulnerable to shock-induced poverty traps. NGOs and international aid agencies can do little to influence the tenure laws in any given country, but these organizations often work in tandem with local government counterparts to implement their projects. Part of the project design can include convincing their local counterparts to take the appropriate steps to increase the legal claim to land. Another option is to make participation a pre-condition for tenure security. A farmer, or community of farmers, who lack recognized ownership of their land would put in their labor, capital, or some combination of those, into the rehabilitation and improvement of their land, and their government would “pay” them for this service with a formal deed. The farmer benefits from increased security. The government benefits from the increased production of the now-improved land, and from a community that does not rely on government assistance.
Risk Preferences and Human Capital
The adoption of a new agricultural technology carries risks and opportunity costs, and for many farmers uncertainty is a major obstacle, but measuring riss preferences is notoriously difficult, especially among poor farmers. Studies measuring the risk preferences of farmers are hardly uniform: they use different utility functions, sometimes in combination with strength of preference functions. Some studies extrapolate risk attitudes from farmers’ indifference to hypothetical lotteries while other studies infer risk attitudes from previous farmer behaviors.93 Much of the variation comes from a disagreement within the agricultural economics community regarding the shape of the utility function;94 however, most studies have one thing in common: they almost all measure the risk preferences of farmers in the developed world.
Aid professionals and development programs will certainly benefit from knowing the risk preferences of targeted communities. Aid agencies should approach the considerable lack of data on risk preferences among the rural poor as an opportunity for adding to the body of knowledge regarding decision-making under uncertainty. Aid agencies can fine-tune these approaches to gather data on poor households, incorporating risk assessments into the beginning stages of any project, and thereby learning if risk is consistent across wealth.
Risk is partly internal, but also influenced by the available resources. A person with more education is more confident in finding alternative income streams should a crop fail and is therefore more likely to participate in risky endeavors. The same principle applies to other human capital characteristics of dominance, such as gender, health, and social status. As we have seen with farm size characteristics, it is often those with the best human capital indicators that are most likely to adopt a technology and benefit from its returns. That risks shutting the most vulnerable — female-headed households, people with little or no education, the elderly, people living with HIV/AIDS, tuberculosis, malaria and other chronic diseases — out of the payoffs of new agricultural technology.
MERET tackles this issue by specifically targeting female-headed households, having mandatory female participation on project governance teams, providing extra FFW to people living with HIV/AIDS and tuberculosis, and encouraging school attendance for girls.95 Targeting those people with the lowest human capital reserves lowers their risk aversion, increasing their likelihood of new technology adoption, which in turn increases food security, income and allows for even more improvements to human capital.
Labor, Credit and Market Access
The labor demands of a new technology must match the labor availability of the targeted community. And it follows that labor-saving techniques will be adopted in areas of labor-shortage, but not in areas of labor-surplus. Similarly labor-intensive technologies will be more popular in communities with labor-surpluses, but not in areas with labor-shortages. In many rural communities labor availability is constrained by existing planting and harvesting cycles, therefore new labor-intensive technologies should not compete with these determined timetables. SRI adoption rates in Madagascar were low because the system required increased labor inputs at the exact time when many farmers were already engaged in other labor-intensive practices. For MERET this would imply that construction projects such as bunds, walls and fences should be scheduled either after harvest or after the sowing, and that introduction of new seeds and fertilizers be introduced during the sowing season.
Access to credit is a good way to overcome some of the financial obstacles to participation, especially in areas where interest rates are so high that they distort future income and make even large returns appear insignificant next to immediate cash. Payment for project participation, either in cash or in kind, is one method for bridging the gap between the need for immediate income and the security of increased financial returns in the future. Projects in areas of sufficient or surplus food production should opt for cash rather than in-kind payments, due to the negative market distortions that free food would cause.
Another option for bridging the credit gap is for aid organizations to provide cheap credit directly to their targeted communities. New financial institutions and project remunerations carry the risk of negative externalities, so aid agencies must carefully weigh the costs and benefits before actually giving money to poor farmers. As MERET is scaled up, WFP will be extending land rehabilitation and degraded land reclamation projects to communities less destitute than where the project currently operates. WFP should keep in mind the effects of food transfers among communities which can already afford food, and either replace FFW with vouchers or cash, or eliminate payments altogether if credit constraints are not a significant obstacle for farmer participation.
Agricultural development is about income generation. In the developing world this means increased production that does not compromise the future productivity of the natural resource base. Many poor communities lack not only access to the means of increased production, they also lack outlets for that increased production. Without external markets able to absorb increased production, excess crops flood the local market and drive down prices. While this is good for consumers, it is bad for producers and acts as a disincentive to produce more in the future, and this in turn acts as a stop against income generation.
Neill and Lee,96 among many, show that proximity to a road is highly correlated with likelihood to adopt a new technology because a road provides farmers with access to the inputs needed to make the technology work, such as fertilizer or pesticides, as well as access to a bigger village or city where they can sell their increased yields. Aid agencies need to provide assurances that farmers will have access to the required inputs should they participate in a new technology, as well as a ready market for their goods. This latter assurance can take the form of a loan to a cooperative of farmers that they can use to buy a truck and ship their crop to larger cities. This would provide both a route to commodity markets and diversify the local economy and create new jobs.
1 Timothy Besley and Anne Case, “Modeling Technology Adoption in Developing Countries,” in New Developments in Development, (Vol. 83, No. 2, May 1993), pg 396-402.
2 Ibid. pg 396.
3 Ibid. pg 396.
4 Ibid. pg 396.
5 Ibid. pg 397.
6 Ibid. pg 397.
7 Ibid. pg 397.
8 Ibid. pg 397-398.
9 Sean Neill and David R. Lee, “Explaining the Adoption and Disadoption of Sustainable Agriculture: The Case of Cover Crops in Northern Honduras,” in Economic Development and Cultural Change, (Iss. 49, 2001), pg 793-820.
10 Gershon Feder and Dina L. Umali, “The Adoption of Agricultural Innovations: A Review,” in Technological Forecasting and Social Change, (Iss. 43, 1993), pg 215-239.
11 The reader may be excused if they interpret the vocabulary gymnastics of the MERET acronym as a typical embodiment of the UN’s design-by-committee approach, but also should take comfort in knowing that meret is also the Amharic word for “land.”
12 Gershon Feder, Richard E. Just and David Zilberman, Adoption of Agricultural Innovation in Developing Countries: A Survey, (World Bank: Staff Working Papers No. 542, 1982), pg 3.
13 Ibid. pg 3.
14 Ibid. pg 3,4.
15 Charles K. Mann, “Packets of Practices: A Step at a Time with Clusters,” in Studies in Development, (Iss. 21, Autumn 1978), pg 73-81. quoted in Gershon Feder, Richard E. Just and David Zilberman, Adoption of Agricultural Innovation in Developing Countries: A Survey.
16 Feder, Just and Zilberman, pg 4.
17 GDP per capita estimate for Ethiopia (2009) is US$900, placing it 213th out of 227 countries surveyed. Source: CIA World Factbook, https://www.cia.gov/library/publications/the-world-factbook/rankorder/2004rank.html.
18 The World Bank estimates Ethiopia’s population (2009) at 82.8 million, placing it third behind Nigeria (154.7 million) and Egypt (83.0 million). Source: The World Bank, World Bank Development Indicators, http://data.worldbank.org/country/ethiopia.
19 United Nations Development Program: Human Development Report 2009, http://hdrstats.undp.org/en/countries/country_fact_sheets/cty_fs_ETH.html, 22 April 2010.
20 Ethiopia has averaged 3 percent population growth from 1970-2009; in that span the population almost tripled, from 28.9 million to 82.8 million. Source: The World Bank, World Bank Development Indicators, http://data.worldbank.org/country/ethiopia.
21 United Nations Development Program: Human Development Report 2009.
22 FAO/WFP “Crop and Food Security Assessment Mission to Ethiopia,” (July 2009).
23 The author’s experience with beneficiary and needs estimation is that political pressure from the federal Ethiopian government routinely revises down the WFP/FAO estimates beneficiary needs by several million.
24 WFP “Mid-Term Evaluation of the Ethiopia Country Program 10430.0, (2007-2011), 22 September 2009, pg 1.
25 Ibid. pg 1.
26 Woldeamlak Bewket, Community-based rehabilitation of degraded lands: an effective response to climate change in Ethiopia, (WFP Research Paper, November 2009), pg iv.
27 Report on the Cost-Benefit Analysis and Impact Evaluation of Soil and Water Conservation and Forestry Measures, (World Food Programme Report, 2005), pg 14.
28 Ibid. pg v.
29 Ibid. pg v.
30 Ibid. pg v.
31 Feder, Just and Zilberman, Adoption of Agricultural Innovation in Developing Countries: A Survey, (1982).
32 Besley and Case, pg 1.
33 Manfred Zeller, Aliou Diagne and Charles Mataya, Market Access by Smallholder Farmers in Malawi: Implications for Technology Adoption, Agricultural Productivity, and Crop Income, (International Food Policy Research Institute: FCND Discussion Paper No. 35, September 1997).
34 Neill and Lee, “Explaining the Adoption and Disadoption of Sustainable Agriculture: The Case of Cover Crops in Northern Honduras.”
35 Peter Arellanes and David R. Lee, “The Determinants of Adoption of Sustainable Agriculture Technologies: Evidence from the Hillsides of Honduras,” (Paper presented at XXV Conference of International Association of Agricultural Economists; August 2003; Durban, South Africa).
36 Keith O. Fuglie and Catherine A. Kascak, “Adoption and Diffusion of Natural-Resource-Conserving Agricultural Technology,” in Review of Agricultural Economics, (Vol. 23, No. 2, Autumn — Winter 2001) pg 386-403.
37 Akinwuni A. Adesina and Jojo Baidu-Forson, “Farmers’ perceptions and adoption of new agricultural technology: evidence from analysis in Burkina Faso and Guinea, West Africa,” in Agricultural Economics, (Iss. 13, 1995) pg 1-9.
38 Christine M. Moser and Christopher B. Barrett, “The Disappointing Adoption Dynamics of a Yield-Increasing, Low External-Input Technology: the Case of SRI in Madagascar,” in Agricultural Systems, (Vol. 76, 2003), pg 1085-1100.
39 Feder, Just and Zilberman, pg i.
40 Ibid. pg 25.
41 Gerald E. Shively, “Poverty, Consumption Risk, and Soil Conservation,” in Journal of Development Economics, (Vol. 65, Iss. 2, August 2001), pg 267-290.
42 Moser and Barrett, pg 1092.
43 Fuglie and Kascak, pg 386.
44 Ibid. pg 397.
45 Ibid. pg 401.
46 Ibid. pg 392.
47 Neill and Lee, pg 809.
48 Feder and Umali, pg 217.
49 Feder, Just and Zilberman, pg 29.
50 David R. Lee, “Agricultural Sustainability and Technology Adoptions: Issues and Policies for Developing Countries,” in American Journal of Agricultural Economics, (Iss. 87, November 5, 2005), pg 1325-1334.
51 Moser and Barrett, pg 1086.
52 S.T. Holden and B. Shiferaw, “Poverty and Land Degradation: Peasants’ Willingness to Pay to Sustain Land Productivity,” in Natural Resource Management in African Agriculture: Understanding and Improving Current Practices, (New York: CABI Publishing 2002), pg 91-101.
53 Paul Mosley and Arjan Verschoor, “Risk Attitudes and the ‘Vicious Circle of Poverty,’” in The European Journal of Development Research, (Vol. 17, No. 1, March 2005), pg 59-88.
54 Mosley and Verschoor, pg 83.
55 Stefan Dercon and Luc Christiaensen, Consumption Risk, Technology Adoption and Poverty Traps: Evidence from Ethiopia, (World Bank: Policy Research Working Paper 4257, June 2007), pg 1.
56 Frederick J. Zimmerman and Michael R. Carter, “Asset Smoothing, Consumption Smoothing and the Reproduction of Inequality Under Risk and Subsistence Constraints,” in Journal of Development Economics, (Iss. 71, No. 2, 2003), pg 233-260.
57 Finis Welch, “Education in Production,” in The Journal of Political Economy, (Vol. 78, No. 1, Jan-Feb 1970), pg 35-59.
58 Feder, Just and Zilberman, pg 32.
59 Fuglie and Kascak, pg 386-387.
60 Mariapia Mendola, “Agricultural technology adoption and poverty reduction: a propensity-score matching analysis for rural Bangladesh,” in Food Policy, (Vol. 32, 2007), pg 372-393.
61 Adesina and Baidu-Forson, pg 2, 5.
62 M.M. Zinnah, J. Lin Compton and A.A. Adesina, “Research-Extension-Farmer Linkages within the Context of the Generation, Transfer and Adoption of Improved Mangrove Swamp Rice Technology in West Africa,” in Quarterly Journal of International Agriculture, (Iss. 32, No. 2, 1993), pg 201-211.
63 Futoshi Yamauchi, Yisehac Yohannes and Agnes Quisumbing, Natural Disasters, Self-Insurance and Human Capital Investment: Evidence from Bangladesh, Ethiopia and Malawi, (World Bank: Policy Research Paper 4910, April 2009), pg 2.
64 Ibid. pg 4.
65 Temesgen Tadesse Deressa, Measuring the Economic Impact of Climate Change on Ethiopian Agriculture: Ricardian Approach, (World Bank: Policy Research Working Paper 4342, September 2007), pg 3.
66 Mendola, pg 391.
67 Report on the Cost-Benefit Analysis and Impact Evaluation of Soil and Water Conservation and Forestry Measures, (World Food Programme Report, 2005), pg 11.
68 United Nations Educational, Scientific and Cultural Organization (UNESCO) and UNESCO/UIS (UNESCO Institute of Statistics), quoted in UNICEF Statistics, http://www.unicef.org/infobycountry/ethiopia_statistics.html, 09 May 2010.
69 Feder, Just and Zilberman, pg 33.
70 Lee, pg 1328.
71 Shively, pg 284.
72 Bekele Shiferaw and Stein T. Holden, “Resource Degradation and Adoption of Land Conserving Technologies in the Ethiopian Highlands: A Case Study in Andit Tid, North Shewa,” in Agricultural Economics: the Journal of the International Association of Agricultural Economists, (Iss. 18, No. 3, 1998) pg 296.
73 Neill and Lee.
74 Rudulph A. Polson and Dunstan S. C. Spencer, “The Technology Adoption Process in Subsistence Agriculture: The Case of Cassava in Southwestern Nigeria,” in Agricultural Systems (Iss. 36, 1991), 65-78.
75 Ibid. pg 77.
76 Moser and Barrett, pg 1089.
77 Ibid. pg 1096.
78 Ibid. pg 1096.
79 Daniel Ayalew Ali, Stefan Dercon and Madhur Gautam, Property Rights in a Very Poor Country: Tenure Security and Investment in Ethiopia, (World Bank: Policy Research Working Paper 4363, September 2007), pg 3.
80 Klaus Deininger, Daniel Ayalew Ali and Tekie Alemu, Assessing the Functioning of Land Rental Markets in Ethiopia, (World Bank: Policy Research Working Paper 4442, December 2007), pg 4.
81 Ali, Dercon and Gautam, pg 4.
82 Ibid. pg 4.
83 Klaus Deininger, Daniel Ayalew Ali and Tekie Alemu, Impacts of Land Certification on Tenure Security, Investment, and Land Markets: Evidence from Ethiopia, (World Bank: Policy Research Working Paper 4764, October 2008), pg 13.
84 Ibid. pg 14.
85 Ibid. pg 15.
86 Deininger, Ali and Alemu, Assessing the Functioning of Land Rental Markets in Ethiopia, pg 2.
87 Ibid. pg 3.
88 Ali, Dercon and Gautam, pg 24.
89 Neill and Lee, pg 817.
90 Lee, pg 1327.
91 Moser and Barret, pg 1091.
92 Guerin and Guerin, pg 563.
93 Joost M.E. Pennings and Philip Garcia, “Measuring Producers’ Risk Preferences: A Global Risk-Attitude Construct,” in American Journal of Agricultural Economics, (Vol. 83, No. 4, Nov. 2001) 993-1009.
94 One of the most popular current theories challenging the standard theory of Expected Utility is Prospect Theory. For further reading, see: Daniel Kahneman and Amos Tversky, “Prospect Theory: an Analysis of Decision Under Risk,” in The Econometric Society, (Vol. 47, No. 2, March 1979) 263-291, Shlomo Benartzi and Richard Thaler, “Myopic Loss Aversion and the Equity Premium Puzzle,” in The Quarterly Journal of Economics, (Vol. 110, No. 1, Feb. 1995), 73-92, Matthew Rabin, “Inference by Believers in the Law of Small Numbers,” in Quarterly Journal of Economics, (Vol. 117, Iss. 3, Aug. 2002) 775-816, and Amos Tversky and Daniel Kahneman, “Belief in the Law of Small Numbers,” in Psychological Bulletin, (Vol. 76, No. 2, Aug. 1971) 105-110.
95 Report on the Cost-Benefit Analysis and Impact Evaluation of Soil and Water Conservation and Forestry Measures.
96 Neill and Lee, “Explaining the Adoption and Disadoption of Sustainable Agriculture: The Case of Cover Crops in Northern Honduras.”