WILLINGNESS TO PAY FOR QUALITY COWPEA SEEDS: EXPERIMENTAL
EVIDENCE FROM NORTHERN GHANA
By
Edward Opoku

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Agricultural, Food and Resource Economics – Master of Science
2017

ABSTRACT
WILLINGNESS TO PAY FOR QUALITY COWPEA SEEDS: EXPERIMENTAL EVIDENCE
FROM NORTHERN GHANA
By
Edward Opoku
High cost of seed relative to grain price, and the lack of actual or perceived quality difference
between available seed products and grain have contributed to the low demand for seeds of
improved varieties for legume crops. This study reports the results of agronomic field
experiments and incentivized experimental auctions conducted in Ghana to address the following
research questions: 1) Are farmers able to perceive quality differences among three types of seed
products that can be used as planting material – recycled grain, certified seed, and quality
declared seed (QDS)? 2) Given the perceived quality difference, what is farmers’ willingness to
pay for these seed products? Results indicate that on average certified seeds performed
significantly better than QDS and the recycled grain. Participants were willing to pay 73% and
20% more for certified and QDS over recycled seed, respectively. This was consistent with
farmers’ observed seed quality difference obtained through their subjective plot rankings.
Consistent with the theory of downward sloping demand curve, there was an inverse relationship
between willingness-to-pay a premium price for quality seed and the number of farmers willing
to pay the premium price. On the supply side, the average cost of producing one kg of certified
seed was estimated to be twice the average cost per kg of producing cowpea grain. The findings
of this study have important policy implications on seed dissemination approaches to reach
farmers across the spectrum based on their willingness-to-pay for quality seed.

ACKNOWLEDGEMENTS

I would like to thank my major advisor Mywish Maredia for her support, insights and
suggestions throughout this research. The contributions of my committee members, Robert
Shupp and John Kerr are greatly acknowledged.
This research was made possible through the support from USAID-funded Feed the Future
Innovation Lab for Collaborative Research on Grain Legumes (under the terms of Cooperative
Agreement No. EDH-A-00-07-00005-00) and the MasterCard foundation Scholars program at
Michigan State University.
Lastly, I would like to also thank Dr. Francis Kusi and all the staff at the Savannah Agricultural
Research Institute (SARI)-Manga Station, Ghana for their support during the field work.

iii

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................................... vi
LIST OF FIGURES ...................................................................................................................... vii
CHAPTER 1: INTRODUCTION ....................................................................................................1
1.1 Background and Motivation ..........................................................................................1
1.2 Study Objective..............................................................................................................3
1.3 Research Questions .......................................................................................................3
1.4 Organization of the Study..............................................................................................4
CHAPTER 2: LITERATURE REVIEW .........................................................................................5
2.1 Seed system Issues for legume Crops ............................................................................5
2.2 Legume Seed System in Ghana .....................................................................................7
2.3 Methodological Approaches for Assessing Seed Demand Existing Practices and Gap 8
2.4 Importance of Building a Sustainable Cowpea Seed System in Northern Ghana .......10
CHAPTER 3: STUDY SETTINGS AND METHODOLOGY .....................................................11
3.1 Criteria for Establishing Experimental Fields ..............................................................11
3.2 Field Experiment (FE) ..................................................................................................12
3.3 Experimental Auction .................................................................................................17
3.4 Collection of Cost of Production Data ........................................................................19
CHAPTER 4: CONCEPTUAL FRAMEWORK AND EMIPIRICAL STRATEGY ...................21
4.1 Conceptual Framework ...............................................................................................21
4.2 Empirical Model ..........................................................................................................23
CHAPTER 5: DATA DESCRIPTION ..........................................................................................26
5.1 Data and Sample Selection ..........................................................................................26
5.2 Sample Characteristics ................................................................................................27
CHAPTER 6: RESULTS ...............................................................................................................30
6.1 Descriptive Statistics on Field Experiment .................................................................30
6.2 Results of Bidding Experimental Auction ...................................................................34
6.3 Cost of Producing Cowpea Seeds ...............................................................................38
CHAPTER 7 : DETERMINANTS OF WILLINGNESS TO PAY FOR QUALITY SEEDS .....41

iv

CHAPTER 8: CONCLUSIONS, MAIN FINDINGS, IMPLICATIONS AND NEED FOR
FUTURE RESEARCH ..................................................................................................................46
8.1 Main Findings ..............................................................................................................46
8.2 Implications for Strategies to Make Quality Seeds Accessible to Different Farmers .48
8.3 Need for Future Research ............................................................................................50
APPENDICES ...............................................................................................................................53
Appendix A: Comparing Cost of Production for Different Seed Types for same Farmer ............54
Appendix B: Specific Input Cost as a Percentage of Total Input and Labor Cost ........................55
Appendix C: Cost of Producing Certified Seeds ...........................................................................56
Appendix D: Cost of Producing Quality Declared Seeds ..............................................................57
Appendix E: Cost of Producing Recycled Seeds ...........................................................................58
Appendix F: Bidding Experimental Auction Script and Consent Form ........................................59
Appendix G: Sheet Auction Bidding Sheet ...................................................................................66
Appendix H: Seed Quality Analysis for Seed Producers...............................................................67
Appendix I: Seed Quality Analysis for Experimental Plots after Harvest.....................................68
REFERENCES .............................................................................................................................69

v

LIST OF TABLES

Table 5.1: Description of the Different Types of Data used in the Analysis .................................27
Table 5.2: Characteristics of Survey Respondents ........................................................................29
Table 6.1: Farmer’s Best Plot Rating at Flowering and Harvesting Stages...................................30
Table 6.2: Farmers Rating of Worst Plot (% of farmers) at Harvest Stage ...................................31
Table 6.3: Characteristics Cited as a Reason for Rating a Seed Type as Best or Worst ...............31
Table 6.4: Seed Quality Test Results before Planting ...................................................................32
Table 6.5: Agronomic and Grain Yield Parameters of the Seed Type across Ten Communities .32
Table 6.6: Yield of Different Seed Types across Field Experiments ............................................34
Table 6.7: Participants Willingness to Pay for Different Seed Types (GHC/kg) ..........................34
Table 6.8: Average WTP a Premium for Seed Type G (Certified) and L (QDS) over Lowest
Ranked Seed Quality (Recycled) ...................................................................................................36
Table 6.9: Relative Cost of Production for Certified and QDS over Recycled Seed ....................38
Table 6.10: Total Cost of Production (GHC) for Different Seed Types (average across all
producers included in the study), Total Production, and Cost per Kg ...........................................39
Table 7.1: Factors that Influences farmers’ WTP for Different Types of Seed Products: Model
Estimation using Household Fixed Effects (Model 1) and without (Model 2)..............................42
Table A1: Comparing Cost of Production for Different Seed Types for same Farmer .................54
Table B1: Specific Input Cost as a Percentage of Total Input and Labor Cost ............................55
Table C1: Cost of Producing Certified Seeds ................................................................................56
Table D1: Cost of Producing Quality Declared Seeds...................................................................57
Table E1: Cost of Producing Recycled Seeds................................................................................58
Table H1: Seed Quality Analysis for Seed Producers ...................................................................67
Table I1: Seed Quality Analysis for Experimental Plots after Harvest .........................................68

vi

LIST OF FIGURES

Figure 3.1a: Experimental Plot 1 ...................................................................................................16
Figure 3.1b: Experimental Plot 2 ...................................................................................................16
Figure 6.1: Average Bidding Price (Ghana Cedi GHC/kg) for Different Seed Types .................35
Figure 6.2: Distribution of Willingness to Pay for the Highest Rated Seed (i.e., Certified Seed)
Relative to Lowest Rated Seed (i.e., Recycled Grain) (Demand Curve) .......................................36
Figure 6.3: Willingness to pay for seed of perceived different quality as a percentage of cowpea
grain price ...................................................................................................................................38
Figure 8.1: Farmers’ Willingness to Pay for Quality Seed Relative to Market Price of Grain,
Certified Seed and Cost of Seed Production: Identification of three Groups of Farmers that
Require Different Approaches to Meet their Seed Needs ..............................................................48
Figure G1: Sheet Auction Bidding Sheet......................................................................................66

vii

CHAPTER 1: INTRODUCTION
1.1 Background and Motivation
The importance of seed in any crop-based production system cannot be overemphasized. Over
the years, a large number of improved varieties (i.e., varieties with genetically improved traits
that are generated by plant breeding research) have been released by both international and
national research institutions. Improving the genetics of planting materials of preferred crops can
greatly contribute to agricultural productivity improvement. However, there is an alarming gap
between the rate of release of improved varieties and their use by farmers (Walker and Alwang
2015, Shiferaw et. al 2008). Full expected gains from the investment in plant breeding research
depends on both the genetic improvement embodied in the seed as well as the existence and
performance of a seed system that can deliver this improved genetic gains to farmers in the form
of good quality seeds (i.e., planting materials). How to build and promote an efficient and
sustainable seed system that can deliver quality seeds of improved varieties of staple crops has
remained one of the major agricultural development challenges in many developing countries
(Maredia et al., 1999; Louwaars and de Boef 2012; Sperling et al. 2014).
Recognizing the importance of quality seed (i.e., planting material), governments, donors and
development organizations, have made significant investments in recent years to scale up efforts
to make seeds of improved varieties available to farmers (McGuire and Sperling, 2015). Many of
these efforts are based on the distribution of subsidized or free seeds (during or after
emergencies) which may not be sustainable in the long run. Hence, there is great interest from
development practitioners in understanding the demand and supply side economics of the seed
system with the goal of finding alternative models that are more market driven and sustainable.
One of the key factors that determines the sustainability of a seed system is the ‘effective

1

demand’ for quality seeds of improved varieties as reflected by the volume and frequency of new
seed purchased by farmers. Quality seeds are characterized by good germination rates, seedling
vigor, and are free from seed-borne diseases and other impurities. These seed quality
characteristics and genetic traits (i.e., tolerance to biotic and abiotic stresses, duration of
maturity, nutritional traits, etc.) are both necessary to realize the productivity gains from
adopting an improved variety. It is true that in the case of self-pollinated crops like cowpea
(Vigna unguiculata), the seed can preserve the genetic quality from one generation to the other.
As a result, farmers don’t have to purchase fresh seeds of an adopted improved variety at the
beginning of each planting season. However, one-time adoption of an improved variety with low
frequency of seed replenishment (either due to non-availability of seeds or farmers’ inability to
perceive and value the importance of seed quality) will not have the desired impact on crop
productivity if the seed of that variety is infected with diseases or its physical quality deteriorates
due to poor post-harvest handling and storage (Harrington, 1972).
For a given variety, production and marketing of quality seed involves taking specific and extra
measures compared to grain production. These include the use of inputs and management
practices to control weeds and pests during seed production, post-harvest processing,
conditioning and treatment, and packaging, and also complying with specific country’s seed
regulatory requirements to be able to sell the product as seed. ‘Seed’ production thus has higher
production and transaction costs than grain production.
The viability of a seed market will thus depend on the co-existence of the following demand and
supply conditions. On the demand side it depends on: 1) whether farmers are able to perceive or
recognize the ‘seed’ product as a quality planting material, and 2) given the perceived quality
difference, whether farmers are willing to pay a premium price for seed compared to grain price.
2

On the supply side, it depends on: 3) whether the price farmers are willing to pay is high enough
to recover the cost of producing quality seed; and 4) whether the quantity and frequency of seed
demanded at that price is large enough to attract suppliers to produce and sell quality seed. To
the best of our knowledge, no rigorous work has been done to systematically examine these
demand and supply side conditions in the context of the seed system of a self-pollinated legume
crop such as cowpea. This gap serves as a motivation for this research.
1.2 Study Objective
The overall objective of this study is to use field experiment and incentivized experimental
auctions to systematically analyze the economics of the demand and supply side characteristics
of the cowpea seed system in the northern part of Ghana.
1.3 Research Questions
This study achieves the stated objective by addressing the following research questions:
a. For a given variety, what is the difference in the performance (as measured by yield and
other characteristics important to the farmers) of cowpea crop across three seed typesCertified, Quality Declared Seed (QDS), and farmer saved cowpea grain when the seeds
are planted and managed by farmers under their own farm conditions?
b. How does the observed differential performance of different types of seeds translate into
farmer’s willingness to pay for quality seed?
c. What factors determine the cost of producing certified seeds, QDS and cowpea grain, and
how do these costs compare to their relative market prices?

3

1.4 Organization of the Study
The thesis is divided into seven chapters. Chapter two focuses on the review of existing
literature and presents major issues confronting seed system development in general, and legume
seed system in Ghana, in particular. The gaps in the literature and motivations for developing a
sustainable seed system are discussed in this chapter.
Chapter three describes the study setting and the study methodology. It describes the criteria
used to establish experimental plots, and explains the methodology, process, and design of the
experimental plots and bidding experimental auctions.
Chapter four describes the conceptual framework and the empirical strategy employed to
estimate farmer’s willingness to pay. A hedonic price model is used to explore the influence of
perceived seed quality on the coefficient of interest.
Chapter five provides description of different types and sources of data used. Summary statistics
of data collected from the farmer survey are also presented in this chapter.
Chapter six presents the results from the field experiment, experimental auction and the cost of
producing different types of seed products.
Chapter seven presents the econometric model estimation results. Determinants and factors that
influence a farmer’s WTP for quality seeds are presented and the results are discussed.
Chapter eight discusses the policy implications of the overall results, limitations of this study,
opportunities for future research, and concludes with recommendations for building a sustainable
seed system.

4

CHAPTER 2: LITERATURE REVIEW
2.1 Seed system Issues for legume Crops
Inefficiencies in the seed system in many Sub-Saharan African countries has contributed to the
low adoption rate of improved seed variety in the region (Tripp, 2001). About 90% of seeds
used by smallholder farmers in Africa is accessed through the informal sector which comprises
of borrowing, farmer-to-farmer seed exchange, on-farm savings, and local grain purchases
(Maredia et. al 1999). The informal seed system is characterized by unstructured and
unregulated activities ranging from seed selection, multiplication and dissemination mostly by
farmers themselves and with no or little involvement of institutional entities. Farmer-to-farmer
dissemination of seed is mostly informed by observing and learning from others (i.e., family and
neighbors). The informal seed system, however, is vulnerable to shocks due to its unstructured
process. For instance, in the case of extreme weather conditions such as flood and drought, there
are higher chances of experiencing seed shortages. Recycling of seeds by farmers also poses
constraints since the availability of seeds depends on the yield and quality of saved seeds.
Quality of seed can be impacted by production practices used (or not used), and can diminish
during storage.
The formal seed system, initiated by governments through the establishment of parastatal seed
companies and research laboratories was a response to address some of the constraints of the
informal seed system. However, these state run corporations faced high costs of production and
distribution challenges, and inefficiencies in meeting the seed needs of a broad range of farmers.
It has focused on a relatively narrow range of crops/varieties, mostly hybrid maize and crops
with relatively high seed demand. Even with the few crops that they produced, the seed quality
was not consistent over time. These challenges coupled with the rolling out of structural

5

adjustment programs across Africa lead to the closure of many parastatal companies. Due to the
way these parastatals operated, it did not create an enabling environment for the private sector to
gradually replace the parastatals in its absence.
There are also other barriers that restricted the private sector from entering into the seed sector
in Africa. These include poor rural infrastructure, inadequate or no seed policy and regulatory
framework to protect the private seed companies. As a result, the governmental and nongovernmental organizations have been working together to strengthen the formal seed system by
providing farmers with free samples of improved variety seeds, and distributing seed during
emergencies and natural disasters. However, emergency seed distribution programs have been
criticized as competing with local seed enterprises and undermining the emergence of private
seed sector instead of supporting them. The emergence of a hybrid (i.e., semi-formal) system
which leverages on characteristics from both the formal and the informal seed sector is gaining
attention. The semi-formal system allows for training and monitoring of farmers to produce
quality seeds for sale to members in their community. The production stage has features similar
to a formal seed system whereas the distribution leans more towards an informal system.
However, this hybrid system is not legally recognized in most countries and there is no seed
certification by a government recognized agency. In Tanzania seeds produced in this system are
called Quality Declared Seeds (QDS), which is also the case in Ghana (Shiferaw et. al 2008).
Like in Tanzania, the QDS in Ghana has been recognized legally, and seed producers need to be
trained before they can produce and sell QDS seed.

6

2.2 Legume Seed System in Ghana
The legume seed system in Ghana, like many other African countries, is faced with low and
inconsistent seed demand and supply, particularly for new improved varieties. Farmers usually
recycle cowpea grain from their own harvest, and use that as planting material. In addition, a
lack of proper seed regulatory frame work contributes to low availability of seed and
consequently, low seed demand. Designated governmental departments responsible for
regulating seed quality in most cases fail to do so. As a result, farmers end up buying counterfeit
(low quality) seed and are left with a reduced incentive to buy fresh certified seeds in subsequent
seasons.
Adopting the seed system evolution framework described by Maredia et al. (1999), the Ghanaian
legume seed system can be classified under phase two of the evolution process. This phase is
characterized by the development of improved varieties of seeds by public research institutes.
The use of inputs (e.g., fertilizer) is limited but increasing. The private sector is gradually
becoming involved in the multiplication and distribution of seeds. There are about 11 seed
companies and 192 registered seed growers in Ghana (Etwire et al. 2013). The Semi-formal seed
system in Ghana also faces the problem of lack of access to foundation seeds, inadequate training
and monitoring as well as market access and distribution challenges. As a result, most of the
QDS producers still operate in the informal farmer-to-farmer mode of seed distribution.
Recently, community projects sponsored by NGO’s in partnership with research institutions and
government extension agents are leading the QDS production and distribution. The role of the
research institute is to supply quality foundation seeds to seed producing farmers and the
extension agents are required to provide training and technical advice to farmers on quality seed
production.

7

2.3 Methodological Approaches for Assessing Seed Demand Existing Practices and Gap
In this sub-section, we review two strands of existing literature that are closely related and
provide the rationale for analytical approaches used in this study.
The first strand of literature looks at how plant breeders and agronomists assess the demand for
genetic attributes. Effective seed demand to a large extent depends on farmer’s ability to link
desired seed characteristics to a particular seed. However, some of these desired characteristics
cannot be physically observed at the time of seed purchase. For example, the genetic make-up of
the seed which enables the seed to be tolerant to drought, resistant to pest and diseases, or have
short or long maturity period cannot be observed. It is in tandem with this gap, that researchers
and NGO’s have adopted the demonstration field experiments as a mechanism to assist farmers
in recognizing the genetic quality of seed variety. Farmer’s inability to link the genetic attributes
to the type of seed variety at the time of purchase is more likely to affect the demand for seed as
a planting material. Over the years, researchers in particular have employed participatory crop
varietal trials to include farmers in the selection of improved varieties (Morris et al. 2004).
Farmers are invited to observe experimental varietal trials and share their opinions on preferred
varietal traits and characteristics. One major short fall of this approach is that the demonstration
trials are managed by researchers, which may not be representative of farmers’ growing
conditions and practices. In this study we follow a similar methodology of conducting field
demonstration trials, but address the aforementioned shortcoming by conducting these trials
under the host farmer’s farming conditions and practices.
The second strand of literature this study draws from looks at measuring demand for attributes
from the perspective of an economist. According to the Lancaster’s theory, consumers do not
derive utility from a good itself, but rather from its characteristics (Lancaster 1966). This implies

8

that the same products with different characteristics provides different satisfaction levels (De
Groote et al., 2011). Consumers are continually making choices among products. Not only do
they not have reliable information about the price of the product, but they also have poor quality
valuation among the products probably because information on product quality is hard to find.
Researchers have used experimental auctions to value consumer’s preference for a new product
or competing attributes in the face of tradeoffs. Experimental auctions have been applied in
various forms to estimate consumer’s Willingness-To-Pay (WTP) for staple crops. This includes
aflatoxin-free maize in Kenya (De Groote et al. 2015) and biofortified yellow and white maize
meal in Kenya (De Groote et al. 2011). These papers focus on consumer’s preference for taste
and how that translates into either a premium or discount they are willing to pay for the staple. In
the case of seed, farmers are interested in both the consumption and production traits of the seed.
Some of these desired characteristics cannot be physically observed at the time of seed purchase.
For example, the genetic make-up of the seed which enables the seed to be tolerant to drought or
resistant to pest and diseases cannot be observed. Waldman et al. (2014) studied both production
and consumption preferences of Rwandan farmers for a common bean variety. They combined
binding experimental auctions with participatory varietal selection to derive farmers’ varietal
preferences. They confirmed that auctions revealed varietal demand more accurately than stated
non-binding preferences. Again, the participatory varietal selections enabled them to understand
how farmers evaluated tradeoffs between crops. In this study we draw from this literature and
design binding auction experiments to assess how farmers’ perception about seed quality
attributes translates into their willingness to pay for quality seed.

9

2.4 Importance of Building a Sustainable Cowpea Seed System in Northern Ghana
Cowpea provides a plant source of protein (Appiah et al. 2011; Ahenkan et al. 1998), bridges the
hunger period during the crop growing season, and is one of the staple crops that is resilient
under low rainfall conditions. It also has a high biological nitrogen fixing ability (Singh et. al
2003) which can reduce the need for farmers to purchase expensive inorganic fertilizers. Given
these characteristics, cowpea is highly suitable for the three Northern regions of Ghana, which
are also considered the poorest in the country (GSS 2014). For these reasons, exploring possible
demand and supply side constraints that farmers face in their cowpea seed adoption decisions can
help inform decisions on designing suitable interventions which can increase cowpea
productivity by improving accessibility of quality seeds to cowpea farmers. Such institutional
innovations in the cowpea seed adoption process can significantly contribute to the general
wellbeing and poverty reduction in the region.

10

CHAPTER 3: STUDY SETTINGS AND METHODOLOGY
3.1 Criteria for Establishing Experimental Fields
One of the objectives of this research is to understand whether farmers are able to perceive
quality differences among three types of seeds that have undergone different quality control
measures during the production, post-harvest processing, and marketing stages. Ultimately, we
are interested in knowing how these perceived seed quality differences translate into relative
differences in farmers’ willingness to pay for these three types of seeds. To examine this, we
conducted agronomic field experiments and incentivized Experimental Auctions in 10 villages in
the Northern part of Ghana.
Unlike the villages in the southern part of Ghana where farmers stay at one place (dwelling) and
move to the outskirt of the villages to farm, most of the villages in the Binduri district in the
Northern Region of Ghana practiced compound farming system, where farmers live on their farm
lands. Host farmers were selected in consultation with the Ministry of Food and Agriculture
(MOFA) extension officers and village leaders. Recommended and selected farmers owned
enough land, were major cowpea decision makers, and were willing to host the demonstration
fields. Another important requirement was the existence of a cordial relationship between the
selected host farmer and the rest of the community members1.
Selected host farmers were briefed on the entire project and their respective roles were
emphasized. The designated fields (or sub-plots) for the Field Experiments (FE) were clearly
demarcated and marked with identification labels representing each seed type.

1

In one of the villages (Azumsapeliga), we had to replace the selected host farmer because he had marital issues some time ago
which resulted in dispute with quite a large proportion of the community. This obviously might affect the attendance and
participation during the field days.

11

3.2 Field Experiment (FE)
The field experiments were established to demonstrate the value of planting different types of
quality seeds of the same cowpea variety, called Songotra. The main reason for selecting this
variety was its genetically improved quality to resist Striga gesnorioides, a major parasitic weed
that has detrimental impact on the productivity of staple crops in Sub-Saharan Africa. The weed
survives by drawing off water and nutrients from the crops for its own growth. It causes
substantive damage to its host crop before emerging from the soil by producing phytotoxins
which are harmful to the host crop (Ouédraogo et al. 2001). For the cowpea breeder collaborator
in Ghana, this experiment also served the purpose of evaluating the genetic qualities of this
variety (i.e., striga resistance) under the on-farm conditions.
The three types of seeds of this variety evaluated in the FEs included certified seed, QDS, and
recycled seeds. The intention was to evaluate the performance of the cowpea crop planted with
three different seed quality, but controlling for the genetics (i.e., same variety) and all other
environmental and management conditions. These field demonstration plots were used to
provide the farmers an opportunity to observe the agronomic performance of the cowpea plants
in terms of yield and other plant characteristics. This observation and first-hand experience was
the basis for conducting an incentivized Bidding Experimental Auction (BEA) for estimating the
relative difference in farmers’ WTP for the different types of seeds representing different quality.
The characteristics of the three seed types used in these experiments are as follow.
•

Certified Songotra Seeds: These are supposedly the highest class and quality of seed
available for farmers as planting material. This type of seed is produced from foundation
seeds that are produced from the breeder seed. The government research program has the
responsibility to maintain the seed stock of both the breeder seed and the foundation seed.
12

Certified seeds are produced by trained seed producers or seed companies under the
required field conditions and have to undergo field inspection and seed quality tests, and
must meet the required standards to obtain official certification from the Ministry of Food
and Agriculture. Only approved seed that meet the seed quality tests (i.e., germination
rate, humidity and purity) are marketed as ‘certified seed’ and sold commercially in
sealed packages with appropriate labeling.
•

Quality Declared Seed: In Ghana, this type of seed is produced by seed producers (either
individual farmers or a farmer group) with guidance from experts but do not go through
formal certification process. QDS producers receive foundation seeds from government
or NGO’s. In most cases, the farmers would have to pay back the cost of the foundation
seed in kind after their harvest. QDS are sold to other farmers living in the same village
or nearby communities. The objective is to solve the travel and search costs that farmers
face in order to acquire quality seeds for planting. Unlike certified seeds, quality declared
seeds are not required to pass purity, germination and moisture content test before being
sold to farmers. QDS are normally cleaned manually and packaged for sale. Another
major difference between certified and QDS is that, quality declared seeds are currently
not legally recognized as a category of seed type in Ghana.

•

Recycled seed: This is the grain saved by farmers from their previous season’s harvest of
cowpea. These are selected after a season’s harvest and stored for future use as seed. This
is the traditional method through which seed has been preserved over the years in Ghana.

The certified seed was bought from a certified seed seller in the market, QDS was acquired from
a breeder who has a private business of producing QDS2 for sale in the district. Recycled seed

2

QDS seeds came in an uncertified bag

13

was purchased from a farmer who had purchased seeds of Sangotra variety and had stored grains
from previous harvest to be used as seed in the following year.
A total of 10 villages were selected and one farmer from each villages was identified to host the
field experiments. The intent was to sub-select eight villages with the best plots to conduct the
field days and bidding experimental auctions. Each farmer hosting the FE was given 2.7 kg of
each seed type. A sample of each seed type was sent to the laboratory for seed purity,
germination and general seed health status test.
The FE was established as a double blind experiment, where neither the farmers (host farmers
and other village farmers) nor the field assistant knew the identity of the three types of seeds
included in the experiment. Technical staff implementing the FE and host farmers only knew the
name of the variety and were also told that there were three types of seeds, namely certified,
QDS and recycled seed to be planted on separate plots, however, the seed packages were labeled
seed type G (which was certified seed), seed type M (which was recycled seed), and seed type L
(which was QDS) and the technical staff and farmers were not told which code was associated
with which seed type. The reason for doing the FE as a double blind experiment was to reduce
any systematic bias on the part of the farmer managing the plot towards or against any preconceived notion of higher and lower quality seed type. The double blind experiment was
expected to also reduce any bias farmers as observers may have towards a specific seed type
based on their prior personal experience or ‘hearsay.’ The same seed codes used in the FE were
used throughout the study to be able to match all the information collected.
A 10m x 20m dimension plot was planted for each seed type in each village. Spatial separation
between plots was also encouraged for easy identification and movement of participants during
field days. Each seed type plot was properly and prominently labeled by the seed codes G, M and

14

L. (see pictures below). The order in which these three types of seeds were planted across the
eight FEs was random. For example, in some fields, the seed was planted in the following
order—M, G, L; in some it was planted in the order – L, M, G, etc. Extension officers and local
collaborators supervised the establishment of the FE in selected villages to ensure uniform
planting rate, not mixing up with plot labels and other practices.
Two field days were organized during the study in eight out of ten villages selected for the FEs.3
All farmers in each village were invited to attend the field days. On the first field day which was
organized during the flowering stage, all farmers who attended were handed a ranking sheet to
rate the overall best plot at the flowering stage. Before the individual ratings, participants
discussed and agreed on important criteria with which they will rate a seed plot as good or bad at
the flowering stage. These desired traits included potential yield/productivity, vigor of plant
growth, plant stand, signs of no disease, and number of flower sets.

3

The two villages that were dropped were located along the river (White Volta), which is prone to flooding during the main
season (July- September). Farmers around this area normally use residual moisture during October to December to grow cowpea.
Since this type of farming was a special case, it was decided to exclude them from the activities involving farmer surveys and
experimental data collection.

15

After the first field day, 35 participants (each representing one household) were randomly
selected from a list of participant households. A survey was administered to these participants
and each received a unique ID tag which they were instructed to bring with them on the second
field day. The second field day was held a week before harvest. This was to enable participant
Figure 3.1a: Experimental Plot 1

Figure 3.1b: Experimental Plot 2

Observe full formed pods to inform their rating decisions. On the second field day, participants
were again asked to rate the best and the worst plot and to give reasons for their choices. After
rating the plots, a bidding experimental auction was carried out on the same day. All participants
who were sampled and surveyed after the first field day but did not show up on the second field
day were replaced with farmers who took part in all the activities in the first field day and were
also present in the second field day.
16

To avoid the chances of the same people (farmers) showing up on two field experiment sites
(i.e., in two neighboring villages), villages were carefully selected to ensure they were located at
some distance from each other. In selecting the villages, we also took into consideration the
number of households in a village. Since the ultimate sampling goal was to select 35
representative cowpea famers, we selected villages with an approximate number of households
between 80 and 120. On the second field day which fell on a week before harvesting and also
coincided with the auction experiment, two FEs were visited per day.
3.3 Experimental Auction
The Bidding experimental auction (BEA) was conducted to determine how much farmers were
willing to pay for the different types of seeds (planting materials) and this was done during the
second field day approximately two weeks after the first field day). This was advantageous since
farmers had gotten the opportunity to evaluate the performance of different types/grades of seeds
planted in the FE, and used this performance observation as the basis for making the decision
regarding their WTP for each seed type. Although farmers did not know yield data during the
BEA, they could easily predict the relative difference in the yield across plots at maturity (based
on their observations of how filled the pods were and how healthy the plants looked relative to
each other).
We followed the Becker-DeGroot-Marschak (BDM) (Becker et al, 1964) method in designing
the BEA. Unlike the classical auction where the highest bidder gets to buy the good, BDM
mechanism is quite unique in the sense that bidders do not bid against each other but rather each
one bids against his/her own self as the winning bid is determined by comparing their own bids
to a randomly generated price.

17

The BDM elicitation mechanism is typically performed using one of two methods – a full
bidding or an endow-upgrade method. In the full bidding method, participants are asked to “bid”
their maximum willingness to pay for a given quantity. A random price is then drawn and
compared to the bid. If the bid is greater than or equal to the randomly drawn price, then the
participant gets to buy the good at the randomly drawn price (not their bid price).
In the endow-upgrade method, the participants are endowed with a given amount of the “lower
value” good and asked to “bid” their maximum WTP to upgrade to an equivalent amount of the
“higher value” good. Again, this bid would be compared to a randomly drawn price and if their
bid is greater than or equal to the randomly drawn price, they would pay to upgrade, but will
only pay the randomly drawn price (not their bid). In this method, the bid itself reveals the
premium given to the “higher value” good. In both methods, the participant is likely to pay less
than their bid (unless the bid and random price are equal) and thus the auctions are theoretically
incentive compatible with regards to eliciting participant’s true WTP. In both these auctions each
participant receives a cash endowment at the beginning with which to either pay for a good or to
upgrade. Each method has its advantages, but the literature (e.g., Lusk and Shogren 2007, and
Alfnes 2009) appears to lean towards using the full bidding method, especially if very similar
products are readily available in the market place.
This study followed the first method explained (i.e., full-bidding method) due to the nature of
“blindness” of the FE and auctions (i.e., farmers were bidding without knowing the type/grade of
seed they are bidding for) and the similarity in the seed variety (Songotra). BDM is also a
convenient method for researchers because it is compatible with both individual and group
administration. However, it is quite expensive (in terms of cost and time) to gather all
participants at one place to execute it.
18

In this study, the group approach was used for the BEA. After farmers had completed the field
visit and evaluated the experimental plots, farmers were asked to gather at one place where
someone from the research team explained the farmers in their local language the BEA exercise
and the steps involved. The script used during this exercise is given in Appendix F. A trial BEA
was implemented first with a bar of soap to demonstrate the procedure to ensure that the bidding
mechanism and the concept of drawing a random price was properly understood by the
participants. Each participant was given GHC 2 (equivalent to $0.57 at the exchange rate at the
time of the exercise) as the endowment for the soap BEA.
During the seed auction, farmers were asked to record their bids on a recording sheet (see
Appendix G) for the three seed types (seed type L, G and M) used to plant the plots observed by
the participants. Each participant was given GHC 10 (about $2.63 at the official exchange rate)
and asked to record their bids ranging from GHC 0 to GHC 9.90 at an increment of GHC 0.10.
After collecting the bidding sheets, a price was randomly generated by the participants for the
seed in three steps. First, three spaces were written on a board with a dot in between the first and
second space. The third digit was filled with zero. Secondly, since the second digit could be a 0,
1, 2, 3, 4, 5, 6, 7, 8 or 9, one participant was asked to roll out a 10-sided die. This second digit
was whatever was rolled unless a 10 was rolled, in which case the second digit was 0. Lastly, the
first digit could also be a 0, 1, 2, 3, 4, 5, 6, 7, 8 or 9. To determine this digit, a 10-sided die was
rolled. If a 1, 2, 3, 4, 5, 6, 7, 8 or 9 was rolled, the first digit was that number. If a 10 was rolled,
the first digit was 0. If a participant’s bid was higher than this randomly generated price, the
participant bought the seed at the random price and kept the balance from their GHC 10
endowment. If their bid was lower than the random price, they did not purchase the seed and
kept the GHC10 endowment. The difference in the bids between the three auctions is interpreted

19

to reveal the premium (or discount) a farmer is willing to pay due to the different quality
attributes as observed/perceived by the farmer in the FEs (i.e., performance of seed type G, L,
and M).
3.4 Collection of Cost of Production Data
As noted before, the three seed types vary in terms of inputs, management practices, and
regulatory supervision, which have different cost implications and ultimately affect the quality of
the final product (i.e., seeds) produced. To gauge these differences in cost of production across
the three types of seeds included in the experiments, a sample of seed producers producing these
different types of seeds were selected from the same area where the study was conducted.
Detailed data on inputs, labor and post-harvest expenses were collected from farmers for the
main cowpea growing season in 2016. Regular visits were made by trained enumerators to
collect the data from the seed producers, who were trained to keep a record of all the input and
labor costs incurred at all stages of the growing season.

20

CHAPTER 4: CONCEPTUAL FRAMEWORK AND EMPIRICAL STRATEGY
4.1 Conceptual Framework
Lancaster (1966, 1971) proposed an alternative theoretical approach to the traditional demand
theory based on the principle that all goods possess characteristics or attributes that are
demanded by the consumers, not the goods themselves. The new consumer demand theory was
based on the assumption that a good in question does not give utility but rather the characteristics
possessed by the good provides utility. Also, collective goods may possess traits different from
those possessed by separate goods. This theory in effect implies that utility can be modelled in
terms of the characteristics of a good.
Our model builds on the framework used by Hoffman and Gatobu (2013), which is rooted in the
Lancaster theory of consumer demand. Consider a model where a farmer derives utility from
growing cowpea using either one of three types of seeds

,

= 1,2,3 (where, 1=certified,

2=QDS and 3=recycled) and other complementary inputs m. Certified and quality declared seeds
can only be obtained through the market whereas recycled seeds are assumed to be available to
farmers from their previous season’s harvest. The seeds possess a vector of attributes ̅= [o, e, c],
where o represents observable qualities such as the presence or absence of foreign materials such
as dirt, stones or dead insects, and the uniformity in the shape and size of the grains, their
appearance, etc., e represents experience attributes which can be known only after planting the
seed, for example, plant vigor, and c, is the credence attributes, which are completely
unobservable characteristics to the farmer such as the genetic make-up of the seed, its
germination rate, moisture content, etc.. When a household uses recycled seed, some elements of
o and e can be controlled. For example, good harvesting and storage practices can reduce the bad
quality elements of o and increase farmer’s confidence in the experience attribute. On the other
21

hand, a household that purchases certified or QDS, can only control for o and to some extent c
but the experience attribute is expected to be always new and unknown to the farmer.
Hoffman and Gatobu (2013) proposed that a reputation effect will transmit information about the
experience attributes.
Holding all other inputs (m) constant, we represent the quality of a seed type planted by a
household as q( ̅ ), q( ̅ ) and q( ̅ ), for certified, QDS and recycled seeds, respectively. Prior to
planting, the expected values of q is dependent on o (observed attributes) and household’s
expectations over e and c. Famers are expected to update their beliefs and perceptions about a
particular seed type after their experience. The quantity harvested from different seed type, ,
,

, which are functions of their respective quality values enter the choice function. If seed

quality is discovered after purchase or after planting to be bad such that there is no grain output
for consumption, sale or planting as seed in the next season, then q( ̅ )=0 J=(1,2,3), then a farmer
is expected to form a lower perception about the seed type and as a result, not willing to pay a
premium for the seed. The household’s first decision problem is to choose a seed type that will
maximize output, holding all other complementary inputs constant.
E

= Y( (q( ̅ )), m)

where j=1, 2,3

(1)

The household then decides how much they are willing to pay for quantities of the seed type
selected in equation 1 above. Willingness to pay is expected to be influenced by the perceptions
formed about seed type after a household’s experience with the seed type. In this study, this
learning process is assumed to be done through a social learning experience, i.e. the field
experiment.

22

The credence (c) and experience (e) attributes of recycled seeds are influenced by the
household’s farm practices. This is the case because in most instances the credence attribute
(e.g., genetic makeup) can be preserved if a farmer observes good farming practices during and
after harvesting. Harvesting and drying under the right temperature, adoption of good storage
practices like hermetically sealed bags (e.g., the Purdue Improved Crop Storage (PICS)
technology) can help preserve the genetics and seed qualities of the grain. But it is totally out of
the farmer’s control for seeds purchased from outside the farm (e.g., certified and QD seeds).
One may argue that it may be better to implement the BEA before planting (in the following
season) when farmers’ demand for seed and their “true” WTP for seed would be revealed.
However, since we are interested in the relative WTP, implementing the BEA at the end of the
growing season still achieved the objective of this study. Thus, we did not include a time
dimension in this model.
The model helps us to predict how farmers’ perception about different seed attributes affects
their willingness to pay for quality seeds. This perception is informed by participant’s
observations of experimental fields. The double-blind nature of the experiment controls for any
pre-conceived perception of the quality of certified, QD or recycled seed, a farmer participant
may have based on his prior experience or credence opinion.
4.2 Empirical Model
Following the estimation model used by De Groote et al. (2016), the seed characteristics of
interest in the case of this study is perceived quality based on the observation of plant
performance in the farmer’s field. At the time of bidding, farmers did not know the identity of
the seed type they were bidding for. However, they had ‘observed’ and ‘experienced’ the
performance of the three types of seeds planted in the field during the flowering and harvest

23

stages, and formed perceptions on the quality of these seeds. This perceived seed quality is
estimated by creating a seed quality rating scale 1 to 3, with 1 being the worst seed plot, 3 being
the best ranked seed plot and 2 representing neither worst nor best seed plot. Each participant
submitted a bid for three seed types.
Since participants were randomly selected for the bidding exercise, we can safely assume that
individual’s effect is not correlated with any of the explanatory variables. A general OLS
specification captures household characteristics and perceived seed quality. Household’s and
individual farmer’s characteristics are assumed to vary across each other, which in turn induce
varying levels of WTP among farmers. Hence including household’s characteristics

will be

appropriate. The household’s characteristics included are demographic, economic and some
behavioral variables (De Groote et al., 2011).
We estimate the relationship between farmer participant i’s willingness to pay for seed type j
(WTPij) as a simple model specification using the farmer fixed effect as specified below
=

+

Where,

+

+

(5)

represents farmer i’s bid for seed type j;

rating for seed type j by farmer i,

represents the perceived seed quality

is the farmer fixed effect, and

is the error term for farmer

i’s WTP for seed type j.
Next we substitute the farmer fixed effect variables with a vector

representing farmer and

household characteristics such as demographic, socio-economic, and behavioral variables. This
allows us to explore the relationships between these characteristics and farmer’s WTP for
different quality seeds (equation 6).
=

+

+

+

(6)
24

Model specifications 5 and 6 are estimated using ordinary least squares (OLS). In both the
models, the coefficient of interest is , which measures the average price premium farmers are
willing to pay for each unit increase in the perceived seed quality rating (i.e., when seed quality
rating changes from worst to neutral to best).
De Groote et al. (2010) showed that WTP increased with the endowment amount. Some
researchers have explored this by giving an endowment that is in value twice the market price
(De Groote et. al., 2014; Morawetz et. al., 2011). In this study, we are interested in the relative
difference in WTP not the absolute WTP for a product. Thus this criticism may not hold for this
study. However, since the market price of cowpea certified seed was widely known, the amount
was carefully set at 25% higher than the market price of cowpea certified seed to avoid low WTP
for a very low amount and overstating of WTP for a very high amount.

25

CHAPTER 5: DATA DESCRIPTION
5.1 Data and Sample Selection
A total sample size of 269 participants was used for the analysis presented in this study after
accounting for attrition. This comprised of 35 farmer observations per village, one person from
each household. The participants in the survey, field evaluations, and BEA were either the major
cowpea decision maker or one of the main decision makers in the household. Attrition problem
was tackled by replacing sampled farmers who did not show up on the second field day with
farmers who were listed in the first field day visit and had showed up on the second field day. A
total sample of 269 cowpea growing farmers participated in the BEA exercise. All the farmers
who participated in the BEA were surveyed either on the first or second field day to collect the
household demographic and other socio-economic characteristics, and their knowledge and
experience with different types of cowpea seeds.
In all the villages where field experiments are established, we also collected yield data at harvest
for all the three seed types (Certified, Quality Declared and Recycled). Samples of harvested
grain from three of these FEs were sent to the lab to test for moisture, purity and germination
rates. As noted before, samples of seeds planted in the FEs of all three types (certified, quality
declared and recycled grain) were also tested at the lab for seed quality (Appendix H). The cost
of seed production data was collected over the same period as the FEs. There were only two
certified seed producers in the district so they were both included as part of the survey. Three
QDS and five grain producers were sampled. The QDS producers were selected from a USAID
funded community seed production program focused on assisting one community member in
each village to produce and sell quality seeds. All the three QDS producers in the district under
the program were included in the study (table 5.1 below). A survey template was designed and
26

administered by enumerators during the 2016 main farming season to collect the cost of
production data from all the seed producers included in the study. Enumerators regularly visited
seed producers to assist them record the inputs and labor costs incurred. Post- harvest expenses
were also captured for certified and QDS producers. Data for cost of producing grains were also
taken from certified and QDS producers to control for farmer fixed effect.
Table 5.1: Description of the Different Types of Data used in the Analysis
Activities

Data Collection

Sample size

method

Number of
Participants by
Gender

Field Experiment

Yield data from all

3 plots x 10

experimental plots and villages

Male - 8
Female - 2

Seed quality test
Field Day 1 (Flowering stage)

Household survey and

269

farmer ranking of seed

Male - 118
Female - 151

plots in 8 villages
Field Day 2 (Harvest stage)

Farmer ranking of

269

seed plots and

Male - 118
Female -151

Experimental Auction
in 8 villages
Cost of seed production

Cost of cowpea seed

Certified – 2

Male - 8

production (survey

QDS – 3

Female - 3

and record keeping)

Recycled - 6

27

5.2 Sample Characteristics
As reported in table 5.2, out of the 269 respondents, 51% were household heads, 46% were
spouse of household heads and the remaining 3% were either sons or daughters of the household
head. The average age of respondents was 42 years with average formal education of 3 years.
There were more female respondents (57%), and compared to men they were slightly older (47
years) with less formal education (on average less than 1 year). Most of the respondents who
participated in the experiments were illiterate (74%), who could not read or write in English or
any other language. Sampled households had an average of 5.6 acres of owned land out of which
they had cultivated 4.9 acres during the 2016 season. The average cowpea yield per household
was estimated to be about 250 kg per acre. About 15% of respondents had purchased or used
certified cowpea seeds which was lower than percentage of respondents who had used Quality
Declared seed (30%). Meaning the experience of using certified and QDS seed was quite low
among the respondents. About 39% of survey respondents identified themselves as early
adopters of a new technology. Also, 29% of the farmers reported that they had used their own
saved seeds as planting material in the last season, 14% had bought grain from the market, 51%
had bought cowpea seeds from the market and 6% had received their seeds from NGO’s and
government support programs.

28

Table 5.2: Characteristics of Survey Respondents
Number of observations

269

Farmer characteristics
Age (years)
Gender (% male)
Percentage of participants who were head of the household
Number of years of education completed
Cannot read or write (%)
Number of years of experience growing cowpeas
Membership in a farmer organization (% yes)
Self-reported adoption behavior (% early adopters of new
technologies)
Household characteristics
Gender of the head = male
HH size
Poverty score (0-100)
> 75% probability of living below $1.25/capita/day (% of farmers)
Have used certified seeds of any crop (% yes)
Have used certified cowpea seed before (% yes)
Have used QDS cowpea seed before (% yes)
Purchases cowpea 'seed' every year or other year (% yes)
Last time purchased seed less than 4 years ago (%)
Source of cowpea seed planted in the last season that was (%)
Saved from own harvest
Purchased as grain from market/others
Purchased as seed from market/others
Received from NGOs/government
Percentage of farmers reported planting an improved variety in the
last season
Average yield of cowpea from the last harvest (kg/acre)
Total land area owned (acres)
Tropical Livestock Units (TLU) owned
Cowpea reported as most important crop on the farm in terms of: (%
of farmers)
Area planted
Inputs used
Source of income
Percentage of farmers who grew songotra variety in the last season
Source: Household Survey August-2016

29

Mean
42
0.43
0.51
3.13
0.74
17.91
0.33

SD
14.1
0.49
0.5
4.74
0.56
14.47
0.47

0.39

0.48

0.85
8.88
29.77
0.01
0.18
0.15
0.30
0.72
0.64

0.34
4.93
11.3
0.1
0.39
0.36
0.46
0.44
0.47

0.29
0.14
0.51
0.06

0.45
0.34
0.50
0.22

0.11
250.29
5.59
4.92

0.31
177.84
3.7
2.48

0.06
0.16
0.73
0.03

0.24
0.37
0.44
0.15

CHAPTER 6: RESULTS
6.1 Descriptive Statistics on Field Experiments
Table 6.1 below presents the results of farmers rating of the best plot at the flowering and harvest
stages. Note that due to the double blind nature of the study, farmers did not know the identity of
the seeds when they were rating the plots. Hence the rating of different seed types are identified
by their plot IDs (i.e., G, L, M). At flowering stage, 90% of farmer’s rated plot type G (certified
seed) as the best (Table 6.1). About 8% and 2% rated plot type L (QDS) and M (recycled) as the
best plot at flowering stage, respectively. A second plot ranking was done one week before
harvest which recorded 95%, 4% and 0.7% of farmers ranking certified, QDS and recycled seed
plots, respectively as the best. It is quite obvious that at the harvest stage, there was a clear
distinction between the three plots resulting in almost all farmers rating certified seed (i.e., plot
G) as the best.
Table 6.1: Farmer’s Best Plot Rating at Flowering and Harvesting Stages
Seed type

Flowering Stage
% of farmer’s (N =269)

Plot G: Certified Seed
Plot L: Quality Declared Seed
Plot M: Recycled Seed

89.93
7.87
2.24

Harvesting Stage
% of farmers (N =268)
95.17
4.09
0.74

Source: Field experiment data, Ghana-2016

At the harvest stage, farmers were also asked to rate the worst plot among the three, and 77%
rated recycled seed plot as worst and quite a substantial percentage rated QDS plot as second
worst (22%) (Table 6.2). Clearly, farmers were confident about rating seed type G (certified) as
the best and least worst, but the same cannot be said about seed types L (QDS) and M (recycled).

30

Table 6.2: Farmers Rating of Worst Plot (% of farmers) at Harvest Stage
Seed type
Plot G: Certified Seed
Plot L: Quality Declared Seed
Plot M:Recycled Seed

% of farmers (N=269)
0.37
21.93
77.70

Source: Field experiment data, Ghana-2016

Farmers gave various reasons for rating a plot as the best or worst. These reasons are categorized
in Table 6.3. About 69% of farmers rated a plot as best because of higher yield, 14.87% rated
best plot with high seed quality criteria. Unhealthy appearance of plants and lower yields were
the major reasons given by famers for rating a plot as worst.
Table 6.3: Characteristics Cited as a Reason for Rating a Seed Type as Best or Worst
(N=269)
Reasons for rating a plot
‘BEST’
At Flowering
1. Yield/Productivity
2. Vigorous
3. No disease
4. Plant Stand
At Harvesting
1. Higher yield
2. Plants look healthy
3. Good seed quality
4. Pods have filled nicely

% of
farmers

Reasons for rating a plot
‘WORST’
At Harvesting
1. Plants look unhealthy
2. Pods have not filled nicely
3. Lower yields
4. Poor seed quality

% of
farmers
39%
15%
38%
8%

69%
7%
15%
9%

Source: Farmers’ ratings recorded on Field Day 1 and 2, Ghana-2016

The data from all the ten communities where the FEs were established were pooled together for
analysis of the mean yield performances of the three seed types and are presented in Table 6.5.
Seed quality test results for the different seed types are presented below in table 6.4

31

Table 6.4: Seed Quality Test Results before Planting
Reference Before planting
Before planting
Before planting

Seed type
Certified
QDS
Recycled

Moisture
Content %
9.9
9.5
9.9

Purity

Germination %

99.9
99.9
88

98
92
98

Source: seed test results from the plant protection and regulatory service department MoFA –Ghana, 2016

Certified and QDS passed the seed purity test of 98% for cowpea by the FAO. Recycled seed
however, had 88% of seed purity test results. The acceptable threshold for moisture content by
FAO is 10% this implies that all the three seed types failed the moisture content test.
Germination rate for certified seeds and QDS were 99.9% for both seed types, which was higher
than the 98% acceptable threshold for cowpea seeds. Recycled seed on the other hand did not
pass the germination rate test (88%).
Table 6.5: Agronomic and Grain Yield Parameters of the Seed Type across Ten
Communities
No. offSeed type
Plants at 2- Plants affected Plants at
Grain
weeks after
by root rot
harvest
type
yield/ha
(25m2)
(25m2)
plants
(kg)
planting
2
(25m )
(25m2)
403
3.5
400
1
1500
Certify seed (G)
403

30

375

2

870

394

90

304

26

455

Mean

400.0

40.5

360

9.4

941

LSD

9.82

17.72

22.5

6.30

286.9

CV%

2.6

46.6

6.7

71.3

32.4

Quality declared
seed (L)
Recycled seed (M)

The means represent data from the study sites of 10 communities
LSD is the Least Significant Difference
CV is the Coefficient of Variation

The parameters considered necessary to bring out differences among the seed types were plant
establishment two weeks after planting, plants affected by root rot disease, the number of offtype plants per seed type, plants that survived till harvesting and grain yield. Other agronomic

32

data including dates to flowering, maturity, plant height, etc. were collected but not reported here
because they were not different among the seed types.
Although the certified seed (plot G) and the quality declared seed (plot L) recorded higher plant
establishment than the recycled seed plots, the differences were not statistically significant.
Across all the ten communities, at two weeks after planting, all the seed types recorded over 90%
establishment, which is considered good.
A significantly higher number of plants in plot M (recycled seed) were severely infected with the
root rot disease. On the other hand, the certified seed (plot G) recorded significantly lower
number of plants with root rot disease. The number of root rot infected plants recorded against
the recycled seed plot was 3 and over 25 times higher than the infected plants in the QDS and
certified seed plots, respectively.
The total number of plants that survived till the time of harvesting ranged between 304 to 400
per plot planted. Certified seed recorded significantly higher number of plant survival than the
other two seed types, the seed type with the least number of plants at harvest was recycled seed.
When the number of plants at harvest were expressed as a percentage over plants that were
established 2 weeks after planting, certified seed plot recorded 99% survival, QDS had 93%
survival rate, while in plots with recycled seeds only 77% of plants survived until the harvest
time.
The recycled seed plot recorded significantly higher number of cowpea plants as off-types
(indicating mixtures of varieties other than Songotra). The off-type plants recorded in the
certified and QDS plots were lower, and did not differ significantly. The number of off-type
plants found in the recycled seed plot were 26 and 13 times higher than those found in certified
and QDS seed plots, respectively.

33

On average, across all 10 villages certified seed plots recorded significantly higher grain yield
per hectare (1500 kg/ha) than yields on both QDS (870 kg/ha) and recycled seed plots (455
kg/ha) (Table 6.5). The seed type with the least grain yield per hectare was recycled seed which
was also significantly lower than average yield recorded for QDS plots. This relative difference
and statistical significance between the grain yields of these three seed types is also observed in
the 8 villages where farmer field days and BEA were conducted (Table 6.6)
Table 6.6: Yield of Different Seed Types across Field Experiments
Harvesting Stage
Seed type
Plot G: Certified Seed
Plot L: Quality Declared Seed
Plot M: Recycled Seed

Yield kg/ha (N=8)
1533.78
975.09
444.67

Standard Deviation
337.85
402.74
156.45

Source: Field Experiment data, Ghana-2016

6.2 Results of Bidding Experimental Auction
On average, farmers indicated they were willing to pay GHC7.19 for one kg seed of type G,
which on average was ranked the best among the three seed types. The WTP for seed type G
(certified seed), on average was higher than that of seed type L (QDS) (GHC 5.27) and seed type
M (recycled seeds) (GHC 4.90). Differences in the WTP between seed type G and seed types L
and M are statistically significant at P=0.01. The difference in the mean prices between seed
types L and M is also statistically significant at P=0.01

34

Table 6.7: Participants Willingness to Pay for Different Seed Types (GHC/kg)
Seed type

Mean

Plot G: Certified Seed
Plot L: Quality Declared Seed
Plot M: Recycled Seed

7.19
5.27
4.90

Standard Deviation

N

2.16
2.11
2.19

269
269
269

Source: Bidding Experimental Auction data, Northern Ghana-2016
Average cowpea grain price =Ghc 2.80/kg

7.2

6

7

8

Figure 6.1: Average Bidding Price (Ghana Cedi GHC/kg) for Different Seed Types

5.3

Average
cowpea
grain price
reported
by farmers
was GHC
2.80/kg

0

1

2

3

4

5

4.9

PLOT

PLOT L

PLOT M

Farmers are on average willing to pay a 73% and 20% price premium for best rated seed type G
(certified) and seed type L (quality declared), respectively, over seed type M (recycled seed),
which was rated ‘worst’ by most farmers during the harvest stage (table 6.8). This implies that
farmers are willing to pay higher prices for higher perceived seed quality based on the observed
agronomic performance of plants corresponding to the three seed types planted in farmers’ fields.

35

Table 6.8: Average WTP a Premium for Seed Type G (Certified) and L (QDS) over Lowest
Ranked Seed Quality (Recycled)
Seed type
Plot G: Certified Seed
Plot L: Quality Declared Seed

Premium farmers are WTP over recycled seed
73.00%
20.14%

Source: author’s estimation from BEA data, Northern Ghana-2016

Figure 6.8 presents farmers’ willingness to pay for the highest quality seed type rated by most
farmers (i.e., certified seed) as a percentage of the WTP for the recycled seed, which was rated
the lowest quality seed by most farmers. The WTP a premium for quality seed (i.e., demand
curve for seed quality) is plotted against the percentage of farmers that are willing to pay that
premium over the price of the recycled seed. As indicated by the downward sloping demand
curve in figure 6.2, the percentage of farmers willing to pay a premium for certified over
recycled seed declines as the price premium goes up. Conversely, more people are willing to
purchase certified seed as the price premium relative to recycled seed goes down. This result is
consistent with the demand theory.

Price of certified seed as percentage
of price of recycled seed

Figure 6.2: Distribution of Willingness to Pay for the Highest Rated Seed (i.e., Certified
Seed) Relative to Lowest Rated Seed (i.e., Recycled Grain) (Demand Curve)
4.5
4
3.5
3
2.5
2
1.5
1
0%

20%

40%

60%

Percentage of farmers

Source: author’s estimation from BEA Northern Ghana-2016

36

80%

100%

Given that cowpea grain is a substitute for ‘seed’ as a planting material, we estimated the price
premium that farmers were willing to pay for different types of seed over the market price for
grain that was reported by farmers. This premium was normalized to GHC1 for grain price to
express the WTP for different types of seed as a percentage of the grain price.
The downward sloping curves for each of the three seed types (certified, QDS and recycled) in
figure 6.3 confirm the inverse relationship between farmers’ WTP a premium for perceived
quality and the quantity demanded as expressed by the number of farmers, each willing to
purchase at least 1 kg of seed of a given type for a given price premium. The downward sloping
demand curves indicate that more farmers are willing to purchase seed as the price premium goes
down and approaches the grain price. In absolute terms, the WTP for different seed types for
about 10-15% of farmers was even below the market grain price. However, in relative terms,
even this category of farmers were willing to pay a relatively higher price for higher rated seed
(certified) than lower rated seed (QDS and recycled). .
On average, a higher percentage of farmers who participated in the BEA are willing to pay a
premium price for higher perceived quality seed. For example, the percentage of farmers
indicating their WTP a premium that is double the grain price was 30% for the seed plot planted
with recycled seed (type M), 50% for the seed plot planted with QDS (type L), and 80% for the
seed plot planted with certified seed (type G). At each level of premium (i.e., points on the Yaxis in Figure 6.3), there are more farmers willing to pay that premium price for certified seed
than for QDS seed, and for recycled seed. This shows that farmers’ WTP for seed is positively
correlated with the perceived quality of seed they observed on the experimental plots.

37

Figure 6.3: Willingness to pay for seed of perceived different quality as a percentage of
cowpea grain price

Seed price as a % of cowpea grain price

6
5
4
3

Average
Cowpea grain
price reported
by farmers
(GHC2.68/kg)

2
1
0
0%

20%

40%

60%

80%

100%

Percentage of farmers
Certified

QDS

Recycled

6.3 Cost of Producing Cowpea Seeds
The estimated total cost of producing a kilogram of certified, QDS and cowpea grain (i.e.,
recycled seeds) were GHC 5.82, GHC 4.08 and GHC 2.86, respectively (table 6.9). The relative
total cost of producing a kilogram of certified seeds is 2.03 times the cost of producing cowpea
grain. QDS on the other hand costs 1.4 times to produce compared to the cost of producing a
kilogram of recycled seed.
Table 6.9: Relative Cost of Production for Certified and QDS over Recycled Seed
Seed Type
Cost/kg (GHC)
Relative difference
Certified Seed (N=2)

5.82

2.03 x recycled seed cost

Quality Declared Seed (N=3)

4.08

1.4 x recycled seed cost

Recycled Seed (N=6)

2.86

--

Source: Cost of cowpea seed production survey, Northern Ghana-2016

38

In the cost of producing certified seeds, labor cost makes up 77% of the total cost and inputs
constituted 15% (table 6.10). Cost of processing, testing and packaging of certified seed (which
only applies to this type of seed) was only 8% of the total cost. Labor cost still was the highest
expenditure item in the production of QDS (75%) and input cost was 25%. There was no
processing and testing cost associated with producing QDS and recycled seeds. This expenditure
trend did not change in recycled seed production. Labor and inputs constituted 78% and 22%
respectively towards total cost of producing a kilo of recycled seed.
Table 6.10: Total Cost of Production (GHC) for Different Seed Types (average across all
producers included in the study), Total Production, and Cost per Kg
Seed type

Input costs

Labor
costs

Certified
Seed (N=2)
Quality
Declared
Seed (N=3)
Recycled
Seed (N=6)

229.25
(15%)
122.98
(25%)

1196.5
(77%)
371.35
(75%)

68.69
(22%)

240.56
(78%)

Postharvest
processing
costs
136.03
(8%)
-

-

Total
cost

Total
production
(kg/acre)

Cost/
kg

1,561.78

268.34

5.82

494.33

121.15

4.08

309.25

108.12

2.86

Source: Cost of cowpea seed production survey Northern Ghana-2016
The percentage in parenthesis are with respect to total cost of production

More specifically, pesticide, and seed (input) cost were the major components of total input cost.
With respect to labor cost, the cost for weeding and harvesting weighed heavily on total labor
cost of producing each seed type. Also, the single most expensive cost item was weeding cost for
certified seeds, labor cost for harvesting for QDS and recycled seeds (see Appendix B for a
breakdown of production costs by different categories). The absolute cost for inputs, labor and
post-harvest were significantly higher for certified seed than for producing QDS and grain. The
total output for certified seed was higher than total output from QDS and grain but not as high as

39

the relative cost difference. Thus, the combination of higher relative cost compared to quantity
produced contributes to the higher cost/kg for certified seed.
It is worth noting that the two certified seed producers included in the sample did not follow all
the requirements of producing ‘certified seed.’ For example, they reported paying no annual
registration fee, which is outlined as one of the requirements for certified seeds producers. This
explains why they also did not report any visits by seed inspectors during the growing season.
Also, they used certified seeds as planting material instead of foundation seeds, which are more
costly, and as per the guidelines should be obtained from the public research station. On both
these counts the cost estimation in table 6.10 may be an underestimation of the cost of producing
certified seed if all the regulatory requirements are followed. Nonetheless, at the end of the
season, the seeds produced by these two producers were sold as ‘certified’ seeds and thus
included as certified seed category in this cost comparison.

40

CHAPTER 7: DETERMINANTS OF WILLINGNESS TO PAY FOR QUALITY SEEDS
A farmer’s decision to purchase a particular seed type depends on the observed, experience as
well as the credence attributes of the seed type, according to the Lancaster demand theory.
However, Farmers are not able to tell by observation, the experience and credence attribute of a
seed type. The field experiment served as a mechanism for farmers to know the experience
attribute of the seed before the auction exercise. Again, we eliminated the time factor from our
analysis and model because we are interested in the relative performance and WTP differences
among the three cowpea seed types. Whether the farmer decides to buy the seed before the
planting season or after harvesting does not really matter in our design.
Holding the genetic makeup of the seed, environmental conditions, time, and management
practices constant, what influences farmer’s WTP for quality seeds? How does perceived seed
quality which is formed from observing the differential performance of the seed types, affects
farmers’ WTP for different quality seed? The purpose of the regression analysis is to explore
these questions. The null hypothesis tested is:
Ho: The relative differences in the WTP for different types of seed products is highly
influenced by the perceived differences in seed quality.
We used Ordinary Least Square (OLS) estimation method with and without the household fixed
effects to explore the relationship between perceived seed quality and WTP for quality seed. In
these models, the dependent variable, WTP by farmer i for seed type G (i.e., certified seed) and
seed type L (i.e., QDS) is expressed as a percentage of that farmer’s bid for type M (i.e., recycled
seeds). In both the regression models standard errors are clustered at the village level. The results
are reported in table 7.1.

41

Table 7.1: Factors that Influences farmers’ WTP for Different Types of Seed Products:
Model Estimation using Household Fixed Effects (Model 1) and without (Model 2)
Dependent variable=Willingness to Pay for 1 kg seed
relative to bid price for seed type M (recycled seed)
Perceived seed quality (1=worst and 3=best)
# of years planting cowpea
1=Belongs to a Farmer Based Organization
Total land area planted to cowpea
1=Has attended formal education
Farmer's age (Years)
1=Male farmer
1=Early adopter of new technologies (self-reported)
1=Can read/write in English and other languages
1=more than 50% of hh income comes from cowpea sales
(self-reported)
1=Household has more than 75% probability of living below
$1.25/capita/day
Cowpea grain price reported by farmers
1=Adopted inputs and new farm practice in the last 3 years
(Self-reported)
1=have used certified seeds of any crop in the past
1=Used own saved seeds for planting in the previous season
1=Purchased as seed/grain from market or others in the
previous season
Importance of cowpea in terms of purchased inputs devoted for
production (self-reported ranking of 3 most important crops)
Cowpea yield per acre in the last season

(1)

(2)

0.302***
(0.021)

0.302***
(0.002)
-0.003
(0.002)
0.028
(0.051)
-0.072**
(0.036)
0.052
(0.072)
0.0003
(0.002)
0.095*
(0.049)
-0.056
(0.045)
0.020
(0.079)
-0.075
(0.047)
0.509***
(0.142)
0.056
(0.050)
-0.070
(0.045)
-0.069
(0.055)
-0.034
(0.062)
-0.052
(0.066)
-0.045
(0.060)
-0.00002
(0.0001)
0.016

Farmer purchases seed at least every two years

42

Table 7.1: cont'd
Dependent variable=Willingness to Pay for 1 kg seed
relative to bid price for seed type M (recycled seed)
Farmer purchases seed at least every two years
Household dummies
Constant
Observations
R2

(1)

(2)

Yes

(0.047)
No

0.669***
(0.028)
799
0.18

0.700
(0.189)
771
0.23

Source: Authors’ estimation from survey and BEA data in Ghana-2016
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

From the summary statistics in chapter 5, we saw that farmers are willing to pay different prices
for different perceived seed quality. Even after controlling for other confounding factors, there is
a positive correlation between perceived quality of seed and farmer’s willingness to pay for one
kg seed of that seed type. As the perceived seed quality increases from a scale of 1=worst to
3=best, farmers are willing to pay 30% more for each incremental quality rating of the seed type
relative to their WTP for the recycled seed. These results are robust and statistically significant.
Total area planted to cowpea has a negative and statistically significant effect on WTP for
quality seeds. This means that as the total area used to plant increases by an acre, the household
is willing to pay 7% less for quality seeds. This could be due to a host of reasons. Quality seeds
requires some amount of complimentary inputs like fertilizer and pesticides to realize the full
potential of the seed. All these complementary inputs come at a cost to the farmer. This might
serve as a disincentive to pay a premium for quality seeds as they increase with the total area
planted to cowpea.
Male farmers are willing to pay 9% more premium for quality seeds than female farmers, and
this estimate is statistically significant at p<0.10. This means that all else equal, male farmers are
43

more willing to pay a premium for perceived higher quality seeds based on their experience of
the seed’s performance in the FE.
Results also indicate that farmers with a higher probability of being poor are willing to pay more
premium for quality cowpea seeds compared to farmers with lower probability of being poor. On
average, farmers with a 75% probability of living below the internationally defined poverty line
are willing to pay about 50% more premium than their counterparts. Poverty score is estimated
based on 10 country-specific indicators of household’s demographic and living conditions that
can be predictors of the probability of a household’s poverty status. Since the poverty score is
considered a proxy for household income and wealth status, this result is surprising. However, it
may point to the fact that poor farmers may be more seed insecure or lack facilities to maintain
quality seed on their farm, and thus value the quality seed and react more positively to the seed
quality than their counterparts.
The relationship between farmer group/association membership and the relative price premium
for seed type predicted by this model is generally aligned with expectation. Respondents who
belong to a farmer based group or association were willing to pay 2.7% more than their
counterparts who do not belong to any group, but this positive association is not statistically
significant. Mostly, government extension agents and some NGOs consider working with farmer
groups in communities as more efficient and cost saving method for promoting technologies and
inputs. They are able to reach out to more farmers through such organizations than through
individual farmer focused extension. It is through such meetings that new technologies are
introduced to farmers. Farmers also get the opportunity to learn from the experiences of other
group members.
Farmers who are already investing a significant amount on inputs for cowpea production are
44

willing to pay less for quality seed relative to farmers who currently spend less on purchased
inputs. Also, given that most quality seeds of improved varieties require the use of other inputs to
realize its full potential, this positive but statistically not significant, correlation between the
willingness to pay a higher price (and thus the ability to access quality seed) and the use of other
complementary inputs can potentially have a productivity boosting effect.
Surprisingly, farmers who had used certified seeds of any crop in the past were willing to pay 7%
less premium for quality seed than farmers who had never used certified seeds. Perhaps, farmers
who have used certified seeds in the past might not have had a positive experience with the seed
quality. . Their lower relative WTP for quality seed may be because they did not see the claimed
seed quality for certified translate into higher yields. Nonetheless, it should be pointed out that
his negative relationship is not statistically significant, and thus cannot be considered robust or
an evidence in support of this claim.
Age had a positive and insignificant effect on the prices that farmers were willing to pay for
quality seeds. Formal education, and years of experience had no effect on the WTP for different
types of seeds. Regular seed buyers, source of seed, and the reported grain purchase price also
did not have any statistically significant effect on the WTP.

45

CHAPTER 8: CONCLUSIONS, MAIN FINDINGS, IMPLICATIONS AND NEED FOR
FUTURE RESEARCH
8.1 Main Findings
This study has highlighted three important results. First, the three types of cowpea seeds
potentially available to farmers as planting material were qualitatively different as reflected in
their yield differences and other agronomic performance indicators of plant growth observed on
the double blind field experimental plots hosted by cowpea farmers. As expected, certified seed
performed better than QDS, which in turn performed better than recycled seeds of the same
variety (i.e., Sangotra). Keeping all else constant, on average, plots planted with certified seed
yielded 3 times more cowpea than plots planted with recycled seed. This is a large and
significant productivity effect from improved seed quality that has not been previously
emphasized in the literature.
Secondly, farmers were able to perceive quality differences between the three types of seed plots
as reflected in the farmer ratings based on plant performances observed at the flowering and
harvesting stages. The perception in relative seed quality differences were highly correlated with
the objective measures of relative performance differences recorded by the research staff such as
yield, number of plants infected by diseases, number of off-types, and number of plants that
survived till harvest. Plots that were ranked as the best plot by most farmers had the highest
yield, least number of disease infected plants or off-types, and most number of plants at harvest.
These were the plots planted with the certified seed (or seed type G). Similarly, plots that were
ranked the worst by most farmers had the lowest yield, highest number of disease infected and
off-type plants, and these were the plots planted with the recycled seeds. Higher perceived yield
and high seed quality (based on observable traits) were the major reasons why a majority of
farmers rated seed type G (certified) as the best plot. Although the results of this study are based
46

on experimental plots established on 10 farms, they are highly encouraging. Sometimes there are
concerns that this type of experimental trials are conducted on research stations or managed by
researchers, which may not be representative of farmers’ conditions. This study was designed to
address this concern by setting up experimental plots which were managed by host farmers under
their conditions in each village.
Third important finding of this study is that, farmers are indeed willing to pay a price premium
for perceived higher quality of seeds. The relative difference in farmers’ willingness to pay a
premium price for different types of seeds was positively correlated with the relative difference
in the farmers’ perceived quality ratings (which in turn were correlated with the actual
performance differences). On average, farmers in this study were willing to pay 73% more for
the highest rated seed (i.e., certified seed) compared to the average bidding price for the lowest
rated seed (i.e., recycled seed). Again the whole analysis was based on comparison of relative
prices rather than absolute WTP.
Lastly, on the supply side, this study documented the costs involved in producing different
quality seeds and found that both in absolute and relative terms it costs more to produce certified
seed than QDS, which in turn costs more to produce than cowpea grain (i.e,, recycled seeds). For
the sample of seed producers included in this study, the estimated cost of producing one kg seed
of certified seed and QDS was, respectively, two and 1.4 times the cost of producing cowpea
grain.

47

8.2 Implications for Strategies to Make Quality Seeds Accessible to Different Farmers
The findings of this study have important implications for seed dissemination strategies and
formal seed sector sustainability. The results from the field experiment, BEA and the cost of seed
production survey are brought together in figure 8.1 to highlight these implications. The
downward sloping curve represents farmers’ WTP for certified seed, which was the highest
quality seed rated by them based on the performance of the experimental plots. This is expressed
as a percentage of premium farmers are willing to pay over the market price of cowpea grain.
This represents the demand side of a potential seed market in the study area.
The three horizontal lines represent three scenarios of potential supply curves in these
communities. The solid line in the bottom represents the market price for cowpea grain, the
middle dotted line is the estimated cost of producing certified seed (which is approximately twice
the cowpea grain price) and the top dashed line is the market price of certified cowpea seed in
the study area (i.e., 2.8 times cowpea grain price).
The intersections of these three horizontal lines with the demand curve potentially divides
cowpea farmers surveyed for this study into three groups represented by the purple, yellow and
green blocks. Farmers within the purple region are willing to pay a premium (relative to the grain
price) that is higher than the relative difference between the market price of certified seeds and
cowpea grain price. This implies that farmers in this category can be potentially served by formal
seed system based on private sector certified seed production and sale. There are approximately
40% of farmers in this study sample in this category.

48

Seed price as percentage of cowpea grain
price (x100)

Figure 8.1: Farmers’ Willingness to Pay for Quality Seed Relative to Market Price of
Grain, Certified Seed and Cost of Seed Production: Identification of three Groups of
Farmers that Require Different Approaches to Meet their Seed Needs
6

5

4

3

Group 1
2

Group 2
1

Group 3

0

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Percentage of Farmers
Bottom solid line - Grain market price
Middle dotted line - Estimated cost of producing certified seed (2 x Grain price) Top dash line Market price of certified seed (2.8 x Grain pr

Farmers in the yellow region are willing to pay a premium that is higher than the cost of
producing certified seed but less than the market price of certified seed. This means that the
premium they are willing to pay (relative to grain prices) is not enough to attract seed producers
to supply quality seeds. About 40% famers in our sample fall in this category. Alternative
models are needed to cater the needs of this group of farmers who are willing to pay a premium
price for seed that are equivalent in quality as certified seeds included in this study, but not high
enough as the market price of certified seed.
The third group of farmers that fall in the green region are farmers (~20%) whose WTP premium
price for quality seed is not enough to recover the cost of producing certified seeds. For a subset
49

of this group of farmers the WTP is even lower than the grain price. This group of farmers
should be the ideal targets for subsidized or free seed distribution efforts if the goal is to promote
the use of quality seeds. The challenge, of course, is to find ways to reach these farmers without
undermining the market-led (or alternative non-subsidy based) approaches that may exist to cater
the needs of group 1 and 2 farmers.
8.3 Need for Future Research
This study was designed to explore the impact of perceived cowpea seed quality on farmers’
willingness to pay a premium for different types of cowpea seed products. The results of this
study provides valuable insights for crop improvement and seed system research
One way to make quality seeds accessible to smallholder farmers is to lower the price of certified
seed, which in turn implies lowering the cost of seed production. Further research and discussion
needs to happen on how to reduce the cost of seed production. Training seed producers in
appropriate methods and technology for seed production, processing, storage and marketing that
can lower seed losses is critical in increasing seed yields and lowering per unit cost. Seed
suppliers and rural shopkeepers must be also provided with basic training in seed business skills
and bookkeeping. The effort to train farmers on ways to improve farming practices will require a
lot of time investment from farmers. However, it is not clear the extent to which farmers are
willing to invest their time for such training. Over the years, researchers have pushed for
involving farmers in seed selection and production to produce preferred foundation seeds and
farmers’ access to these seeds thereafter. This seed accessibility should be complemented with
the right recognition of seed breeders for their good work done.
A second issue that needs further research and policy dialogue within Ghana is the regulatory
role of the government in building a sustainable seed system. One important question that needs
50

to be addressed is how can farmers be assured of the seed quality (credence attribute) without
making the regulatory process costly and cumbersome? How to ensure that the contracts between
seed sellers and farmers (as seed purchasers) are honored?
In most developing countries, NGOs play a major role and are a main source for the delivery of
legume seeds to smallholder farmers. A key issue with their involvement is sustainability of their
efforts, and the negative impact they can potentially have on the survival of for-profit business
models. Thus, it is now widely recognized that instead of directly playing a major role in seed
delivery themselves, NGOs should be supporting local seed producers and distributors in ways
that promote their sales or decrease their costs. However, as shown in this study, NGOs and
government led efforts may still be needed to meet the needs of a small target number of farmers
who fall on the lower end of the WTP demand curve. For example, about 20% of farmers in our
study sample offered bids for seeds (relative to grain price) that were lower than the cost of
producing quality seed or even the grain price. This may be a reflection of their lack of
affordability or their lack of appreciation for quality seed as a planting material. Hence a seed
system that is based 100% on the principle of full cost-recovery might not work if the goal is to
reach all types of farmers (including this marginal group of farmers) with quality seeds. A
subsidy based approach might still be necessary to reach farmers with low affordability to pay
for seed. There is also a need for the public extension programs to increase farmer’s knowledge
and awareness about the importance of ‘quality’ seed to increase the demand for seeds.
Although the average WTP for quality seeds was higher than the grain price and also got a
favorable ranking for preferences from farmers for quality seeds, it is important to note that there
are other ingredients needed to building a sustainable seed system. Factors such as the logistics
of seed distribution, distance a farmer will have to travel to get these seeds, how the identity and
51

quality of seed is preserved throughout the value chain may influence the functioning of a seed
system. More research that combines farmers’ WTP for seed with a host of other supply side
factors is needed to better understand the conditioning factors to build a sustainable and efficient
seed system.
Finally, it is important to outline a few limitations of this study and the need for future research
to address them. First, the fact that farmers did not know the seed type they were bidding for at
the time they submitted their bids could partly explain why some farmers were inconsistent with
their decisions and choices. Since the main interest in this study was to find the relative price
premium/discount associated with the perceived seed quality, the study is not able to say much
about the absolute WTP for different seed types. Perhaps, doing incentivized experiments with
actual seed types that farmers can buy in the market may be useful in capturing the absolute
WTP for each seed type. Second, the BEAs were designed to solicit bids from farmers for one
kg of seed, which does not allow the estimation of quantities of seed demanded at that bid.
Future studies based on this methodology should also collect data on the quantities farmers are
willing to purchase at their bid price and the frequency of purchase to be able to gauge the size of
the demand for quality seeds.
Lastly, one of the important contributions of this study is the systematic collection of cost of seed
production data across the three types of seed qualities. However, the sample size for estimating
these costs of seed production was small. More studies documenting the cost of producing
certified, QDS and grain legume seeds in different settings are needed to derive generalizable
results and policy implications.

52

APPENDICES

53

Appendix A: Comparing Cost of Production for Different Seed Types for same Farmer
Table A1: Comparing Cost of Production for Different Seed Types by Same Farmers
Total Cost GHC
Farmer ID

Certified

QDS

Farmer 1
Farmer 2
Farmer 3
Farmer 5
Farmer 6
Farmer 7
Farmer 8
Farmer 9
Farmer 10
Farmer 11

8.04
3.59
-

2.33
2.33
7.13
-

Recycled
7.10
3.83
1.73
1.71
4.67
1.30
2.75
1.69
2.90

Source: Cost of cowpea seed production survey Northern Ghana-2016
10 farmers producing different seed types were survey during the 2016 major farming season in northern Ghana.

54

Appendix B: Specific Input Cost as a Percentage of Total Input and Labor Cost
Table B1: Specific Input Cost as a Percentage of Total Input and Labor Cost for Each Seed
Type.
Input:
Seed cost
Pesticide
Labor:
Weeding
Harvesting

Certified

QDS

Recycled

62%

68%

37%

45%

24%

40%

36%
30%

24%
35%

25%
48%

Source: Cost of cowpea seed production survey Northern Ghana-2016

55

Appendix C: Cost of Producing Certified Seeds
Table C1: Cost of Producing Certified Seed
Input
seed cost
fertilizer cost
pesticide cost
herbicide cost
Labor cost
plowing w animal
cost
plowing w tractor
cost
planting cost
replanting cost
fertilizer application
cost
herbicide application
cost
fungicide application
cost
pesticide application
cost
weeding cost
harvesting cost
threshing cost
winnowing cost
transport cost
drying cost
bagging cost
Post-harvest
Quantity harvested
processing cost
packaging cost
testing cost -- fixed
total cost
total cost per kg
Observations

count
2
2
2
2

min
126.00
0.00
45.00
0.00

mean
141.75
0.00
87.50
0.00

sd
22.27
0.00
60.10
0.00

max
157.50
0.00
130.00
0.00

2

0.00

75.00

106.07

150.00

2
2
2

0.00
56.00
0.00

75.00
70.00
24.50

106.07
19.80
34.65

150.00
84.00
49.00

2

0.00

0.00

0.00

0.00

2

0.00

10.00

14.14

20.00

2

0.00

0.00

0.00

0.00

2
2
2
2
2
2
2
2
0
2
2
2
2
2
2
2

20.00
140.00
226.00
41.00
24.00
0.00
42.00
5.00

137.50
317.50
311.00
52.00
34.00
25.00
52.00
13.00

166.17
251.02
120.21
15.56
14.14
35.36
14.14
11.31

255.00
495.00
396.00
63.00
44.00
50.00
62.00
21.00

225.00
36.12
15.48
60.00
1312.46
3.60

295.00
53.22
22.81
60.00
1561.78
5.82

98.99
24.18
10.36
0.00
352.60
3.15

365.00
70.32
30.14
60.00
1811.11
8.05

56

Appendix D: Cost of Producing Quality Declared Seeds
Table D1: Cost of Producing Quality Declared Seeds
Input Cost
seed cost
fertilizer cost
pesticide cost
herbicide cost
Labor cost
plowing w animal
cost
plowing w tractor
cost
planting cost
replanting cost
fertilizer application
cost
herbicide application
cost
fungicide application
cost
pesticide application
cost
weeding cost
harvesting cost
threshing cost
winnowing cost
transport cost
drying cost
bagging cost
Post-Harvest
Quantity harvested
processing cost
packaging cost
testing cost -- fixed
total cost
total cost per kg
Observations

N
4
4
4
4

Min.
16.00
0.00
1.92
0.00

Mean
77.50
2.50
39.23
3.75

SD
98.08
5.00
36.85
7.50

Max
224.00
10.00
80.00
15.00

4

15.00

50.00

60.14

140.00

4
4
4

0.00
8.00
0.00

0.00
34.25
0.00

0.00
29.92
0.00

0.00
77.00
0.00

4

0.00

1.50

3.00

6.00

4

0.00

2.75

5.50

11.00

4

0.00

2.00

4.00

8.00

4
4
4
4
4
4
4
4

0.00
8.00
0.00
0.00
0.00
0.00
0.00
0.00

22.50
66.25
80.00
14.00
7.00
0.00
3.50
0.00

23.85
72.87
110.04
28.00
14.00
0.00
7.00
0.00

56.00
162.00
238.00
56.00
28.00
0.00
14.00
0.00

4
4
4
4
3
3
4

15.00
0.00
0.00
0.00
107.00
2.34

141.75
0.00
0.00
0.00
494.33
4.08

159.79
0.00
0.00
0.00
456.07
2.65

360.00
0.00
0.00
0.00
997.00
7.13

57

Appendix E: Cost of Producing Recycled Seeds
Table E1: Cost of Producing Recycled Seed
Input Cost
seed cost
fertilizer cost
pesticide cost
herbicide cost
Labor cost
plowing w animal
cost
plowing w tractor
cost
planting cost
replanting cost
fertilizer application
cost
herbicide application
cost
fungicide application
cost
pesticide application
cost
weeding cost
harvesting cost
threshing cost
winnowing cost
transport cost
drying cost
bagging cost
Post-Harvest Cost
Quantity harvested
processing cost
packaging cost
testing cost -- fixed
total cost
total cost per kg
Observations

count
10
10
10
10
0

min
5.00
0.00
7.20
0.00
.

mean
28.80
7.52
28.87
3.50
.

sd
24.59
10.50
24.72
7.47
.

max
84.00
27.00
90.50
20.00
.

10

0.00

4.90

11.45

35.00

10
10
10

0.00
0.00
0.00

28.93
5.00
24.00

19.50
15.81
16.01

60.00
50.00
48.00

10

0.00

1.70

3.65

10.00

10

0.00

2.50

4.95

15.00

10

0.00

3.40

5.58

14.00

10
10
10
10
10
10
10
10
0
10
10
10
10
9
9
10

0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
.
3.00
0.00
0.00
0.00
157.20
1.30

20.60
24.20
69.90
16.40
5.30
4.55
3.50
0.00
.
108.55
0.00
0.00
0.00
309.25
2.86

19.27
33.48
53.52
19.95
9.31
13.04
5.10
0.00
.
61.27
0.00
0.00
0.00
131.75
1.27

63.00
90.00
171.00
56.00
28.00
41.50
14.00
0.00
.
195.00
0.00
0.00
0.00
573.50
4.67

58

Appendix F: Bidding Experimental Auction Script and Consent Form
WILLIGNESS TO PAY FOR QUALITY COWPEA SEEDS: SCRIPT FOR
CONDUCTING BIDDING EXPERIMENTAL AUCTIONS
NOTE: All text in italics are instructions for the enumerator. All text not in italics must be read
to the farmer.
This experiment/survey will be performed at field days in 8 villages in northern Ghana. Each
village has 1 field experiment (FE) and the field days will be run in all of these 8 villages.
During each field day, 35 farmers (who attended the first field day and were surveyed) will
participate in a willingness-to-pay (WTP) auction experiment. A FE consists of one field split
into 3 plots. All of the plots contain the same variety of cowpea, but were planted using different
qualities of seed – Certified seed (CS) Quality Declared seed (QDS) or recycled Seed (RS). The
plots are labeled M, G, and L and farmers and extension agents do not know (and should not be
told) which quality of seed was used for which plot.
When signing in farmers, make sure they are one of the 35 chosen and give them a name tag with
their number on it. After a brief welcome to the field day and running through the criteria and
plot ranking exercise – including a question regarding WTP per kg for each plot (). the attending
farmers will be divide into 2 groups (1-17, 18-35)It is required that only one person/household
participate. The script below is for the enumerator and helpers running the WTP auctions.
Step 1:Introduction/consent
The enumerator will introduce his or herself and read the consent script to the farmers and
record their verbal consents to participate.
CONSENT STATEMENT Hello, my name is (INTRODUCE YOURSELF). I am assisting
the Savannah Agriculture Research Institute (SARI) from Ghana and Michigan State University
(MSU) from the US in conducting a study to document farmers’ willingness to pay for
different types of cowpea seeds. As part of this study, we will be conducting an auction of
cowpea seeds in which you will be asked to bid a price you are willing to pay for one kg of
cowpea seed of different types. We will explain the rules of this game in detail, and will also
play a practice game before we start the seed bidding game. Your participation in this game will
not involve any out of pocket cost for you. The game will take approximately 60 minutes. Your
participation is voluntary. Your refusal to participate or to withdraw from the study carries no
penalty or loss of any benefits.
59

The results of your participation in the game will be used to document what cowpea growers are
willing to pay for different types of cowpea seed, when available; which will help us to generate
recommendations that may benefit cowpea production in this region.
Do you have any questions about this study for me?
<Enumerator: pause and respond to any questions raised, then continue with the following
statement>.
By continuing to be part of this group, you indicate your willingness to voluntarily
participate in the bidding game.
Step 2: Overall description of Experiment
ENUMERATOR:
Ok, thank you for being willing to participate. To begin with, let me give you an overall
description about what we will be doing. We are interested in getting an idea about how much
you would be willing to pay for the seed quality that was used to grow each of the 3 plots that
you looked at earlier. To make your decisions more realistic, we are going to give you 10 GHC
that you can use to purchase a one kg bag of one of the seeds used to grow one of the plots.
But before we do that, we would like to do a practice auction where we will give each of you 2
GHC in order to bid on purchasing a bar of soap like this one.
Hold up bar of soap
Let’s do the practice auction first, and then we will explain more about the seed auction, ok?
Do you have any questions? Should we begin
Step 3: Practice Auction
The enumerator will begin explaining the practice auction.
ENUMERATOR:
Ok, so for this practice auction each of you will be given 2GHC to bid on one bar of soap. Unlike
in most auctions, or in auctions you may have participated in in the past, in the type of auction
we will be using, it is possible for everyone to win and thus everyone might purchase a bar of
soap using part or all of their 2GHC.
Let me step through how you bid and how we determine who wins and buys a bar of this soap.
First, we will hand out a bidding sheet like this one.
Hold up bidding sheet.

60

On this bidding sheet you will write down the maximum amount you would be willing to
pay for this bar of soap (in increments of 0.10 GHC). Once everyone has done this, we
will collect the bidding sheets and move on to determine how many of you win and buy a
bar of soap.
To determine who wins we simply choose a random price between 0 and 2GHC – we will
explain how in a moment.
If the price you bid is greater than or equal to this random price, then you win, BUT
you pay the random price – not what you bid. This means that if you win, you pay a
lower price for the soap than you bid (unless the random price is the same as your bid).
On the other hand, if the price you bid is less than this random price, then you do not
purchase the soap and you can keep the money.
If you win, we will give you a bar of soap and the remaining amount of your 2GHC; that
is, 2GHC minus the random price.
For example, if you bid GHC1.30 and the random price is GHC1, then you would pay
GHC1 for the soap and get it, along with the remaining GHC1.
If you do not end up buying the soap, you do not spend any of your GHC2 buying soap
and we will give you GHC2.
So, for example, suppose that “name an enumerator1 in the room” bids 2, I bid 1.7 and
“name an enumerator2 in the room” bids 0.5. Now suppose that the random price is
1.5…in this case, enumerator1 would buy the soap, but would pay 1.5GHC, not his /she
bid of 2GHC. He/she would get a bar of soap and 2-1.5=0.5GHC. I would also buy the
soap and pay 1.50 (my bid was 1.7) so I would also get a bar of soap and 2-1.5=0.5GHC.
Enumerator2 would not buy the soap since his/her bid of 0.5 is less than 1.50 so he/she
would just get 2GHC.
Are there any questions?
We will determine the random price as follows
Enumerator: Write on a board two spaces, with a dot in between the first and the second
spaces as in:
_____ . _____ _____
The last digit will always be a zero.

61

The second digit can be be a 0, 1, 2, 3, 4, 5, 6, 7, 8 or 9…we will roll this 10-sided die to
determine which it is…the second digit will be whatever is rolled unless a 10 is rolled. If
a 10 is rolled the second digit will be a 0.
The first digit can be a 0, 1 or 2. We will roll this 4-sided die to determine which it is…if
the die comes up 1 or 2, the first digit will be whatever is rolled. If the die comes up 3,
the first digit will be a 0. If the die comes up 4, we will roll again.
Overall, we will end up with one of the following numbers: 0, .10, 0.20, 0.30, 0.40, 0.50,
0.60, 0.70, 0.80, 0.90, 1.00, 1.10, 1.20, 1.30, 1.40, 1.50, 1.60, 1.70, 1.80, 1.90 or 2.00
right?
Are there any questions so far?
Before we hand out the practice round bidding sheets, let me explain the best strategy in
this type of auction. The BEST thing to do is to bid the MAXIMUM amount you are
willing to pay. This is because it is very likely you will actually pay LESS if you win.
However, bidding less than what you would be willing to pay might mean that you miss
out on buying the soap at a price lower than you would be willing to pay.
Similarly, bidding more than what you would be willing to pay might mean that you end
up having to pay more for the soap than you really want to. For example, if you are
willing to pay a maximum of 1 GHC, but you bid 2 GHC and the random price ends up
being 1.5 GHC, then you would pay 1.5GHC – more than you were willing to!
Overall, your best strategy is to bid the MAXIMUM amount you are willing to pay.
Ok, let’s go ahead and hand out the bidding sheets.
[Hand out bidding sheets]
Ok, go ahead and write down your number (from your name tag – this helps us keep track
of who to pay how much) and your bid for a bar of soap. Please do not talk with others
until we have collected the bids.
[Collect bidding sheets, making sure that bids and numbers are entered and legible and
that the bid is in 0.1GHp increments (i.e., 1.35 is not a valid bid).]
Ok, now let’s go ahead and determine the random price.
[Determine random price as outlined above while writing it down on board. A helper
should record this number on one of the bidding sheets so we have this information. We
can allow farmers to flip coin/role die as long as it is tossed sufficiently to make it
62

random.]
Ok, so this is the price (say the random price) – if you bid more than or equal to this
price, you buy a bar of soap at this price (say the random price). If your bid was less than
this price (say the random price) you will not buy a bar of soap, but will receive the
2GHC.
[It might be a good idea to briefly say “if you feel comfortable sharing, raise your hand if
you bought a bar of soap. “ If they are willing to reveal, they can even say how much
they bid.]
Ok, so we will pay you and give you the soap (if you bought one) after we do the seed
auction.
Step 4: Seed Auction
The enumerator will begin explaining the seed auction.
ENUMERATOR:
Ok, so hopefully you have a better idea about how this seed auction will operate. It will
be very similar to the practice auction you just did, except for a few things:
First, you will be bidding to purchase a one kg bag of the seed that was used to plant the
plots in the field experiment you just looked at. Specifically, you will be making 3 bids –
one for each plot (labeled G, M, and L). HOWEVER, even though you are bidding for
each type, ONLY ONE type will actually have a random price determined and will be
bought/sold. You will not know which type is available until after you bid, so you should
bid as if each one might be the one chosen.
Second, instead of 2GHC, we are giving you 10GHC to use to bid. Just as before, any
amount you do not use to purchase seed, will be given to you after we are done.
Third, the random price can be between 0 and 9.90GHC and will be determined as
follows:
Enumerator: Write on a board three spaces with a dot in between the first and second
spaces as in:
_____ .______ _____
As before, the last digit will always be a zero.

63

The second-to-last digit and the first digit can be a 0, 1, 2, 3, 4, 5, 6, 7, 8 or 9. We will
roll this 10 sided die….if it comes up 1, 2, 3, 4, 5, 6, 7, 8 or 9, it will be that number, but
if it comes up 10, it will be a 0.
Overall, we will end up with a number between 0 and 9.90GHC in 0.10GHC increments?
As before, each number is equally likely.
Are there any questions?
Ok, before we hand out the bidding sheets, let me just remind you that your best strategy
is to bid the MAXIMUM amount you are willing to pay for each seed quality.
Remember, since we are only going to determine a random price for ONE of the seed
qualities, you do NOT need to try and spread your 10GHC across the three seed qualities
– in fact you can bid 10GHC for each quality and not have to worry about spending more
than 10GHC.
Any final questions?
As before, please do not talk with others until we have collected the bids.
[Hand out seed bidding sheets.]
Ok, go ahead and write down your ID number (from the card) and bids for all three seed
qualities. Remember that this is for a 1 kg bag of the seed quality used to plant the
indicated plot (G, M or L). Also, please keep bids to 0.10GhC increments.
[Collect bidding sheets, making sure that bids and numbers are entered and legible and
that all bids are in 0.10GHC increments.(i.e., 9.48GHC is not a valid bid)
Ok, so now we will reveal which seed quality was selected for today’s auction. For this
group, the seed type is [G, M or L as previously determined].
Ok, so now that we know which quality, let’s go ahead and determine the random price.
[Determine random price as outlined above while writing it down on board. A helper
should record this number on one of the bidding sheets so we have this information. We
can allow farmers to flip coin/role die as long as it is tossed sufficiently to make it
random.]
Ok, so this is the price – if you bid more than or equal to this price, you won and will buy
a 1 kg bag of this quality seed at this price. If your bid was less than this price you will
not buy seed, but will receive the 10GHC.

64

Ok, so we will call you up one or two at a time to give you the seed/soap if you bought
them and however much we owe you in GHC.
Thank you and please do not discuss this with the other group of farmers until they have
completed the auctions.
Thank you!

65

Appendix G: Sheet Auction Bidding Sheet
Figure G1: Sheet Auction Bidding Sheet

Seed Auction Bidding Sheet
Number on
Participant Card
Participant Name
Enter the most you are willing to pay.
Your Bid for
Plot M Seed

GHC
Enter the most you are willing to pay.

Your Bid for
Plot G Seed

GHC
Enter the most you are willing to pay.

Your Bid for
Plot L Seed

GHC

Seed Type Selected: M

G

L

Random price: _____________________

Village Name
Village ID
Name of the farmer hosting this FE
Participant purchased the seed

Yes

Total Cash Owed

10 – Random price =

Quantity of seeds willing to buy (kg)
at this price

66

No
GHC

10 GHC

Appendix H: Seed Quality Analysis for Seed Producers
Table H1: Seed Quality Analysis for Seed Producers
Reference ID
005
005
002
001
001
003
004
011
007
010
008
006

Seed type
Grain
QDS
Certified
Grain
Certified
QDS
QDS
Grain
Grain
Grain
Grain
Grain

M.C %
8.2
8.2
9.8
10.2
8.3
8.3
9
8.9
8.8
8.4
8.4
8

Pure
99.9
99.9
99.8
99.9
99.5
99
99.8
99.8
98.5
99.3
99.2
99.2

67

Inert
0.1
0.1
0.2
0.1
0.5
1
0.2
0.2
1.5
0.7
0.8
0.8

Germination
%
0
33
87
0
0
11
24
80
27
21
26
0

Appendix I: Seed Quality Analysis for Experimental Plots after Harvest
Table I1: Seed Quality Analysis for Experimental Plots after Harvest
Reference
community

Seed type

M.C %

Azumsapelega
Azumsapelega
Azumsapelega
Tansia
Tansia
Yalugu
Bansi
Koluku

Certified
QDS
Recycled
QDS
Recycled
QDS
Recycled
Certified

Pure
seed
99.9
99.9
99.9
99.4
99.9
99.9
99.6
99.8

9.5
9.3
9.2
9.2
9.1
9.3
9.2
9.2

68

Purity
Inert
0.1
0.1
0.1
0.6
0.1
0.1
0.4
0.2

Germination %

42
80
24
21
24
80
12
51

REFERENCES

69

REFERENCES

Ahenkora, K., Dapaah, H.A. and Agyemang, A., 1998. “Selected nutritional components and
sensory attributes of cowpea (Vigna unguiculata [L.] Walp) leaves”. Plant Foods for Human
Nutrition, 52(3), pp.221-229.
Appiah, F., Asibuo, J.Y. and Kumah, P., 2011. “Physicochemical and functional properties of
bean flours of three cowpea (Vigna unguiculata L. Walp) varieties in Ghana”. African Journal of
Food Science, 5(2), pp.100-104.
Becker, Gordon M., Morris H. DeGroot, and Jacob Marschak. "Measuring utility by a single‐
response sequential method." Behavioral science 9.3 (1964): 226-232.
Shiferaw, B.A., Kebede, T.A. and You, L., 2008. Technology adoption under seed access
constraints and the economic impacts of improved pigeonpea varieties in Tanzania. Agricultural
Economics, 39(3), pp.309-323.
Colson, Gregory, Jay R. Corrigan, and Matthew C. Rousu. "The Impact of Perceived Prices on
Willingness to Pay in Experimental Auctions." Journal of Agricultural & Food Industrial
Organization 8.1 (2010).
Corrigan, Jay R., and Matthew C. Rousu. "Posted prices and bid affiliation: evidence from
experimental auctions." American Journal of Agricultural Economics 88.4 (2006): 1078-1090.
Corrigan, Jay R., and Matthew C. Rousu. "Testing whether field auction experiments are demand
revealing in practice." Journal of Agricultural and Resource Economics (2008): 290-301.
De Groote, Hugo, Simon Chege Kimenju, and Ulrich B. Morawetz. "Estimating consumer
willingness to pay for food quality with experimental auctions: the case of yellow versus fortified
maize meal in Kenya."Agricultural Economics 42.1 (2011): 1-16.
De Groote, H., Chege, C.K., Tomlins, K., Gunaratna, N.S., 2014. Combining Experimental
auctions with a modified home-use test to assess rural Consumers’ acceptance of quality protein
maize, a biofortified crop. Food Quality Pref. 38, 1–13.
Diallo, Marthe. "Diagnostic Study of Farmers Seed Quality and Variety Preference in Southern
Mali." Order No. 3381112 Michigan State University, 2009. Ann Arbor: ProQuest. Web. 3 Apr.
2016.
Egbadzor, Kenneth Fafa, et al. "Farmer participation in selection within segregating populations
of cowpea in Volta Region, Ghana." Agriculture & Food Security 4.1 (2015): 1.—Participatory
plant breeding.
70

Feuz, Dillon M., et al. "US consumers' willingness to pay for flavor and tenderness in steaks as
determined with an experimental auction." Journal of Agricultural and Resource Economics
(2004): 501-516.
Ghana, Republic of, Statistical Service Department. 2014. Economic Survey, 2014, Accra.
Government printing press.
Harrington, James F. "Seed storage and longevity." Seed biology 3 (1972): 145-245.
Hoffmann, Vivian, and Ken Mwithirwa Gatobu. "Growing their own: Unobservable quality and
the value of self-provisioning." Journal of Development Economics 106 (2014): 168-178.
Kormawa, P. M., C. I. Ezedinma, and B. B. Singh. "Factors influencing farmer-to-farmer transfer
of an improved cowpea variety in Kano State, Nigeria." Journal of Agriculture and Rural
Development in the Tropics and Subtropics (JARTS) 105.1 (2004): 1-13.
Lancaster, Kelvin J. "A new approach to consumer theory." The journal of political economy
(1966): 132-157.
Langyintuo, Augustine S., and Catherine Mungoma. "The effect of household wealth on the
adoption of improved maize varieties in Zambia." Food policy 33.6 (2008): 550-559.
Lusk, Jayson L., and Jason F. Shogren. Experimental auctions: Methods and applications in
economic and marketing research. Cambridge University Press, 2007.
Maredia, Mywish K., et al. increasing seed system efficiency in Africa: concepts, strategies and
issues. No. 54578. Michigan State University, Department of Agricultural, Food, and Resource
Economics, 1999.
Maredia, M., Byerlee, D., Pee, P., 1998. Impacts of food crop improvement research in Africa.
SPAAR Occasional Paper Series No.1, World Bank, Washington, DC. Available
at: http://www.worldbank.org/afr/aftsr/impaper.pdf (accessed on March 11, 2008).
McGuire, Shawn, and Louise Sperling. "Seed systems smallholder farmers use." Food Security
(2015): 1-17.
Morris, M. L., & Bellon, M. R. (2004). Participatory plant breeding research: opportunities and
challenges for the international crop improvement system. Euphytica, 136(1), 21-35.
Morawetz, U.B., De Groote, H., Kimenju, S.C., 2011. Improving the implementation of
experimental auctions in Africa: Theory and evidence. J. Agric. Res. Econ. 36, 263–279.
Munthali, Grace Timanyechi. "Assessing Farmers' Willingness-to-Pay for Improved Common
Bean Seed Varieties in Malawi: A Case Study of Kasungu and Dedza Districts." Order No.
1595869 University of Pretoria (South Africa), 2013. Ann Arbor: ProQuest. Web. 3 Apr. 2016.

71

Nelson, Phillip. "Information and consumer behavior." Journal of political economy 78.2 (1970):
311-329.
Ouédraogo, J. T., et al. "Identification of AFLP markers linked to resistance of cowpea (Vigna
unguiculata L.) to parasitism by Striga gesnerioides." Theoretical and Applied Genetics 102.6-7
(2001): 1029-1036.
Singh, B.B., Ajeigbe, H.A., Tarawali, S.A., Fernandez-Rivera, S. and Abubakar, M., 2003.
“Improving the production and utilization of cowpea as food and fodder”. Field Crops
Research, 84(1), pp.169-177.
Tobin, J., 1958. Estimation of relationships for limited dependent variables.
Econometrica 26 (1), 24–36. http://dx.doi.org/10.2307/1907382.
Waldman, Kurt B., John M. Kerr, and Krista B. Isaacs. "Combining participatory crop trials and
experimental auctions to estimate farmer preferences for improved common bean in
Rwanda." Food Policy 46 (2014): 183-192.

72