You are on page 1of 8

Ecological Economics 69 (2010) 23942401

Contents lists available at ScienceDirect

Ecological Economics
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e c o l e c o n

Analysis

Farmers' preferences for crop variety traits: Lessons for on-farm conservation and
technology adoption
Sinakeh Asrat a,1, Mahmud Yesuf b,, Fredrik Carlsson c,2, Edilegnaw Wale d,3
a

International Food Policy Research Institute (IFPRI), P.O.Box 5689, Addis Ababa, Ethiopia
Kansas State University, 337B Waters Hall, Manhattan, KS 66506-6925, United States
Department of Economics, School of Business, Economics and Law, University of Gothenburg, P.O. Box 640, 405 30 Gothenburg, Sweden
d
Department of Agricultural Economics, School of Agricultural Sciences and Agribusiness, PBag X01 Scottsville 3209, Pietermaritzburg, South Africa
b
c

a r t i c l e

i n f o

Article history:
Received 19 January 2009
Received in revised form 12 June 2010
Accepted 6 July 2010
Available online 30 July 2010
Jel classication:
Q18
Q51
Q57
Keywords:
Crop biodiversity
Choice experiment
Crop variety
Random parameter logit

a b s t r a c t
Although in-situ conservation is increasingly considered an efcient way of conserving plant genetic
resources, little is known about the incentives and constraints that govern conservation decisions among
small farm holders in developing countries. Using a choice experiment approach, we investigated Ethiopian
farmers' crop variety preferences, estimated the mean willingness to pay for each crop variety attribute, and
identied household-specic and institutional factors that governed the preferences. We found that
environmental adaptability and yield stability are important attributes for farmers' choice of crop varieties.
Farmers are willing to forego some extra income or yield to obtain a more stable and environmentally
adaptable crop variety. Among other things, household resource endowments (particularly land holdings
and livestock ownership), years of farming experience, and contact with extension services are the major
factors causing household heterogeneity of crop variety preferences. Based on our experimental results, we
derived important policy implications for on-farm conservation, breeding priority setting, and improved
variety adoption in Ethiopia.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Farmers, plant breeders, gene-bank managers, and crop scientists
draw on diverse crop genetic resources to innovate, support, and benet
society at large (Smale, 2006). Biodiversity is an important component
of ecological systems (e.g., Heal, 2000; Tilman and Downing, 1994;
Tilman et al., 1996), and its loss can have adverse effects on the
functioning of these systems, including impairment of their capability to
produce (e.g., Loreau and Hector, 2001; Naeem et al., 1994). This is
typically the case in agriculture. Crop genetic resources are natural
assets that are renewable, but also vulnerable to losses from natural or
human interventions (including disruptions caused by droughts, oods,
or wars) and to the gradual process of social and economic changes. Loss
of diversity in local seeds, a major source of planting material, threatens
future breeding activities to the extent that those varieties are inputs to
develop varieties that are adaptable to biotic and a-biotic stress factors

Corresponding author. Tel./fax: + 1 785 532 6925.


E-mail addresses: sinak12@yahoo.com (S. Asrat), myesuf@ksu.edu (M. Yesuf),
Fredrik.carlsson@economics.gu.se (F. Carlsson), walee@ukzn.ac.za (E. Wale).
1
Tel.: + 251 11 6172500.
2
Tel.: + 46 31786 4174.
3
Tel.: + 27 33 260 5410.
0921-8009/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecolecon.2010.07.006

(e.g. moisture, temperature, disease, pests, and soil quality). It also


reduces the available pool of genetic materials for breeding to enhance
productivity and ensure environmental stability. A number of economic
studies have also noted that the diversity of crop varieties can boost
agricultural productivity (Di Falco and Chavas, 2006; Smale et al., 1998).
The main challenge faced by policy makers in developing countries is
how best to conserve crop genetic diversity while fullling the growing
demand for food production and ecological services. In the literature,
maintaining diverse plant varieties in farmers' elds (now known as insitu conservation) is increasingly considered an effective means of
conserving plant genetic resources (Benin et al., 2003). However, onfarm conservation of crop diversity, a sub-set of in-situ conservation,
poses obvious policy challenges in terms of the design of appropriate
incentive mechanisms and possible trade-offs between conservation
and productivity. Maintaining traditional varieties of crops on farmers'
elds also has opportunity costs (Wale, 2008). Smale et al. (2003) also
noted one fundamental problem that affects the design of policies
meant to encourage on-farm conservation: crop genetic diversity is an
impure public good, meaning that it has both private and public
economic attributes. Farmers maintain traditional varieties only to the
extent that the varieties fulll their private demands. The public benets
beyond farmers' concerns are under-valued. Due to the quasi-public
good nature of traditional crop varieties, on-farm conservation outcomes could still be sub-optimal. Thus, farmers' own conservation

S. Asrat et al. / Ecological Economics 69 (2010) 23942401

activities, also called de facto conservation, should still be supplemented


by public conservation efforts.
One of the attractive features of in-situ conservation is its ability to
maintain evolutionary processes and farmers' indigenous knowledge
that cannot be protected in gene banks. However, even if in-situ
conservation is believed to be an effective means of conserving crop
genetic resources, the degree of success is highly dependent on
individual farmers' decisions, which, as noted above are mainly
governed by traditional varieties addressing farmers' private concerns. Thus, it is important to understand the farm-level incentives
and constraints for in-situ conservation. This study contributes to the
literature by providing insights into farmers' preferences for crop
variety attributes, using a choice experiment approach in a typical
developing country setting Ethiopia. Choice experiments are hardly
applied in crop diversity studies. There are several reasons for using a
survey-based approach instead of relying on actual behavior. The
most important is the absence of well-functioning markets for crop
varieties (seeds) and their attributes due to high transaction costs,
limited information and other forms of market failure in developing
countries including Ethiopia. For this reason, the market fails to value
the desirable traits of farmers' varieties that are stored and exchanged
more informally through local networks and value systems. Due to the
lack of market data, Jabbar (1998), for example, used a hedonic
pricing method to attach values to the different attributes of animal
genetic resources in southern Nigeria.
The rest of the paper is organized as follows. Section 2 provides
background on the current state of crop biodiversity and crop
production in Ethiopia, with special reference to the two major cereals
in the country: teff, and sorghum. Section 3 describes the details of the
choice experiment design and administration of the survey. Section 4
presents the econometric approach, and Section 5 discusses the
empirical results. Finally, Section 6 concludes the paper.
2. Crop Biodiversity and Agricultural Production in Ethiopia
As in many other developing countries, agriculture is the mainstay
of the Ethiopian economy, accounting for 85% of all employment, 40%
of gross domestic product (GDP), and 90% of export earnings.
However, the agricultural GDP and per capita cereal production
have been declining over the last 40 years, with a cereal yield that has
been stagnant at only 1.2 metric tons per hectare (World Bank, 2005).
Despite huge investments and extension programs to promote
improved seeds, the use of improved seeds4 is very lowonly 35%
of Ethiopia's cultivated agricultural area is covered with improved
seedsleaving a great proportion of the farm households to depend on
traditional varieties (World Bank, 2005). The low rate of adoption of
improved varieties is often attributed to a number of socio-economic,
natural/environmental, and institutional factors (Bezabih, 2003;
Degnet et al., 2003; Legesse, 2003; Chilot et al., 1996; Kidane and
Abler, 1994; Yohannes et al., 1990). Wale and Yallew (2007) have
argued that the lack of tness of variety technology attributes to
farmers' needs and circumstances is the major factor hampering
technology adoption in Ethiopia.
Given its agro-ecological diversity and high altitudes, Ethiopia is
both the center of origin and a center of diversity for many crops,
including sorghum, teff (Eragrostis abyssinica), coffee (Arabica), and
ensete (Ensete ventricosum). Sorghum and teff are the two major
cereals grown in the country, with teff being the primary staple food
crop, occupying 22% of all cultivable land (CSA, 2008). As a source of
staple food for many parts of the country, teff is primarily grown to
prepare injera (Ethiopian bread), porridge, and some native alcoholic
4
Improved seeds are genetically uniform crop varieties channeled to farming
communities through agricultural extension packages after developed either by formal
public breeding programs or imported from abroad. Most of the improved seeds are
designed mainly to increase agriculturalproductivity and improve food security.

2395

drinks. The straw is mainly used for animal feed. Sorghum, the major
crop second to teff and grown all over the country, contributes about
1520% of Ethiopia's total cereal production (CSA, 2008). The crop is
used for many purposes, such as food, animal feed, fuel, house
construction, and fences. Ethiopia holds ex-situ 4% of the world's
sorghum genetic stock (FAO, 1998 and Hawkes et al., 2000). The crop
exists in tremendous variety throughout the areas of sorghum
production in Ethiopia (Gebrekidan, 1979; de Wet and Harlan,
1971). However, this diversity is potentially threatened by the
dissemination of genetically uniform varieties of improved varieties
channeled to farming communities after being developed by formal
public breeding programs. This has already been the case, for instance,
for wheat in the central highlands of the country where widespread
adoption of improved wheat varieties has resulted in crop diversity
loss (Yifru and Hammer, 2006).
The National Biodiversity Strategy and Action Plan (NBSAP) of
Ethiopia claims that agricultural intensication is potentially the major
cause of loss of agricultural biodiversity in the world, and particularly in
Ethiopia (FDRE, 2005). The report argues that replacing traditional crop
varieties with high-yielding varieties that are dependent on high levels
of agricultural inputs can result in genetic erosion of resilient native
varieties. In Ethiopia, the loss is sometimes due to displacement of the
food crops and their traditional varieties by other more rewarding cash
crops such as khat (Catha Edulis) (Wale, 2004). The trade-offs (positive
productivity outcomes and loss in traditional varieties) have to do with
the step-by-step, cumulative and invisible loss of traditional varieties of
crops due to their displacement by uniform improved seeds and/or
other cash crops (Wale et al., 2009). It is, thus, both a challenge and an
opportunity for Ethiopia to design conservation policies that enable its
agriculture-based economy to make the best use of its crop diversity. As
said earlier, a loss of crop diversity implies a big threat to the livelihoods
of millions of smallholders who depend on local seeds as their major
source of planting material. This loss is also a threat to future prospects
of crop variety development as the portfolio of traditional varieties of
crops are the fundamental inputs for crop breeding that are adaptable to
local and harsh climate and soil conditions (Wale, 2004). Thus,
sustainable agriculture will need concerted public effort to ensure the
continuous survival of traditional varieties of crops along with farmers'
indigenous knowledge that evolves with them. This calls for public
conservation initiatives (such as on-farm conservation) that require
designing incentive mechanisms based, among other things, on
opportunity costs farmers face when maintaining traditional varieties
(Wale, 2008). To this effect, understanding farmers' preferences and
driving forces behind crop variety choices will remain to be crucial. The
remaining parts of this paper will deal with the research methodology
applied, the results of the exercise and their implications.
3. Survey Design and Study Sites
In this study, we employed a choice experiment approach to
evaluate farmers' preferences for the various attributes of crop
varieties. In a choice experiment, individuals are given a hypothetical
setting, and then asked to choose their preferred alternative (usually
repeatedly) from several alternatives in a choice set. Each alternative
is described by a number of attributes that take on different levels.5 In
our case, the farm households were given choice sets with three
different alternative crop varieties. The most important crop variety
attributes and their levels were identied in consultation with experts
(crop breeders and researchers with hands-on experience and
practical knowledge of the relevant variety attributes), by reviewing

5
For detailed reviews on the choice experiment method, see Alpizar et al. (2003)
and Louviere et al. (2000), for example. Although, applications of choice experiments
to crop diversity are limited, there are some applications of this technique to livestock
genetic resources (see for example Scarpa et al., 2003a,b).

2396

S. Asrat et al. / Ecological Economics 69 (2010) 23942401

Table 1
Attributes and attribute levels used in the choice experiment.
Attributes
Sorghum
Producer's price
Productivity
Environmental adaptability
Yield stability

Teff
Producer's price
Productivity
Environmental adaptability
Yield stability

Denition

Attribute levelsa

The amount of money the farmer earns by selling 100 kg of harvested


sorghum of a particular sorghum variety
Average production (in 100 kg) harvested per hectare from planting a
particular sorghum variety
Whether the variety is resistant or tolerant to environmental stress
factors such as poor soil, poor rainfall, and frost
Whether the variety gives stable yield year-after-year, despite occurrences
of crop disease and pest problems in a scenario of no drought and frost

(1) ETB 110, (2) ETB 150, (3) ETB 200

The amount of money the farmer earns by selling 100 kg of harvested teff
of a particular teff variety
Average production (in 100 kg)harvested per hectare from planting a
particular teff variety
Whether the variety is resistant or tolerant to environmental stress factors
such as poor soil, poor rainfall, and frost
Whether the variety gives stable yield year-after-year, despite occurrences
of crop disease and pest problems in a scenario of no drought and frost

(1) 210 ETB, (2) 270 ETB, (3) 330 ETB

(1) 14, (2) 19 (3) 25


(1) The variety is adaptable (resistant); (2) the variety is not
adaptable (nonresistant)
(1) The variety gives stable yield year-after-year; (2) the variety
gives variable yield year-after-year

(1) 8, (2) 15 (3) 20


(1) The variety is adaptable (resistant); (2) the variety is not
adaptable (nonresistant)
(1) The variety gives stable yield year-after-year; (2) the variety
gives variable yield year-after-year

Note: ETB = Ethiopian birr; ETB 8.93 = US$ 1 at the time of the experiment (June/July 2007).
a
The levels for these attributes are set based on the Zone's minimum, average, and maximum values of producers' price and productivity of the crops during the last decade.

previous studies and historical data, and by identifying the most


important seed selection criteria put forward by a focus group of
surveyed households. This is explained in more detail towards the end
of this section. The variety traits are identied based on historical facts
and with carefully crafted focus group discussions that have involved
eld extension workers, experienced farmers, village leaders, agricultural researchers working in the areas (mainly crop breeders),
district agricultural ofcers etc. The principal investigator was
involved in this exercise. The expert consultations and intensive
discussions made with farmers reveal that yield levels, yield stability,
environmental adaptability, and selling unit price are the most
important crop variety attributes for these food crops. Based on this,
the experiment was conducted using these variety attributes and for
the two major cereals grown in Ethiopia, sorghum and teff. A full
description of the attributes and the levels of each attribute are
presented in Table 1.
In our sites, as in many developing country settings, where
production decisions are mainly subsistent-oriented and production
and consumption decisions are non-separable, productivity attributes
become important considerations for all farm households (netbuyers, net-sellers, and the self-reliant). On the other hand, the
price attribute is most relevant for farm households with better access
to output markets and those selling most of their agricultural
products. Thus, as we shall conrm or disprove in Section 4, we
expect that price and productivity attributes play a distinct role in
governing farm households' preferences for different crop varieties
and in designing appropriate incentive-based policies.
Given the attributes and their levels as presented in Table 1,
different choice sets were constructed using a cyclical and fractional
main effect design principle (Bunch et al., 1996).6 Each household
made nine choices: there were three alternatives in each choice set.
Separate choice sets were presented for teff and sorghum varieties.
Table 2 presents an example of a choice set for sorghum.
Thus, the farmer had to choose between one of the alternatives,
and there was no possibility of opting out. There are several reasons
for why we did not include an opt-out alternative. First, we are not

A cyclical design is a straightforward extension of the orthogonal approach. First,


each of the alternatives from a fractional factorial design is allocated to different choice
sets. Attributes of the additional alternatives are then constructed by cyclically adding
alternatives into the choice set, based on the attribute levels. The attribute level in the
new alternative is the next higher attribute level to the one applied in the previous
alternative. If the highest level is attained, the attribute level is set to its lowest level.

going to try to measure the purchase of different varieties, instead our


focus is on estimating the marginal willingness to pay for the
attributes of the variety, and, therefore, it is not necessary to include
an opt-out alternative (Carlsson et al., 2007). Second, in this situation
farmers have to buy some variety and therefore forcing them to make
a choice is not putting them in an awkward situation. The main
drawback of not including an opt-out in this setting is if subjects that
would have preferred the opt-out alternative are making different
trade-offs between the attributes than the other subjects.
We carried out the experiment in two peasant associations (PAs)7
in the northeastern part of the country (the North Wollo Zone of the
Amhara Regional State) in June and July of 2007, and adopted
stratied multi-stage sampling to identify zones, districts, PAs,
villages, and farm households. All sampled households are located
in two PAs of the Guba Lafto district, North Wollo, Amhara. The
villages are located in temperate agro-ecology and have an average
annual rainfall of 630970 mm per year and a mean temperature of
1520 C. In addition, they have experienced recurrent drought over
the past decades. Teff, sorghum, and maize are among the most
important food crops in both study sites.
To check the relevance of the choice experiment questions in the
context of the local conditions, farmers' expectations, and level of
understanding, the questionnaires were pre-tested on 16 farmers (8
from each PA). The pre-test results were discussed with the
enumerators and necessary changes were made following farmers'
responses. In total, 131 farmers were selected. Of these, 66 were
randomly selected and presented with choice sets for the sorghum
variety and the remaining 65 were presented with the teff choice sets.
Enumerators used the local language and choice cards to present the
various choice sets. Overall, 1179 choices were elicited from the
surveyed households. To complement the experimental data, a
separate survey was conducted to collect data on socio-economic
characteristics. Table 3 reports the basic descriptive statistics of the
explanatory variables considered in this paper.
4. The Econometric Approach
Since farmers' preferences are observed in terms of their choices,
we employed a random utility framework to analyze the responses to
the different choice sets (McFadden, 1974). Assuming a linear indirect
7
A peasant association is the smallest representation of social units in rural Ethiopia
often comprised of 400 to 500 peasants.

S. Asrat et al. / Ecological Economics 69 (2010) 23942401


Table 2
Example of a choice situation.
Assuming that the following sorghum varieties were your ONLY choices, which
one would you prefer to plant?
Sorghum variety
characteristics

Sorghum
variety 1

Sorghum
variety 2

Sorghum
variety 3

Producer's price (for 100 kg)


150
200
110
Productivity per ha (in 100 kg)
14
19
25
Environmentally adaptable
Yes
No
Yes
Stable-in-yield
No
Yes
No
I would prefer to plant sorghum variety 1, sorghum variety 2, or sorghum variety 3.
(Please tick () one option.)

utility function, the utility for alternative i in choice situation t for


farmer h is given by:
Vith = Ait + priceit + ith ;

where Ait is the attribute vector, (not including the price attribute),
is the corresponding parameter vector including an alternative
specic constant, priceit is the price attribute for alternative i, is
the marginal utility of money, and ith is an error term. The probability
that individual h will choose alternative i can be expressed as:
n
o
Pith = P Ait + priceit + ith N ajt + pricejt + jith ;ji :

From this specication, the mean marginal WTP (willingness to


pay) for a certain attribute is given by the ratio of the attribute
coefcient to the marginal utility of income (Hanneman, 1984). In the
analysis, we employed a random parameter logit model where the
non-monetary attributes are random normally distributed. We can
then write utility as:

h Aith + ith : 3
Vith = priceit + Aith + ith = priceit + Aith +
Thus, each individual's coefcient vector, , is the sum of the
h . The stochastic part of
population mean and individual deviation
h Aith + ith , is correlated over alternatives, which means that
utility,
the model does not exhibit the IIA (independent of irrelevant
alternatives) property. This is the rst reason for choosing a random
parameter model. The second reason is that we want to capture
unobserved heterogeneity. Let tastes, , vary in the population with a
distribution with density f j, where is a vector of the true
parameters of the taste distribution. If the 0 s are IID (independent and
identically distributed) type-I extreme value, we have a random
parameter logit (RPL) model. We assume that the randomly distributed
parameters are constant across the choice situations for each individual.
This reects an underlying assumption of a stable preference structure
for all individuals over the choice experiment (Train, 1998). Since the
choice experiment is relatively small and simple, this seems to be a
realistic assumption. The assumption of a xed price coefcient is made
for two reasons. First, keeping at least one parameter facilitates
estimation. Second, the distribution of the marginal WTP is then simply
the distribution of the random parameter.
First, we estimated one model for each crop with only the attributes
of the experiment. Then, we estimated two models, where a number of
socio-economic characteristics interact with the attributes in order to
capture observed heterogeneity. The models are estimated with
simulated maximum likelihood, using Halton draws with 500 replications.8 Although the experiment was generic, we included alternative
specic constants that are allowed to vary with the choice sets, since we
wanted to test whether any factors other than the attributes themselves
affected the choices. We have tested the hypothesis of different price
coefcients for farmers with different income levels, but in all cases we
8

See Train (2003) for details on simulated maximum likelihood and Halton draws.

2397

could not reject the hypothesis of equal coefcients. We therefore


estimate the models restricting the price coefcient to be constant
across subjects.
5. Results and Discussions
The results of the RPL models (with and without socio-economic
characteristics) are presented in Tables 4 and 6, respectively. Note
that the producer price variable is expressed in 100 birr, and the
productivity attribute in 100 kg. We begin by commenting on the
results of the models without socio-economic characteristics.
All the attribute parameters are highly signicant and have the
expected signs. Thus, farmers care not only about the productivity of
the crop varieties but also about environmental adaptability and yield
stability. The estimated standard deviations are also signicant and
sizeable, indicating that we captured unobserved heterogeneity with
the random parameter specication. The alternative specic constants
are also signicant. Since the experiment was generic, this indicates
that factors other than attribute levels affected behavior. As can be
seen this behavior is less prevalent for the latter choice sets in the
experiment since the coefcients of the alternative specic constants
are larger for the rst three choice sets. With this caveat in mind, we
move on to investigate the marginal WTP for the attributes. We report
two sets of estimates of marginal WTP. The rst is the unconditional
marginal WTP, which is simply the ratio of the respective attribute
coefcients to the price coefcient (Hanneman, 1984). The second is
the individual level marginal WTP. The individual level WTP is
obtained from the individual level parameter estimates that are based
upon the actual respondent choices (see e.g. Train, 2003). Table 5
presents the results. Note that the attributes for environmental
adaptability and yield stability are binary variables, and hence they
can be directly compared. The marginal WTP for the productivity trait
is calculated as the WTP for an increase in productivity by 100 kg per
hectare. The WTP levels for productivity is well within the bounds of
producers' price of 100 kg of teff (ETB 210-300) or sorghum (ETB 110200) showing our elicited WTP estimates are reasonable. The individual
level mean estimates are similar to the conditional (population) means.
The farmers in the experiment had a higher WTP for environmental adaptability than for yield stability. This is consistent with
farmers' ranking of seed selection criteria in Appendix A.1 where only
about 11% and 24% of the farm households indicated productivity and
marketability as their major seed selection criteria for sorghum and
teff, respectively. This compared to more than 60% of the farm
households that indicated factors related to environmental adaptability and yield stability as their major preferred criteria to select
their seed varieties for both crops. One reason for this could be the
frequent droughts and other environmental stress factors such as soil
degradation being experienced by most farmers over the last several
decades. Comparing productivity with the adaptability and stability
attributes, we see that there is strong preference for adaptability and
stability. For example, for teff, increasing yield stability is equivalent to
increasing productivity by 1300 kg per hectare. For sorghum, the
preferences are even stronger mainly because this is a crop common
in environmental stress areas. The high WTP for environmental
adaptability and yield stability, compared to productivity, points to
farmers' strong preference for resilient crop varieties. In countries like
Ethiopia, where crop production is mainly rain-fed and often subject
to various natural calamities, production risk is an important
consideration when making production decisions. The ever increasing
deterioration of the natural environment to support crop production
and lack of resources and technology for farmers to address
production shocks have increased the importance of environmentally
adaptable and yield stabile traits. If one is to consider the lack of
harmony of technology traits with farmers' concerns this is the major
reason for precarious adoption of improved varieties (mainly highyielding varieties) which are generally believed to be less resilient to

2398

S. Asrat et al. / Ecological Economics 69 (2010) 23942401

Table 3
Descriptive statistics of sampled farm households.
Variable

Description

Household characteristics
Male
Household size
Experience
Off-farm work
No. of dependents
Agricultural output surplus
Drought frequency

= 1 if the household head is male and zero other wise


Number of household members who share the same food stock
Farming experience of the household head in years
= 1 if at least one member works off-farm and zero other wise
Number of dependents with no labor or money contribution in the household
= 1 if the household is a net-seller of agricultural outputs and zero other wise
The number of times the household faced drought problems during the last ten years

Farm and livestock characteristics


Land shortage is major problem

Mean

= 1 if the household head considers land shortage to be the primary problem and
zero other wise
Total land size operated by the household
Total land size per household member
Total value of livestock (including poultry and bee hives) currently owned by the
household

Total land size (in hectares)


Land size per capita
Livestock value (in ETB)

Access to infrastructure and extension services


Average distance to household servicesa (in minutes)
Participate in extension programs
Experience in extension programs

0.90
5.38
25.38
0.32
1.15
0.27
2.45

Std. dev.

2.04
11.64
1.45
1.709

0.64
0.75
0.15
5016.5

Average walking distance to basic infrastructure and services


= 1 if the household has been participating in the agricultural extension program
and zero other wise
Years of participation in agricultural extension program

48.24
0.70
4.14

0.52
0.088
4745.5

27.07

5.226

Services include electricity, piped water, telephone, primary school, secondary school, all weather roads, and irrigation. Respondents were asked to specify the walking distance
(in minutes) to each type of service, and an average walking distance to services was then calculated for each respondent.

environmental hardships (Wale and Yallew, 2007). This result has


important implications not only for in-situ conservation of crop
genetic resources but also for crop varietal technology adoption. It
demonstrates how important the environmental adaptability and
yield stability attributes are in motivating farmers to participate in any
in-situ conservation effort. It also points that these traits have to get
enough attention in future breeding activities. Comparing sorghum
and teff, the results show that farmers attach stronger WTP for
productivity, environmental adaptability and yield stability traits of
teff than the respective traits in sorghum.
Table 4
Results of random parameter logit estimates for choice of crop variety, standard errors
in parentheses.

Mean parameters
Alternative 1 (Set 13)

Sorghum

Teff

Coefcient

Coefcient

Alternative 2 (Set 13)

0.795
(0.409)
0.062
(0.230)
2.012

0.565
(0.11)
0.485
(0.041)
1.67

Alternative 2 (Set 49)

(0.736)
1.276

(0.635)
0.906

Environmental adaptability

(0.651)
1.892
(0.291)
0.321
(0.039)
5.537

(0.534)
1.115
(0.221)
0.322
(0.042)
6.297

Yield stability

(0.818)
4.839

(1.053)
4.231

(0.903)

(0.934)

0.149
(0.033)
3.269

0.172
(0.028)
4.469

(0.613)
2.892
(0.630)
66
594
0.578

(0.910)
2.858
(0.589)
65
585
0.557

Alternative 1 (Set 49)

Producer's price in 100 birr


Productivity in 100 kg

Standard deviation parameters


Productivity
Environmental adaptability
Yield stability
Number of respondents
Number of choices
Pseudo R2

***, **, and * denote signicance at 1%, 5%, and 10%, respectively.

It is likely that there are large heterogeneities of preferences across


farm households. Any in-situ conservation effort should take these
heterogeneities into consideration. Additional information on the
factors contributing to the private value that farmers assign to crop
varieties may help to identify a strategy for ensuring the conservation of
the crop genetic resources that are embodied in landraces while at the
same time minimizing the costs (Brush and Meng, 1998). To account for
observed heterogeneity of preferences across farm households, we also
estimated models where a set of socio-economic characteristics were
interacted with the attributes. Due to potential multicollinearity
problems, it was not possible to include all interactions between the
explanatory variables and the four crop variety attributes. The results of
the RPL model are presented in Table A2 in the Appendix. What we will
focus on here is instead the estimated marginal WTPs. A number of
comparisons can be made, but we focus on two. For the binary
household variables, we present the marginal WTP for each category of
the variables. Table 6 presents the results of the marginal WTP estimates
for all the binary variables. For simplicity, we only report the
unconditional WTP estimates i.e. WTP is estimated as the ratio of the
respective attribute coefcients to the price coefcient.
The results show that highly productive teff varieties are valued the
most by female-headed households and net-sellers of agricultural output
while productive sorghum varieties are valued highly by net-buyers of
agricultural output. This suggests that net-sellers are more interested in
the marketable crop, teff, while sorghum, which is less marketable but
yields on average more harvest per hectare, is preferred by net-buyers.
This is also reected in their preference for environmental adaptability
Table 5
Mean marginal willingness to pay (in Ethiopian Birr) for each variety attribute by crop,
standard errors in parentheses.
Sorghum

Productivity
Environmental
adaptability
Yield stability

Teff

Unconditional

Individual
level

Unconditional

Individual
level

16.97
(2.76)
292.59
(51.06)
255.75

17.49
(0.64)
293.52
(16.41)
249.86

28.88
(5.79)
564.84
(124.07)
379.51

27.85
(1.49)
554.92
(39.85)
360.04

(56.32)

(13.66)

(104.26)

(23.42)

***, **, and * denote signicance at 1%, 5%, and 10%, respectively.

S. Asrat et al. / Ecological Economics 69 (2010) 23942401


Table 6
Marginal WTP for different household types, standard errors in parentheses.
Sorghum
Productivity

Env. adaptability

Yield stability

Female
Male
No off-farm
Off-farm
Net buyer of
agricultural output
Net-seller of
agricultural output
Female
Male
No off-farm
Off-farm
Net buyer of
agricultural output
Net-seller of
agricultural output
Female
Male
No off-farm
Off-farm
Net buyer of
agricultural output
Net-seller of
agricultural output

Teff

11.78 (5.00)
20.17 (3.59)
17.21 (3.31)
25.13 (5.30)
19.44 (3.62)

35.85
29.04
29.89
28.47
27.65

18.87 (4.31)

33.04 (7.85)

(15.19)
(5.79)
(6.25)
(7.12)
(5.88)

196.86 (97.99)
320.02 (61.85)
317.43 (65.11)
277.99 (73.39)
316.72 (63.39)

274.89
702.17
695.87
652.55
641.90

280.00 (76.84)

761.61 (239.77)

252.91 (114.78)
260.07 (66.76)
258.11 (69.65)
262.59 (84.50)
257.14 (68.86)

722.96
493.25
465.41
576.45
503.81

265.30 (87.09)

(308.07)
(198.93)
(192.00)
(248.43)
(190.07)

(353.03)
(178.12)
(173.60)
(235.37)
(186.62)

507.30 (213.13)

***, **, and * denote signicance at 1%, 5%, and 10%, respectively.

trait for the two crops, where net-buyers indicated less preference
towards a non adaptable sorghum variety than net-sellers, and the
opposite holds for the same trait in teff. Farmers with no off-farm job
opportunities value environmentally adaptable sorghum and teff
varieties more than what farmers with off-farm job opportunities do.
This is mainly because farm households with no members working offfarm rely highly on their agricultural produce for food and ber and are
more likely to be highly risk-averse towards choosing crop varieties that
are less resistant to environmental stress factors such as poor soil and
poor rainfall. There is also a signicant difference between male- and
female-headed households in terms of their WTP for the different traits of
teff and sorghum, where male-headed households show more preference
to environmentally adoptable traits for both teff and sorghum than
female-headed households.
Next we present the WTP estimates for the continuous household
variables. In order to compare across characteristics, we estimate the
changes in WTP due to a standard deviation change in the household
variables (Table 7).
These results reveal that differences among farm households in
terms of household characteristics, endowments, and degree of access
Table 7
Change in WTP for a standard deviation change in household variables.
Sorghum

Teff

Productivity

Experience
3.09 (1.92)
1.44 (2.93)
Total land per capita
2.19 (1.85)
12.96 (4.76)
Livestock value
0.12 (2.16)
1.68 (2.32)
Drought frequency
1.02 (1.86)
8.49 (3.31)
Expr. agric. extension
3.11 (1.81)
2.79 (2.52)
Household size
0.59 (2.05)
2.75 (3.87)
Env. adaptability Experience
67.40 (39.56)
97.62 (62.42)
Total land per capita 84.39 (38.03)
72.09 (95.58)
Livestock value
6.22 (42.98) 181.30 (76.34)
Drought frequency
5.42 (35.32)
57.87 (63.57)
Expr. agric. extension
9.51 (28.53)
264.62 (152.20)
Household size
20.40 (41.75)
256.57 (152.27)
Yield stability
Experience
19.33 (40.16)
11.68 (56.57)
Total land per capita
25.49 (37.89)
96.69 (96.66)
Livestock value
61.17 (46.89) 127.41 (66.01)
Drought frequency
13.24 (37.55)
97.60 (66.29)
Expr. agric. extension 70.38 (32.77)
265.57 (157.34)
Household size
26.97 (44.34)
90.84 (138.58)
***, **, and * denote signicance at 1%, 5%, and 10%, respectively.

2399

to agricultural extension do affect farmers' private valuation of crop


variety traits. The results show that the highly productive teff varieties
are valued the most by households with larger per capita land
holdings and farm households with more drought experience, and
less by medium-sized farm households and those with moderate
experience with drought. The demand for environmental adaptability
also varies across crops. The preference for environmentally adaptable
sorghum varieties is stronger for households with more years of
farming experience and is signicantly weaker for households with
larger per capita land holdings. Farm households owning more
livestock and small families and households with less experience in
agricultural extension had a weaker preference for environmentally
adaptable teff varieties. The results also show that those farm
households that have participated in the agricultural extension for a
long time derive the lowest positive utility from more stable sorghum
varieties. On the other hand, farm households owning more livestock
and less experience with agricultural extension package derive the
lowest positive utility from more stable yielding teff varieties.
In general, there appears to be heterogeneity in preferences for
crop varieties, especially across large and smallholder farmers and
farmers with high and low values of livestock. Larger farm households
and households with a high value of livestock gave more weight to
productivity than to environmental and yield stability attributes than
did smallholders and households with a low value of livestock. This
result is very intuitive, in that larger farm households are less riskaverse and are ready to adopt crop varieties that are more productive
but less resilient and less stable in terms of yield. Households with a
higher value of livestock also prefer more productive crop genetic
varieties. Their livestock asset is giving them the leverage to make
riskier production decisions in terms of choice of crop varieties.
6. Conclusions and Policy Implications
Given the growing concern for food insecurity and adverse effects
of long-term climate changes, crop biodiversity is an important asset
for both increasing agricultural productivity and minimizing the
downside risk of adverse climate change. However, due to the lack of
appropriate incentive mechanisms to implement in-situ conservation
strategies, crop biodiversity is subject to irreversible losses. This is a
recurrent phenomenon as development interventions (meant to
address productivity and growth) hardly internalize their external
impacts on crop diversity. If policy makers are to supplement their exsitu conservation efforts with in-situ conservation strategies, understanding of farmers' incentives and constraints on their choice of crop
varieties is of paramount importance. This study employed a choice
experiment approach to investigate farmers' preferences for crop
variety attributes and identify the most important socio-economic
forces driving these preferences. Farmers revealed strong preferences
for environmental adaptability for both teff and sorghum. Yield
stability was also more important than increased productivity. These
ndings may explain the low adoption rates of high-yield variety
seeds in Ethiopia over the last several decades. The fact that farmers
attach sizeable values to both environmental adaptability and yield
stability traits of sorghum and teff points to the need for supplying a
crop genetic variety with additional attributes of resilience to harsh
environmental conditions, rather than a breeding strategy that solely
targets enhanced agricultural productivity.
The results also reveal that there are signicant differences among
farm households in their private valuations of crop variety traits based
upon household characteristics, resource endowments, and level of
access to agricultural extension. There are signicant differences
between smaller and larger farmers, and between households with
low and high values of livestock.
These results have important implications for on-farm conservation, contextual variety development, breeding priority setting, and
targeted diffusion of improved varieties in Ethiopia.

2400

S. Asrat et al. / Ecological Economics 69 (2010) 23942401

First, the farm households, which now attach the highest values to
attributes already embedded in traditional crop varieties would
maintain those varieties de facto. To enhance compliance, these farmers
have to be targeted in future on-farm conservation activities. This,
however, does not imply that where de facto conservation is occurring,
there is no need to design external (conservation) incentives for these
varieties. Weitzman (1993), for instance, has shown that it might still be
desirable to assign conservation priorities to species/varieties not yet at
risk. Ensuring that a particularly unique species/variety never becomes
threatened may, therefore, require external incentives.
Second, understanding farmers' varietytrait preferences also informs
decision makers about the variety attributes that have to be considered
for on-farm conservation. For instance, more experienced farmers and
small farm holders with smaller livestock assets are affected the most
when they have to forego teff and/or sorghum varieties with better yield
stability and environmental adaptability. They are, therefore, less likely to
cooperate with on-farm conservation activities that deny them varieties
with these attributes unless they get equivalent compensation. The
results also suggest that as farmers' economic well-being improves,
reected by the level of their incomes and assets, farmers' preferences for
crop variety traits shift from environmental adaptability and yield
stability, and thus favoring the conservation of productive and marketable varieties. This can run counter to crop varieties that are more
environmentally adaptable and stable. At a national level, this result
suggests that as the country's agriculture develops, there will have to be a
conservation policy shift in favor of those varieties that have been out of
farmers' choice in the course of agricultural development.9
The third important policy implication relates to the area of
varietal technology adoption. For agricultural technologies to be
successful, their attributes should address farmers' concerns. Clearly,
understanding farmers' varietytrait preferences informs public
decision making in varietal technology adoption. For instance, according
to the results, in order to target the variety demands of poor, vulnerable,
and segmented farmers, the variety attributes of environmental
adaptability and yield stability should be prioritized over productivity
traits.
The fourth policy implication is in the area of breeding priority
setting. Given that farmers' varietyattribute preferences determine
both their propensity to use improved varieties and the chance of using
them successfully, breeding should satisfy the demands of different farm
household types classied according to resource endowments, preferences, and constraints. To this end, analyzing farmers' varietyattribute
preferences will help target farmer demands in the making of a
technology. For instance, this study has found that farmers attach the
highest private value to the environmental adaptability trait, followed
by the yield stability attribute of both sorghum and teff. The national
institution primarily dealing with crop breeding programs in Ethiopia,
namely the Ethiopian Institute of Agricultural Research (EIAR) and other
donor and collaborative institutions should therefore prioritize these
attributes in their direct or supportive breeding programs if they are to
address the demands of Ethiopian smallholders.

Acknowledgements
The authors gratefully acknowledge the nancial support to
administer the survey from the Genetic Resources Policy Initiative
(GRPI)-Ethiopia project of the Biodiversity International and the
Environment for Development initiative at the University of Gothenburg and the Sida (Swedish International Development and Cooperation Agency). The authors also acknowledge Ekin Birol at IFPRI for
her comments on the earlier version of the paper and the anonymous
referees for their constructive comments.

We thank one of the anonymous referees for this interesting insight.

Appendix A
Table A1
What is the most important factor (criteria) you take into account in your seed selection?
Teff
Frequency
Early maturing
Good price and acceptance in
the market
Resistance to disease
Good for injera, local drink
tella, and bread making
Accepted by the land (soil) conditions
and suitable for local environment
Good productivity
Multipliable
Long seed size, and aggressive
in growth
Resistant to drought
Good straws for livestock
Other factors
Total

Sorghum
Percent

Frequency

Percent

11
10

8.4
7.6

10
2

7.7
1.6

21
1

16.0
0.8

28
2

21.4
1.5

3.9

6.1

22
5
22

16.8
3.8
16.8

13
5
28

9.9
3.8
21.4

7
18
9
131

5.3
13.7
6.9
100

2
21
12
110

1.5
16
9.2
100

Table A2
Results of random parameter logit model with socio-economic characteristics for choice
of crop variety.
Sorghum

Teff

Coefcient

S.e.

Coefcient

0.780
0.015
2.191
1.356
1.691

0.424
0.000
0.009
0.013
0.005
0.010
2.082
0.119
0.667
16.304

0.434
0.549
0.386
0.241
0.520
0.247
0.754
1.939
0.696
0.692
0.955
0.582
0.299
1.211
0.247
0.125 10.970
6.644
2.983
8.032
5.657
3.213
7.787
6.124
0.084
0.082
0.174
0.003
0.001
0.003
0.072
0.017
0.078
0.353
1.547
0.562
0.008
0.004
0.006
0.016
0.078
0.030
0.007
0.007
0.006
0.017
0.017
0.024
0.064
0.065
0.080
1.639
5.176
3.558
0.068
0.098
0.059
1.225
0.525
2.178
7.051
8.606
11.848

0.023
0.047
0.040

0.157
0.305
0.120

0.435
0.535
0.636

0.179
0.582
0.358

1.579
1.450

0.955
1.798

Mean parameters
Alternative 1 (Set 13)
Alternative 1 (Set 49)
Alternative 2 (Set 13)
Alternative 2 (Set 49)
Producer's price
Productivity
Environmental adaptability
Yield stability
Productivity male
Productivity experience
Productivity off-farm work
Productivity total land per capita
Productivity livestock value
Productivity drought frequency
Productivity expr. agric. extension
Productivity household size
Productivity agric. output surplus
Env. adaptability male
Env. adaptability experience
Env. adaptability off-farm work
Env. adaptability total land
per capita
Env. adaptability livestock value
Env. adaptability drought frequency
Env. adaptability exper.
agric. Extension
Env. adaptability household size
Env. adaptability agric.
output surplus
Yield stability male
Yield stability experience
Yield stability off-farm work
Yield stability total land per capita
Yield stability livestock value
Yield stability drought frequency
Yield stability exper. agric. extension
Yield stability household size
Yield stability agric. output surplus

0.168
0.621

0.344
1.208

0.121
0.034
0.076
4.926
0.224
0.115
0.298
0.222
0.138

1.781
2.783
0.071
0.012
1.288
1.345
7.309 11.543
0.170
0.306
0.324
0.902
0.134
0.638
0.363
0.559
1.309
0.042

Standard deviation parameters


Productivity
Environmental adaptability
Yield stability
Number of respondents
Number of choices
Pseudo R2

0.110 0.036
1.866 0.603
2.393 0.640
57
513
0.610

0.033
4.442
4.999
0.142
0.005
0.134

0.182
3.124
3.006
51
531
0.603

***, ** and * denote signicance at 1%, 5%, and 10%, respectively.

S.e.

3.503
0.057
2.097
11.102
0.151
0.608
0.377
0.862
1.972

0.040
1.103
1.902

S. Asrat et al. / Ecological Economics 69 (2010) 23942401

References
Alpizar, F., Carlsson, F., Martinsson, P., 2003. Using choice experiments for non-market
valuation. Economic Issues 8 (1), 83109.
Benin, S.M., Smale, B., Gebremedhin, B., Pender, J., Ehuil, S., 2003. Determinants of cereal
crop diversity on farms in the Ethiopian highlands. Environment and Production
Technology Division Discussion Paper, 105. International Food Policy Research
Institute, Washington, DC.
Bezabih, E., 2003. Determinants of multiple technology adoption: dening adopters and
empirical analysis. Ethiopian Journal of Agricultural Economics 5 (12), 122.
Brush, B., Meng, E., 1998. Farmers' valuation and conservation of crop genetic resources.
Genetic Resources and Crop Evolution 45, 139150.
Bunch, D., Louviere, J.J., Andersson, D., 1996. A comparison of experimental design
strategies for choice-based conjoint analysis with generic-attribute multinomial
logit models. Working Paper. University of California-Davis, Graduate School of
Management, Davis, CA, USA.
Carlsson, F., Frykblom, P., Lagerkvist, C.J., 2007. Consumer benets of labels and bans on
GM foods choice experiments with Swedish consumers. American Journal of
Agricultural Economics 89, 152161.
Central Statistical Agency (CSA), 2008. Agricultural Sample Survey 2007/08 (2000 E.C.):
Area and Production of Crops (private peasant holdings). Meher Season, Addis Ababa.
Chilot, Y., Shapiro, B.I., Mulat, D., 1996. Factors inuencing adoption of new wheat
technologies in Wolmera and Addis Alem areas of Ethiopia. Ethiopian Journal of
Agricultural Economics 1, 6384.
de Wet, J.M.J., Harlan, J.R., 1971. The origin and domestication of Sorghum Bicolor.
Economic Botany 25, 128135.
Degnet, A., Belay, K., Aregay, W., 2003. Adoption of high-yielding maize varieties in
Jimma zone: evidence from farm-level data. Ethiopian Journal of Agricultural
Economics 5 (12), 4162.
Di Falco, S., Chavas, J.-P., 2006. Crop genetic diversity, farm productivity, and the
management of environmental risk in rain-fed agriculture. European Review of
Agricultural Economics 33, 289314.
FAO, 1998. The State of the World's Genetic Resources for Food and Agriculture. FAO,
Rome, Italy.
FDRE, 2005. National Biodiversity Strategy and Action Plan. Institute of Biodiversity
Conservation, Addis Ababa, Ethiopia.
Gebrekidan, B., 1979. Sorghum genetic resources in Africa. Ethiopian Journal of
Agricultural Sciences 1 (2), 108125.
Hanneman, M., 1984. The discrete/continuous model of consumer demand. Econometrica 52, 541561.
Hawkes, J.G., Maxted, N., Ford-Lloyd, B.V., 2000. The Ex-Situ Conservation of Plant
Genetic Resources. Kluwer Academic Publishers, New York.
Heal, G., 2000. Nature and the Marketplace: Capturing the Value of Ecosystem Services.
Island Press, New York.
Jabbar, M.A., 1998. Buyer preferences for sheep and goats in southern Nigeria: a hedonic
price analysis. Agricultural Economics 18, 2130.
Kidane, A., Abler, D.G., 1994. Production technologies in Ethiopian agriculture.
Agricultural Economics 10, 179191.
Legesse, D., 2003. Empirical analysis of duration of herbicide adoption in teff-based
farming system of West Shewa Zone. Ethiopian Journal of Agricultural Economics 5
(12), 122.

2401

Loreau, M., Hector, A., 2001. Partitioning selection and complementarity in biodiversity
experiments. Nature 412, 7276.
Louviere, J.J., Hensher, D.A., Swait, J.D., Adamowicz, W.L., 2000. Stated Choice Methods:
Analysis and Applications. Cambridge University Press.
McFadden, D.L., 1974. The measurement of urban travel demand. Journal of Public
Economics 3, 303328.
Naeem, S., Thompson, L.J., Lawler, S.P., Lawton, J.H., Woodn, R.M., 1994. Declining
biodiversity can affect the functioning of ecosystems. Nature 368, 734737.
Scarpa, R., Drucker, A., Anderson, S., Ferraes-Ehuan, N., Gomez, V., Risopatron, C., RubioLeonel, O., 2003a. Valuing animal genetic resources in peasant economies: the case
of the Box Keken creole pig in Yucatan. Ecological Economics 45 (3), 427443.
Scarpa, R., Kristjanson, P., Ruto, E., Radeny, M., Drucker, A., Rege, J.E.O., 2003b. Valuing
indigenous farm animal genetic resources in Africa: a comparison of stated and
revealed preference estimates. Ecological Economics 45 (3), 409426.
Smale, M., 2006. Valuing Crop Biodiversity, On-farm Genetic Resources, and Economic
Change. CAB International, Wallingford, UK.
Smale, M., Hartell, J., Heisey, P.W., Senauer, B., 1998. The contribution of genetic
resources and diversity to wheat production in the Punjab of Pakistan. American
Journal of Agricultural Economics 80, 482493.
Smale, M., Meng, E., Brennan, J.P., Hu, R., 2003. Determinants of spatial diversity in
modern wheat: examples from Australia and China. Agricultural Economics 28,
1326.
Tilman, D., Downing, J.A., 1994. Biodiversity and stability in grasslands. Nature 367,
363365.
Tilman, D., Wedin, D., Knops, J., 1996. Productivity and sustainability inuenced by
biodiversity in grassland ecosystems. Nature 379, 718720.
Train, K., 1998. Recreation demand models with taste differences over people. Land
Economics 74, 230239.
Train, K., 2003. Discrete Choice Methods with Simulation, 2nd ed. Cambridge University
Press.
Wale, E., 2004. The economics of on-farm conservation of crop diversity in Ethiopia:
incentives attribute preferences and opportunity costs of maintaining local
varieties of crops. PhD Thesis, University of Bonn, Germany.
Wale, E., 2008. Computing opportunity costs of growing local varieties for on-farm
conservation: the case of sorghum in Ethiopia. Ecological Economics 64 (3),
603610.
Wale, E., Yallew, A., 2007. Farmers' variety attribute preferences: implications for
breeding priority setting and agricultural extension policy in Ethiopia. African
Development Review 19, 379396.
Wale, E., Chishakwe, N., Lewis-Lettington, R., 2009. Cultivating participatory policy
processes for genetic resources policy: lessons from the Genetic Resources Policy
Initiative (GRPI) project. Biodiversity and Conservation 18 (1), 118.
Weitzman, M.L., 1993. What to preserve? An application of diversity theory to crane
conservation. Quarterly Journal of Economics CVII, 363405.
World Bank, 2005. Ethiopia: well-being and poverty in Ethiopia the role of
agriculture and agency. Report no. 29468-ET. World Bank, Washington, DC.
Yifru, T., Hammer, K., 2006. Farmers' perception and genetic erosion of tetraploid
wheats landraces in Ethiopia. Generic Resources Crop Evolution 53, 10991113.
Yohannes, K., Gunjal, K., Cofn, G., 1990. Adoption of new technologies in Ethiopian
agriculture: the case of Tegulet-Bulga District, Shewa province. Agricultural
Economics 4, 2743.

You might also like