Investigating consumer preferences of rice

Economic analysis of consumer
based attributes for rice in
Benin
Saneliso Mhlanga
Department of Agricultural Economics
McGill University
May 2010
A thesis submitted to McGill University in partial fulfillment of the requirements of the
degree of Master of Science
Saneliso Mhlanga, 2010
Abstract
Consumers are becoming more aware of the quality attributes of different
commodities found in the market and are choosing products that closely match
their tastes and preferences. The consumer behavior model postulated by
Lancaster (1966), says that products are consumed for the characteristics they
possess, other than the product itself, and are associated with consumer
preferences/utility. For example, in the case of rice, the quality characteristics
(attributes) have important price based implications in terms of incentives for
producers and consumers. This study empirically analyzes the relationship
between price and product attributes towards consumer‟s choice for rice in Benin,
using the hedonic pricing approach and discrete choice modeling at the household
level. The results of econometric estimation indicate that consumers pay a
premium for grain size, aroma, color, wholeness and cleanliness of grain and
convenience attributes across the different regions studied. Consumers (rural and
urban) prefer parboiled and imported rice over domestic rice and raw rice.
Country of origin was found to influence rice preference indirectly through
perceived quality. Socioeconomic factors are not important in consumer
purchasing decisions. The results from this study suggest that both domestic and
imported rice varieties have positive and negative implicit prices this emphasizes
the importance of quality based attributes in future breeding programs to make
domestic rice more competitive to imported rice.
ii
Résumé
Les consommateurs sont de plus en plus conscients des caractéristiques des
différents produits rencontrés sur le marché et choisissent des produits qui
correspondent le mieux à leurs goûts et préférences. Les caractéristiques de la
qualité du riz ont des implications importantes pour les producteurs, les
chercheurs et les décideurs politiques. Cette étude analyse la relation entre le prix
et les attributs des produits et des préférences des consommateurs pour le cas du
riz au Bénin, en utilisant l'approche des prix hédoniques et la modélisation des
choix discrets et les données sur les ménages. L'analyse est basée sur la nouvelle
théorie du comportement du consommateur de Lancaster (1966), qui postule que
les produits sont consommés plus pour les caractéristiques qu'ils possèdent que le
produit lui- même et sont associées à des préférences/utilité des consommateurs.
Les résultats indiquent que les facteurs les plus importants sont la taille du grain,
l'arôme, les corps étrangers, la couleur, le temps de cuisson et la présence de
brisures. Les résultats montrent également que les consommateurs sont en mesure
de différencier le riz local du riz importé et arrivent à exprimer leurs préférences
sur la qualité. Bien que le riz soit déficitaire au Bénin et que la recherche sur les
variétés à haut rendement continue d'être privilégiée, l'amélioration de la qualité
du riz basée sur les caractéristiques sus identifiées devra it permettre d‟amoindrir
la concurrence entre les riz importé et local. Les décideurs politiques doivent
examiner les incitations réelles et les contraintes à l'amélioration de la qualité au
niveau des exploitations rizicoles.
iii
Acknowledgements
I would like to thank Professor Anwar Naseem for sharing valuable suggestions
and for providing many papers, to enrich my understanding of the topic. I am
grateful to Dr M. Ngadi and Dr A. Diagne (Mr Midingoyi and all staff at
WARDA) and Dr Dodds for rendering their valuable time and providing me with
data and questionnaire translation. I also appreciate the staff of the department of
Agriculture and Natural Resources (especially Prof John Henning and Marie J
Kubecki) for giving me all the necessary assistance during my studies. I am
deeply indebted to my family (mum, late father, brothers and sisters and daughter
Takunda) and friends especially Lina Picon and Enrique Calfucura and Harriet
Okronipa for their love and support, they provided me the energy to attain this
study. Above all I thank my almighty God for giving me the vision and endurance
to complete this interesting research.
iv
DEDICATION
To my family, my Mother Mrs Mhlanga and my lovely daughter Takunda.
v
ACRONYMS
ADRAO
Centre du Riz pour l‟Afrique or WARDA
FCFA
CFA Franc, currency used in formerly French ruled Africa
NERICA
New Rice for Africa
WARDA
Africa Rice Center
COO
Country of origin
GMO
Genetically modified products
vi
Table of Contents
ABSTRACT ................................................................................................................................................. II
RÉSUMÉ .................................................................................................................................................... III
ACKNOWLEDGEMENTS ......................................................................................................................... IV
DEDICATION ............................................................................................................................................. V
ACRONYMS .............................................................................................................................................. VI
CHAPTER 1: IN TRODUCTION ................................................................................................................. 1
1.1 O VERVIEW......................................................................................................................................... 1
1.2 MOTIVATION FOR THE RESEARCH ........................................................................................................ 4
1.3 O VERVIEW OF BENIN RICE MARKET ..................................................................................................... 5
1.4 RESEARCH OBJECTIVES...................................................................................................................... 11
CHAPTER 2: LITERATURE REVIEW ...................................................................................................... 13
2.1 I NTRODUCTION................................................................................................................................ 13
2.2 CHOICE THEORY ............................................................................................................................... 13
2.3 METHODS FOR VALUING NON-MARKET COMMODITIES....................................................................... 14
2.4 CHOICE MODELS............................................................................................................................... 17
2.5 HEDONIC PRICE FUNCTION................................................................................................................ 20
2.6 THE DISCRETE CHOICE MODEL ........................................................................................................... 21
2.7 RELATED STUDIES ............................................................................................................................. 22
2.7.1 COUNTRY OF ORIGIN EFFECT .......................................................................................................... 25
CHAPTER 3: METHODOLO GY AND DATA ......................................................................................... 27
3.1 O VERVIEW....................................................................................................................................... 27
3.2 THEORETICAL MODEL ....................................................................................................................... 27
3.3 MODEL 1: HEDONIC PRICE FUNCTION............................................................................................... 30
3.3.1 EMPIRICAL ESTIMATION OF HEDONIC PRICE FUNCTION.................................................................... 31
3.4 MODEL 2: DISCRETE CHOICE MODEL ................................................................................................. 34
3.4.1 EMPIRICAL ESTIMATION OF DISCRETE CHOICE ................................................................................. 36
3.5 BENIN RICE DATA 2006.................................................................................................................... 37
CHAPTER 4: RESULTS AN D DISCUSSION ........................................................................................... 44
4.1 I NTRODUCTION................................................................................................................................ 44
4.2 HEDONIC PRICE FUNCTION................................................................................................................ 44
4.2.1 DEPENDENT VARIABLES ................................................................................................................. 44
4.2.2 EXPLANATORY VARIABLES.............................................................................................................. 46
4.3 DISCRETE CHOICE MODEL.................................................................................................................. 48
4.3.1 DEPENDENT VARIABLE................................................................................................................... 48
4.3.2 I NDEPENDENT VARIABLE................................................................................................................ 50
4.4 EMPIRICAL RESULTS OF THE HEDONIC PRICE FUNCTION ...................................................................... 53
4.5 EMPIRICAL RESULTS OF THE DISCRETE CHOICE MODEL ........................................................................ 58
CHAPTER 5 CONCLUSION AND SUMMARY ...................................................................................... 64
5.1 O VERVIEW....................................................................................................................................... 64
5.2 RECOMMENDATIONS........................................................................................................................ 66
5.3 LIMITATIONS AND FUTURE RESEARCH ................................................................................................ 69
REFERENCES............................................................................................................................................ 70
APPENDICES............................................................................................................................................ 76
vii
APPENDIX 1 BACKGROUND INFORMATION ............................................................................................... 76
APPENDIX 2 DESCRIPTIVE STATISTICS....................................................................................................... 77
APPENDIX 3 DISCRETE CHOICE MODEL REGRESSION RESULTS..................................................................... 78
APPENDIX 4 CORRELATION TABLES........................................................................................................... 82
APPENDIX 5 DISCRETE CHOICE MODEL REGRESSION RESULTS..................................................................... 83
APPENDIX 6 QUESTIONNAIRE .................................................................................................................. 86
List of tables
Table 3.1 Expected relationship between price and rice attributes
Table 3.2 Definition of variables
Table 3.3 Distribution of respondents in Sample
Table 3.4 Sample and population comparison
Table 4.1 Rice mean price and standard deviation
Table 4.2 Summary statics for Hedonic model
Table 4.3 Distribution of respondents by rice type and region
Table4.4 Summary statics for discrete choice model
Table 4.5 Results of hedonic model
Table 4.6 Results of discrete choice model
33
37
40
41
46
47
49
50
54
60
List of figures
Figure 1.1 Evolution of rice consumption and imports in Benin
Figure 1.3 Price of rice in Benin in 2002
Figure 3.1 Map of Benin
7
9
39
viii
Chapter 1: Introduction
1.1 Overview
Globalization of food markets has triggered concern about food quality, standards and
preference. Around the world consumers are becoming more aware of the quality
attributes of different commodities found in the market and are c hoosing products that
closely match their tastes and preferences. Developed countries implement high safety
and quality standards to alleviate consumer health and environmental concerns often,
at the expense of low quality products from importing countries (Mergenthaler, 2009).
This poses new challenges for producers in developing countries as they have to adapt
to these emerging standards to ensure high food quality standards, especially for the
export market, but also to meet the preferences of domestic consumers.
Consumers often express their preferences for product quality by paying a premium
for the product with the desired characteristics. These premiums give producers an
incentive to improve product quality and quantity consequently enhancing the welfare
of the consumer and the producer. Consumers face tradeoffs in their purchasing
decisions since income is limiting and choices are numerous. When making choices,
consumers must combine budget constraints and preferences. Budget constraints are
determined by both the income of the consumer and the relative prices of products.
On the other hand products are differentiated based on the characteristics they
possess. It is assumed that once consumers recognize their peculiar preferences they
are prepared to pay more for variants that are better suited to their tastes (Rachmat,
2006). Better market information on consumer preferences will assist rice producers
in making informed decisions to produce appropriate rice qualities and maximize their
returns.
The recent economic growth resulting in increased incomes in Benin have reflected in
changing food consumption preferences among the consumers, towards eating more
rice than root and tuber crops (USAID, 2008). In Benin rice is produced primarily as a
cash crop although a significant proportion of the harvest is retained for household
consumption. It is no longer viewed as a luxury good but as a major source of
calories. It is important to small scale farmers for meeting immediate household
consumption requirements and as a source of income. In Benin there is evidence of
rejection of domestic rice and preference for imported rice because local rice is of low
quality (ADRAO, 2006). Benin consumers are willing to pay a premium for better
quality imported rice at the expense of low quality local rice. The emerging demand
trend (that is import rice is viewed as superior to domestic rice) provides both threats
and opportunities for local producers. On one hand it offers new opportunities for
improving quality and adding value to rice products which can facilitate better market
power and higher margins. On the other hand if no action is taken the domestic
producers can be driven out of business (Grunert, 2003). It is imperative that research,
policies and investment efforts be directed toward improving quality of local rice.
Consumer behavior theory is intended to identify consumer variables, explain
relationships between variables and specify cause and effect outcomes from variable
interactions (Kivetz & Simonson, 2002). Consumer behavior theory states that
consumers evaluate a product based on intrinsic and extrinsic attributes. Intrinsic
attributes (characteristics integrated in the physical product) include factors such as
2
the wholeness of grains, taste, aroma, amount of damaged and discolored grains and
kernels, milling level and amount of foreign matter (Koasa-ard, 1991). Extrinsic
attributes collectively known as credence (acceptance, confidence) attributes include
GMO free food, organic food, pesticide free food and country of origin (COO).
Several studies have been carried out on food standards and quality in developed
countries and very few if any in developing countries. Studies have been carried out
on consumer valuation of environmental and quality attributes For example beef
(Belcher, 2007), apples (Baker, 1999), pork (Ubilava, 2009; Melton, 1996) and rice
(Tomlins, 2004; Rutsaert, 2009; Horna, 2005 and Dalton, 2003. Recently there has
been emerging interest among policy markers on country of origin as one of the
credence attributes. Several studies conducted on effect of country of origin (Chung,
2009; Loureiro, 2005; Scarpa, 2005) suggest that reference to the country of origin
of a product to its label influences consumer‟s perception regarding its quality
concurrently with specific product attributes. In other words country of origin is not
per se the defining attribute but consumer‟s valuation of certain attributes in a product
that are associated with a certain country.
Inappropriate technologies, unsuitable management techniques, and lack of
knowledge on grain harvesting, drying, storage and milling often result in quality
deterioration and low market price of rice. Rice quality deterioration can be in the
form of damaged and yellow grain, incomplete milling, discoloration, impurities or
undesirable odor and taste. Post harvest handling methods have a tremendous impact
on production costs, quality and therefore price of agricultural produce. Studies show
that, post harvest losses account for about 16% of production losses(qualitative and
quantitative) , therefore the use of efficient modern post harvest technologies can
3
considerably reduce losses and ensure high quality standards in food production
(Kormawa &Toure, 2005). Preventing losses and improving efficiency is also
essential to concurrently reduce poverty and improve food security and farm incomes
through the introduction of time and labor saving techniques.
There is very little research on consumer food quality preferences in developing
countries and, since research findings from developed countries cannot be transferred
to other countries because culture, taste preferences and perceptions differ within
countries (Grunert, 2003). In this context, the present study analyzes the relationship
between price and product attributes and consumer preferences for the case of rice in
Benin using the hedonic pricing approach and discrete choice modeling and
household data. The analysis is based on the new consumer behavior theory originally
coined by Lancaster (1966), which postulates that products are consumed for the
characteristics they possess, other than the product itself and are associated with
consumer preferences/utility. Cost effectiveness requires targeting research at
characteristics that meet consumer‟s tastes and preferences (Langyintuo, 2004)
Producers and processors will be more likely to adopt storage and postharvest
technologies that improve the characteristics consumers‟ value. Understanding
consumer needs and valuation of these attributes is essential to the producer and for
designing effective policies towards improved rice production.
1.2 Motivation for the research
Rice has become an increasingly important crop in Africa, with imports into the
continent accounting for more than a third of world trade in rice (FAO, 2006).
Industrialization, urbanization, population growth, income growth and accordingly
4
change in consumer preferences have over the past three decades drastically increased
the demand for rice in West Africa (WARDA, 2008). High incomes in low income
countries translate into high demand however, as income grows, consumer tastes and
preferences change. Initially demand for cereals grows at the expense of root and
tuber crops subsequently the surplus income is used to purchase better quality grain
(income effect). In 2001-2005 the per capita consumption of rice in West Africa was
estimated to be growing at an average rate of 6.55% per annum while, production was
expanding at the rate of 5.06% per annum (FAO, 2008). Per capita per annum
consumption was averaging 18kg between 2001 and 2005 (with some countries
especially in West Africa averaging more than 45kg/per capita/annum). West Africa
is the focal point of rice production and consumption; it is the largest producer and
importer of rice in sub-Saharan Africa. The share of rice in cereal consumption has
increased from 15% in 1973 to 26% in 1998 (ADRAO, 2006). The continent relies on
the international market to satisfy 39% of rice consumption needs (FAO, 2006).
Major cities especially in coastal countries are swamped with import rice. Imports
account for 40-45% of total rice supply in West Africa and cost about US$1.5 billion
annually (WARDA, 2007). Importation of rice is placing a heavy burden on scarce
foreign currency resources of the region, the most impoverished of the world.
1.3 Overview of Benin rice market
Benin is a small coastal country located in West Africa. It has an estimated population
of 8.7 million and a population growth rate of 3.14% (WARDA, 2009). Its economy
is underdeveloped and dependent on cotton production and agriculture, (which
accounts for about 40% of the gross domestic product (GDP) and two thirds of the
country‟s labor force) and regional trade. Production is dominated by small scale
5
farmers (subsistence) with limited access to markets, capital, mechanical implements
and production inputs (Ereinstein et. al 2003).
At the national level Benin‟s agriculture sector is relatively diversified with
production based on tubers (yams and cassava), maize, millet, rice and sorghum.
Cereals only account for 37% of the total calorie intake (USAID, 2008).In general
maize and cassava are grown throughout the country, while millet yams and sorghum
are produced in the north. Although rice is not a traditional staple, its consumption has
grown markedly over the past few decades. This has been attributed to increased
urbanization and the presence of import rice. The rice market is segmented, local rice
is grown milled and consumed exclusively in the north by subsistence farme rs and has
poor milling quality. Small quantities of import rice can be found in a few major cities
in the north. Generally very little if any local rice is consumed or traded in the south
where two thirds of the population reside (USAID, 2008).Domestic rice is not
available in the three major cities of the south region.
Rice is produced primarily for the market although a significant proportion of the
harvest is retained for household consumption. It is estimated that 66 307 farmers out
of a total population of about 7 million were producing rice in 2002 (WARDA, 2009).
In 2002 it is estimated that the country produced 52 thousand metric tonnes of rice
and a total of 953 thousand metric tonnes of course grain (cereal crops). The
livelihood of most households is centered on crop production it is however
supplemented with off- farm income and livestock sales (WARDA, 2009).
6
Population growth, income growth and accordingly change in consumer preferences
have over the past three decades drastically increased the demand for rice in Benin
(WARDA, 2007). While the increase in production has surpassed population growth,
a rapid growth in consumption, augmented with high income growth has elevated rice
to an important staple food (FAO, 2006). In 2001-2005 it is estimated that production
grew at 7.13% while per capita per annum consumption grew at3 2.7%, nevertheless
the country was only 23% self sufficient in rice production (WARDA, 2008).
According to the World Bank (2006) aggregate imports costs for wheat, rice, maize
and fertilizer were estimated at US$ 120 million and rice accounted for US$45
million. In 2005 rice was ranked second after petroleum on the list of top ten imports
(WARDA, 2008).
Figure 1.1 Evolution of rice consumption and imports in Benin
Benin imports and consumption
quantity (000) tons
250
consumption
200
150
100
import
50
2007
2004
2001
1998
1995
1992
1989
1986
1983
1980
1977
1974
1971
1968
1965
1962
year
0
Source USDA 2008
Figure 1.1 above shows that rice consumption has been increasing since the 1970s.
The country experienced a rapid increase in consumption in the 1990s following the
7
devaluation of the FCFA in 1994. The devaluation of the FCFA improved the
profitability of rice production, but rice remains less profitable than cotton (Adjovi,
2006). In addition the government removed import tariffs and put in place subsidies
and other controls, which increased the flow of rice imports into the country
(presently the country imports more than 70% of aggregate consumption). Domestic
rice is characterized by high breakage rate, impurities, is of uneven quality and is
usually sold in bulk or unbranded 5kg bags at a lower price than import rice
(WARDA, 2008).
There is a huge discrepancy in terms of price and quantity consumed when comparing
import and local rice in the Benin rice market. Small scale traders and middlemen
import several varieties of foreign rice. Prices for imported rice are differentiated
according to quality, with higher quality varieties receiving a premium on the
domestic market. The importation of cheaper Asian rice has resulted in a decline of
the share of United States rice imports into the local market (USAID, 2003). The
major port of entry is the Port Autonome of Cotonou, located in the South region of
the country. A study by ADRAO (2006) found that in the South and the Centre of the
country import rice has a market share of 40 to 60%. The study also found that in the
Northern parts of the country domestic rice is predominant (supply of local rice is 2 to
11% more than that of import rice). These variations are credited to location
(proximity to the harbor means intense competition with import rice), type of
consumers (urban consumers purchase more import rice because of quality
preferences and higher incomes) and the transport network (import rice does not get
to remote areas inaccessible by road). Figure 1.5 below shows the price differences
8
between local and import rice in the Benin market. The Figure 1.2 below suggests that
in all regions consumers pay a higher premium for import rice than local rice.
Figure 1.2 Price of rice in Benin, 2002 (FCFA) 1
price FCFA
Rice price in Benin (FCFA)
450
400
350
300
250
200
150
100
50
0
domestic price
import price
region
Source: ADRAO/INRAB/PAPA, 2006.
Benin has made significant progress towards increasing rice production and
productivity through area expansion and the adoption of improved and high yielding
varieties. In 2001-2005 yield per hectare grew at 7.24% while cultivated area
increased at a rate of 0.11% (FAO, 2005). Historically, research on rice quality
improvement has largely focused on genetic plant improvement and very little post
harvest handling and management. Studies reveal that 70% of production and yield
growth can be attributed to land expansion whereas only 30% is a result of technology
advancement and capital investment (FAO, 2005). This poses the challenge of finding
1
NB: 1US$ = 453 FCFA
9
new ways of increasing productivity and grain quality as land and water are becoming
scarce due to population growth and environmental problems.
The country has made significant progress in genetics through the development of
improved varieties that are believed to have the potential of spearheading an African
green revolution. NERICA (New Rice for Africa) was developed by plant breeders at
the Africa Rice Center (WARDA). The rice variety is a crossbreed between African
rice varieties already adapted to the local environment and Asian varieties that have
high yielding potential (WARDA, 2009). NERICA is a short season, nutrient rich,
drought tolerant variety and high yielding variety. It has better grain quality, taste,
milling characteristics and less grain breakage compared to traditional varieties
(WARDA, 2005). It has the potential of competing with import varieties in terms of
productivity and grain quality. Grain quality is determined by several factors which
include rice variety, production and harvesting conditions, post-harvest handling,
milling and marketing techniques. Physical characteristics of a commodity relate to
the appearance of the product while chemical characteristics influence cooking
quality. The adoption of NERICA varieties has allowed farmers to double crop;
NERICA takes 3 months to ripen compared to six months for the parent species which
has resulted in increased output and farmers do not have the capacity to handle such
huge output resulting in greater post-harvest losses (Diagne, 2007). Thus there is need
for investment in technologies that can improve grain quality through post-harvest
handling methods. The adoption rate of NERICA seems to be mow; an impact
assessment report carried out in Benin in 2003-2005 indicates that 500 hectares were
cultivated to NERICA out of a total of 28 828 hectares cultivated with rice (WARDA,
2005).
10
1.4 Research objectives
The values consumers place on product attributes can be derived through observing
consumer purchase and demand trends. The overall objective of this survey is to study
and understand the factors that influence consumer‟s purchasing decisions and what
they value in a product in this case rice. This study will investigate consumer
valuation of rice attributes (a non market good) using market price of rice as an
indicator. The survey will analyze how consumer preferences are influenced by
country of origin as a function of price, socioeconomic characteristics (personal) and
intrinsic attributes of the product. Understanding consumer choice is critical for
developing effective policies and to help local farmers tailor their products to suit
consumer tastes (in that way they can obtain high premiums accordingly improving
farm incomes), therefore reducing rural poverty and inevitably increasing the
productivity and competitiveness of domestic rice to import. The specific objectives
of the study are to
1. Estimate implicit values of rice grain characteristics using market price as a
measure of consumer preferences under budget constraints.
2. Evaluate the relative importance of country of origin in determining consumer
choice.
3. Examine variations in consumer tastes across regions, between rural and urban
markets and across income classes.
These objectives will be achieved with the aid of a hedonic pricing model and a
discrete choice model as well as household data obtained from a 2002 household
survey conducted by the Benin Ministry of Agriculture in collaboration with
WARDA.
11
The goal of this survey is to provide rice producers and researchers with essential
information to develop cost effective technologies to enhance rice quality. When the
cost of producing high quality varieties is reduced consumer welfare is enhanced in
that they obtain improved quality at a lower price. Then again research on product
quality allows producers to improve quality so they can capture quality premiums and
raise the value of rice sold. Studies have found that quality improvement can also
result in production efficiency and an increase in overall production.
This thesis is divided into five chapters, the introductory chapter is followed by
chapter two which reviews literature on valuation of non market goods and choice
modeling and concludes with a summary of studies that have applied non market
valuation to agricultural products. Chapter three briefly explains the research
methodology and the data. Chapter four presents the results and discussion. Chapter
five presents conclusions and limitations and recommendations of the survey.
12
Chapter 2: Literature review
2.1 Introduction
Consumer behavior theories together with models derived from such theory are
intended to address the concepts of consumer preference, utility maximization and
demand functions. It is generally agreed that individual choices are influenced by
habit, inertia, experience, advertising, peer pressure, environmental constraints,
accumulated opinion, household and family constraints (Louviere, 2000). Consumer
theory postulates that consumers make choices not on the simple marginal rate of
substitution between goods (traditional approach) but based on preferences for
attributes of these goods (the new consumer demand theory). The new consumer
theory proposes that consumers make choices based on preferences for attributes of
goods. However, attributes of goods are not always valued in real markets.
Economists have over the years developed methods for estimating the value of nonmarket goods and services which include stated preference and revealed preference
methods. The purpose of this chapter is to review methods of evaluating non- market
commodities and their application to agricultural commodities with the aim of
identifying the best method for meeting research objectives outlined in chapter one.
2.2 Choice theory
Lancaster (1966) proposed a new approach to consumer demand theory which asserts
that consumers derive utility from attributes of goods rather than the goods itself. In
other words it is product characteristics that explain why consumers prefer certain
products to others in the same category. He departed from the traditional approach
that the goods are the direct object of utility. This theory has been applied to market
13
and transport economics, environmental economics and evaluation of non-market
goods. The model is deterministic in that the consumer is assumed to choose the
alternative with the highest utility. The choice problem is viewed as the selection of
one object from a given set of alternatives on each trial. However, product quality
characteristics are non- market goods their economic value is not revealed in market
prices. Economists have devised innovative indirect measurement methods of
assigning monetary value to non- market goods which include revealed and stated
preference techniques. These methods rely on observable behavior to decide how
much something is worth to an individual even though it is not traded in markets
(Louviere, 2000). Both revealed and stated choice theory are based on the premise of
a utilitarian consumer who bases preference on the characteristics a good possesses
other than on the good per se (Lancaster, 1966). These methods predict a consumer‟s
choice by determining the relative importance of various attributes in consumer‟s
choice set. Of great interest is the implicit value the consumer places on each one of
the characteristics (Ara, 2003).
2.3 Methods for valuing non-market commodities
Revealed preference and stated preference methods are the most common methods
used to value consumer preferences of agriculture products.
The revealed preference method has been appraised for its use of market prices to
depict the prevailing market equilibrium. This method uses a number of techniques
which include hedonic pricing and the travel cost method among others. Hedonic
pricing was initially intended to capture willingness to pay for variation in property
values. It is well accepted in estate transactions that location is fundamental to
14
property prices and proximity to desirable environmental amenities is considered as
part of location (Loomis, 2001). Statistical analysis allows economists to disentangle
the portion product price differential due to product characteristics by comparing the
market value of product which differ only in quality characteristics. Economists may
assess the implicit price of commodity attributes by observing the behavior of buyers
and sellers. The travel cost method estimates visitor‟s valuation of recreation sites, it
is assumed that visitors incur economic costs in the form of time and travel expenses
to visit a site.
The revealed preference method only considers existing alternatives to the good in
question as observables and essentially embodies existing technological constraints. It
is a true representation of the socio-economic and cultural constraints of the decision
maker and portrays the real market, which renders it as a highly reliable and valid
method. Moreover it yields one observation per respondent at e ach observation point
(Carson, 2000). The theoretical drawback of the method is that it assumes a perfect
market with perfect competition and perfect information in an open market. Most
food commodity markets tend to approximate the conditions of perfect competition
because there are many buyers and sellers. Without government regulated product
quality standards and grades market information is compromised (Unnevehr, 1992)
In stated preference valuation one asks people the context of a hypothetical market
how much they are both willing and able to pay for commodity characteristics that are
not valued in the market. It involves the creation of a hypothetical market, description
of the market items, reason why payment is needed and payment method. Specific
methods include contingent valuation, conjoint analysis (rank, rate), and contingent
15
choice (rank, rate). The contingent valuation method is a survey or questionnaire
based approach, which allows valuation of a wider variety of non- market goods than
is possible with any of the indirect techniques. It is the only method currently
available for the estimation of non-use values (Louviere, 2000). Stated preference
methods have been criticized for being unrealistic and not offering proper incentives
so that consumers would reveal their true preference. Literature shows individuals
overstate the amount they are willing to pay in hypothetical settings as compared to
when real money is used and there are real budget constraints (Ara, 2003).
Experimental auctions with real money and real products are more re liable. The
outcomes are close to true willingness to pay, but auctions are difficult to organize
and require more time and resources (Louviere, 2000). This method can include
existing and proposed goods and attributes, and can be reliable if respondents
understand and respond honestly to the questions asked. Unlike revealed preference
method this method can yield multiple observations per respondent at each
observation point.
Stated preference methods are more useful for policy making with respect to new
products, food safety, and certification and labeling. Where markets of a product do
not exist stated preference techniques employ a survey instrument in which a
hypothetical market for the item being valued is created (Carson, 2000). On the other
hand revealed preference methods are more useful for home purchases, job
acceptance and estimating recreation valuation. Combining the two methods brings
better results although, stated preference method are more controversial in non-market
valuation of goods.
16
2.4 Choice models
The two models that are typically used to value product attributes are the hedonic
model (Rosen, 1974) and the multinomial logit model (McFadden 1974).
The discrete choice model adopts the random utility function
associated with a
discrete alternative that is interpreted as the maximum utility attainable by the
consumer, given his budget constraint and a fixed set of alternatives j. Alternative j
has a vector of attributes
and the random utility function
is a function of
income and prices including the price of alternatives. The sources of randomness in
the utility function are unobserved variations in taste and in the attributes o f
alternatives (Edmeandes, 2005). Some models assume that the decision rule is
stochastic (complete knowledge of the problem). Other models assume that the
decision rules are deterministic which introduces uncertainty (impossible for the
analyst to observe some complex human behavior characteristics). The transportation
mode choice model by MacFadden (1974) is stochastic in that the consumer‟s
probability of choosing each alternative is functionally related to its utility. Stochastic
models allow for a mix of brands to be chosen. The Luce (1959) model assigns each
alternative a probability to be chosen instead of identifying one alternative as the
chosen option. In both models the choice problem is viewed as the selection of one
object from a given set of alternatives on each trial. The model was initially applied to
transportation sector but has since been applied to a number of disciplines including
agriculture.
Rosen (1974) proposed a method of interposing a market between buyers and sellers
through hedonic modeling. He proposed a two stage approach that first estimates the
17
hedonic price function, and then establishes the demand and supply function. Rosen
asserts that producers receive returns for serving the needs and wants of consumers by
tailoring their goods to embody final characteristics desired by consumers. Ladd and
Suvannunt (1976) extended the Rosen model by developing the consumer good model
(CGCM) to test the hypothesis that the retail price of a particular good is the sum of
that good‟s yield of individual characteristics multiplied by the marginal implicit price
of each attribute.
Ratchford (1979) challenged Lancaster, and Ladd and Suvannunt‟s assumption of
finitely divisible goods by adding some restrictions to the consumption matrix, and
adopting a specific form of the consumer utility function. He argued that in the real
world choices do not involve combining fractional amounts of various brands into a
desired set of attributes rather they involve selecting one brand from several
alternatives (each brand has a specific set of attributes). He then developed the model
of brand preference model. Epple (1987) expanded the model by estimating the
marginal price and the marginal cost from consumer budget constraints and producer
cost functions. Initially the model was applied to the demand side and not the supply
currently it is widely applied to both consumption and production. The findings of
Ratchford (1979) are sustained by a study conducted by Holcomb (1997) who
examined the marginal implicit prices of rice attributes on a number of rice varieties
and found that no single variety contained the optimal combination of all attributes
(color, texture, stickiness, aroma, flavor, aftertaste and moisture) but observed that
each variety has positive and negative marginal implicit prices.
18
The concept of a good‟s price being a function of its various quality characteristics
has over time been applied to an array of commodities and sectors. The first known
application of Hedonic pricing to agriculture products was conducted by Waugh
(1929). He studied the influence of quality factors (color, size and uniformity) on
vegetable prices and found that market prices of vegetables vary with physical
characteristics that the consumer associates with quality. Since then, the hedonic
method has been applied to different agriculture products. Dalton (2003) studied
factors affecting introduction of new rice varieties in West Africa, drawing upon input
characteristics and consumer good characteristics. He found that a combination of
production and consumption characteristics (plant cycle length, plant height, grain
swelling and tenderness) best explains willingness to pay for new upland rice
varieties. In addition the study found that yield was not a significant attribute in
determining farmer willingness to pay for new varieties yet it has served as the
defining factor for promoting new varieties for official release.
In contrast Horna et al (2005) used conjoint analysis to estimate farmer preferences
for rice seed of new varieties in Nigeria and Benin. They concluded that potential
yield was an important factor together with other variety attributes such as tillering
capacity, size of grain and height of the plant and days to maturity. Edmeades (2005)
used farm gate prices to assess the impact of variety improvement to banana
production and found that quality of fruit, bunch size and fruit size affect price. She
argues that hedonic models should include both consumption and production
attributes. Although this study focuses on the consumption side, inference is made on
both consumption and production factors. The underlying principle being that a high
quality rice variety is dependent on production factors as well as consumer preference
19
and demand. Consumers respond to services rendered by a product through purchases
on the other hand, the producer determines the characteristics of the product. The
producer needs knowledge on the way the consumers react in the market and by
varying product characteristics he can affect purchases. If research can allow
producers to improve quality, then producers can capture quality premiums and raise
the value of rice sales (Unnevehr, 1992).
2.5 Hedonic price function
The hedonic model assumes that there is a continuous function relating the price of a
good to its attributes (the hedonic price function). Hidano (2002) defines the hedonic
approach as a method of ascertaining the value of or the pleasure felt from attributes
of a good. In contrast to conventional economic evaluation, where the value of a good
is calculated for the whole of the good, the hedonic approach regards a product as a
set of attributes and considers the value of a good as a function of each attribute of
that good as coined by Lancaster (1966). The value of an attribute is called an implicit
price because it cannot be observed in a real market. Hedonic analysis records prices
paid at purchase, observable and unobservable attributes of product and estimates
implicit payment for attributes with a multivariable regression method. Rosen (1974)
asserts that producers tailor their goods to embody final characteristics desired by
consumers and receive premiums for them. These premiums arise from specialized
production achieved by specialization and government policy. This method has been
applied to a wide range of economic issues ranging from durable goods (houses,
amenities) to non durable goods like agriculture products.
20
Hedonic pricing draws from revealed preference to estimate the value consumers
place on non- market goods and is based on utility maximization theory. This method
has traditionally been used for the valuation of public goods and environmental
goods. More recently it has been applied in market research for private goods like
houses, automobiles and food productsamong other commodities.
2.6 The discrete choice model
A discrete choice model is a mathematical function which predicts an individual‟s
choice based on the utility or relative attractiveness of competing alternatives. The
logit function is a common mathematical form used in discrete choice modeling. The
model generally includes characteristics of the individual (age, gender, income) and
relative attributes of competing choices (Ben-Akiva, 1985). The model estimates the
probability that an individual chooses a particular alternative. The probability that a
person chooses a particular alternative is determined by comparing the utility of
choosing that alternative to the utility of choosing other alternatives. Neoclassical
economic theory assumes that each decision maker is able to compare two or more
alternatives in a choice set using a preference indifference operator which is supposed
to be exhaustive, finite and transitive (Bierlaire, 1997). The model assumes that
consumer utility functions, market shares and substitution patterns depend on
differentiated product characteristics that are observed or unobserved by the analyst.
The economic justification for including unobserved product differentiation is that the
researcher typically does not observe all the product characteristics that are relevant to
consumer‟s choice (Manski, 1997). Researchers use discrete choice models to study
consumer demand and to solve problems of pricing and product develop ment.
Discrete choice models take on many forms including binary logit, binary probit,
21
multinomial logit, conditional logit, multinomial probit, nested logit, generalized
extreme value models, mixed logit and exploded logit.
2.7 Related studies
Scholars have investigated the relationship between product quality, nutritional
characteristics and prices of different food products. In general food quality attributes
can be classified into four groups namely: sensory attributes (taste, smell,
appearance), health attributes (nutritional benefits), process attributes (organic against
GMO) and convenience attributes (time and energy saving) (Grunert, 2003) Hedonic
models have been used to derive implicit values of product characteristics including
rice, cotton, wool, wheat, grapes, wine, pork, tomatoes, asparagus, vegetables and
beef among others. Apart from hedonic models, techniques such as conjoint analysis
and choice models have been used to examine consumer preference for food products.
In the first known application of hedonic pricing to agricultural products Waugh 1929
gathered data on the prices of vegetables like asparagus from the Boston market in
1927 with the aim of explaining the determinants of the price differences for the
average price of a bundle of asparagus. He found that the price of asparagus was
correlated with the length of the green portion (asparagus with eight inches of green
portion was 8.5 cents higher than that with five inches). Drawing him to the
conclusion that Bostonians, place more importance on the green part of asparagus.
Schnettler (2009) suggests that the effect of the country of origin means consumers
use a product‟s origin as an attribute related to its quality. This is supported by the
findings of Hara (2000) who estimated a hedonic function for the Japan rice market.
He found that consumers pay a premium for domestic certified rice, pesticide free and
22
fertilizer free rice in comparison with import rice. Tomlins (2004) investigated
consumer preferences and acceptability of domestic and import rice in Ghana. He
found that consumers prefer import raw and parboiled rice to domestic rice, and that
acceptability was influenced by location and gender. Rutsaert (2009) used vickery
second price auctions to compare consumer willingness to pay for Senegal rice and
Thai rice and found that consumers are willing to pay up to 80% more for local rice
than import rice. He concluded that bids are influenced by taste whereas sociodemographic factors are not important.
Duff (1991) asserts that changes in consumer demand are a result of changes in taste
and income. As incomes increase consumers are able to substitute more preferred for
less preferred foods, the assumption being that consumers desire to improve their
diets. Grazia (2007) found that organic food was highly valued by consumers due to
perceived benefits to the environment and consumer‟s health. Ara (2003) valued
organic rice in the Philippines and found that consumers were concerned with health
risk and the farm environment and certification of products.
Juliano et al (1992) found that in Hong Kong consumers pay high premiums for long
grain, whole grain, flaky and soft textured rice. Italians prefer chalky grain with
harder gel. Germans are particular with the level of processing and packaging type
and types of outlets. In Thailand consumers prefer soft but flaky rice, percentage of
broken grain and kernel length are more important than presence of impurities. They
concluded that consumer preferences depend on historical and socio-cultural factors
and that families in which both spouses work and spend significant time commuting
have more demand for convenience foods. Baker (1999) used conjoint analysis to
23
evaluate consumer responses to hypothetical apple products which include price,
quality and pesticide use levels and health risk. He found the market is comprised of
four segments of consumers that is; those who have a strong preference for food
safety whereas some exhibit a more balanced desire for all product characteristics,
others are extremely price sensitive and others have a strong preference for product
quality. Mishili (2007) investigated consumer preference of cowpea grain quality
characteristics in West and Central Africa and found that consumers pay a premium
for large cowpea grains and discount damaged grains. Dalton (2004) derives a
hedonic model formulation based on the model of the agricultural household. Tests
for the statistical relevance of consumption attributes using experimental data and
concludes that rice breeders should consider post-harvest attributes in addition to
production traits. Important determinants include income level, taste of product and
market price.
Hassine-Belghith (2009) studied the association between exporting and product
quality and found that exporting results in quality upgrading and more efficient use of
resources. The results also showed that there is a positive relationship between
product quality and productivity growth. Although food security takes the central role
at national level for a self insufficient country, a country like Benin with a
comparative advantage in rice production can take advantage of the opportunities
presented by the emerging consumer preferences for import rice by improving
product quality to penetrate the export market. A combination of good quality
products and sound agricultural policies can improve product quality and production
efficiency. Recent trade liberalization efforts by the World Trade Organization could
present real opportunities to producers. Trade may play a key role in encouraging
24
rural development, promoting the modernization of the traditional sector and
enhancing product quality and technical efficiency (Juliano, 1992).
2.7.1 Country of origin effect
Country of origin is one of the many credence attributes for which there is emerging
international interest among policy markers (Ehmke, 2007). In general food quality
attributes are divided into two that is intrinsic characteristics (color, grain size, taste)
and extrinsic characteristics (price, country of origin and information on production).
Recent labeling policies in developed countries place new focus on origin labeling for
a variety of food products. Several studies state that reference to country of origin of a
product on its label influences consumer‟s perception regarding its quality (country
specific) yet the magnitude depends on the product category (product specific) as well
as on specific product attributes (attribute specific) (Chryssochoids, 2007). Thus
there is need to obtain information on consumer preferences for origin.
Results from studies in Europe and United States show consumers prefer own country
of origin in meat products (Loureiro, 2005). Umberger (2003) found United Sates
consumers ranked country of origin above organic production and food safety
inspection. Scarpa (2005) and von der Lans (2001) found that Italians prefer home
grown grapes, table oil and oranges to imports. Chung (2009) studied the Korean beef
market and found that consumers are willing to pay almost three times more for
domestic beef than they pay for imported beef. He argues that consumers are more
concerned with country of origin and prefer organic beef more than conventional
beef. Ehmke (2008) found that consumers in China, France, Niger and the United
25
States have a strong own country preference which she attributes to ethnocentric
tendencies. Grunert (2003) classifies consumers into two, those who place high
emphasis on traditional food qualities (rational approach to food choice), known
products and production methods and non innovative methods of preparing meals.
While the second group is composed of individuals who are innovative oriented
(social and adventurous approach).
Unlike Asian and American rice, there are no pre-established market standards for
rice grain or quality properties in Benin (Horna, 2005). When food standards are not
regulated by a government agency, quality premiums should reflect consumer‟s
valuation of characteristics (Unnevehr, 1992). Consumers are at the beginning of the
value chain and information flows back to retailers, manufacturers and farmers. If
retail price premiums are transmitted back to the farmer through the market system
then market participants have the incentive to improve quality (Toquero, 1992).
26
Chapter 3: Methodology and Data
3.1 Overview
Consumer valuation of rice attributes is of great importance to the rice value chain
and actors involved in production, processing and marketing. This chapter gives a
detailed account of the theoretical framework on analyzing the behavior of Benin rice
consumers in the choice of specific rice attributes within the context of country of
origin. Furthermore the research determines the implicit prices for the choices of rice
consumers in Benin across different regions. For this purpose, reduced equations
derived out of the consumer choice model and the hedonic price functions were used.
The structural, reduced form model is described briefly followed by the description of
the empirical data used in the study.
3.2 Theoretical model
The major assumptions of the consumer demand model as stated by Ladd (1966) is
that under perfect competition, consumers have complete information about the
quality characteristics of goods. This hypothesis holds especially true among food
commodity markets with many buyers and sellers. The major premise, upon which
“consumer‟s choice set” under perfect competition operates, holds if it includes all
possible alternatives (finite in nature) along with mutual exclusiveness. However, in
rural, agricultural markets with no regulations and standards (existence of weak,
incomplete or missing markets), the consumers may not reflect their choices based on
perfect market equilibrium conditions (Sadoulet & Janvry, 2009). Often the price
denoted in these imperfect markets could be derived from their shadow prices. In
other words, the price of the commodity in question is indirectly determined through
27
their choice attributes, which would be reflected as premium price associated with the
attributes. Thus this would reflect the choice set or the preferences of the consumer
involved.
The new consumer demand theory asserts that desirable quality characteristics of a
product relate to the production process that is not represented by the product itself. In
other words the good per se does not give utility to the consumer but it possesses
characteristics from which the consumer derives utility (Lancaster, 1966). Product
characteristics explain why consumers prefer certain products to others in the same
category. For example, rice growers and traders (exporters) usually establish certain
criteria to grade their commodities. Grading is based on key physical properties, that
is consumption qualities and credence attributes. Physical quality comprise of length
of grain, grain color/whiteness, percent of broken, moisture level and foreign matter
among others. Cooking quality consists of aroma, taste and stickiness. Credence
characteristics (fair trade practices) include use of pesticides in production, organic
food compared to conventional food and country of origin (Unnevehr, 1992). While
consumers may easily identify physical characteristics, it is not possible for them to
determine the chemical characteristics until the rice has been cooked; thus consumers
have to rely on proxies such as brand name and place of origin to represent
unobserved characteristics (Holcomb, 1997; Hidano, 2002). In the Benin rice market
consumers make their decisions and choices based on physical characteristics alone as
product labeling is either weak or nonexistent (Horna, 2005). This often results in
market failure as the consumers lack sufficient information to make decisions prior to
purchase of products. It should be noted that each rice variety available in the market
has distinct aroma and flavor characteristics that are determined partially either by
28
genetics or post- harvest handling practices. In order to precisely define and interpret
measures of quality characteristics it is necessary to understand the role of genetics
and other factors that determine quality. For example grain color (whiteness) is based
on the degree of milling; whereas grain size is a function of rice variety. In this
research study, in order to assess the demand and supply for rice (both domestic and
or imported), a consumer choice model based on attributes is estimated empirically.
The major assumption here is that consumers value „varieties with desirable
attributes‟ than an „ordinary variety or varieties with inferior qualities, to attain
maximum utility in terms of taste. The higher value of the premium variety is also in
turn reflected in its higher production costs-mainly post harvesting packaging and
labeling. Dalton (2004) in his attempt to characterize attribute based hedonic pricing
model advocates for the inclusion of both consumption and production characteristics
to obtain a holistic understanding of consumer behavior. This study also estimated the
choice of the consumers based on „country of origin‟ (domestic versus imported rice)
with regard to prices and other attributes using a discrete choice model.
Choice modeling predicts consumer‟s behavior by determining the relative
importance of various attributes in consumer‟s decision making process (Ara, 2003).
Discrete choice modeling reveals the relationship between the probability of choosing
an alternative and its attributes along with its alternative choices. A simple logit
model has been used to empirically determine consumer choices that provide
estimates of the utility or value consumers place on different alternatives (McFadden
1974).
29
In hedonic pricing, econometrically implicit prices are estimated by the stepwise
regression analysis where product price is regressed on characteristics of the good,
using Ordinary Least Squares method (Rosen, 1994). Here the method relates the
probability of a ranking of consumers to the attributes of the variety.
3.3 Model 1: Hedonic price function
Lancaster theory postulates that consumers maximize a utility function
where
is defined as the total amount of characteristics obtained by the consumer. Here
a choice bundle represents the utility derived indirectly through its attributes. This
could be derived as a linear function of consumption technology that relates the vector
of characteristics to the quantity of commodity consumed
where
is the
quantity of commodity consumed. The consumer is assumed to choose quantity that
maximizes utility subject to the consumption technology (the consumption technology
is the seed or genotype) and the budget constraint (Lucas, 1975). This is represented
as
Max
s.t.
and
is the market price of the commodity and
is consumer‟s income. The major
assumption is that the consumer is limited to choosing one integer unit of
from
among the various brands of the product i.e. the maximum attainable utility could be
derived from consuming only one brand at a time among the given choices
(Ratchford, 1979).
30
Griliches (1961) asserts that when a product has multiple brands, price differentiation
occurs mainly due to quality characteristics. Thus he expresses price
of a set of quality attributes
as a function
and some additional small and random factors measured
by the disturbance term .
(3.1)
If the sample is large enough quality characteristics can be represented with dummy
variables that take on the value one if the product possesses the particular attribute
and zero if it does not. Moreover proxy characteristics can be used to represent some
basic dimensions which may be difficult to measure if the two characteristics are well
correlated. One can then derive the average contribution of each quality characteristic
to the price of the product (Griliches, 1961)
3.3.1 Empirical estimation of hedonic price function
In this study, we have made assumptions regarding the consumer choice of rice
attributes, based on Griliches (1961) model of quality change and further deduc ed for
empirical estimation. It is assumed that there is a functional relationship between the
good‟s price
and its characteristics vector
in the form of the equation
.
The reduced form of the hedonic price function for empirical estimation takes the
form
(3.2)
This equation can be rewritten as follows with each
representing every one of the
eleven rice attributes as defined in table 3.1;
(3.3)
31
Where
is observed market price of rice and
is the stochastic error term.
, the
dependent will vary for the different rice characteristics. The dependent variables
would explain variance in the rice price and the parameter
gives the implicit value
of rice grain characteristics.
The attributes considered for evaluation in this research have been listed on table 3.1.
It should be noted that the hedonic price function is neither a demand nor a supply
function but simply expresses implicit product price (Rosen, 1974). Ratchford (1979)
states that there is no priori rule about the inclusion of quality characteristics but
quality characteristics should be observable and economically relevant to consumers.
Whilst it is important to include all attributes into the model care should be taken to
identify correlation within attributes as multi-co- linearity among variables inflates the
standard errors. A large sample can be used to offset multi-co- linearity (Wooldridge,
2006). The appropriate hedonic functional form would be fitted after subjecting
through box-Cox transformation.
In order to get attributes value, during the survey, the consumers were asked to rank
each type of rice variety they consume, on a scale of 1 through 5 with 1 being
abundant presence of the characteristic and 5 least present (refer to appendix 4,
questionnaire). To avoid the problem of dummy trap, the ranking of attributes on the
scale have been further reduced or collapsed into two classes as shown in table 3.1.
The dummy variable trap arises when too many dummy variables describe a given
number of groups. In this study the analysis is mostly dependent on categorical,
numerical variables are not sufficient to justify inclusion of many dummy variables.
32
Table 3.1 Hypothesized relationship between price and rice attributes
Grain characteristic
Regression
Hypothesized
measure
effect
1-5 (very few to
1= absent
positive
high)
0 = present
2. Color
1-5 (white to yellow)
1= white 0=yellow
positive
3. Rate of
1-5 (low to high)
1= low 0=high
positive
1-5 (long to short)
1= long 0=short
positive
1-5 (cooks fast to
1= fast 0=slow
negative
1-5 (very sticky to
1= sticky 0=not
negative
not sticky)
sticky
7. Taste
1-5 (good to bad)
1= good 0=bad
positive
8. Aroma
1-5 (good to bad)
1= good 0=bad
positive
9. Grain size
1-5 (long to short)
1= long 0=short
positive
1-5 (swells to does
1= swells 0=does
negative
not swell)
not swell
1-5 (tender to hard)
1= tender 0=hard
1. Impurities
Survey ranking
breakage
4. Shape of
grains
5. Ease at
cooking
6. Cohesion
after cooking
takes long)
after cooking
10. Swelling
capacity
11. Grain texture
negative
Prices in cross-sectional data are assumed to reflect quality effects, as well as
consumer tastes and producer costs. Use of unadjusted cross-sectional prices may lead
33
to potential distortions associated with quality effects, which increase with the
heterogeneity of aggregated commodities (Holcomb, 1997). Other price variations
may be due to outlet type (wholesale, retail, supermarket), region (rural, urban), price
discrimination, services purchased, seasonal effects and quality differences caused by
heterogeneous commodity aggregates.
3.4 Model 2: Discrete choice model
Ladd (1976) found that consumer demand for a product is a function of household
income, product prices and product characteristics. Higher income households pay
more for high quality products, while low income households would pay less for the
same product with basic characteristics. In this study, a bi- variate choice (logit) model
is used to estimate the impact of household and product characteristics on consumer
demand for domestic and or imported rice in Benin. Logistic regression is used to
predict the dependent variable (whether one purchases imported rice or domestic rice)
on the basis of continuous and categorical independent variables (rice attributes and
consumer socioeconomic factors) to determine the point of variance in the dependent
variable explained by the independent variable.
Random utility models assume that the decision maker has a perfect
discrimination capability, however the analyst is supposed to have incomplete
information and therefore uncertainty must be taken into account. Uncertainty is
a result of unobserved taste variations, measurement errors and instrumental
variables (Manski, 1997). The utility is modeled as a random variable to reflect
this uncertainty. The transportation model (McFadden, 1974) demonstrates that
34
if unobservable utility is independently and identically distributed then i n the
binary response model interest lies primarily in the response probability
)
Where
(3.4)
the conditional indirect utility functions, and f and d represent
import and domestic rice
= vector containing all attributes
= the dependent variable is a binary choice variable
Assuming a cumulative normal distribution then we can define the error term
and the parameter
(3.5)
(3.6)
The bi-variate choice model is designated by dummy variables
for each
alternative represented as follows:
(3.7)
Where,
is the utility that individual is associating with an alternative (either
import or domestic rice),
is the deterministic part of utility and
is the
stochastic part capturing uncertainty. The alternative with highest utility is
supposed to be chosen. Therefore probability that import rice will be chosen is
given by the related observable variable
If
defined as follows
or
(3.8)
Otherwise
In accordance with utility maximizing behavior, consumers choose the
alternative with the most desired set of attributes.
35
3.4.1 Empirical estimation of discrete choice
This research assumes consumer choice of product based on country of origin
(McFadden, 1974). Given a set of explanatory variables
the log odds
of the outcome can be expressed as a linear combination of the explanatory
variables and
the estimated parameter of the explanatory variables.
is the
utility function obtained from purchasing import rice. The equation takes the
form;
(3.9)
Where
i.e. probability of purchasing import rice
. A greater value of
when
implies a greater probability for the event to take place,
approaches infinity,
for the event to occur. When
(zero). When
and domestic rice
approaches 1(one) indicating a high likelihood
approaches negative infinity
approaches 0
equals zero the probability is 0.5 implying a 50/50 chance for
the event to occur. The characteristics considered for evaluation in this research have
been listed on table 3.2.
36
Table 3.2 Definition of variables
Variable name
Definition of variable
Dependent variable (to
Origin of rice categorical dummy 1 = import rice, 0 =
represent origin of rice)
domestic rice
Income
Annual income of household FCFA
Price
Price of rice FCFA/kg
Location
Location of household categorical dummy 1 = rural, 0 =
urban
Region
1= South, 2= Central, 3=North-East, 4=North-West
Rice attributes
Rice characteristics as indicated in table 3.1 (includes both
consumption and production attributes)
The previous section described empirical models to be used in estimating consumer
choices and implicit value of attributes within the new consumer demand framework.
The following section gives a brief description of the data set collected from all four
regions of Benin in 2006 by WARDA in collaboration with other organizations.
3.5 Benin rice data 2006
Data was collected by WARDA in collaboration with the Ministry of Agriculture
from ten of the twelve administrative departments (provinces) of the country namely
Alibori, Donga, Atacora, Borgou, Collines, Zou, Mono, Couffo, Littoral, Ouémé
37
(excluding the Atlantique and Plateau). For the purpose of this survey the ten
administrative zones are grouped into four regions namely South (includes Mono,
Couffo, Littoral and Ouémé), Central (that is Zou and Collines), North-East (Borgou
and Alibori) and North-West (Atacora and Donga).
The Southern region houses the port of Cotonou, this region is characterized by good
roads and transport network. Due to its proximity to the port rice in this region faces
intense competition from import rice (ADRAO, 2006). The Central region also faces
high competition with import rice, because of a good transport network, but the
competition is not as stiff as in the southern region. The North East is well known for
cotton production and produces 42% of total national rice production (ADRAO,
2006). The North West faces very little if any competition with import rice.
A total 546 consumer households from both rural and urban areas were interviewed
with the aid of a structured questionnaire. Sample selection was partly stratified in
that it targeted households on the basis of location that is the administrative zones
(North West and East and South and Central Benin) and partly random in that
households were randomly selected. The primary objective of data collection was to
investigate postharvest activities and estimate losses at every level of the food chain
(through interviewing producers, millers, traders and consumers) and to identify types
and quality attributes of rice preferred by consumers. This study will only focus on
the consumption side with the aim of investigating consumer preferences. The map
below shows the area of study.
38
Figure 3.1 Map of Benin
Source: Institut de la statistique Benin (www.CITYPOPULATION.DE)
39
Table 3.3 below shows the distribution of the sample. The data exhibits the
consumption and production inseparability that is non recursive nature of the
consumption behavior as a significant number of consumers are also producers of
rice. About 66% of the respondents are located in rural areas while 34% dwell in
urban areas.
Table 3.3 Distribution of respondents in sample
Region
Total
location
frequency
percent
North east
153
Rural
99
64.7
urban
54
35.3
Rural
46
38.9
urban
72
61.1
Rural
89
91.8
Urban
8
8.2
Rural
124
69.7
urban
54
30.3
North west
Center
South
118
97
178
Data source: ADRAO/ WARDA 2006
40
Table 3.4 Comparison of sample to Benin 2002 census population
Region
Total population
Urban
Urban
population
population
%
of
Urban
as population
as
total % of sample
population
population
South
2 250 495
1 173 877
51%
35.3%
Centre
1 135 877
287 573
25%
61.1%
Northeast
1 245 264
428 126
34%
8.2%
Northwest
899 479
305 947
34%
30.3%
Total
6 769 914
2 630 493
32%
34%
Source: Institut de la statistique Benin (www.CITYPOPULATION.DE)
The sample of respondents per region does not correspond with the proportion of
inhabitants per region in the population. Table 3.4 above shows urban population in
the South, Northeast and to a lesser extent the Northwest has been underrepresented.
For example in the South urban population makes up 51% of total population but in
the sample urban consumers make up 30% of respondents (the majority (69%) of
respondents are from rural areas). Urban population in the central region is
overrepresented; nevertheless overall the proportion of urban respondents to rural
respondents almost matches with the population. This representation means that the
results might fail to reveal preference differences within re gions, thus it might be
difficult to interpret intra and inter regional differences.
41
The survey gathered household socioeconomic household data, like age, sex, and
marital status, level of education, location, income, and household size. Consumption
data focused on rice consumption patterns including types of meals, frequency of
meals, and rice quality characteristics. Whereas rice quality characteristics have an
effect on rice as was stated in the preceding chapters, household characteristics
influence household consumption choices, expenditure patterns and demand for
products. These factors also help classify households by income categories (low,
medium, high) and location (rural versus urban).
Consumers were also asked to identify the most frequently consumed and preferred
type of rice out of the following options;
Domestic rice
Import rice
Other factors included in the analysis
Household annual income
Age, sex and education level of household head
Location – categorical dummy variable 0= urban, 1=rural
Region – the 12 administrative departments will be classified into four regions
namely the South, Centre, Northeast and Northwest.
Price – dependent variable for hedonic model and independent for discrete
choice model
42
Although some regions have been overrepresented whereas others have been
underrepresented, in general the data is potentially representative of all consumers
across geographic and socioeconomic distinctiveness.
43
Chapter 4: Results and Discussion
4.1 Introduction
This chapter presents findings of the hedonic pricing and the discrete choice model
performed on Benin household rice consumption data set of 2002. The first section of
the chapter gives a brief description of the construction of the dependent and
independent variables that affect household decisions and their choices. In the final
section of this chapter results from the econometric analyses i.e. results of both
hedonic pricing model and logistic regression are presented with detailed
interpretation to derive relevant policy implications.
4.2 Hedonic price function
This section estimates consumer‟s values of quality attributes (mostly consumption
based) such as presence of foreign matter, discoloration, rate of broken grain, grain
shape, ease of cooking, grain cohesion, taste, aroma, grain size, swelling capacity and
texture. Specifically this section examines the relationship between price and rice
quality attributes. The economic value of a good is revealed by the consumer‟s
willingness to pay for the good.
4.2.1 Dependent variables
The dependent variable for the hedonic function is market price. Market price was
obtained from respondents, who expressed it as an average of all rice varieties
purchased by household. The results of an independent samples t- test indicate that
there is a statistically significant difference between the mean price of rice for rural
44
and urban areas (t=-6.84 and p =0.000). In urban areas the mean price of rice is
statistically significantly higher (315.96) to (300.18) in rural areas. This can be
attributed to the fact that people living in cities have comparatively higher incomes
than rural people and in part due to high consumption of import rice by urban
consumers. The results also suggest that there is a statistically significant difference
between the mean price for imported and domestic rice (t=-9.7937 and p=0.0000). In
other words imported rice has a statistically significantly higher mean price (300.19)
compared to domestic rice (345.87) which can be interpreted as signaling quality
differences between the two types of rice. On average imported rice is priced 22%
South, 5% Central, 15% North East and 12% North West more than domestic rice.
The price difference is outstanding in the Southern region most likely because of its
proximity to the port, which is the point of entry for imports. The results of the
krustal wallis test and Anova test indicate that mean price is not statistically
significantly different among the four regions (p=0.51 and 0.27 respectively). Price
differences from one region to another depend on marketing costs and household
income which influences demand for grain properties (Chodhury, 1992).
45
Table 4.1 Rice price mean and standard deviations FCFA/kg
Region
South
Central
North-East
North-West
All regions
Household location
Rural
Urban
312.13
321.15
319.98
304.99
315.97
(71.99)
(46.89)
(55.88)
(58.46)
(59.45)
388.81
322.05
349.09
337.03
350.09
(101.23)
(72.26)
(66.44)
(73.52)
(79.53)
272.93
312.22
300.38
300.65
300.19
(47.12)
(40.76)
(54.06)
(59.43)
(53.18)
351.52
329.74
352.63
340.77
345.87
(87.92)
(53.91)
(56.61)
(74.42)
(71.90)
Rice origin
Domestic
Import
4.2.2 Explanatory variables
Consumers were asked to rate each rice variety according to quality characteristics
that are believed to affect rice price (see table E.4 appendix 6). Quality is measured
as a categorical variable (1-5, 1=good 3=average 5=bad) with one representing
abundant presence of the assumed desirable effect and five otherwise. Rating of
attributes is subjective in that it is not expressed relative to any standard measure. In
the regression model the categorical variables (1-5) were transformed into a dummy
variable that is value 1-3 = 1 and 4-5 = 0.
46
Table 4.2 Summary statistics for dependent and independent variables Hedonic model
Variable
Definition of variable
Range
Mean
SD
Price
Average price of rice
150-700
327.37 68.70
Impurities
Presence of foreign matter
1-5 (1=few, 5=high)
1.9
1.00
Color
Whiteness of rice
1-5 (1=white 5=off)
1.9
0.88
Rate of breakage
Presence of broken grain
1-5 (1=low 5=high)
2.0
0.96
Grain shape
Shape of grain
1-5 (1=long 5=short)
1.7
1.01
Ease of cooking
Cooking time (minutes)
1-5 (1=long 5=short)
2.0
0.79
Grain cohesion
Grain cohesion after
1-5 (1=sticky 5=not)
2.8
1.19
cooking
Taste
Flavor palatability
1-5 (1=good 5=bad)
1.8
0.75
Aroma
Smell/perfume
1-5 (1=good 5=bad)
2.5
1.13
Grain size
Size of grain after cooking
1-5 (1=long 5=short)
2.6
1.02
Swelling capacity
Capacity to enlarge
1-5 (1=good
1.9
0.89
2.1
0.83
5=does not swell)
Texture
Feel/consistency
1-5 (1=tender
5=hard)
Product labeling is the main source of information on a product. A typical product
label includes the name of the product, net quantity, name and address of distributor,
list of ingredients, shelf life, storage and cooking instruction and nutritional
information. In the Benin rice market normally we would not find labeling for rice
that distinguishes one variety from the other. Hence consumers mostly make their
purchase decisions based on physical quality alone rather than intrinsic or
47
unobservable qualities such as varietal distinctiveness. Nearly all respondents
indicated that there is a clear distinction between local rice and import rice. Table 4.2
presents consumer subjective valuation of attributes. The results suggest that
consumers prefer rice with little if any impurities, few broken grains, long grain,
cooks fast, separate grain (not sticky) and are not so concerned with the aroma, grain
size after cooking and texture. The proposed choice based decision model described
below would further determine the key attributes and their relationship to consumer
purchase behavior.
4.3 Discrete choice model
This section examines consumer‟s preferences for product country of origin, quality
attributes and price. In addition to intrinsic product characteristics, we investigate
whether inclusion of extrinsic product characteristics and consumer socioeconomic
characteristics might reveal direct and indirect effects of country of origin.
4.3.1 Dependent variable
The outcome variable of this model is a dichotomous variable defined based on the
origin of rice i.e. either the consumer prefers imported or domestic rice. The country
of origin variable takes the value of 1 if the respondent states that household
purchases import rice and 0 otherwise.
48
Table 4.3 Distribution of respondents by region and rice type (number)
Type of rice
South
Central
Northeast
Northwest
Domestic
55
75
107
89
Urban
3
4
27
52
Rural
52
71
80
37
Imported
154
79
140
106
Urban
48
6
58
71
Rural
106
73
82
35
Chi square test results indicate that there is a statistically significant relationship
between the types of rice (imported versus domestic) consumed and geographic
location of the household (region). Chi square with three degrees of freedom =
24.7024 and p =0.000. The results seem to suggest that in all regions imported rice is
more preferred to domestic rice although the magnitude of coefficients varies across
regions. Regression results for domestic rice purchased in urban areas in the Southern
and Central regions and import rice purchased in the Central urban region have to be
interpreted with caution because the sample size of 3, 4 and 6 respondents/households
respectively is much smaller than the statistically acceptable sample size of 30. This
obstacle made it difficult to compare urban consumer preferences across regions, but
meaningful inferences can be made for other regions and for the rural areas.
49
4.3.2 Independent variable
Table 4.4 Summary statistics for dependent and independent variables discrete model
Variable
Variable definition
Range
Mean
S. Dev
Income
Household annual
18 000 to 4 664 000
512 229.6 556 860
income
Origin
Rice origin
0=local 1=import
0.6
0.49
Location
Household location
0=rural 1=urban
0.33
0.47
Region
Administrative zone
1=South 2=Central
2.53
1.12
1=male 2=female
1.18
0.38
3=Northeast
4=Northwest
Sex
Gender of household
head
Age
Age of household head
19 to 85
42.09
11.49
Married
Marital status
0= not married
0.82
0.39
0.61
0.49
3.18
2.18
1=married
Education
Level of education
0=non 1=primary,
secondary, college
Children
Number of children per
0 to 16
household
The results show that consumers purchase 88% of their consumption; the remaining
proportion is made up of proceeds from subsistence production and emergency food
donations by non-governmental organizations. Rice is purchased from urban and rural
retail shops, wholesale (urban and rural) shops and local millers. Urban markets are
50
the major rice suppliers, urban markets account for about 40% of market share (total
rice purchases) compared with about 28% for village/rural markets.
The majority of households (83%) are headed by males compared to 16% female
headed households. The majority of respondents (82%) are married, 15% are single or
widowed or divorced. Marital status has important implications for household
allocation of roles according to gender, which is a common phenomenon in a rural set
up of a developing country, bearing in mind that consumers are also producers.
Women and girls are more involved in rice production while men and boys usually
engage in other farm and off- farm activities (Horna, 2005). The study did not solicit
specific roles of household members. The average age of the household head is 42
while the age range is between 19 and 85 years. A large number of the respondents
(40%) have not received any formal education; this is consistent with UNICEF (2004)
report that 41% of adults in Benin are illiterate. The remaining 26% have attended
elementary school, 27% have gone up to secondary or college level, while 6% have
attained tertiary education.
As per consumer economics theory individuals choose a consumption set that closely
matches their preferences subject to a budget constraint. A result of the MannWhitney test suggests that mean income is not statistically significantly related across
the four regions. Chi square with three degrees of freedom = 1.115 and p = 0.7734.
Although not statistically significant, on average income is highest in the South region
because the capital city of Benin (that is Porto Novo) is located in this region and
another large densely populated city of Cotonou. Other large cities are located in the
Central and Northeast region. It is noteworthy (even if not statistically significant)
51
that in the Central and Northwest region rural income is higher than urban income on
average and in the Northeast average rural and urban income is almost equal. This can
be due to the fact that in the North cotton production is a major source of livelihood
and rural development (WARDA, 2007). Cotton production is the country‟s largest
foreign currency earner contributing 50 to 70% of national export earnings per year.
The North east region produces about 75% of total annual cotton production
(ADRAO, 2006).
In the survey, the consumers were asked if they would consider increasing the amount
of rice they consume were their income augmented. Nearly 60% of consumers
responded affirmatively (if income increases they would increase rice consumption)
and the remaining claimed that they were already meeting their consumption
requirements. A minority (28%) stated they would switch to better quality rice if their
income is augmented. This seems to suggest that the rice quality preferred by
consumers is available in the market, perhaps the quantity is insufficient or price is a
restraining factor. The results also show that average per capita per annum
consumption is about 67kg. The WARDA (2007) rice trends report shows that per
capita consumption increased from 5.3kg in 1961 to 45.7kg in 2005. A survey
conducted by ADRAO (2006) illustrates that there are peak and lean periods of
consumption, the difference between the WARDA and survey average may be a result
of the timing of data collection. The results also demonstrate that 51% of the
respondents consume more rice now than they used to before (although a time period
was not specified). This could be related to the effect of income growth and
urbanization in Benin.
52
The average number of children per household is three nevertheless it ranges from 0
to 16 children. The average number has been calculated across all consumers, which
also included single, divorced and widowed individuals with or without children and
polygamous households with many children. Polygamy is socially accepted in Benin
(Horna, 2005).
4.4 Empirical results of the Hedonic price function
The model assumed a linear relationship between the predictors and the outcome
variable. OLS linear regression assumes that the relationship between the predictors
and the outcome is linear. Residual versus predictor variable plots did not indicate
non- linearity. The problem of non linearity means that the residuals have stronger
dependence, not only between them but also with explanatory variables. When there
is a perfectly linear relationship among the predictors the regression model estimates
of the coefficients become unstable and standard errors are inflated (Wooldridge,
2006). The variance inflation factors VIF test was used to detect for co linearity of the
predictors. The results of the VIF test show that there is no evidence of correlation
within variables. The Breusch-Pagan test was used to test for heteroscedacity; the Ftest results confirm that the independent variables are jointly significant thus there is
no evidence of heteroscedacity in three (south, centre and northwest) of the four
regions of Benin. Although there is evidence of heteroskedacity in the Northeast
region, the OLS estimators are consistent and unbiased but they might not be efficient
(Wooldridge, 2006). The Box-Cox parameter corresponds to no transformation at all.
In all four regions (South, Centre, Northeast and Northwest) of Benin applications
cannot be rejected at the conventional five percent levels, indicating that there is no
53
evidence that we need to transform the model. The Ordinary Least Squares
assumption of normality, linearity and constant variance are met by this model.
Table 4.5 Empirical results of hedonic model
Variable
All regions
South
Centre
Y = price
N=786
N=204
N=152
Foreign matter
4.23
5.39
-24.66**
(0.63)
(0.31)
(-2.16)
Color/whiteness
-6.15
21.94
1.16
(-1.01)
(1.29)
(0.11)
Rate of broken
17.56***2
17.90
10.60
(2.93)
((1.35)
(0.71)
Grain size
12.36**
26.77**
10.44
(2.05)
(2.03)
(0.71)
Time to cook
-5.48
3.43
-7.23
(-0.88)
(0.16)
(-0.64)
Grain cohesion
1.49
0.68
5.00
(0.31)
(0.06)
(0.56)
Taste
10.26
-26.52
-20.33
(1.56)
(-1.36)
(-1.29)
Aroma
3.66
41.39***
-8.65
(0.71)
(3.22)
(-0.89)
After cook size
-8.97*
-76.11***
10.35
(-1.83)
(-6.92)
(1.17)
Swelling
-6.42
34.22*
-12.69
(-1.09)
(1.93)
(-1.10)
Texture
2.30
9.46
-1.58
(0.41)
(0.60)
(-0.15)
Parboil
28.33***
5.18
39.39***
(4.23)
(0.35)
(3.77)
Origin
31.89***
41.86***
20.75**
(5.03)
(2.86)
(2.01)
Income
0.00***
7.95
0.00***
(2.96)
(0.07)
(2.86)
Location
31.20***
49.92***
-2.99
(5.92)
(3.93)
(-0.19)
South region
-5.99
(-0.69)
Central region
12.74*
(1.68)
Northeast
17.00***
region
(2.69)
Constant
227.77*** 214.84*** 282.46***
(19.23)
(6.05)
(13.18)
2
R
20%
43%
13%
Breusch-pagan
0.0000
0.0000
0.0016
2
Northeast
N=239
26.69***
(2.56)
-22.82***
(-2.71)
14.57
(1.64)
4.62
(0.60)
-2.09
(-0.24)
0.13
(0.02)
-6.48
(-0.75)
-9.36
(-1.26)
20.47***
(2.75)
-21.30***
(-2.68)
3.86
(0.50)
33.85***
(3.16)
33.09***
(3.18)
5.17
(0.99)
20.01***
(2.75)
Northwest
N=191
-2.05
(-0.14)
-8.04
(-0.62)
15.98
(1.14)
10.58
(0.72)
-17.25
(-1.49)
8.85
(0.82)
14.92
(1.10)
9.93
(0.91)
9.07
(0.78)
-2.91
(-0.24)
-5.27
(-0.45)
36.03***
(2.31)
35.91***
(2.64)
0.00
(1.44)
30.20***
(2.81)
264.11***
(17.46)
32%
0.1321
219.81***
(7.72)
11%
0.0000
T statics in parentheses p value ***<0.01, **<0.05 and *<0.1
54
The expected relationship of the explanatory variables to price (refer to table 3.1) was
based on the premise that Benin consumers prefer parboiled rice to raw rice. In
general the coefficient signs presented in table 4.5 match with the expected signs in
table 3.1; although the relationship between price and attributes varies within regions.
The results displayed in table 4.5 indicate that importance of characteristics varies
within regions nevertheless on the whole consumers from all regions prefer parboiled
rice (except for the South region). On average consumers pay a premium of 28.33
FCFA for parboiled rice compared to raw rice. The results demonstrate that there is a
positive relationship between price and geographic location. In other words urban
consumers pay higher premiums for better quality rice (31.20 FCFA) except for the
central region. The results suggest that consumers in the central region pay a discount
for better quality. There is need to note that first of all this figure is statistically
insignificant and secondly the number of observations (10 urban consumers for both
domestic and urban rice) is too small thus this result is not representative of the
population. In appendix three we have presented regression results that compare rural
and urban consumer preferences as well as consumer preference for imported and
domestic rice. For further information regarding preferences variations in each region
between urban and rural consumers and rice types please refer to appendix 3. The
coefficient estimate for rice origin dummy variable is statically significant and
positive indicating that consumers pay higher premiums for imported rice compared
to domestic rice. The results suggest that on average consumers pay 31.89 FCFA
more for imported rice. By definition of hedonic pricing consumers pay premiums for
better quality and discount low quality. In other words the results indicate that
imported rice is preferred because it is of higher quality than locally produced rice. A
55
study by ADRAO (2006) found that traders and middlemen only sell imported rice
because it sells fast and is more profitable.
The most important attributes for consumers in the South region are grain size, rice
origin and aroma. The results indicate that consumers in the south region prefer long
grain imported rice with a pleasant aroma. Although most of the attributes do not have
a statistically significant relationship with price, nearly all attributes have a positive
relationship with price excluding taste and size of grain after it has been cooked. This
means that consumers are willing to pay more for rice with less impurities, less
broken grain, with long grains, cooks fast, white (compared to off white or
yellow/brown) and sticky tender grains. The results also suggest that urban consumers
pay higher premiums (49.92 FCFA) for better quality rice compared to rural
consumers. The variable parboiled is not statistically significantly related to price
which could imply that consumers are not so concerned with parboiled rice. This can
be partly explained by the high degree of urbanization in the South. Literature
suggests that as both women and men participate in the labor force, their demand for
food that saves time and energy increases (Juliano, 1992). Parboiled rice takes more
time and energy to prepare in comparison to raw rice.
The presence of foreign matter is an important factor in purchasing decision in the
central region, households pay a discount of 24.66 FCFA for cleaner rice i.e. rice
with less foreign impurities. This might indicate that consumers are more sensitive to
price than grain quality, this is somewhat justified by the fact that only 6% of the
consumers from this region live in urban areas. Literature suggests that incomes are
usually lower in rural areas in comparison with urban areas. Consumers pay a
56
premium for parboiled rice (39.39 FCFA) and imported rice 20.75 FCFA. The
difference between parboiled rice and raw rice relate s to nutrition, convenience and
intrinsic attributes. Raw rice cooks fasts and when it is milled the bran layer is
removed resulting in a polished white kennel. On the other hand hydrothermal
processing of parboiled rice presents a distinctive aroma, a pale yellowish color, a
typical flavor. The grains have a hard texture even after cooking and do not stick
together after cooking (Juliano, 1992). Some consumers claim that parboiled rice is
more filling than raw rice. Tomlins (2005) found that in Ghana consumers preferred
imported parboiled rice in comparison with locally produced parboiled rice because of
its peculiar quality. This demonstrates that differences in processing have an impact
on consumer preferences and acceptability of rice grain. Dalton (2004) gives
emphasis to the inclusion of production traits as well as post-harvest attributes in rice
breeding.
In the Northeast region degree of cleanliness (presence of foreign material), color and
size of grain after cooking and swelling capacity are important factors. Consumers
pay a premium of 26.69 FCFA for rice with fewer impurities, 20.47 for long grains,
33.85 FCFA parboiled rice and 33.09 FCFA for imported rice. Consumers discount
white rice (22.82 FCFA) and grains that swell during cooking (21.30 FCFA). These
characteristics indicate that consumers have a strong preference for parboiled rice.
In the Northwest region all grain attributes are not statistically significant except for
rice origin and parboiled rice variables. About 60% of domestic rice is produced and
consumed in the north region (ADRAO, 2006). This region does not face a lot of
competition with imported rice because of inaccessibility (bad transport network)
57
Consumers face tradeoffs in their purchasing decisions, choosing parboiled r ice
means foregoing some positive consumption traits contained by raw rice varieties.
Consumers pay a premium of 39.80 for parboiled rice and 35.30 for imported rice.
In summary the results demonstrate that consumers are highly responsive to rice
quality characteristics. Each variety of rice found in the market has its own distinct
characteristics determined either by genetics or postharvest handling or both.
Consumers are faced with tradeoffs in their purchasing decisions, choosing one
variety from another means the consumer has to forego desirable characteristics of the
alternative. The results also illustrate that the rice market is segmented, which
explains preference differences from one region to another. On one hand segments of
the market seem to have a strong preference for rice quality characteristics on the
other hand some consumers are extremely sensitive to market price. Prices in cross
sectional data are assumed to reflect quality effects as well as consumer tastes and
producer costs. In other words this analysis considers both the consumption and the
production side. The results also suggest that the effect of country of origin is a very
important factor in household purchasing decision. Consequently the following
section examines the relationship between country of origin and household
socioeconomic factors as well as intrinsic attributes of rice and price.
4.5 Empirical results of the discrete choice model
The aim of this section is to examine how consumers evaluate a product when they
are aware of its country of origin. Logistic regression has been used to analyze this
relationship. The model does not assume a linear relationship between the
independent variables and the dependent and does not require variables to be linearly
58
distributed and does not assume homoscedacity. Nonetheless, it requires that
observations be independent and that the independent variables be linearly related to
the logit of the dependent (Garson, 2009). The slope of the coefficient is interpreted as
the rate of change in the logarithm odds as the explanatory variable
changes.
59
Table 4.6 Coefficients of the discrete choice model
Variable
All regions
South
Centre
Northeast
Northwest
Y = origin
N=785
N=204
N=152
N=238
N=191
Foreign matter
1.94***3
0.75
3.85***
3.88***
2.24***
(5.81)
(0.95)
(2.97)
(3.95)
(3.01)
2.06***
3.37***
2.19**
1.70***
(6.65)
(2.30)
(3.47)
(3.29)
(2.53)
1.99***
2.33***
3.01**
2.06***
2.29***
(6.70)
(3.13)
(2.32)
(3.36)
(2.90)
-2.65***
-4.25***
-4.70***
-1.74***
-4.15***
(-7.18)
(-3.80)
(-3.17)
(-2.57)
(-3.91)
1.07***
2.26**
2.71**
0.11
1.24**
(3.36)
(1.97)
(-2.62)
(0.15)
(2.06)
-0.74***
-0.56
-2.31***
-0.58
-0.77
(-2.97)
(-0.80)
(-2.62)
(-1.06)
(-1.48)
-1.49***
0.24
-0.96
-1.51*
-1.73**
(-3.99)
(0.23)
(-0.70)
(-1.75)
(-2.41)
-0.27
1.85**
0.28
-1.51***
-0.33
(-1.06)
(2.28)
(0.39)
(-2.62)
(-0.59)
1.38***
0.65
-1.51
-2.67***
-2.35***
(-4.22)
(0.63)
(-1.51)
(-3.42)
(-3.23)
0.01***
0.03***
0.02***
0.01***
0.01**
(5.53)
(3.79)
(2.45)
(3.14)
(2.32)
-0.02**
-0.00
-0.04*
-0.03
-0.04*
(-2.46)
(-0.04)
(-1.64)
(-1.30)
(-1.75)
-4.32***
-15.61***
-9.44**
-4.62**
2.69
(-4.33)
(-3.64)
(-2.50)
(-2.09)
(0.32)
Pseudo R2
50%
64%
57%
65%
53%
Log likelihood
-265.19
-44.38
-45.18
-57.18
-62.31
Chi 2(20)
0.0000
0.0000
0.0000
0.0000
0.0000
Color/whiteness 1.96***
Rate of broken
Grain size
Time to cook
Grain cohesion
Taste
After cook size
Swelling
Price
Age
Constant
3
Z statics in parentheses p value ***<0.01, **<0.05 and *<0.1
60
The results indicate that in all regions consumers pay a premium for imported rice
(both parboiled and raw rice). The results suggest that choice of imported rice over
domestic rice is based on rice quality characteristics. Literature suggests that country
of origin is a key explanatory variable of willingness to pay a price premium; it
influences product preferences directly and indirectly through perceived quality
(Loureiro, 2000). A rice variety has a higher probability of being chosen if it contains
little to no foreign matter, very few broken grains, has white grain, short grains and
cooks fact. When consumers are satisfied with grain quality they are prepared to pay a
premium for that variety. This presents evidence of price differentiation based on
grain quality characteristics. These results are consistent with findings by Griliches
(1961) that products with multiple brands differentiate price according to product
attributes. The results show that socioeconomic characteristics of the individual are
not important. Rutsaert (2009) studied consumer preference in Senegal and found that
socio-demographic factors were not important in determining consumer purchase
decisions. It is worth mentioning that except for the Northeast region location o f
household (rural/urban) is not a statistically important factor in purchasing decisions.
Some variables have not been included in table 4.6 above please refer to appendix 5
for a full list of variables. Appendix 5 presents both the coefficients of the regression
and the odds ratio.
In the Southern region imported rice is positively related to grain color, wholeness of
grain (less broken grain), ease of cooking, income and price however, it is negatively
related with grain size. In other words consumers prefer white rice (compared to
yellowish rice), short grain and wholegrain (a low proportion of broken grains).
Consumers are also concerned with convenience attributes that is the length of
61
cooking (time to cook). The results suggest that higher income consumers
discriminate on rice quality compared to low income consumers who may be more
concerned with price than quality. Duff (1991) found that an increase in consumer
income results in increased demand for higher quality food. Better quality rice fetches
a higher price and sells quickly in both urban and rural areas. The results illustrate that
choice of imported rice over domestic rice is based on consumption attributes and
price and not the characteristics of the individual. A survey conducted by ADRAO
(2006) demonstrates that the South region is dominated by imported rice due to its
proximity to the port. This poses stiff competition for low quality domestic rice. From
a policy perspective this implies that policy makers and p lant breeders have to come
up with policies and rice varieties that will improve local rice for it to be a viable
enterprise.
In the central region consumers prefer imported rice because it has less impurities
(foreign matter), has whiter grain, less broken grain, has short grains, cooks fast, and
its grains are not sticky. The most important household characteristics are age and
education level. The results indicate that consumers who are young and educated are
most likely to purchase imported rice (than domestic rice). Young and educated
consumers are more concerned with convenience attributes (energy and time saving)
compared to older less educated consumers who are usually conservative. Grunert
(2003) classifies consumers into two gropus that is those who value traditional food
and others who are innovative and adventurous.
In the Northeast purchase decisions are based on the degree of cleanliness of the grain
(less impurities), color (white), proportion of broken grains, grain size (short), taste,
62
swelling capacity (does not swell) and texture (hard). Married consumers are more
likely to purchase imported rice compared to single, widowed or divorced. This is
probably related to income that is married people are more likely to have higher
incomes. High income households pay a premium for high quality attributes.
In the Northwest region consumers are concerned with presence of foreign material,
rice color, proportion of broken grain, grain size (short), cooking time (cooks fast),
taste and swelling capacity (does not swell). Younger consumers are more cognizant
of quality characteristics in comparison with older consumers. There is evidence that
in some parts of the northern region consumers prefer domestic rice (see appendix 5)
partly due to inaccessibility and in part due to duality in production. This region is
located very far from the port and the road network is not good, which makes it
inaccessible; the cost of transporting rice is not worth the effort (ADRAO, 2006). A
greater proportion of domestic rice is produced in this region and because of duality in
production most farmers retain the bulk of their production for household
consumption (ADRAO, 2006)
To sum up the results indicate that rice quality characteristics are the most important
factor in purchasing decisions. Consumer preference of imported rice is based on
quality attributes. Household characteristics have a limited effect on household
purchasing decisions. The results illustrate that both domestic and imported rice have
positive and negative implicit prices. Food and agriculture policy is expected to play
an important role to cope with structural changes in food consumption.
63
Chapter 5 Conclusion and Summary
5.1 Overview
This study applies the hedonic pricing approach and the discrete choice model in
estimating consumer valuation of rice attributes. Both models are derived within the
framework of the new approach to consumer theory proposed by Lancaster (1966)
which states that utility of the product could be derived from its attributes. The
hedonic model assumes that utility maximizing consumers consider a set of quality
characteristics other than price in making their purchasing decisions. Market price
data is used to examine consumer‟s valuation of rice quality characteristics. In
addition to intrinsic characteristics, the study also analyzed the importance of country
of origin, household characteristics and price on determining consumer choice.
Moreover the study examined variations in consumer tastes across regions, between
rural and urban markets and across income groups. A total of 546 rural and urban
households were evaluated in this study.
The implicit prices obtained from the hedonic pricing model together with logit
coefficients obtained from the discrete choice model indicate that rice attributes are
important determinants of consumer preferences. The results illustrate that consumers
differentiate on rice varieties based on sensory attributes that include rice type
(parboiled or raw), grain size, taste, smell, color, wholeness and cleanliness of grain.
Specifically consumers pay a premium for parboiled rice varieties, white grain, good
smell and taste, fewer impurities, fewer broken grain. Consumers discount for grain
size (prefer short grain to long grain) and grain cohesion (prefer non sticky grains).
Some of the characteristics seem to be contradicting, for example parboiled rice is
64
characterized by a distinctive yellowish color, a distinct taste and smell. The attributes
outlined above illustrate that consumer preferences vary across households (rural or
urban), income classes and regions. The results demonstrate that some consumers
especially in the South region prefer raw rice (because it cooks fast and saves energy)
while consumers in other regions have a strong preference for parboiled rice.
Among all factors contributing to price differences (between domestic and imported
rice) results of the hedonic price function illustrate that country of origin is the most
important factor across all regions. Logit coefficients show that country of origin
influences consumer preference indirectly through perceived quality. Consumers
prefer the aroma, grain size, color, taste, texture and swelling capacity of imported
rice. Inspection of the results revealed that consumer attribute preferences vary
considerably across the four regions studied. Differences across the regions for
preferred attributes convey the high degree of social and economic heterogeneity
among the sites. In theory household characteristics such as level of education,
income and age affect household decision making and choice. Surprisingly this study
suggests that household characteristics have very little if any impact in determining
household purchasing decisions. The only characteristics that were found to be
important are income and age. Younger consumers and high income consumers were
found to be more concerned with quality characteristics. The strong evidence for high
value of country of origin is consistent with many previous studies in literature
(Tomlins et al 2005; Rutsaert et al 2009; ADRAO 2006)
Findings indicate that Benin should concentrate government and private resources on
improving rice grain quality. Plant breeders at WARDA have already done a
65
tremendous job by developing NERICA rice varieties that have the ability to compete
with imported rice grain in terms of yield and grain quality. However increasing
emphasis on higher value rice varieties to meet the changing diets of consumers
indicates the need to focus renewed attention on postharvest systems. Unacceptably
high losses due to poor handling and lack of appropriate infrastructure have reduced
economic benefits to small producers (Diagne, 2006). Postharvest activities are an
integral part of the food production system. Policy makers and researchers have to
work together to promote best practices for post harvest handling and management,
focusing on a broad spectrum of operations and stakeholders in traditional and
modern marketing systems. To meet the goal of delivering high quality food to
consumers consequently making domestic rice more competitive in the global market.
5.2 Recommendations
The rice postharvest system requires improvement in the use of resources for research
and development, in particular postharvest handling practices. Postharvest handling
practices influence rice sensory characteristics. Different rice properties (aroma,
flavor, color) appeal to different consumers across the regions. To meet consumer
needs for rice with specific sensory attributes government policy must provide
incentives for farmers and private entrepreneurs to invest in more efficient processing
technologies. The producer is only motivated to improve product quality if the
premium is sufficient to pay the increased cost of growing a superior product (Waugh,
1929). Production of good quality milled rice starts at the farm with good quality
seed, good crop care for uniform growth and grain size. Factors that affect grain
quality such as mixing varieties, heat discoloration, contamination, insect damage in
storage, fissuring in drying and breakage at milling can be controlled in post
66
production operations. The lack of appropriate technologies, technical and
management skills results in poor quality milled rice leading to economic losses
(FAO, 2002). When food losses are minimized both food security and farm incomes
increase and this is of vital importance for small and medium farme rs in developing
countries.
The rice variety is chosen by the farmer, cultivated and subjected to a series of
postproduction activities (e.g. harvesting, threshing, cleaning and drying) before being
sold at the farm gate level. Research has shown that postharvest handling techniques
affect grain quality. For instance delayed and poor drying of grain results in
deterioration of grain quality result in a higher proportion of broken grain and
discolored grain (Wedgewood, 1992). It is at farm level that the variable elements of
rice grain quality are determined.
To the best of our knowledge there is no standard system of grading in the Benin rice
market. Market participants currently use proxy indicators such as country of origin,
physical and chemical characteristics because of imperfect market information. There
is much scope for further analyzing the impact that formal grades and standards could
have on household purchasing decisions and market efficiency.
Although Benin is at present deficient in rice and while higher yields continue to be
the prime objective of research, it is believed that improvement in rice quality will
become more important in abating competition between domestic rice and import rice.
Poor quality rice is sold at a lower price (as it is considered as an inferior good) and is
also difficult to market (WARDA, 2006). If farmers do not get an attractive price, rice
67
production is discouraged and will naturally decline. Hence market price will be an
important factor in attaining self- sufficiency in rice.
The study raises questions about the ability of the marketing system to transmit
quality incentives from consumers to farmers. At farm gate level farmers sell rough
rice while the consumer purchases milled rice. The results indicate that price has a
direct positive correlation with certain quality characteristics. In other words quality
of milled rice is highly dependent on the quality of rough rice purchased by the mills.
In a perfectly competitive market, consumer preferences would be transmitted back
through the pricing system so that the derived demand at each level in the commodity
chain would be directly related to consumer preferences (Duff, 1992). However the
actual performance of rice markets does not conform to the perfectly competitive
model and hence it is not necessarily true that the market will provide adequate
incentives to produce high or higher grain quality (Wedgewood, 1992). Policy
markers have to decide whether to correct for marketing inefficiencies through market
based instruments or government intervention. If improved grain quality is to be
achieved consistent with consumer preferences, it is essential that farmers have the
incentives to supply good quality rough rice to the market.
The government will be confronted with the goal of maintaining producer incentives
and keeping prices at tolerable levels for consumers in line with self sufficiency and
price stabilization, while encouraging production efficiency. Consumers want a steady
supply of good quality rice at reasonable prices while farmers want the highest
possible price for their products.
68
5.3 limitations and future research
Some potential future improvements to the analytical procedures and to our overall
understanding of consumer preference include including both consumption and
production attributes. This study focused on consumer preferences, but the duality of
production would require policies that encompass both demand and supply. The goal
of the producer is to maximize production subject to a technology constraint and a
cost function. Investigating producer preferences for seed together with consumer
preferences for grain quality will give a holistic understanding of the Benin rice
market.
Inspection of the results revealed that parameter estimates found to be statistically
important varied considerably among the four regions. To assess the acceptability of
various rice varieties by consumers, a variety by variety evaluation would be helpful.
For example literature suggests that NERICA rice can compete with imported rice. It
is worthwhile to further investigate consumer preferences of specified NERICA
varieties and Specified imported rice attributes using conjoint analysis.
There is evidence that rice production efficiency is partly a result of market
efficiency. Producers have limited market power, millers and traders/middlemen
decide on prices. As a result producers do not have an incentive to improve rice
quality, because the farm gate price might not reflect on rice quality characteristics. It
is worthwhile to further investigate producer preference for rice grain qualities that is
both consumption and production using farm gate price instead of market price. The
results from such a study results can then be compared with results of consumer
preferences based on market price (like this current study) to understand the
inefficiencies in the rice market.
69
References
Adjovi, N. R. A. (2006). The economics of rice production systems in South Benin:
potential, constraints and perspectives. Available at www.cabi.org.
ADRAO/INRAB (2006). AMELIORATION DE LA QUALITE ET DE LA
COMPETITIVITE DU RIZ LOCAL AU BENIN. Africa Rice Center (WARDA).
Ara, S. (2003). Consumer willingness to pay for multiple attributes of organic rice: A
case study in the Philippines. Annual Meeting, August 16-22, 2003, Durban, South
Africa, International Association of Agricultural Economists.
Baker, G. A. (1999). "Consumer preferences for food safety attributes in fresh apples:
Market, consumer characteristics and marketing opportunities." Journal of agricultural
and resources economics v24 (1): pp80-97.
Belcher K W, Germann A E, & Schmutz J k, (2007). "Beef with environmental and
quality attribute: Preferences of environmental group and general population
consumers in Saskatchewan, Canada." Agriculture and human values v24(iss3):
pp333-42.
Carson, R. (2000). "Contingent valuation: A user's guide." Environmental science and
technology. vol 38(iss 4).
Chryssochoidis G, K. A., & Perreas P (2007). "Ethnocentric beliefs and country of
origin (COO) effect: Impact of country, product and product attributes on Greek
consumer's evaluation of food products." European Journal of Marketing 41(11-12):
1518-1544.
Chung, C., Boyer, T, Han, Sungill (2009). "Valuing Quality Attributes and Country of
Origin in the Korean Beef Market." Journal of Agricultural Economics, Vol. 60, No.
3, 2009, 682–698 vol 60(No 3): 682-698.
Chung, C., Boyer, T, Han, Sungill (2009). "Valuing Quality Attributes and Country of
Origin in the Korean Beef Market." Journal of Agricultural Economics, Vol. 60, No.
3, 2009, 682–698 vol 60(No 3): 682-698.
Dalton, J., T, (2003). A hedonic model of rice traits: Economic values from farmers in
West Africa. Annual Meeting, August 16-22, 2003, Durban, South Africa,
International Association of Agricultural Economists.
de Janvry, A., E, Sadoulet (2009). "Agricultural growth and poverty reduction:
Additional evidence." oxford University press.
Diagne A, S. M., Diawara S, Diallo AS, AB Barry. (2007). "Evaluation de la
diffusion et de l‟adoption des variétés de riz NERICA en Guinée. Miméo ADRAO,."
Presented as a paper at the 2nd Conference of the AAAE, 19–22 August 2007.
Edmeandes, S. (2006). A hedonic approach to estimating the supply of variety
attributes of a subsistence crop. IFPRI discussion paper 148.
70
Ehmke, M., D, &J, L Lusk, & W, Tyner (2008). "Measuring the relative importance
of preferences for country of origin in China, France, Niger and United states."
International Association of Agricultural Economists 38(3): 277-285.
Epple, D. (1987). "Hedonic prices and implicit markets: Estimating demand and
supply functions for differentiated products." Journal of political economy vol 95(1).
FAO (2004). "Fact sheet No 5 on International Year of Rice” Food and Agriculture
Organisation Rome, Italy. Available on www.rice2004.org Accessed December 2009.
FAO (2004). "International Trade in Rice, Recent Developments and Prospects."
Food and Agriculture Organisation, Rome, Italy World Rice. Research Conference
2004 Tsukuba, 5-7 November 2004.
FAOSTAT (2005). "Food insecurity in the world." Food and Agriculture
Organization Statistics Database accessed July to December 2009.
FAO (2006). "International Rice Newsletter." Food and Agriculture Organization vol
55: Rome, Italy Accessed August to November 2009.
FAO (2008). "The state of food and agriculture 2008." Food and Agriculture
Organization Rome, Italy accessed August to December 2009.
Garson, G. D. (2009). "Event history analysis: Statanotes North Carolina State
University."
Grazia, A., & T, de Magistris (2007). "The demand for organic foods in the South of
Italy: A discrete choice model." Food policy vol 33.
Griliches, Z. (1961). "Price indexes and quality change: Studies in new methods of
measurement." Harvard press university Cambridge. MA.
Grunert G K (2005). "Food quality and safety: Consumer perception and demand."
European review of Agricultural Economics 32(3): 369-391.
Hara, M. (2000). Measuring positive externalities in Japanese rice production: A
hedonic price analysis. Ecological Economics. New York, Rensselaer Polytechnic
institute. D Phil.
Hassine-Belghith, N. B. (2009). "Exporting, Technical Efficiency and Product
Quality: An Empirical Analysis ofthe Agricultural Sector in the Mediterranean
Countries." Journal of Development Studies vol 45(Iss 5): 769-788.
Holcomb, R. (1997). Evaluating the effects of rice quality attributes on conumer
preferences and rice demand. Agriculture Economics. Texas, Texas A & M
University. Phd thesis.
Horna, J., D, & M, Smale, & M, von Oppen, (2005). "Farmer willingness to pay for
seed related information: Rice varieties in Niger and Benin." Environmental and
Development Economics 12: 799-825.
71
IRRI (2008). "Rice solutions: Why are prices high." International Rice Research
Institute database Accessed September-December 2009: http://ricelib.irri.cgiar.org.
IRRI (2008). "Why is it happening." International Rice Research Institute database
Accesses August to December 2009.
IRRI (2009). "Rice policy - World rice statistics- trends in the rice economy."
International Rice Research Institute database http://ricelib.irri.cgiar.org.
Juliano, B., O, & C, M, Perez, & M, Kaosa-ard (1992). Grain quality characteristics
of export rices in selected markets. in the Book Consumer demand for rice grain
quality. D. B. Unnevehr L J, Juliano B O, IRPRI and IDRC.
Kivetz R, S. I. (2002). "Self-control for the righteous: Toward a theory of precpmmitment to indulgence." Journal of Consumer Research 29.
Koasa-ard, M., & B, O, Juliano (1991). Assessing quality characteristics and price in
selected international markets. Consumer demand for rice grain quality. D. B.
Unnevehr L J, Juliano B O, IRRI/IDRC.
Kormawa P, T., A, Diagne A, Kebbeh M, (2005). "Global rice trade: Dynamics,
policy conflicts and strategies in Africa." Africa Rice Center (WARDA) Benin.
Ladd, G., Zober, M (1977). "MODEL OF CONSUMER REACTION TO PRODUCT
CHARACTERISTICS." Journal of Consumer Research vol 4(Iss 2): 89.
Ladd G W , V., Suvannunt (1976). "A model of consumer goods characterstics."
American Journal of Agricultural Economics 58(3): 504-510.
Lancaster, K. J. (1966). "A new approach to consumer theory." Journal of political
economy 74(2).
Langyintuo, A., S, & G, N, Toukam, & L, Murdock, & J, Lowerberg-De boer, & D, J,
Miller, (2004). "Consumer preferences for cowpea in Camerroon and Ghana."
Journal of Agricultural Economics 30(3): 203-213.
Loomis J, G., Helfand (2001). "Environmental policy analysis for decision making."
Kluwer Academic publishing.
Loureiro, M., L, & Umberger, W, J, (2005). "Assessing consumer preference for
country of origin labeling." Journal of Agriculture and applied economics vol 37(1).
Louviere, J., J, & Hensher, D, A, & Swait, J, D (2000). Stated choice methods:
analysis and applications. Cambridge, Cambridge University Press.
Lucas, R. B. E. (1975). "hedonic price Functions." Economic Inquiry 13.
Luce, R. D. (1959). "Individual choice behaviour: A theoretical analysis." New york
Wiley.
72
MacFadden, D. (1974). "Econometric models for probabilistic choice among products
" Journal of business vol 53(no 3).
Manski, C., F, (1997). "Nonparametric analysis of randomized experiments with
missing covariate and outcome data." working paper.
Melton B.E (1996). "Consumer preferences for fresh food items with multiple quality:
Evidence from an experimental auction of pork chops." American journal of
Agricultural economics v78(4): 916-23.
Mergenthaler M, Weinberger K, &, Qaim M, (2009). "Consumer valuation of food
quality and food safety attributes in Vietnam." review of Agricultural Economics
v31(2): 266-83.
Mishili, F., J, & J, Fulton, & et al (2007). "Consumer preferences for quality
characteristics along the cowpea value chain in Nigeria, Ghana and Mali." working
paper 06-17 Purdue University, College of Agriculture.
Rachmat, R., & R, Thahir, & M, Gummert, (2006). "The empirical relationship
between price and quality of rice at market level in west Java." Indonesian Journal of
Agricultural Science 7(1): 27-33.
Ratchford, B. T. (1979). "Operationalizing Economic Models of Demand for Product
Characteristics." Journal of Consumer Research vol 6(1): 76.
Rosen S (1974). "Hedonic prices and implicit markets: product differenciation in pure
competition." journal of political economy v82(1).
Rutsaert, P., Demont, M., Ndour, M. & Tollens, E. (2009). Competitive rivals:
Willingness to pay for Senegal river valley versus imported rice. II Workshop on
valuation methods in Agro- food and environmental Economics Experimental
Auction: Theoretical background and empirical applications. Barcelona.
Samuelson, P. (1938). "A Note on the Pure Theory of Consumer's Behaviour " J-stor
Vol. 5 (No. 17): 61-71.
Scarpa, R., & G, Philippidis, & F, Spalatro, (2005). "Product country images and
preferences heterogeneity for Meditteranean food products: A discrete choice
framework." Agribusiness vol 21(3).
Schettler, B., Danilo Ruiz1, O, Sepúlveda, J, Sepúlveda, M, Denegri, (2009).
"Importance of origin in rice purchasing decisions in Talca and Termus, Chile."
AGRARIA 36(2): 239-248.
Tomlins, k., L, & J, T, Manful, P, Larwer, L, Hammond, (2005). "Urban consumer
preferences and sensory evaluation of locally produced and import rice in West
Africa." Food quality and preference 16: 79-89.
73
Toquero, Z. (1991). Consumer demand for rice grain quality. In rice grain marketing
and quality issues: Selected papers from the International Rice research Conference
37-46. Los Banos Phillipines, IFPRI.
Ubilava, D., & Foster, K, (2009). "Quality certification vs product traceability:
Consumer preferences for information attributes of pork in Geogia." Food policy
34(3): pp305-10.
Umberger, W., J, & D, M, Feuz, & C, R, Calkins, & M, Sitz, (2003). "Country of
origin labeling of beef products: US consumer perceptions." Journal of food
distribution research.
Unnevehr, L., & Duff, B, & Juliano, BO (1992). "CONSUMER DEMAND FOR
RICE GRAIN QUALITY." International Rice Research Institute, Manila Philippines
and International Development Research Center, Ottawa, Canada (1992).
USAID (2003). "Rice Production Systems in Nigeria: A Survey. West Africa Rice
Development Association. September 2003. PN-ADB-852."
USAID (2004). "Building on Successes in African Agriculture. International Food
Policy Research Institute. April 2004. PN-ADA-154." International Food Policy
Research Institute Available at www.ifpri.org.
USDA (2008). "Trends in the rice economy." Global Agricultural Supply and
Demand: Factors Contributing to the Recent Increase in Food Commodity Prices/
WRS-0801Economic Research Service/USDA Available at www.ers.usda.gov.
Varian, H. (1992). Microeconomic analysis. New York, Norton.
von der Lans, I., A, &k, van Ittersum, A, De Cicco, & M, Loseby, (2001). "The role
of the region and EU certificates of origin in consumer valuation of food products."
European review of agriculture economics vol 28(iss 4).
WARDA (2005). "Rice Policy and Food Security in sub-Saharan Africa. Proceedings
of a workshop held on 7–9 November 2005, Cotonou, Benin. Cotonou, Benin: Africa
Rice Center (WARDA). 418 pp."
WARDA (2008). "Rice trends in Sub-Saharan Africa." Africa Rice Center (WARDA)
Cotonou, Benin available at www.warda.org.
WARDA (2009). "Africa rice factbook." Africa Rice Center (WARDA) Benin
available at www.warda.org.
Waugh, F., V, (1929). "Quality as a determinant of vegetative prices: A statistical
study of quality factors influencing vegetable prices in the Boston wholesale market
market." Columbia University Press, New York.
Wooldridge, J. (2006). Introductory Econometrics: A modern approach. Michigan,
South -Western.
74
World. Bank (2006). "World Bank database." Accessed August to December 2009.
75
Appendices
Appendix 1 background information
Rice production in Benin in 2006
Rice ecologies
Cultivated area
Production
Average yield
Number of
hectares
tonnes
tons/hectare
farmers
Irrigated
2 883
18 464
6.4
2 883
Upland
5 766
11 078
1.9
23 064
Lowland
20 180
44 312
2.2
40 360
total
28 828
73 854
2.6
66 307
The evolution of consumption and production in Benin, 1961-2007
450000
400000
350000
300000
250000
200000
production
150000
consumption
100000
50000
year
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
0
Source FAO, 2007
76
Appendix 2 Descriptive statistics
Rating attributes by origin (dummy variable 1=abundance, o otherwise)
Variable
Impurities (1=fewer, 0=abundant)
Color (1=white, 0=yellowish)
Broken grain (1=less, 0=abundant)
Grain size (1=long, 0=short)
Cooking time (1=cooks fast, 0=slow)
Cohesion (1=sticky grain, 0=not sticky)
Taste (1=good, 0=bad)
Aroma (1=good, 0=bad)
After cooking size (1=long, 0=short)
Swelling (1=swells, 0=does not)
Texture 1=tender, 0=hard)
Domestic
0.48
0.52
0.44
0.87
0.60
0.5
0.82
0.43
0.45
0.85
0.56
Import
0.93
0.89
0.86
0.74
0.87
0.39
0.83
0.64
0.37
0.72
0.79
Rating attributes by region (dummy variable 1=abundance, o othe rwise)
Variable
Less impurities
Color (white)
No broken grain
Grain size (long)
Cook time (fast)
Cohesion (sticky)
Taste (good)
Aroma (good)
Size after cook (long)
Swelling
Texture (tender)
South
0.83
0.83
0.69
0.79
0.87
0.34
0.82
0.56
0.47
0.89
0.81
Central
0.75
0.68
0.79
0.90
0.79
0.51
0.92
0.59
0.47
0.84
0.76
Northeast
0.70
0.70
0.64
0.73
0.73
0.48
0.79
0.47
0.38
0.67
0.59
Northwest
0.73
0.75
0.68
0.82
0.66
0.44
0.8
0.63
0.31
0.71
0.66
Household mean annual income per region (numbe r in brackets)
Region
South
Central
Northeast
Northwest
Rural
397 620.92
(153)
478 943
(142)
508 747.44
(156)
641 440
(70)
Urban
969 647.06
(51)
427 000
(10)
517 976.47
(85)
437 398.37
(123)
Total
540 627.45
(204)
475 526.32
(152)
512 002.49
(241)
511 403.11
(193)
77
Appendix 3 discrete choice model regression results
Regressions for the South region
Variable
South
all
N=208
Impurities
8.39
(0.52)
Color
18.30
(1.13)
Broken grain
14.57
(1.23)
Grain size
26.01**
(2.11)
Cooking time
8.10
(0.53)
Cohesion
1.11
(0.06)
Taste
-26.16
(-1.45)
Aroma
41.47***
(3.43)
After cooking
-73.97***
size
(-7.15)
Swelling
29.41*
(1.83)
Texture
9.93
(0.44)
Origin
44.46***
(3.23)
Location
51.21***
(4.42)
Constant
223.72***
(11.35)
47%
South
urban
N=51
-69.19
(-0.96)
3.01
(0.04)
50.24
(1.34)
16.79
(0.51)
27.48
(0.30)
6.63
(0.21)
-126.13*
(-1.71)
120.11***
(3.69)
-129.14***
(-4.84)
26.98
(0.49)
118.17*
(1.98)
dropped
South
urban
import N=48
2.78
(0.03)
dropped
urban
South rural
N=158
12.34
(0.76)
19.96
(1.28)
5.36
(0.41)
22.87*
(1.70)
11.27
(0.60)
1.97
(0.19)
-12.60
(-0.73)
25.91**
(2.14)
-61.63***
(-5.62)
25.01
(1.49)
-3.11
(-0.22)
48.29***
(3.54)
rural
331.07***
(4.02)
45%
228.26***
(11.82)
32%
156.34
(1.01)
43%
34.81
(0.89)
17.13
(0.50)
73.47
(0.77)
3.73
(0.12)
-60.35
(-0.71)
115.17***
(3.54)
-135.01***
(-5.00)
39.63
(0.72)
123.79**
(2.08)
import
urban
The above table offers the comparison between urban and rural consumers and
consumer preferences between domestic rice attributes and imported rice attributes. It
is worth noting that in the South region domestic rice is not consumed at all in the
urban areas.
78
Regressions for the Central region
Variable
Impurities
Color
Broken grain
Grain size
Cooking time
Cohesion
Taste
Aroma
After cooking
size
Swelling
Texture
Origin
Location
Constant
Central all N= Central
153
urban N=10
-24.97**
(-2.01)
-3.80
(-0.25)
17.03
(1.50)
0.12
(-0.01)
-15.82
(-1.31)
3.53
(0.38)
-11.98
(-0.74)
-6.00
(-0.53)
3.91
(0.47)
0.35
(0.04)
8.92
(0.84)
28.26***
(2.43)
9.12
(0.52)
329.83***
(16.87)
10%
Dropped
Dropped
-5.55
(-0.00)
Dropped
-23
(-0.13)
17.5
(0.10)
Dropped
-53.33
(-0.24)
30.83
(0.14)
-17.5
(-0.10)
Dropped
108.33
(0.77)
Urban
305.5
(1.12)
-1.83
Central rural
domestic
N=71
-35.23***
(-2.95)
13.37
(1.23)
18.99*
(1.70)
33.69
(1.31)
-21.75*
(-1.88)
-10.53
(-0.97)
1.69
(0.07)
4.00
(0.35)
13.83
(1.39)
47.68**
(2.74)
27.61**
(2.52)
domestic
Central
rural
import N=73
rural
Rural
230.34***
(5.79)
15%
372.00
(12.32)
0.09
10.00
(0.32)
-31.95
(-0.82)
54.16
(1.64)
-16.12
(-0.62)
-20.24
(-0.46)
22.91
(1.48)
13.59
(0.45)
1.78
(0.11)
-5.98
(-0.38)
-21.56
(-1.28)
-52.12**
(-2.27)
Import
Urban import and domestic regression results omitted because number of observations
is statistically insignificant. The results suggest that consumers in the central region
prefer domestic rice to imported rice but then again our interpreted is limited because
urban areas are underrepresented in the sample.
79
Regressions for the Northeast region
Variable
Northeast all
N=236
Impurities
23.15**
(2.25)
Color
-23.92***
(-2.75)
Broken grain
15.66*
(1.84)
Grain size
9.15
(1.23)
Cooking time
3.36
(0.34)
Cohesion
0.37
(-0.02)
Taste
-8.37
(-0.96)
Aroma
-5.21
(-0.92)
After cooking
21.77***
size
(2.99)
Swelling
-19.59**
(-2.50)
Texture
1.49
(0.10)
Origin
31.29***
(3.11)
Location
21.22***
(3.27)
Constant
298.379***
(25.03)
29%
Northeast
urban N=85
-8.34
(-0.40)
-11.21
(-0.68)
63.92***
(2.69)
11.14
(0.80)
2.34
(0.12)
-21.32
(-1.65)
-8.08
(-0.48)
20.94
(1.37)
36.70***
(2.64)
-13.88
(-0.92)
-10.29
(-0.59)
13.59
(0.55)
urban
Northeast
rural N=162
34.42***
(2.92)
-28.11***
(-2.77)
2.89
(0.30)
10.17
(1.10)
-2.38
(-0.22)
12.47
(1.58)
-4.80
(-0.47)
-19.74**
(-2.33)
17.81**
(1.99)
-19.71**
(-2.15)
-0.27
(-0.03)
36.59***
(3.18)
rural
Northeast
rural
domestic N=80
22.84*
(1.79)
-12.15
(-0.95)
14.19
(1.24)
-13.21
(-0.96)
8.03
(0.59)
21.96**
(2.05)
-32.25**
(-2.15)
-9.49
(-0.79)
21.51**
(1.99)
-38.43**
(-2.52)
22.27**
(2.04)
Domestic
306.65
(13.58)
28%
302.59***
(22.35)
27%
322.96***
(17.90)
22%
Rural
The results suggest that rice characteristics of domestic rice are preferred to those of
imported rice. The results indicate that rural consumers are more responsive to quality
attributes in comparison with urban consumers.
80
Regressions for the Northwest region
Variable
South
all
N=189
Impurities
0.18
(-0.05)
Color
-9.85
(-0.65)
Broken grain
10.84
(0.82)
Grain size
-0.43
(0.27)
Cooking time
-10.29
(-1.74)
Cohesion
7.05
(0.60)
Taste
9.99
(1.02)
Aroma
14.70
(1.56)
After cooking
12.56
size
(0.81)
Swelling
-10.56
(-0.65)
Texture
-9.14
(-0.50)
Origin
30.55**
(2.66)
Location
24.06**
(2.71)
Constant
288.09***
(13.76)
13%
South
rural
N=72
-24.36
(-1.26)
21.08
(1.25)
-9.10
(-0.52)
27.48
(1.20)
-21.03
(-1.45)
12.69
(0.93)
19.89
(1.19)
13.21
(0.98)
40.55**
(2.40)
-46.19***
(-2.73)
-19.53
(-1.23)
61.95***
(3.55)
Rural
South urban South
rural
N=123
import N=35
13.70
-41.77
(0.67)
(-0.66)
-21.06
110.60***
(-1.18)
(3.18)
26.61
75.52*
(1.33)
(1.94)
-3.54
68.14**
(-0.19)
(2.28)
-24.33
-21.16
(-1.56)
(-0.95)
-3.08
27.40
(-0.21)
(1.11)
25.30
32.23
(1.33)
(1.53)
25.12*
-34.33
(1.70)
(-1.34)
-5.67
117.36***
(-0.37)
(3.80)
13.29
-49.77**
(0.83)
(-2.11)
5.35
0.53
(0.35)
(0.02)
17.06
Import
(0.91)
Urban
Rural
278.77***
(10.64)
27%
287.81***
(10.51)
0
164.34**
(2.46)
38%
The results suggest that consumers prefer imported rice to domestic rice . Rural consumers
pay high discounts and premiums compared to urban consumers maybe because on average
rural incomes are higher than urban incomes.
81
Appendix 4 correlation tables
Variable correlation tables
Variable
VIF
1/VIF
Marital status
2.19
0.46
Sex
2.17
0.46
Origin
1.98
0.51
Impurities
1.70
0.59
Presence of broken grain
1.58
0.63
Color
1.45
0.69
Time to cook
1.45
0.69
Texture
1.39
0.72
Aroma
1.33
0.75
Taste
1.27
0.79
Swelling capacity
1.21
0.83
Grain size
1.20
0.83
Grain size after cooking
1.19
0.84
Grain cohesion
1.14
0.88
Education level
1.14
0.88
Parboil
1.13
0.88
Income
1.11
0.90
Age
1.10
0.91
Location
1.07
0.94
Mean VIF
1.41
Co linearity of predictor variables was checked using the VIF command. If a variable
has a tolerance value less than 0.1 (i.e. greater than 10) it means that the variable
could be considered as a linear combination of other independent variables.
82
Appendix 5 discrete choice model regression results
Coefficients of the discrete choice model
Variable
South region
N=204
Impurity
0.66
(0.375)
2.18**4
(0.015)
2.19***
(0.002)
-4.14***
(0.000)
2.71***
(0.018)
-0.29
(0.684)
0.26
(0.808)
0.39
(0.682)
1.87**
(0.025)
0.39
(0.678)
-1.11
(0.166)
4.07***
(0.007)
0.028***
(0.000)
-0.28
(0.112)
-0.44
(0.675)
-10.96***
(0.000)
61%
142.75
-45.48
Color
Broken grain
Grain size
Cooking time
Cohesion
Taste
Aroma
After cook size
Swelling
Texture
Income
Price
No. of children
Location
Constant
Pseudo
Chi
Log likelihood
4
Central region
N=152
2.91***
(0.006)
3.19***
(0.000)
2.15**
(0.038)
-3.91***
(0.002)
1.98*
(0.094)
-1.88**
(0.013)
-2.09*
(0.097)
1.01
(0.146)
-0.12
(0.860)
-0.82
(0.360)
-0.12
(0.892)
-7.45
(0.445)
0.019**
(0.021)
-0.031
(0.839)
-0.43
(0.691)
-6.94**
(0.028)
54%
113.79
-48.40
Northeast
region
N=239
3.92***
(0.000)
2.13***
(0.001)
2.08***
(0.001)
-1.81***
(0.007)
0.15
(0.836)
-0.66
(0.227)
-1.79**
(0.042)
0.49
(0.377)
-1.42***
(0.010)
-2.78***
(0.000)
1.43**
(0.014)
-4.62
(0.277)
0.015***
(0.002)
-0.068
(0.382)
1.33**
(0.028)
-5.97***
(0.002)
65%
212.27
-57.51
Northwest
region
N=193
2.15***
(0.003)
1.53**
(0.016)
2.37***
(0.003)
-4.09***
(0.000)
1.22**
(0.029)
-0.91*
(0.070)
-1.92***
(0.007)
0.13
(0.805)
-0.26
(0.631)
-1.90***
(0.003)
0.034
(0.950)
-4.50
(0.389)
0.009**
(0.028)
-0.054
(0.627)
-0.023
(0.963)
-0.89
(0.600)
50%
134.32
-65.87
Z statics in parentheses p value ***<0.01, **<0.05 and *<0.1
83
Odds ratio of discrete choice model
Variable
Impurity
Color
Broken grain
Grain size
Cooking time
Cohesion
Taste
Aroma
After cook size
Swelling
Texture
Income
Price
South region
Central region
N=204
N=152
North east
region N=238
North west
region N=193
2.06
48.15***
66.12***
8.37***
(0.90)
(2.93)
(4.32)
(2.97)
8.91**
31.26***
9.90***
5.85***
(2.37)
(3.37)
(3.34)
(2.65)
10.36***
19.79**
8.04***
10.36***
(3.07)
(2.30)
(3.37)
(2.93)
0.015***
0.010***
0.18**
0.015***
(-3.79)
(-3.19)
(-2.55)
(-3.79)
19.69**
11.55*
1.15
4.031**
(2.38)
(1.87)
(0.19)
(2.34)
0.789
0.12**
0.48
0.39**
(-0.31)
(-2.45)
(-.26)
(-1.80)
1.18
0.51
0.19*
0.14***
(0.15)
(-0.48)
(-1.84)
(-2.62)
1.45
1.93
1.61
1.23
(0.41)
(0.87)
(0.83)
(0.39)
7.26**
1.18
0.22***
0.75
(2.23)
(0.23)
(-2.56)
(-0.51)
1.54
0.27
0.06***
0.12***
(0.43)
(-1.24)
(-3.46)
(-3.09)
0.25***
0.67
4.56**
1.13
(-1.62)
(-0.40)
(2.42)
(0.20)
1.00***
1.00
0.99
0.99
(1.65)
(0.08)
(-1.10)
(-0.61)
1.029***
1.03***
1.015***
1.01**
(3.83)
(2.59)
(3.06)
(2.41)
84
No. of children
Location
Parboiled
Age
Education level
R2
5
0.679** 5
1.13
0.96
0.98
(-1.97)
(0.68)
(-0.52)
(-0.16)
0.644
0.39
4.91**
1.04
(-0.39)
(-0.76)
(2.42)
(0.07)
13.38**
0.71
0.69
0.66
(2.50)
(1.81)
(-0.43)
(-0.48)
1.02
0.96*
0.97
0.96
(0.59)
(-1.67)
(-1.31)
(-1.61)
1.13
1.59 *
1.03
1.02
(0.65)
(1.81)
(0.20)
(0.14)
65%
57%
66%
52%
T statics in parentheses p value ***<0.01, **<0.05 and *<0.1
85
Appendix 6 questionnaire
CONSUMER Questionnaire
Department ( province):…………………Town or City……………...…..
Ward or Borough:.………………………
Village:……………:..Neighbourhood...................
Name of the head of household: . . . . . . . . . ……... . . :…… . . . . . . . .Person
responding ….. ……….. . . . . . .
Ethnicity...............................................
Enumerator‟s
name:…………………………………………………………………………..………
…
Survey Number:………………... Date :
. . . …. - ………. . . . – 2005
1. Characteristics of the household
Name
Family, First
A Marital
Sex
ge
status
Kinship
1=head of household,
2=Spouse of head of
household,
3=son/daughter of head of
household,
4=nephew/niece,
5=father/mother of head
or head’s spouse,
6=brother, sister,
7=brother-in-law, sister-inlaw,
Kinship
to head of
household
Sex
1=male
2=female
Yea
Level
rs living of
in the
educatio
neighbo n
urhood
Ma
in
liveliho
od
activity
Seco
ndary
activity
Statut matrimonial
1=married
2=bachelor/spinster,
3=widow(er),
4=divorced
86
Freq
uency of
sickness
over the
last 12
months *
8=labourer
9=protégé,
10=other (specify)
Activities
Education level
1=agriculture,
2=livestock production,
3=Housework,
4=merchant,
5=craftsman,
6=laborer,
7=none,
8=student/pupil,
9=other (specify)
1= Primary (specify) ,
2= High school (specify),
3= College/University
(specify) ,
4= Koran studies,
5=none,
6= Literate,
7=other (specify)
87
E.2. Rice preparation and list of dishes
Please provide us with a list of meal menus (dishes) you prepare and what types of rice are required?
List of dishes
Type of rice
required
C
ode
Quality (code)
N
ame
Br
oken/
wh
ole
Coo
king
method
(code)
Parboiled
O
rigin
Cooking time
Time
(min)
Quantity
L
U
Whe
n the rice
is
prepared
(code
)
Daily
household
rice
consumption
k
g
U
L
k
g
Frequency and time of the
year when the dish is most
often prepared
Frequency and time of the
year when the dish is least
often prepared
B
egins
B
egins
E
nds
Frequen
cy of
preparation
(per month)
E
nds
Frequen
cy of
preparation
(per month)
88
Mode de cuisson :
1. Coal/Charcoal
2.Wood
3.Gas
4. Hot plate
5. Pressure-cooker
6.Sawdust
7. Other (specify)
Quality:
Broken/whole —
Parboiled--Origin
1. Whole
1. Yes
1. Local
2. Broken
2. No
2. Impor ted
3. Mixed
Time when prepared:
1breakfast
2. Lunch
3. Super
4. Breakfast & lunch
5. Breakfast & super
6. Lunch & super
7. Breakfast, lunch & super
Period: 1=January
2=February
3=March
4=April
5=May
6=June, 7=July
8=August
9=September
10=October
11=November
12=December.
89
E.3. Quality and types of rice consumed.
Types of
Do
rice consumed in you know
the region
this type
of rice ?
1=ye
C
Nam s
ode e*
2=no
End
Start
End
Start
End
Start
Have
Frequency and
Frequency and
Frequency and
Origin
Qu
you
periods of
periods of
periods of consumption
1=own antity
consumed consumption when consumption when when rice stocks are
production purchas
this type of rice is in abundance little rice is available empty
2=purc ed (if a
rice in the
hase
purchas
last 12
3=gift e was
months ?
4=vario made)
(1=yes,
us
2=no)
Frequ
Frequ
Frequen
L
ency
ency
cy (month)**
U g
(month)**
(month)**
Frequency of
purchases (code)
(if a purchase
was made)
Sit
Price of
e of
rice***
purchas
e
(code)
(if
a
purchas
e was
k
A
L
Em made)
(
(
bun- ittle pty rice
FCFA FCFA
dance avail. stocks
/
/
L
k
U)
g)
Top 3
criteria, in
order, for
the
selection
of this
type of
rice
(code
)
1
2
3
4
5
6
7
8
9
1
0
* Give the local name, the Fr ench name, and its description
** Give the number of times rice was consumed per month ; r efer to the previous page’s codes.
***Price (per kg) at wh ich the housewife buys this type of rice at the market. If she does not buy this type of rice, but consumes it, she is to indicate the price (per kg) at wh ich
this type of rice is sold at the market.
90
Top 3
criteria, in
order, for
the selection
of this
supplier
(code)
(if a
purchase
was made)
LIST OF CODES
Frequency of
purchase (code):
1. Daily
2. Weekly
3. Monthly
4. Annually
4. Other (specify)
Selection criteria (code):
1. No foreign matter
2. Whiteness
3. Rate of breakage
4. Shape of grains
5. Ease of cooking
6. Grains very sticky after
cooking
7. Grains not sticky after
cooking
8. Taste
9. Aroma (perfume)
10. Keeping qualities
11. Swelling capacity
12. Hard texture
13. So ft texture
14. Price
Si te of
purchase (code):
1. Store
2. Urban or
regional ma rket
3. Village
ma rket
4. Weekl y
ma rket
5. Specialty
s tore outside the
ma rket
Supplier selection
criteria (code) :
1. Distance/proximity
2. Reputation
3. Easy de payment
(credit)
4. Acquaintance or
friend
5. Kinship (family
member)
6. Price
7. Availability of the
product
8. Type of rice sold
9. Cleanliness of the rice
sold
91
E.4. Rating of types of rice* know by the housewife. For each of the criteria below, a score should be given for the corresponding type of rice according
to the codes given in the table immediately below.
Type of rice
Code
Name
Foreign
materials/hulls
Whiteness
Rate of
breakag
e
Shape of
grains
Scores for selected criteria (CODE)
Cohesion
Taste
Ease of
after cooking
cooking
Aroma
(perfum
e)
Keeping
qualities
Swelling
capacity
Texture
*This is not only the types of rice listed and known by the housewife, but also those mentioned by the housewife and not (necessarily) appearing on the list drawn up by the enumerator. The
enumerator must ask the individual being interviewed to provide a score for each of the criteria listed below. The scores vary from 1 to 5, as indicated in the table below, with each score (code)
corresponding to a single level of appreciation. A value of 9 is to be given in the case where the housewife does not know what score to give for a given criterion..
Scoring each criterion
C
Foreign
materials/hulls
1
Absent/very
few
Few
ode
2
3
Average
Whiteness
Rate of
breakage
Very white
Very
low
White
Low
Average
Avera
Shape of
grains
Ease of
cooking
Long and
slim
Long
and fat
Average
Very
quick
Quick
ge
4
High
Poor
Avera
Cohesion after
cooking
Very sticky
Sticky
Short/ro
Aroma
(perfume)
Very
good
Good
Very
good
Good
Average
ge
High
Taste
Avera
Very high
Very poor
Very
Broken
Does
Does
high
9
Does not
Does not
Very
long
Long
Swelling
capacity
Texture
Very
good
Good
Very
tender
Tender
Average
Average
Average
Short
Poor
Hard
Very
short
Does
Does
not swell
Does
Very
hard
Does
ge
Slow
Weakly sticky
Poor
und
5
Average
After
cooking
keeping
quality
Very
slow
Does
Not sticky
Does not know
Very
poor
Does
No
aroma
Bad
smell
Does
92
know
know
not know
not know
not know
not know
not know
not know
not know
not know
93
5. Do you always buy rice from the same supplier?
1. No (*)
2. Often (*)
3. Always
(*)
From how many regular suppliers do you have: . . . . . . . . . . .[ . . . . . . . . . .]
6. Can you tell upon buying it whether rice is local or imported?
1. Yes
2. No
7. If so, how do you recognize imported rice (code), give up to 3 criteria?
1. Price
2. No foreign materials
3. Cleanliness (absence of bran)
4. Whiteness
5. Number of broken kernels
6. Shape of kernels
7. Ease of cooking
8. Cohesion of kernels after cooking (very sticky)
9. Cohesion of kernels after cooking (notsticky)
10. Taste
11. Aroma (perfume)
12. ‘Shelf-life’ after cooking
13. Swelling capacity
14. Hard texture
15. Soft texture
7b. What type of rice do you prefer (code)?
1 Local parboiled
2 Local non-parboiled
3 Imported parboiled
4 Imported non-parboiled
Why: . . . . . . . . . . . . . . . . . . . . ……………………………………………
..................................................................
8. Do you have your own rice field?
1. Yes
2. No
94
9. If so, what type of cropping method do you use (code)?
1. Upland rain-fed rice
2. Plains rice
3. Lowland non-irrigated rice
4. Lowland irrigated
5. Mangrove rice
10. Are you the rice field’s:
1. Owner
2. Renter
3. Simple beneficiary (loan)
11. What is the production period (month) . . . . . . . . . . . . . . . . . . . . . . . . . . . ……………………...
12. How many cropping seasons are there per year?…………………………………………………..
13. What is the field plot’s production used for?
Use
LU
Quantity
kg
Total production
Consumption by self
Sold
Given away
Seed bank
Other
95
14. Do you receive gifts/donations of rice?
1. Yes
2. No
Na
me
Quality (see codes for
question 2)
Whol
Parb
Ori
e/
oiling
gin
broke
n
Quantity
U
nit
L
U
Su
pplie*
Loc
ation
k
g
Type 1
Type 2
Type 3
Type 4
Type 5
Type 6
1=producer, processor, village merchants; 2=NGO; 3= Japanese donation, 4= Politicians ; 5=Other
(specify)
15. Were your income to increase, would you eat more rice?
1. Yes
2. No
16. Would you change the quality of the rice you consume?
1. Yes
2. No
If so, what other qualities of rice would you consume : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
............................................................... ...
17. What is your monthly income?……………………………..FCFA
18. What is your annual income?……………………………..FCFA
96