<%BANNER%>

Assessing Consumer Willingness to Pay for Malawi Organic Coffee

Permanent Link: http://ufdc.ufl.edu/UFE0042276/00001

Material Information

Title: Assessing Consumer Willingness to Pay for Malawi Organic Coffee Evidence from a Consumer Survey
Physical Description: 1 online resource (117 p.)
Language: english
Creator: Nkana, Fiskani
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: coffee, consumer, malawi, organic, pay, premium, price, to, willingness
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Abstract The study assessed Consumer WTP for organic coffee in Malawi using data collected through a household survey from three major cities of Malawi (Blantyre, Lilongwe and Mzuzu) using the CVM and CE methodologies. Major determining factors influencing consumer preference and WTP for organic coffee were also analyzed. CVM data was analyzed using an OLS model while CE data used a Conditional Logistic Model. Based on CVM, about 40% of the sample was WTP an average price of MK816.75 per 250g of organic coffee translating into a price premium of MK164.75 per 250g of organic coffee representing about 25% price premium over the average market price for conventional coffee of MK652 per 250g. Taking into consideration the whole sample, participants were WTP an average price of MK599.66 per 250g of organic coffee which is lower than the average market price of conventional coffee by about 8%. Only four variables were significant in influencing WTP and these are: actual price paid for coffee, being of 60 years and older, being in the high income group of over MK322, 000 per month, and an attitudinal variable depicting whether an individual thinks that organic products may offer more of some nutrients than their conventional counterparts. Based on CE, people were WTP an average price of MK1, 444.38 per 250g of organic coffee translating to a price premium of MK 792.38 per 250 g of organic coffee which represents over 100% price premium over the average market price of conventional coffee. The significant variables that influenced the probability of choice of coffee were ?Organic? representing the method of production of coffee, ?Price of coffee?, and ?None? representing that individuals did not choose either organic or conventional coffee. High price premiums were registered for organic coffee mainly due to health related issues. The majority of the sample opted for government subsidies in organic coffee for increased accessibility. Our results show that there exists a potential niche market for organic coffee in Malawi.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Fiskani Nkana.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Gao, Zhifeng.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042276:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042276/00001

Material Information

Title: Assessing Consumer Willingness to Pay for Malawi Organic Coffee Evidence from a Consumer Survey
Physical Description: 1 online resource (117 p.)
Language: english
Creator: Nkana, Fiskani
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: coffee, consumer, malawi, organic, pay, premium, price, to, willingness
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Abstract The study assessed Consumer WTP for organic coffee in Malawi using data collected through a household survey from three major cities of Malawi (Blantyre, Lilongwe and Mzuzu) using the CVM and CE methodologies. Major determining factors influencing consumer preference and WTP for organic coffee were also analyzed. CVM data was analyzed using an OLS model while CE data used a Conditional Logistic Model. Based on CVM, about 40% of the sample was WTP an average price of MK816.75 per 250g of organic coffee translating into a price premium of MK164.75 per 250g of organic coffee representing about 25% price premium over the average market price for conventional coffee of MK652 per 250g. Taking into consideration the whole sample, participants were WTP an average price of MK599.66 per 250g of organic coffee which is lower than the average market price of conventional coffee by about 8%. Only four variables were significant in influencing WTP and these are: actual price paid for coffee, being of 60 years and older, being in the high income group of over MK322, 000 per month, and an attitudinal variable depicting whether an individual thinks that organic products may offer more of some nutrients than their conventional counterparts. Based on CE, people were WTP an average price of MK1, 444.38 per 250g of organic coffee translating to a price premium of MK 792.38 per 250 g of organic coffee which represents over 100% price premium over the average market price of conventional coffee. The significant variables that influenced the probability of choice of coffee were ?Organic? representing the method of production of coffee, ?Price of coffee?, and ?None? representing that individuals did not choose either organic or conventional coffee. High price premiums were registered for organic coffee mainly due to health related issues. The majority of the sample opted for government subsidies in organic coffee for increased accessibility. Our results show that there exists a potential niche market for organic coffee in Malawi.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Fiskani Nkana.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Gao, Zhifeng.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042276:00001


This item has the following downloads:


Full Text





ASSESSING CONSUMER WILLINGNESS TO PAY FOR MALAWI ORGANIC
COFFEE: EVIDENCE FROM A CONSUMER SURVEY




















By

FISKANI ESTHER NKANA


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR MASTER OF SCIENCE DEGREE

UNIVERSITY OF FLORIDA

2010


































2010 Fiskani Esther Nkana

































To my father and mother, I dedicate this to you!









ACKNOWLEDGMENTS

I thank my parents for instilling discipline in me and making me who I am today.

You still remain my greatest role models. Many thanks to your prayers!

I appreciate the support rendered by my sisters; Lydia, Irene, Tawonga and my

brother Temwanani.

My heartfelt appreciation goes to Herbert, my fiance; I appreciate your support

towards my studies during the entire period.

I thank my Supervisor Dr. Zhifeng Gao, he had a lot of things to handle but he still

set aside some time for this work. He was quite a great resource. I also acknowledge

Dr. James Sterns who was the member of my committee.

I thank the Ministry of Agriculture and Food Security for letting me participate in

this program and for their support during my studies most especially during research

work.

I would like to thank USAID for awarding me this scholarship. This goes to both

the DC and the Malawi Local Office. Specifically, my appreciation goes to Martin Banda,

Jean Msosa-Maganga and the entire team at USAID Malawi. In the same regard, I

recognize the support by Martin Kanjadza of the American Embassy-Education

Department.

I also thank Dr. Walter Bowen, Dr. Burkhadt, Jessica Herman, Marti and their

entire team at the University of Florida for their support.

My heartfelt appreciation goes to Mr. Peter Njikho of the Coffee Association of

Malawi (CAMAL) and all its subsidiaries for their assistance in sharing necessary data

for this research. In addition, I thank all the households that I interviewed under this









study for their cooperation and honesty in giving out the data. Lastly but not least, I

thank all my friends for their moral support.

Above all, I thank my God for his abundant grace during my studies, May his name

be always praised!










TABLE OF CONTENTS


page

A C KNOW LEDG M ENTS ......... ............... ............................................. ............... 4

LIST O F TA B LES .......... ..... ..... .................. ............................................. ...... .. 8

LIS T O F F IG U R E S .................................................................. 9

LIST OF ABBREVIATIONS ....... .......... .......... ....................... 10

A BST RA C T ............... ... ..... ......................................................... ...... 12

CHAPTER

1 INTR O D U CT IO N ............................................................................................. 14

General Problem .................... .......... .......... ......... 14
Specific Problem ................ .......... .................. 18
P project O objectives ................................. ........... ............................... 19
Testable Hypotheses ......... .......... ............................... 19

2 LITERATURE REVIEW .................. ........................... 20

O overview of O organic A agriculture ................................................................. 20
Previous Studies on Consumer Preference and WTP for Organic Products .... 22
Previous Studies on Consumer Preference and WTP for Non Organic
Products .................. ......... ..... ...............32
Methods of Eliciting Consumer WTP .............................. ....... 36
The Contingent Valuation Method (CVM)......................................... ........... 37
The Choice Experiment (CE)........................... .................... 39
Experimental Auction ............... .... ........................ 40

3 RESEARCH METHODS AND DATA .............................. ....... 42

Theoretical Model ........................ ........... ........ 42
Model Estimation ........................... .......... ......... 43
Model Estimation under CVM ................. ............... ........ 43









M ode l E stim atio n under C E ..................................................... ... ................. 4 8

4 DA TA C O LLEC TIO N .................................................. 49

D e s ig n o f C V M ................ .................................. .................................... 5 3
D design of the C E ............................................. 55

5 RESULTS OF THE EMPIRICAL ANALYSIS ...................................... ............... 58

Descriptive Statistics........................ ........ 58
W willingness to Pay for O rganic Coffee .......................................... ............... 64
Motivation for the WTP for Organic Coffee ....... ....... ...... .................. 67
Support towards Organic Coffee Production ...... ........ ..... ............... 68
Empirical Analysis of Data from CVM ...................................... ............... 69
Empirical Analysis of Data from CE ...... .. ........................................ ...... 75
D descriptive Statistics ......... ....... ............ ......................... ............... 76
Results from Conditional Logit Model...................................... ........ .. 76
Comparison of Results from CVM and CE................................... .... ........... 80

6 CONCLUSION ............. ......... ................... ..................... 82

Study Lim stations ............... ............... ......... ..................... 84
Recommendations and Further Research ........................................ 85


APPENDIX

A SURVEY INSTRUMENT (VERSION A)...................................... ... .................. 87

B SURVEY INSTRUMENT (VERSION B)..................................... ............... 98

LIST O F R EFER ENC ES ................................................................. ............... 110

BIOGRAPHICAL SKETCH ............... .... ......................... 117









LIST OF TABLES

Table page


1-1 Total annual sales for tobacco in Malawi............... ..................................... 15

4-1 Sample representativeness in terms of the demographic structure of the
population of Malawi (gender, age and religion)....... .... ................................ 51

4-2 Price levels for coffee in MK/250 grams (Version B) ................ ............ .. 57

5-1 Summary for descriptive statistics ....................... ................. ... ............. 59

5-2 Expressed price for organic versus actual price paid for coffee...................... 66

5-3 Expressed price for organic versus actual price paid for coffee...................... 67

5-4 Percentage of the sample per motivation factor for positive WTP ................... 68

5-5 Percentage of the sample per motivation factor for negative WTP........ ........ 68

5-6 Support for certification of organic coffee production.................. ............ 69

5-7 Estim ated O LS m odel................................................ .............. 72

5-8 Estim ates for conditional logistic m odel................................... ..................... 76

5-9 Comparisons of WTP price premiums between CVM and CE........................ 81









LIST OF FIGURES

Figure page


4-1 Choice set in choice experiment........................ ... ........................... 57

5-1 Frequency distribution of consumer WTP for organic coffee........................... 65

5-2 Average market price for conventional coffee and WTP for organic coffee........ 65

5-3 Frequency distribution of tax ............................................ ............... ............ 70

5-4 Frequency for choice of coffee ............................................ ..... 76









LIST OF ABBREVIATIONS

BDM Becker-De Groot-Marschak's

CAMAL Coffee Association of Malawi

CE Choice Experiment

CIA Central Intelligence Agency

COOL Country of Origin Labelling

DC District of Colombia

DDT Dichlorodiphenyltrichloroethane

EA Enumeration Area

ETEI Emissions Trading Education Initiative

GDP Gross Domestic Product

GM Genetically Modified

GMO Genetically Modified Organisms

GOM Government of Malawi

IFOAM International Federation of Organic Agriculture

MCCCI Malawi Confederation Chambers of Commerce and Industry

MCPCU Mzuzu Coffee Planters Cooperation Union

MK Malawi Kwacha

MoAFS Ministry of Agriculture and Food Security

MSCE Malawi School Certificate of Education

MT Metric Tonnes

NOAA National Oceanic Aviation Administration

NOP National Organic Program

OLS Ordinary Least Squares









SAS

TCC

UNCTAD

UNIMA

US$

USA

WHO

WTA

WTP


Exchange Rate = MK150/US$


Statistical Analysis Software

Tobacco Control Commission

United Nations Conference on Trade and Development

University of Malawi

United States Dollar

United States of America

World Health Organization

Willingness to Accept

Willingness to Pay









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Master of Science Degree

ASSESSING CONSUMER WILLINGNESS TO PAY FOR MALAWI ORGANIC
COFFEE: EVIDENCE FROM A CONSUMER SURVEY

By

Fiskani Esther Nkana

August 2010

Chair: Zhifeng Gao
Major: Food and Resource Economics

Tobacco is a major cash crop for Malawi; however its performance is currently

dwindling mainly due to anti-smoking lobby by WHO. One of the potential alternatives to

tobacco is coffee, the seventh largest export crop in Malawi and the second most

commonly internationally traded commodity.

However, consumers are becoming more sensitive to the type of coffee they

consume. Taste characteristics, label of origin and other unobservable credence

attributes (e.g. organic) are becoming of great concern to consumers. It is against this

background that demand for organic products is continuously growing worldwide. This

study therefore aims at assessing Consumer WTP for organic coffee in Malawi and to

determine important factors that influence consumer preference and WTP for organic

coffee.

Data were collected from 129 participants through a household survey from three

major cities of Malawi (Blantyre, Lilongwe and Mzuzu) using the CVM and CE. CVM

data was analyzed using an OLS model while CE data used a Conditional Logistic

Model.









Based on CVM, about 40% of the sample was WTP an average price of MK816.75

per 250g of organic coffee translating into a price premium of MK164.75 per 250g of

organic coffee representing about 25% price premium over the average market price for

conventional coffee of MK652 per 250g. Taking into consideration the whole sample,

participants were WTP an average price of MK599.66 per 250g of organic coffee which

is lower than the average market price of conventional coffee by about 8%. Only four

variables were significant in influencing WTP and these are: actual price paid for coffee,

being of 60 years and older, being in the high income group of over MK322, 000 per

month, and an attitudinal variable depicting whether an individual thinks that organic

products may offer more of some nutrients than their conventional counterparts.

Based on CE, people were WTP an average price of MK1, 444.38 per 250g of

organic coffee translating to a price premium of MK 792.38 per 250 g of organic coffee

which represents over 100% price premium over the average market price of

conventional coffee. The significant variables that influenced the probability of choice of

coffee were 'Organic' representing the method of production of coffee, 'Price of coffee',

and 'None' representing that individuals did not choose either organic or conventional

coffee. 'Organic' had a higher economic impact than 'Price' signifying that consumers'

preference for coffee was mainly based on the method of production rather than price.

High price premiums were registered for organic coffee mainly due to health

related issues. The majority of the sample opted for government subsidies in organic

coffee for increased accessibility. Our results show that there exists a potential niche

market for organic coffee in Malawi.









CHAPTER 1
INTRODUCTION

General Problem

Malawi is situated in the southeast of Africa. It is bordered by Zambia to the

northwest, Tanzania to the northeast and Mozambique to the east, south and west of

the country. The economy of Malawi is agro-based. The agricultural sector in the

country employs about 80% of the labor force, contributes over 80% of foreign

exchange earnings and accounts for 39% of the Gross Domestic Product (GDP)

(MoAFS, 2010).

The agricultural sector in Malawi is dualistic in nature. It has the smallholder sub-

sector, which contributes more than 70% to the agricultural GDP, and the estate sub-

sector, which contributes less than 30% to agricultural GDP (MoAFS, 2010). The main

food crops that are grown include maize, cassava, and sweet potatoes, while the main

cash crops grown are tobacco, sugar, tea, coffee, macadamia nuts, and cashew nuts

(MoAFS, 2010).

Tobacco is the major cash crop for Malawi and the major foreign exchange earner

as well. Being a country without mineral resources, the crop is normally called the

"green gold." It is sold as an export crop through multinational companies such as

Limbe Leaf and Alliance One including others. The crop accounts for 60% of the

country's exports, contributes about 13% of the GDP and 23% of the country's tax base

(Jaffee, 2003). Therefore, the crop has been crucial for economic growth of the country.

Nevertheless, tobacco production is currently dwindling due to a number of

factors. For instance in 2007, according to the Tobacco Control Commission (TCC) of

Malawi, total production was about 111 million tonnes against 155 million tonnes in









2006. In terms of foreign exchange earnings, the country registered export earnings of

about US$160 million in 2006 against US$162.1 million in 2005. Generally, production

and foreign exchange levels have been erratic for the past 10 years. (Refer to Table 1-1

for detailed estimates). Among others, this is highly attributed to antismoking campaigns

led by public health activists with support of the World Health Organisation (WHO). To

that effect, there are fears amongst stakeholders in the agricultural sector that this

development is likely to result in a lot of producers abandoning the crop for other more

lucrative ones.

Table 1-1. Total annual sales for tobacco in Malawi
Year Volume (Tones) Realization (US$)
1995 130,181 201,562,572
1996 141,662 237,755,361
1997 158,113 248,406,791
1998 133,996 178,451,093
1999 134,386 186,784,038
2000 159,869 164,734,418
2001 124,669 143,880,881
2002 138,181 163,114,209
2003 134,326 144,061,678
2004 180,181 347,179,018
2005 145,267 162,061,730
2006 155,098 160,110,819
2007 110,715 195,547,819
2008 194,708 471,583,387
Total 2,020,352 3,190,547,231
Source: TCC

In this regard, the Government of Malawi (GOM) has recently laid down a number

of strategies aimed at coming up with alternative crops with potential to replace tobacco

as a major exchange earner for the country.









One of the policies is crop diversification which is principally implemented as a risk

management strategy. In complement, GOM devoted its efforts to revitalize production

and marketing of crops with high potential for growth such as coffee, cotton and others

whose performance have been worsening in the recent past.

For instance, Arabica coffee which is the major type of coffee grown in Malawi is

the seventh largest export crop for the country and remains an essential source of

income for farmers. On average, it has an estimated production of over 4,176 mt

annually (MCCCI, 2009). It is grown by both estate and smallholder farmers. The

smallholder farmers are concentrated in the highland areas of the northern region of

Malawi particularly in the districts of Chitipa, Rumphi, Mzimba and Nkhata-Bay. In the

Southern region, the industry is dominated by large scale farmers most especially in

Thyolo and Chiradzulu districts. Contrary to its poor performance in the past, the coffee

sub-sector has seen great improvement in its performance in recent years. For

example, in the northern region alone production increased from 2,250 mt in 2007 to

2,600 metric tonnes in 2008 and there are indications of continued improvements in the

industry (Chirwa et al., 2008).

Coffee production areas in Malawi have favourable climatic conditions with

altitudes ranging from 1,000 to 2,500 metres above sea level. It is for this reason that

Malawi coffee has a very fine flavour with a balanced body and acidity. In addition,

Malawi coffee is gaining popularity both domestically and globally. One of the most

popular brands sold is 'Mzuzu coffee'. The brand has gained both international and local

recognition. According to reports, the sale price of Mzuzu coffee has in general gone up

registering a price premium of up to 47%. Malawian coffee is exported to The









Netherlands, Germany, South Africa, Switzerland, Japan, Australia, United States of

America (USA), Italy, among others (MCCCI, 2009).

On the other hand, the world consumption of coffee is projected to increase from

6.7 million tonnes in 1998-2000 to 6.9 million tonnes in 2010 by 0.4% annually (FAO,

2003). In the meanwhile, consumers are becoming more sensitive to the type of coffee

they consume. They are specifically conscious about search, experience and credence

attributes of a product (as defined by Nelson (1970), and Darby and Karni (1973)). An

example of a search attribute is colour; that for experience attributes is taste while that

for credence attributes could be the type of production used to produce a particular

product (e.g. organic or fair trade production). According to a number of market studies,

consumers pay much attention to these attributes mainly due to health concerns

associated with the products as well as environmental and social justice concerns

associated with their production or marketing methods. It is for this reason that the world

industry for organic products has been growing to meet the growing demand for food

with special attributes. Evidently, in 2007, the global retail sales of organic products

increased to US$41.6 billion against US$ 23 billion in 2002. Despite this alarming

increase in demand, the organic industry still remains undersupplied worldwide (Wilier,

2009).

Organic coffee has established a niche market within the market for organic foods.

The primary markets for the product are North America and Europe. Currently, the two

regions account for about 97% of the global sales for organic products (Willer, 2009). It

is generally believed that because such countries have more affluent people with higher

purchasing power, consumers in these countries are willing to pay high price premiums









for expensive products. In general, consumers in developed countries are willing to pay

an average of 15% to 25% premium for organic coffee alone (Wilier and Yussefi, 2007).

With the current performance of coffee production in Malawi coupled with adoption

of Organic Agriculture in most countries in sub-Saharan Africa e.g. South Africa, Kenya,

Tanzania, Uganda and others, there is a high potential for production of organic coffee

in Malawi. Like in the other countries, organic coffee is likely to find niche markets in

both, the domestic and international markets, thereby potentially increasing farmers'

incomes and national incomes as well. Because coffee is the second most economically

important commodity in the world after oil, (Pendergrast, 2006), there is a high

possibility that the marketing of the crop will be sustainable and hence become one of

the most suitable candidates replacing tobacco as the major foreign exchange earner

for Malawi.

Specific Problem

The critical question for the Government of Malawi to answer before supporting

production of organic coffee is whether markets exist for the crop. Particularly, can

Malawi organic coffee attract higher price premiums from domestic consumers as

compared to conventional coffee? If so, what are the possible factors that would

influence the consumer preferences that may determine the market segmentation of the

organic coffee market? This study attempts to provide answers to these questions and

thus offer valuable information to Malawian government policy makers seeking suitable

substitutes for tobacco as major foreign exchange earners.

If the study shows higher Willingness to Pay (WTP) for organic coffee, it will

suggest that there exists a clear niche market for organic coffee in the country. Policy

makers will therefore be advised accordingly to promote production of organic coffee so









as to meet the potential demand. The potential growth of the domestic market for the

crop will therefore be seen as a step towards targeting international markets that have

relatively the highest level of demand for organic products in general. In complement to

organic coffee, other crops with great potential for growth could also be promoted. Such

a crop diversification strategy is likely to replace tobacco as the major foreign exchange

earner for the country.

Project Objectives

The overall objective of this study is to assess the potential demand for organic

coffee in Malawi. This will be achieved through the following specific objectives:

1. To assess consumer preference and WTP for organically produced coffee versus
conventional coffee.

2. To determine important factors that influence consumer preference and WTP for
organic coffee.

Testable Hypotheses

The following hypotheses will be tested:

1. Because of the perceived benefits associated with product attributes such as
method of production (e.g. organic production), consumer WTP and preference for
organic coffee is likely to be higher than that of conventional coffee across
consumers.

2. Heterogeneous preferences exist among consumer and thus the WTP for organic
coffee will vary across consumers' socio-demographic factors e.g. income, gender,
age, level of education, including their perceptions.









CHAPTER 2
LITERATURE REVIEW

Overview of Organic Agriculture

Organic Agriculture is increasingly being practised in more than 141 countries of

the world. There are also strong assumptions that uncertified organic production is

being practised by more countries (Yussefi et al, 2007). Currently, about 32.2 million

hectares of land are being subjected under organic production, representing 0.8% of the

total land for agriculture worldwide and the rate is estimated to be increasing (Wilier,

2009). The regions with the largest land under organic agriculture are Oceania, Europe

and Latin America. The rate of increase is mainly a response to the global market for

organic food which is rapidly increasing and mostly constitutes affluent countries

(Sahota, 2007).

As already highlighted in Chapter 1, global sales for organic food increased from

US$ 23 billion in 2002 to US$ 41.6 billion in 2007 (Willer, 2009). Although this is the

case, the market for organic products still remains undersupplied because of

underproduction of organic food globally. To that effect, many consuming countries are

relying on imports but the supply is still insufficient.

The demand for organic products is concentrated in North America and Europe. In

2005, sales of organic products in North America were about US$14.9 billion,

representing 45% of the world's generated revenue. The sales of North America were

dominated by the USA mainly due to the National Organic Program (NOP) of 2002

which propelled growth in the production of organic foods in the country. In recent

years, Europe has overtaken North America as the largest consumer of organic foods

and drinks mainly due to the appreciation of the Euro against the US dollar. Revenue for









organic products in Europe is about 75% of world's total revenue from organic food and

beverages, (Sahota, 2007). The major countries include Germany, United Kingdom,

France and Italy. Other emerging markets are Denmark, Sweden and the Netherlands.

It is thus clear that there is a great disparity between production and consumption

of organic foods and products in the world. An additional threat to the market's

sustainability is that consumption remains concentrated in Europe and America. There

are thus fears that a slight change in consumption patterns in these two regions would

likely cause a significant impact on the world production and trade trends. For instance,

if these countries decide to stop importing organic products, there is likely going to be

oversupply of the products in the producing countries thereby depressing their prices in

those countries and the world at large. As such, organic producers are advised to

develop their own domestic markets for their organic products rather than rely on export

markets only. Among others, this would be used as a risk management strategy in their

business of organic production and so sustain organic production worldwide (Sahota,

2007).

It is partly against this background that organic production is increasing in several

African countries where domestic markets are also opening up. Certified production is

mainly being practised in Uganda, Tanzania, Ghana, Ethiopia, Kenya and Zambia. The

main certified crops produced are: Fresh Vegetables, Bananas, Citrus Fruits, Coffee,

Tea, Cocoa, Sugar, Cotton and others (Elzakker, 2007). Although a large proportion of

the production is geared for export markets e.g. North America, Europe and a little bit in

Japan, regional markets have recently opened up. The major ones constitute the

Republic of South Africa and the Gulf area.









Many studies have concluded that people with higher disposable incomes are the

largest spenders on organic food (Sahota, 2007). This conclusion is therefore a threat to

the sustainability of domestic markets of organic food in countries with relatively less

affluent people. However, a number of studies have concluded otherwise. Among

others, Rodriguez et al. (2007) ascertain that the relationship between income and

consumers' WTP is very controversial. There exists a relationship between WTP and

income among some segments of the market while in other segments the relationship

does not exist. Other demographic variables tend to influence WTP for organic products

e.g. education, consumer perceptions, age, price, religion, gender (Rodriguez et al,

2007; Zepeda and Li, 2007; Peterson et al., 2008; Engel, 2008). This current study also

attempts to assess consumer preference for organic coffee in a domestic market,

particularly to determine whether consumers in Malawi are willing to pay a higher

premium for organic coffee than the conventional one.

Previous Studies on Consumer Preference and WTP for Organic Products

A number of studies have been conducted on consumer preference and WTP for

organic food, of these a few have been done on organic coffee. This section will

highlight findings of studies on organic foods in general; those for non organic products

and services will be highlighted in the next section. The non organic products include

beef, fruits, pharmacist services, medication for Alzheimer's diseases, and ground water

protection. From the literature review, we expect to gain general knowledge on

consumer preference and WTP for organic foods.

Rodriguez et al, (2007) conducted a study on WTP for various organic foods in

Argentina. The project was specifically aimed at estimating consumers' WTP for organic

products available in the Argentinean domestic market with the view of providing useful









evidence to the government to support promotion of such crops, regulating processes

and labelling programs. Data were collected from both organic and non organic food

consumers using the Contingent Valuation Method (CVM). A Binomial Multiple Logistic

Regression model was used to estimate the parameters of the targeted products

(Regular Milk, Leafy Vegetables, Whole Wheat Flour, Fresh Chicken and Aromatic

Herbs) by Maximum Likelihood. The dependent variable of the model was WTP for a

particular product and the independent variables were 'organic price premium', 'income

Level', 'risks and quality attributes perceptions' and 'socio-demographic characteristics'.

Based on the notion that quality has become a key concept in Demand Theory

(Lancaster, 1966; Antle, 1999) among others, the results of the study confirmed this as

it turned out that Argentinean consumers were willing to pay price premiums of 6% to

200% in order to acquire the better quality products (Rodriguez et al, 2007). Based on

the empirical analysis, there was a significant relation between consumer income and

the WTP for the organic products in question. The other major factors determining the

willingness to pay were 'scarce availability' and 'price' of the products that were seen as

a hindrance to consumer access to products thereby acting as a threat to expansion of

the domestic market in Argentina.

Another study on WTP for organic food was conducted by Millock et al. (2002). In

his study he used both panel and survey data. Organic food was identified as a product

with the 'Danish State Label.' The food attributes used in his study included

environmental concerns, animal welfare, and food safety/healthy concerns. A

comparison was made between results drawn from the use of CVM and those from

observed WTP. Based on the results, participants valued the attribute of 'avoidance of









chemicals' the highest. However, ordering of the valued attributes did not differ at all

across organic product types. The study conducted both in store interviews and in store

experiments on purchases of organic products. The questionnaire that was used had

four sets of questions: on purchase habits and food culture (choice of store, important

product characteristics, statements on risks from eating certain foods), questions on

organic food production (identification of the Danish Organic label, statements on

organic production and its effects), questions on habits and environmental attitudes (use

of recycled toilet paper, aluminium foil, membership of environmental associations,

statements on consumer's role in environmental protection) and finally questions on

WTP for four different products (milk, rye bread, potatoes and minced beef). According

to the study, the majority of the sample was willing to pay more than the stated

conventional market price for the products. About 59% of the sample was willing to pay

more for the organic milk, 48% for potatoes, 51% for rye bread and 41% for minced

beef. Specifically, the price premium for organic milk was 32.1%, 40.2% for organic

potatoes, 23% for rye bread and 18.5% for minced beef. A logistic maximum likelihood

was estimated (defined as willingness to pay for all four organic products). It was found

that about 32% of the sample was indeed willing to pay more for all products.

In the same study using the actual purchase data to measure revealed WTP,

about 55% of the sample were willing to pay more for organic milk, 35% were willing to

pay more for organic rye bread, 14% of the sample were willing to pay more for organic

potatoes while 6% were willing to pay more for minced beef. Results from the two

methods found that elicited (stated) WTP is overestimated compared to the revealed

(real) WTP and in this study the practise was dominant in milk. Surprisingly, for the









other products (organic rye bread, organic potatoes and organic minced beef),

consumers were actually paying more than their stated WTP for the products. For future

studies, the team expressed interest in modelling individual household's consumption of

organic foods with demographical variables as independent factors such as income,

geographic location, age, etc. This research focused on this area as one way of

understanding the consumers' WTP for organic products, most especially the

determining factors behind consumers' behaviour in organic coffee consumption.

In a related development, Didier and Lucie (2008) measured consumer's WTP for

organic and fair trade chocolates. It was mainly aimed at measuring consumer

preferences and WTP for organic and/or Fair Trade Labels. Two methods of data

collection were used; experimental auction and survey. An experiment was used in

order to measure the actual consumer WTP as the method creates a real bidding set up

that reduces any social desirability bias (Noussair et al., 2004). Specifically, the Becker-

DeGroot-Marschak's (BDM) mechanism was used for data collection. On the other

hand, the survey was used to collect information to measure elicited WTP. Selection of

the chocolates by consumers was made on two criteria: hedonic characteristics and the

price level of the product. The bidding was done in 13 sessions each containing three

stages. The first stage required that participants taste a bar of chocolate without any

information of the chocolate; the participant thus attributed a hedonic rating for the

chocolate and declared his/her WTP for the chocolate tasted. This was done in order to

evaluate consumers' preferences and their WTP based on their liking of the product

(e.g. taste). On the second stage the consumers declared their WTP for each chocolate

based on information on labels as a way of determining their WTP for the label(organic









and fair trade) independent of liking. The last stage involved tasting the chocolates

tasted from stage 1 but with comprehensive information. Two of the chocolates had

organic and fair trade labels while the other two had neither organic nor fair trade labels.

This was done in order to analyse the evolution of arbitrages between the hedonic

evaluation of chocolates and the information provided. The results of the study

specifically for stage 3 indicate that the organic and fair trade labelled chocolates

received the highest bids compared to the standard ones confirming that the information

on product attributes provided in this stage influenced consumer's WTP for the products

positively. This study also conducted a consumer typology according to the valuation of

the organic and fair trade labels. In this regard, the sample was divided into three

clusters based on their WTP for chocolates using comprehensive information provided

for the chocolates. Cluster 1 represented about 42% of the total sample; it had the least

number of women (about 63% against 71% in the total sample), it included students and

people without an occupation with average age of 35 years. Generally, the group

constituted mainly non consumers and occasional consumers of organic products. The

second cluster had 41% participants of the total sample; it had about 71% women, had

people with average age of 45 years with professions as Executives, commercials

sector workers and retired officers. Members of this group were regular consumers of

organic and fair trade products. The last group had the least participants representing

only 17% of the sample; it had about 88% women with people aged 32 years on

average without specific professions. This group constituted participants who consume

organic products moderately but fair trade products more occasionally. According to the

study, the first cluster registered the lowest WTP for the chocolates; the second cluster









registered twice as much WTP as the previous cluster while the third cluster registered

the highest WTP. These results clearly show that consumers' WTP for organic food

varies across consumers' demographical factors e.g. age, gender, education,

occupation etc. The study also compared the variation of consumers' WTP for the

chocolates across the 3 groups. The findings showed that the additional information of

the chocolates provided to the participants had varying influences over the valuation

process. It was also noted that consumers in cluster 1 were more sensitive to price than

the 'organic and fair trade label', those in group 2 were sensitive to the labels but

without any conditions while those in cluster 3 were sensitive to labels but conditioned

on taste. Lastly, a comparison test was carried out to assess the factors that motivated

the consumers in the valuation process. The results showed that group 1 were more

influenced by taste and health issues associated with the labels and not the

environmental and the social concerns associated with them, to the contrary consumers

in group 2 were mainly influenced by the environmental and the social concerns

associated with the 'organic and fair trade labels' while consumers in group 3 were

similar to those in group 2 apart from the fact that they considered taste as the main

motivation factor as well. In conclusion, according to the results nearly half of the

sample was insensitive to the 'organic and fair trade' labels. This proportion of the

sample was mainly motivated by price in its choice of the chocolates, then taste and

health related issues and lastly the environmental and social factors associated with the

chocolates. The remaining proportion was mainly motivated by the environmental and

social concerns associated with the labels as they were able to value the products to

20% to 30% of the product price. Although the environmental and social concerns were









some of the motivational factors, some consumers based their WTP on the liking of the

products. The labels just enhanced the valuation process and varied a lot across

consumers. The study therefore concluded that consumers were not ready to pay more

for organic and fair trade products and so the market for such should not be

overestimated.

Wikstr6 m (2003) measured WTP for sustainable coffee (organic and fair trade

certified) in Sweden. He also made an attempt to determine the underlying factors for

the choice of sustainable coffee. In his study he used choice experiments as a method

of collecting data and conducted data analysis using a binary probit model. The analysis

was based on the neoclassical demand theory. The study targeted 100 respondents

who were required to make a choice between a number of alternatives of coffee

provided to them in the choice experiments. In the end, the analysis had a total of 900

observations since each participant was required to make 9 choices. Results of the

study showed that there was higher WTP for the organic certified coffee as compared to

the fair trade coffee, although the monetary attribute of the coffee had a significant

impact on the consumer utility. The implication of this conclusion is that consumers

were willing to pay higher price premium for the two types of coffee; the choice was

made at a minimum cost. On the other hand, social demographic factors and the

attitude factors were significantly influential in the consumer choice of the coffee. The

regular consumers of coffee were less likely to buy organic and fair trade coffee due to

their high premiums, and consumers who were aware of environmental concerns,

health benefits and other benefits of 'sustainable' coffee were more likely to buy the

coffee than those without the knowledge. The results therefore concluded that there









was an existence of a market for both certified and fair trade certified coffee in Sweden

as consumers were willing to pay high price premiums for the products under the study.

In this light, recommendations were made to organizations in the industry to consider

lowering the prices of these products as one way of expanding the market shares for

their brands. In addition, organizations were encouraged to incorporate health benefits

of the coffee in their marketing campaigns as about 20% of the respondents based their

choice of the two types of coffee on the aspect of reduced levels of chemical

substances in the products.

The Commission for Environmental Cooperation in 1999 conducted a consumer

demand study on Mexican shade -grown coffee. Shade grown coffee is most of the

times referred to as organic because it is generally grown naturally, it does not use

heavy chemicals. The main purpose of the study was to assess the potential market for

Mexican Shade grown coffee in USA, Canada and Mexico. Data collection was done

through personal surveys and focus group discussions where individuals even

conducted taste tests of Mexican Shade grown coffee compared with other brands.

Based on the results, 22%, 42% and 50% of the people interviewed in USA, Canada

and Mexico respectively were willing to pay $1 more per pound for the coffee.

Consumers were motivated to register WTP for the coffee because of its environmental

friendliness. Quality and taste were also key factors in determining consumer choices.

As such, the study recommended that emphasis should be done on quality and taste of

the Mexican shade grown coffee in marketing campaigns.

Loureiro and Hine (2002) also assessed consumer preference and WTP for local

(Colorado grown); organic and GMO free potatoes as one way of discovering their









potential niche markets. The study also focused on determining factors (socio-

demographic factors and quality characteristics) of consumer response to a particular

attribute of the products in question. Revealed consumer preference data were

collected in a survey and the analysis was done using a multiple bounded probit model.

The results showed that the locally grown potatoes had the highest WTP estimate of

9.37 cents while the organic and GMO free potatoes had 6.64 cents and 5.55 cents

respectively. The social demographic variables and quality characteristics had different

effects on WTP for the three attributes. For organic and GMO free potatoes, consumers

who were sensitive to freshness and nutrition registered higher premiums; WTP was

negatively related to age. In addition, there was also a negative relationship between

WTP and the number of children per household. Even though the WTP was the highest

for locally growth potatoes, the only variable that had statistically significant positive

relation with WTP was consumers' concern for nutrition. Overall, the results indicated

that there was a potential niche market for the locally grown potatoes in Colorado.

Engel (2008) calculated consumer WTP for major organic products (wine and fruit

juice) in South Africa. He used the CVM in data collection. The study used the binary

logit model to analyze consumer decision to purchase organic food or not and the

ordered logit model to analyze the determinants of WTP for organic wine and fruit juice.

Based on the results, the significant socio-demographic variables influencing the

decision to purchase organic food were age, marital status (being married) and level of

education. Age positively influenced the decision to purchase organic food; marital

status had a negative influence over the decision to buy and level of education had a

positive influence over the decision to buy. The results from ordered logit model









demonstrated that age, language (Afrikaans), head of household and citizenship

significantly affected consumer WTP for organic products. Specifically, 'being of

younger and middle age' and head of the household had a positive effect on WTP;

language was negatively related to WTP as the majority of Afrikaans speakers have low

disposable incomes. Overall, South Africans were willing to pay bid values of $0.25,

$0.37 and $0.49 more for organic fruit juice compared to conventional fruit juice.

For organic wine, the significant independent variables were age (younger and

older), Afrikaans, English or language other than Afrikaans, English or Xhosa as the

home language and Christian faith. Age was positively related to WTP, Afrikaans and

home English speakers were negatively related to WTP and being Christian was

positively related to WTP.

Peterson et al. (2008) made a contribution to the study of demand for non food

organic goods. The study assessed consumers' WTP for various attributes of wool

products (gloves) made in USA and Australia. The attributes used for the research

included country of origin, environment focused (organic and pro-environment), animal

focused, and price. In their study, choice experiments were used to collect data and

data analysis was based on the conditional logit model. According to results; consumers

were willing to pay $1.20 more for a pair of USA wool glove compared to a pair of

acrylic ones and WTP for Australian wool glove was $0.25.

It was noted that the pro-environmental label was valued more than the organic

label by 14 cents. This could be attributed to the relatively low recognition of organic

clothing than food by the participating consumers. In addition, consumer preferences for









the gloves varied by socio-economic and psychographic characteristics e.g. gender,

age, income, education attainment, location and beliefs of animal rights.

In a study aimed at assessing household's WTP for "green" goods (organic cotton

sportswear), Casadesus-Masanell et al. (2009) used sales data to elicit consumer

revealed preferences and WTP. The study showed that consumers were willing to pay

significant premiums for organic cotton clothes regardless of the related costs

associated with the apparel.

Recently, Hustvedt and Bernard (2008) examined consumer WTP for three

credence attributes of organic socks made from cotton and corn. The attributes

assessed were: origin (imported, US and Texas), type, and production method

[conventional, organic and non-genetically modified (GM)]. Data were collected through

experimental auctions and were analyzed using a Tobit regression model. Bidding in the

first round was conducted without information about the credence attributes while in the

second round respondents were provided with various attribute information. The model

included demographic variables as possible factors determining WTP for the attributes.

According to the results, consumers were willing to pay the highest premium of $1.86

for organic socks, which was slightly higher than the premium for non-GM socks.

Regarding the effect of demographics, females were less willing to pay for the U.S.

fibers than men, and Hispanics were less willing to pay for organic or non GM fiber.

Among others, the study concluded that there is a potential market for organic garments

in USA, which is in line with the results of Casadesus-Masanell (2009).

Previous Studies on Consumer Preference and WTP for Non Organic Products

Methodologies used in consumer preference studies are vital to the analyses of

the current study. In this regard, this chapter highlights previous studies on consumer









WTP for non organic products. The major aim was to explore the methodologies used in

pursuing these studies, which later assisted in the selection of appropriate data

collection and analysis methods that are most suitable to the research objectives.

Umberger et al. (2002) assessed the consumer preference and WTP for domestic

corn-fed beef against international grass-fed beef. The study targeted two locations in

the USA, Chicago and San Francisco. Data were collected through panel taste testing

and experimental auctions (fourth-price Vickrey Auction). The taste test was conducted

to elicit consumer's preference over beef flavour between the two types of beef and the

auction was conducted to elicit consumer's WTP for the preferred steak. The collected

data were analyzed using two types of models. Data on preference were analyzed using

a multinomial logit model based on random utility theory. The dependent variable was

'flavour preference' (0 for consumers preferring corn fed beef over grass-fed, 1 for those

indifferent and 2 for those preferring grass-fed beef over the corn-fed beef), the

independent variables included: location, age, gender, ethnic, income, education, family

size and other factors representing characteristics of the consumer. In the analysis of

data on WTP, an OLS regression model was used. The dependent variable was 'bid

difference' (the difference between the bid prices between the beef types) and the

dependent variables were the same as that used in the logit model. The study found

that on average, consumers preferred the domestic steak on all sensory qualities and

they were willing to pay a 30.6% premium for corn-fed beef. To be specific, about 62%

of the participants were willing to pay an average premium of $1.61 more per pound for

the corn-fed beef, 23% were willing to pay a premium of $1.36 more per pound for the

grass-fed beef and only 15% of the consumers were indifferent. These results show









that there exist respective niche markets for the two types (corn-fed and grass-fed) of

beef as well as beef with country of origin labelling. The demographic factors such as

age, ethnicity, beef knowledge and quality grade were seen to have some influence

over the flavour preference. However, these factors did not have any influence over the

bid difference. It was thus difficult to predict the type of consumers willing to pay for the

product they prefer.

Mabiso (2005) estimated the WTP for Country of Origin Labelling (COOL) for

American fresh apples and tomatoes and established the major determining factors for

the WTP. In the study, experimental auctions (Vickrey-fifth bid sealed price) were used

for data collection. The study used the double hurdle probit model for analysis of the

data. The findings indicated that 99% of the consumers were willing to pay $0.49/lb

more for apples labelled 'USA grown', 72% of the participants were willing to pay

$0.48/lb more for tomatoes labelled 'USA grown' as compared to identical ones without

the labels. The demographic and psychographic variables such as food quality

perceptions and consumers' location had significant relationships with WTP.

Numerous studies on valuation of goods and services have been done in the

health sector as well. Dong-Churl (2000) measured WTP for pharmacists' services

directed toward reducing the risk of medication-related problems. The study also

attempted to determine factors that have a significant influence on WTP. Like most of

the studies highlighted above, Dong collected data using the CVM. Data analysis was

based on logistic regression and semi log regression models. Overall, there was WTP

for the pharmacists' services. For instance, the mean WTP for pharmacy services

ranged from $4.02 to $5.48 per prescription. Of the factors used in the regressions,









magnitude of risk reduction had an influence on WTP; income was positively related to

WTP although it was not statistically significance.

Werner et al. (2002) examined primary caregivers' WTP for medication for

Alzheimer's diseases. Data were collected using two methods. The first one was

through experimental auctions and the second approach used a questionnaire with

open ended questions. The data collected were analyzed using an econometric model

with WTP as the dependent variable and psychological factors, social-demographic

factors and other characteristic variables as independent variables. According to the

results, the mean WTP for the treatment was $188.45 (using open ended questions)

and some independent variables such as income, age, cognitive status and periods of

caring for the sick had a significant impact the WTP.

Aulong and Rinaudo (2008) assessed population WTP for ground water protection

in the Upper Rhine Valley. The valuation was elicited based on two scenarios (restoring

drinking water quality and eliminating all traces of polluting substances). They used the

standard contingent valuation method and analyzed the data using three models. The

logit model was used to assess whether or not participants were willing to pay for

proposed scenario; a linear regression which excluded some of the variables (protest

answers) was used to elicit stated WTP while the Tobit model was used to capture the

same but included the variables capturing protest answers. Based on the results, 62%

of the respondents were willing to pay for the first scenario at a mean WTP of $59.6 per

household while 52% were willing to pay for the second scenario at a mean WTP of

$107.72 per household. It was also noted that some of the independent variables used

in the three models were statistically significant. For the linear logic model such









variables included 'realism of the scenario,' 'number of children in the household,'

'income' and the 'number of known polluting substances.' For the other models (the

linear regression and the Tobit model), the statistically significant variables were

'knowledge of the water bill,' 'income,' 'concerns about groundwater pollution,' 'leisure'

and 'use and non use values of groundwater.'

Gao and Schroeder (2009) investigated the effects of additional beef steak

attributes on consumer WTP in two different US markets. They used Choice Experiment

(CE) and analyzed the data using Random parameter logit models. The survey had four

questionnaires; two of the questionnaires were aimed at collecting data to test the effect

of additional attributes when cue attributes exists while the others tested the effect of

additional attributes when no cue attributes are available. Results from both sets of

questionnaires showed consistent results of the effect of additional attribute information

on consumer WTP. Based on the results, response of additional attribute information

was twofold. In some instances, WTP for the most important attributes decreased when

consumers were provided with additional attribute information whilst in certain instances

it was the opposite, WTP was positively related with additional information of the most

important attributes of the study. It was thus concluded that the varying WTP for the

attributes was conditioned on the relationship existing between the attributes and the

additional ones.

Methods of Eliciting Consumer WTP

In reviewing literature, there exist a number of techniques that are used in

estimation of WTP for products. This includes the CVM, Experimental Auctions,

Hedonic Pricing models, Conjoint Analysis, and others. Amongst these, the CVM has

proved to be the most widely used method in many market research studies. However,









the Experimental Auctions are more reliable and are currently being used in market

research the most. Below is a detailed summary of some of the methods.

The Contingent Valuation Method (CVM)

The CVM is one of the "stated preference" models used to elicit consumer

preferences and WTP for products. It is used to attach monetary value to products most

especially when their markets do not exist. The valuation is based on the change in

attributes of a particular product such as prices and quality. Consumer preference for

the product is therefore assessed based on the monetary value attached to it. The

valuation process is also extended to services. The method thus creates a hypothetical

market situation for those goods. Through the valuation process, the data collected

forms what consumers are willing to pay for a particular product.

In CVM, the valuation of the products is done using a questionnaire which is

administered through mail, telephone and face-to-face interviews. The survey

instrument used offers the respondents an opportunity to make an economic decision

on the non or market goods (Rahmatian, 2005). The valuation process is therefore

contingent upon the simulated market presented to the respondents. Product valuation

is done through bidding. The bidding takes different formats e.g. open ended questions,

bidding game, payment card, dichotomous choice questions and randomized card

sorting. In open ended questions, participants disclose their WTP without the use of a

starting bid level. The bidding game uses a number of discrete choice questions but one

open ended WTP question and also provides a starting bid value. In payment cards,

visual aids bearing product monetary values for attribute changes are used while the in

dichotomous choice questions researchers use yes or no questions on whether









consumers are willing to pay for a particular product at a certain price. In addition, this

format uses additional follow up question specifying lower bid levels.

Carson et al. (1994) documented the advantages of CVM. According to him, CVM

is a flexible tool for product valuation, it is easy to apply and cost effective. Aulong et al.

(2008), Werner et al. (2002), Dong-Churl (2000),Rodriguez et al. (2007), Engel (2008),

Millock et al (2002) and others used the method in their respective studies as

highlighted in the literature review above.

However, being hypothetical methods of eliciting consumers' willingness to pay,

the CVM has got a number of flaws. The major one is of response bias, which mainly

emanates from the use of open ended questions. Mitchell et al. (1989) reported biases

on the use of open ended questions mainly due to high non response rates. In addition,

participants overstate their preferences, which most of the times is different from their

actual purchase behaviour. Many consumers state high WTP but are less willing to pay

the exact amount during actual purchases.

Nevertheless, based on Rahmatian (2005), CVM are more reliable when one is

using test-retest (conducting CVM on a different sample of the same population

overtime) or when convergent validity checks are employed. This compares results

obtained from CVM with other methods e.g. CE, travel cost or hedonic. These

precautionary measures ensure that results based on CVM are more reliable even with

the presence of hypothetical biases.

After assessing the reliability of the CVM, the National Oceanic Aviation

Administration (NOAA) (1993) recommended that the CVM should incorporate the

following:










"1.The use of face-to-face interviews,
2. The use of WTP as opposed to WTA,
3. Provision of comprehensive information about a product to be valued,
4. The need to remind consumers of the budget constraints they are subjected to
in the course of the valuation process,
5. Inform the participants the possible substitutes of the product under valuation,
6. Need to use probing to ensure that respondents understand issues being
asked."

This study adopted these recommendations during its implementation.

The Choice Experiment (CE)

Like the CVM, CE is also a stated preference method based upon the Lancaster's

utility model of consumer economics (Lancaster, 1966). In this method, individuals are

given a hypothetical setting and asked to choose their preferred alternative among

several alternatives in a choice set and they are usually asked to perform a sequence of

such choices.

Each alternative is described by a number of attributes or characteristics. A

monetary value is included as one of the attributes, along with other attributes of

importance, when describing the profile of the alternative. Thus, when individuals make

their choice, they implicitly make trade-offs between the levels of the attributes in the

different alternatives presented in a choice set.

According to Alpizar et al. (2001), there are four major steps that need to be

followed when designing Choice Experiments and these are:

"1. Definition of attributes, attribute levels and customisation,
2. Experimental design,
3. Experimental context and questionnaire development, and
4. Choice of sample and sampling strategy."

Within the stated preference models, the CE is currently being widely used.

Among others this is highly due to the fact that their use reduces some of the potential









biases created by the use of CVM. More information is elicited from each respondent

compared to CVM and it allows for the possibility of testing for internal consistency.

Based on our literature review, Gao and Schroder (2009), Peterson et al. (2008)

and Wikstrom (2003) used CE in their respective studies.

Experimental Auction

Unlike the other two models, Experimental Auction is a revealed preference

model. It is used to capture revealed (actual) WTP for a particular attribute of a product

as it creates a real market auction bidding environment. The Vickrey sealed-bid,

second-price auction is the most commonly used experimental auction model. This

requires participants to submit written bids of a particular product in a real auction

environment (Friedman et al., 1994). In a sealed bid, second price auction, bids are

ranked from highest to lowest. The highest bidder wins the bid and purchases the

product at the second highest price. Unlike the CVM and the CE, it is advantageous in

the sense that it is designed to reveal true preferences; the use of real money for

bidding in additional to other factors like repetitive bidding ensures reliability of results

from this methodology. The method also reports less bias by non responses (Fox et al.,

1995). However, the major flaw of the Experimental Auction is that it is very expensive

to implement. Werner et al (2002), Mabiso (2005), Umberger et al. (2002), Hustvedt and

Bernard (2008), Didier and Lucie (2008) and others used Experimental Auctions in

order to estimate WTP in their studies (details in literature review above).

It is very clear according to published studies that revealed preference models are

preferred to stated preference models because of their reliability. However, due to the

target (sample) of this study and resource constraints the CVM and the CE were used

as methods of data collection. CVM is flexible and easy to use and most importantly









they are easier to implement in developing countries than industrialized countries

(Whittington, 1998). The main purpose of multiple valuation techniques is for convergent

validity of the estimates. If the results from the two models (CVM and CE) converge, this

is likely to give policy makers the confidence to reliably base their decisions on the

results.

The analysis of data collected from CVM will be based on an Ordinary Least

Squares Regression Model while that of CE will be based on a Conditional Logistic

Model.









CHAPTER 3
RESEARCH METHODS AND DATA

Theoretical Model

The focus of the current study is on assessing consumers' WTP for Malawi

organic coffee. Estimation of consumers' preference and WTP of products is based on

consumer cognitive demand theories and random utility models. The random utility is

based on the assumption that individual utility is a function of observable product

attributes, individual characteristics and an unobservable random component. Such as

U ,J = ',JX + ............................. ................................ ......... (3 -1)

Where Xi is a row vector of independent variables. These variables could

represent characteristics specific to the individual and also attributes of the choices

subjected to the consumer; 3ij is a vector of estimated coefficients; and Eij is an error

term. It is assumed that the error term is independently and identically distributed with

certain distribution (Greene, 1998). Based on this framework, it is also assumed that a

consumer chooses the attribute combination or a product that gives him or her the

maximum utility. The CVM and the CE that the study employed are in tandem with

random utility theory and Lancaster's theory of utility maximization. According to

Thurstone (1927), Random Utility Theory specifically explains the way a consumer

makes his choices out of a set of choices provided to him. On the other hand, the

Lancaster Theory asserts to disaggregate utilities of products into utilities derived from

respective attributes (Lancaster, 1966). In the CVM, the error term is usually assumed

to be normally distributed while in the CE, it is normally assumed to follow an extreme

maxima value distribution which will result in Conditional Logit Models (Hoffman and

Duncan, 1988).









Model Estimation

As already mentioned above, the current study used two models based on the

type of data collection procedure. Ordinary Least Squares Model was used to estimate

consumer WTP using data from CVM while the Conditional Logistic Model was used to

analyze data collected through the CE method.

Model Estimation under CVM

Within the framework of Random Utility Theory, WTP is estimated using Ordinary

Least Squares Regression Models among other models. The theory's assumption of

independence and normal distribution of the error terms is thus in line with one of the

Gauss Markov assumptions of the OLS which also asserts that the error term be

normally distributed. In the estimation of WTP for beef from USA and Argentina,

Umberger et al. (2002) used an OLS model. She also investigated the impact of socio-

demographic variables on consumer taste preferences and WTP. The current study

used a similar model as below:

Y, = B'jX + .................................... ......................... ..... (3-2)

Where:

Y = the bid price premium (the difference between the maximum price an

individual is willing to pay for Malawi Organic Coffee and the average price of

Malawi Conventional coffee)

Xi represents the regressors as follows:

Actprice = the actual price consumers pay for coffee at domestic market

Demographic Variables:

Female= gender (1= Female)









Age= (1= 15-24 years, 2=25-39 years, 3=40-59 and 4=60 years and older)

Education = (1=at least degree education, 2= at least college certification but no

degree, 3= Secondary School qualification (M.S.C.E) and 4=Elementary

School certificate)

Income= (1=MK66, 000 and below, 2= MK67,000-MK321,000 and 3=MK322,000

and above)

Denomination=(l =Presbyterian, 2= Catholics, 3=Anglican, 4=Pentecostals, 5=

Sevethday Adventist, 6=Muslims and 7= Other denominations)

Attitudinal Variables:

OrgFertr = if one believes that organic coffee is grown without the use of fertilizers

(1=strongly agree, 2=agree, 3=uncertain, 4=disagree,5=strongly disagree)

Orgchm = if one believes that organic coffee is grown without the use of pesticides

or chemicals (1=strongly agree, 2=agree, 3=uncertain, 4=disagree,

5=strongly disagree)

Orgnat = if one believes that by buying organic products you are supporting

natural and healthiest way to grow crops (1=strongly agree, 2=agree,

3=uncertain, 4=disagree,5=strongly disagree)

Orgsup = if one believes that by buying organic coffee you are supporting farmers

(1=strongly agree, 2=agree, 3=uncertain, 4=disagree,5=strongly disagree)

Orgris = if one believes that by drinking organic coffee there is lower risk of

ingesting chemicals (1=strongly agree, 2=agree,

3=uncertain,4=disagree,5=strongly disagree)

Orgnut = if one believes that organically grown food may offer more of some









nutrients than conventional counterparts (1=strongly agree, 2=agree,

3=uncertain, 4=disagree, 5=strongly disagree)

Orgtas = if one believes that organically grown food have better taste than

conventional counterparts (1=strongly agree, 2=agree, 3=uncertain,

4=disagree, 5=strongly disagree)

Model 3-2 is the original model with 13 independent variables. Out of the 13

variables, 12 are discrete variables namely; 'female,' 'age,' 'education,' 'income,'

'denomination,' 'Orgfert,' 'Orgchm,' 'Orgnat,' 'Orgsup,' 'Orgris,' 'Orgnut,' and 'Orgtas.'

'Price' is the only continuous variable in the model. Running the model in its original

form would create problems most especially in interpretation of the estimates of the

coefficients of the polytomous variables (those taking more than two levels). This is so

because the intervals of the levels may not be standard except for the attitudinal

variables whose intervals are assumed to be standard in our study. Dummy coding of

these variables was therefore essential in order to address this problem. In dummy

coding, each of the polytomous variables apart from the attitudinal variables was made

binary, thus taking the value of either 1 or 0. Instead of having 13 independent

variables, this modification resulted in having a total of 27 independent variables in the

model. Nineteen (19) of these were dummy variables, seven were discrete (polytomous)

while only one was continuous.

Traditionally, the model with 27 regressors as explained above is normally run by

dropping one of the dummy variables per each group of the polytomous variables in

order to set it as a base group. Among others, this addresses the problem of dummy

variable trap and so makes the model easy to estimate. In this case, the intercept of the









estimated model assumes the value of all five of the base groups. Interpretation of the

estimates of the coefficients of the dummy variables in the model is thus done in

comparison to the intercept. However, it becomes so complicated to make such

interpretations in a situation whereby the base groups are more than one as one is

laboured to remember all of them when making the comparisons, among other things.

According to Jauregui (2007), the best approach is therefore to perform an effect coding

of the dummy variables. With the effect coding, the intercept takes the value of the

average household of the sample instead of a particular base group hence

interpretation of results becomes easier. For the sake of illustration, let's take 'income'

variable, which has three levels according to model 3-2. In effect coding the first stage

requires that the variable be decomposed into three dummy variables representing each

of the three levels and this is represented in equation 3-3 below.

Y = ao + D +a2D2 a3D3 ............. ........ ................. (3-3)

Where y = dependent variable, Di are the dummy variables representing the three

income categories and ai are estimates of the coefficients of the dummy variables.

The process also requires that the original variable 'income' be weighted in such a

way that the sum of their respective coefficients is equal to zero at the mean of the

dummy variables as in Equation 3-4:

a D + a D2 a D = 0 ................................. .... .................... (3-4)

where D, represent the mean of the respective income dummy variables

This implies that

a3 = (D, /D 2 )-a2(D2 /D3 )........ .................... ...... .............. (3-5)

If Equation 3-5 is inserted in Equation 3-3, the following is yielded:









Y = ac + aD, +a2D -(a, D D3 +a2 D2 ID3 )D3 ............. .............. (3-6)

This eventually is transformed into:

Y = ao + a,(D -D3 *D ID3)+a2(D2 -D3 *D2D3 )...............................(3-7)

let (D1 -D3 *D1 /D3 ) = Dlncomei representing the restricted variable

for the first dummy of 'income,'

and (D2 -D3 *D2 ID3) = Dlncome2 representing the restricted variable

for the second dummy of 'income.'

Model 3-7 is what is eventually run as the model with restricted variables. The

dummy variable of the third income category (D3) is what has been dropped from the

original model to avoid dummy variable trap. When all the discrete variables are at their

means, the restricted variables which in our illustration are 'Dlncomel' and 'Dlncome2

equal zero (0). This is what transforms the intercept to represent the average household

of the sample.

The process was replicated to the other polytomous variables and we finally run a

model containing all variables as outlined in Model 3-2. Instead of having thirteen (13)

variables as in Model 3-2, we eventually had 23 (27 4 = 23 variables) in our final

model. The variables add up to 23 because we factored out 4 variables, each from the

set of the polytomous variables to avoid dummy variable trap. It should therefore be

noted that (Female) was not restricted because it is not a polytomous variable;

interpretation of its coefficient will therefore be in comparison to its base group (Male)

and not the average household. The attitudinal variables were not restricted as well.

Interpretation of these variables will therefore be done similarly to continuous variables









based on the assumption that their respective intervals (1 for strongly agree,..., 5 for

strongly disagree) are constant as explained above.

Model Estimation under CE

The second model that was estimated in our study is the Conditional Logistic

Model. The model is based on the Lancaster's theory of utility maximization. Based on

this theory, a consumer chooses the product that maximizes his or her utility. The

probability that an alternative j is chosen among J alternatives is:

Prob(U,, > UJ) for all other k j (Greene, 1997).

Assuming U, = V + ,

The random component Eij is independent and identically distributed following an

extreme maxima value distribution. In this case, the probability of an alternative j can be

chosen as below:


Prob[Yi=choice j]= e
Ye

Where Yi is a random variable that indicates the choice made by the ith individual

and is equal to 0,1,..., J depending on consumer preference; V#i is the utility an

individual obtained and is determined by individual specific characteristics and product

attributes. In our current study, consumer utility is assumed to be a function of price and

the method of production of a product such as:

= /Po+APx +P2Xj2

Where Xjl is the price of coffee, Xj2 is the dummy of organic coffee.









CHAPTER 4
DATA COLLECTION

Data collection was conducted using two stated preference methods, namely:

CVM and the CE. The multiple valuation techniques were used mainly for convergent

validity of the estimates from the two models. In order to conform to the

recommendations of NOAA (1993), face-to-face household interviews were conducted

in data collection with an aim of reducing hypothetical bias when one is using CVM.

According to Cochran (1977), a formula for determining a sample size expressed

as a percentage is;

(t2 )(P)(q)
n = (t)( )() ................... ............................................................. (4-1)
(j2)

where t2 = the standard deviation score that represents the probability level of a

variable of falling within a confidence interval when the variable is normally

distributed

(p)(q) = Variance

j2= confidence interval

The following are the results after incorporating our data variables into the formula:

(1.962)(.5)(.5)
n7=
(.052)

n=384

The probability level and confidence interval of 1.96 and 0.05 respectively were used as

these are the commonly used estimates and normally accord estimation process

efficiency. The variables making up the variance represent the proportion of consumers

and non consumers of coffee according to our study. Since it was difficult to source the









specific estimates for these proportions, Czaja and Jonny (1995) recommends that a

50% proportion for each is ideal. We thus needed to collect a sample size of about 384

to represent the target population of our study. However, due to budgetary constraints,

time and other factors, the study managed to collect a sample size of 129. This is still a

significant figure considering that it is still a large sample and it was randomly collected.

The survey therefore targeted 129 participants in the three main cities of Malawi.

These are: Blantyre, Lilongwe and Mzuzu in the Southern, Central and Northern regions

respectively. The sample was randomly selected using the Systematic Random Sample

Method in order to reduce response biases. This was based on a sampling frame

collected from the Malawi National Statistics Office (NSO) of the 2008 Census

Household list. According to the methodology used by NSO in conducting surveys, the

country is divided into clusters and then further broken into Enumeration Areas (EAs).

Maps of these clusters including their respective EAs were used to locate the

households that were selected in the sample. Selection of households was done in such

a way that diverse income categories of the Malawi population be represented in our

sample. This was done by first dividing our target population into three major clusters

representing the low, middle and high income groups. For instance, households of

people with high income levels were selected from the low density clusters while those

of people with low incomes were selected from the high density clusters. Households of

participants of the middle class were selected from the clusters in the middle of the two.

It should however be noted that collection of data on household incomes was not limited

to the three main income groups, ten (10) categories of household income data were

collected guided by the three main categories (Refer Appendices A and B). This









information was used for descriptive statistics. On the other hand, the ten categories

were compressed into three income brackets that were eventually used for estimation of

the OLS.

The survey targeted people of at least eighteen (18) years of age although the

minimum age captured in the survey instrument used as per Appendices A and B is

fifteen (15). This was done so, because the study adopted the age categories normally

used in surveys conducted in Malawi. It should be emphasized that only those with a

minimum age of 18 were targeted in the survey.

Apart from the age, there were no other restrictions in terms of the characteristics

of an individual so long as they were able to speak either the local language or English.

The survey targeted both consumers and non consumers of coffee. The non consumers

were interviewed most especially to get their perceptions on organic coffee. Table 4-1

presents some of the demographic structure of our sample in relation to the population

of Malawi to show the sample's representativeness to the population. This will be in

terms of gender, age and religious affiliation.

Table 4-1. Sample representativeness in terms of the demographic structure of the
population of Malawi (gender, age and religion)
Demographic Category Sample (%) Population (%)
Variable
Gender Male 32% 49%
Female 68% 51%
Denomination Christians 98% 80%
Muslims 2% 13%
Age 15-64 years 98% 52%
65 years and older 2% 3%
Source: For Population figures Gender Census Report 2008, and Denomination and Age CIA World
Fact book. For Sample figures Author's Analysis









Our sample estimates do not converge with the population estimates although

they both portray a similar pattern. The distribution of gender is skewed towards

females for both the survey sample and the population. Sixty-eight percent (68%) of our

sample are women while 51% of the population are females. This could be highly

attributed to the fact that the interviews were mostly conducted during working hours

when most men were on duty. In addition, most men insisted that their wives be

interviewed in situations were both were available to make sure that there was a good

rapport since all interviewers were also females. In terms of denomination, both

structures show that the country is dominated by Christians. About 98% of our sample

were Christians while 80% of the Malawi population are Christians. Among others, this

could be due to the fact that the Muslim population is not very significant in the cities

(Blantyre, Lilongwe and Mzuzu) that the survey targeted. The dominant denominations

were Presbyterians and Catholics representing 36% and 20% of the sample

respectively. This is in line with CIA World Fact book which states that 80% of

Christians in Malawi constitute the major denominations of Presbyterians and Catholics

with the former taking a higher percentage. According to age distribution, our target

population were people of at least 18 years. Since the comparison is done with

categories provided by the CIA World Fact book; this leaves greater proportion of our

sample in the age category (15-64 years). However, on the elderly population (65 years

and older), the estimates are close to each other. About 2% of our sample were

participants belonging to this age group which is close to 3% of the population of

Malawi. Based on this comparison, our sample qualifies to be representative of the

Malawi population.









Design of CVM

The CVM is a survey instrument used to obtain preferences of respondents in

monetary values for changes in the price or quality of a particular good or services

(Engel, 2008). The current study used a well structured questionnaire to collect

information from participating consumers. In order to save on time both CVM questions

and CE questions were captured on the same instrument (Refer Appendices A and B

for detailed survey instruments). The survey instruments had five main parts, in all parts

discrete questions were asked except for the one that captured the WTP for organic

coffee. The first part had questions aimed at collecting information related to consumer

consumption pattern of coffee. Participants were asked whether they drink coffee or not,

in what quantities they take the coffee (how many cups per day) and when they

normally take the coffee (breakfast, lunch, supper or in between meals).

The second part contained questions that were aimed at collecting participants'

information on their perception of organic products or organic agriculture in general. In

this section a number of sentiments related to organic products and production were

read out and the consumer was required to either strongly agree or just agree or

disagree or strongly disagree. There was also an option of 'uncertain' for those who

were not sure.

The third part was an open-ended question that asked consumers to provide their

WTP for organic coffee. Before giving out the WTP estimate, a definition of organic

products/production was read out to participants to make sure they have the knowledge

of the coffee to be valued. Two packets of coffee were then shown to the participants;

one was conventional while the other was organic by assumption (this was done

because Malawi does not grow organic coffee yet). The packaging was done in such a









way that all coffee attributes were held constant i.e. brand, quantity, taste, country of

origin, etc., apart from type of production and price. Participants were then asked to

assume they were in a grocery shop to make a coffee purchase. They were then asked

to bid for the organic coffee through the 'bidding game' format of the CVM that the

survey adopted. The bench mark price that was used in the bidding process was the

average market price for conventional coffee in Malawi which was MK 652 per 250g.

The average price was calculated using retail coffee prices collected from the

Consumer Association of Malawi. The survey did not use the actual price paid for coffee

on the domestic market as reported by the survey participants, as the bench mark price

because there was a possibility that some of the coffee purchased could be organic

though imported. The WTP question that was asked by participants during the bidding

session was; "if the conventional coffee cost MK652 per 250 grams, how willing are you

to pay for the organic coffee?" Consumers' preference for organic coffee was thus

revealed through the calculated price premium for organic coffee (Expressed WTP for

organic coffee minus average market price for conventional coffee (MK 652 per 250 g).

If the price premium was positive (if a participant was willing to pay more than MK 652

for organic coffee), the implication was that he preferred organic to conventional and if

the price premium for a participant was negative (he/she was willing to pay less than

MK 652 for organic coffee), the implication was he preferred conventional coffee to

organic. Consumers were also required to give out the reasons to back up their

willingness to pay for the organic coffee. These reasons acted as their motivation for

their preference for coffee. After giving out the reasons, the participating consumers









were asked to provide information of the actual price they pay for coffee on the

domestic market.

The fourth category of the questionnaire had the CE questions whose details will

be elaborated in the subsequent section of CE design. The last part had discrete

questions aimed at capturing socio-demographic information of the participants. These

included age, marital status, income, level of education, number of children under 18

years staying in their household, occupation (whether they work with an environmental

related organisation) and denomination.

Lastly consumers were asked to choose an approach that would contribute

towards the production of organic coffee in Malawi from the three approaches given

(leaving every cost of production to producers, through government subsidies, or

through consumer taxes), if one chose through taxes he/she was required to estimate

the rate of tax he would desire to contribute.

Design of the CE

As already highlighted, the fourth part of the questionnaire had questions of the

CE. The survey had two types of questionnaires (Versions A and B), in which the order

of the products in a choice set, and the order of the choice sets were different to avoid

certain types of order effects. Each participant was required to answer one version of

the questionnaires.

Design of the CE was based on the fractional factorial design in SAS to maximize

the D-efficiency. In the choice experiment, each respondent was asked to choose

between 'Organic Coffee' and 'Conventional Coffee' at corresponding price level. In

addition, there was a third option for 'None' to cater for those who neither preferred any

type of coffee. Due to the fact that organic products are relatively new to Malawian









consumers, Organic Coffee with lower price was not treated as one dominant choice.

Therefore, the choice experiment included some sets of choice options with Organic

Coffee having prices lower than Conventional Coffee counterpart. The final choice

experiment composed of 13 sets of choice options with the D-efficiency of 63%. The

lower D-efficiency resulted from the use of 7 level real coffee prices in the CE design.

The prices used in the CE were prices for the major brands of coffee for the past four

months (May to August 2009) in Malawi. Less price level may increase the D-efficiency

of the CE design, but the real prices we used make the choice scenario facing

respondents more realistic. In addition, the learning efforts of the respondents were

reduced, which may be more important than small improvement in the statistical

efficiency. In order to eliminate the potential order effect, the order of the Organic and

Conventional Coffee were changed after reaching a certain choice set in the choice

experiment. For instance, the order was changed after reaching the 7th choice set under

Version A and it changed after the 6th choice set under Version B. Refer Figure 4-1 for

an example of a choice set in the CE. An example of the combination and ordering of

prices is as in Table 4-2.








Please choose a 250 grams packet of coffee as you are shopping in the market, or
choose None option if you are not satisfied with both coffees.
Organic Conventional

Coffee Coffee None


MK720/ 250 g MK660/250 g

13 13 13


Figure 4-1. Choice set in choice experiment



Table 4-2. Price levels for coffee in MK/250 grams (Version B)
Organic coffee Conventional coffee
795 729
795 720
729 795
729 485
720 699
720 660
Conventional coffee Organic coffee
795 699
699 699
720 660
485 485
475 485
729 475
475 475


I Pllllllllle









CHAPTER 5
RESULTS OF THE EMPIRICAL ANALYSIS

This chapter presents empirical analyzes of our data. The presentation will be

twofold; the first section will highlight results based on CVM and the second one will

include those from the CE methodology. In each respective section, descriptive

statistics will first be presented before results from the models.

It should be noted that amongst the 129 individuals that were interviewed in the

survey, nobody refused to divulge any information that was required. As such, there

were no incomplete questionnaires hence 100% response rate. This could be due to the

use of face-to-face interviews that gave enough room for probing, clarification of

questions, and others. One refusal was encountered but this was addressed by

selecting the appropriate alternative household using the random sampling method that

was employed under the study. As already explained under the section of data

collection, the survey collected five categories of information from the respondents, the

next section outlines descriptive statistics of our sample.

Descriptive Statistics

This section gives out the major summary of descriptive statistics of the sample

(Refer to Table 5-1 for details).

The survey interviewed a total of 129 participants in the three major cities of the

country. About 31.78% of the respondents were from the city of Blantyre while Lilongwe

and Mzuzu cities had the same percentage of about 34.11% of the sample.

Out of the 129 respondents, 31.78% were male, while the majority about 68.22%,

were female. This could be attributed highly to the fact that the population distribution of

Malawi is skewed towards women and secondly the interviews were conducted during









working hours when most men were at work. This is based on the fact that most women

in developing countries are unemployed due to low levels of literacy when compared to

men among other things.

Table 5-1. Summary for descriptive statistics
No. Name of Variable category No. of Percentage of
variable participants sample


1. Gender


2. Marital Status






3. Children under
18 years old
(multiple
answer)






4. Age


Male
Female
Married
Divorced
Single
Other (divorced
and widowed)
Under 2 years




2 to 5 years
6 to 12 years
13 to 18 years
None
15 to 19 years
20 to 24 years
25 to 29 years
30 to 34 years
35 to 39 years
40 to 44 years
45 to 49 years
50 to 54 years
55 to 59 years
60 to 64 years
65 years and older


41
88
68
11
43
7


31.78%
68.22%
52.71%
8.53%
33.33%
5.43%


21.71%


32.56%
43.41%
63.57%
18.6%
17.83
18.6%
15.5%
14.73%
7.75%
7.75%
3.1%
5.43%
4.65%
3.1%
1.55%









Table 5-1. Continued
No. Name of
variable
5. Education


I


No.
participants
2


of Percentage of
sample
1.55%


2.33%


Variable category

Completed post-
graduate degree
Completed
university
undergraduate
degree
Attended university
undergraduate
Completed college
degree
Completed college
diploma
Attended some
college
Some post
secondary
technical school
Completed high
school certificate
(e.g. MSCE)
Attended some
high school (e.g.
MSCE)
Completed
elementary/primary
school
Attended some
elementary/primary
school


22.48%


22.48%


8.53%


5.43%


2.33%

3.1%

14.73%

11.63%

5.43%









Table 5-1. Continued
No. Name of Variable category No. of Percentage of
variable participants sample


6. Income

























7. Occupation(if
one works
with an
environmental
related
organization)

8. Denomination


MK 15,000
below
MK16,000 to
66,000
MK67,000
MK117,000
MK118,000
MK168,000
MK169,000
MK219,000
MK220,000
MK270,000
MK271,000 to
321,000
MK322,000
MK423,000
MK424,000
MK525,000
MK526,000
above
Yes







No
Presbyterians
Catholics
Anglicans
Pentecostals
Seventh Day
Adventist


and 20


15.5%


MK 44

to 19


to 16


to 6


to 2


MK 2


34.11%

14.73%

12.4%

4.65%


1.55%

1.55%

1.55%

5.43%

8.53%

12.4%


to 2

to 7


and 11


113
47
26
2
23
14


87.6%
36.43%
20.16%
1.55%
17.83%
10.85%









Table 5-1. Continued
No. Name of Variable category No. of Percentage of
variable participants sample


9. City



10. Drink coffee


11. Frequency of
coffee
consumption


12. Drinking
(multiple)


Muslims
Other
denominations
Blantyre
Lilongwe
Mzuzu
Yes


3
14


2.33%
10.85%


41
44
44
110


One cup a day


Two cups a day
Three to five cups a
day
Six to ten cups
More than ten cups
a day
Other
time Breakfast


31.78%
34.11%
34.11%
85.27%


14.73%


37.27%


44.55%
5.45%


14
100


Lunch
Dinner
In between meals


12.73%
90.9%


5.45%
8.18%
56.36%


As already alluded to, the survey targeted people of at least 18 years of age.

According to our sample the age group that registered the largest number of

respondents was the '20 to 24 years' representing 18.6% of the sample, then the '15 to

19 years,' '25 to 29 years', '30 to 34 years' representing 17.83%, 15.5% and 14.73%,









respectively of our sample. The 65 years and older age category registered the least

number of respondents at about 1.55% of the sample.

In terms of education, our target group included individuals with at least an

Elementary or Primary School qualification. According to our sample, the majority were

individuals with the highest Secondary School qualification (M.S.C.E) and those who

attended Secondary School but have Junior Certificates of Education qualification

(J.C.E) both representing 22.48% of the total respondents. About 14.73% of the sample

completed their College Diploma while 11.63% have College Certificates. The category

of individuals with post graduate school qualification had the least number of

respondents representing just 1.6% of the total sample.

We had 10 income categories in the survey with 'MK15, 000 (=US$100) and

below' as the minimum and 'MK 525,000 (=US$3500) and above' as the maximum

group of monthly net earnings. The income category with the highest number of

respondents was the 'MK16,000 (=US$106.67) to MK66,000 (=US$440)' representing

34.11% of the sample followed by the minimum income group 'MK15,000 (=US$100)

and below' representing 15.5% of our sample then the 'MK67,000 (=US$446.67) to

MK117,000 (=US$780)' followed by the 'MK118,000 (=US$786.67) to MK168,000

(=US$1,120)' representing 14.73% and 12.4% of the total sample respectively. The

'MK271, 000 (=US$1,806) to MK 321,000 (=US$2,140)' and the 'MK 322,000

(=US$2,146.67) to MK 423,000 (=US$2,820)' had the least number of respondents both

representing 1.55% of the total sample.

Malawi has a variety of denominations and quite a number of them were

represented in our sample. The proportion of Presbyterians were the highest in all the









cities representing about 36.43% of the total sample followed by Catholics who

represented 20.16% of the sample. The least were Anglicans with a representation of

only 1.55% of the sample.

Based on the sample only 12.4% of the total respondents work with organizations

dealing with organic farming, food safety and environmental related against 87.6% who

are not associated with such type of organizations.

In terms of coffee consumption pattern which is very crucial in this study, the

majority of the respondents drink coffee representing 85.27% of the total sample against

only 14.73% who do not drink coffee at all. Of the majority, most people take two cups

of coffee a day representing 44.55% of the coffee consumers and mostly at breakfast

and in the evening. About 12.73% of the coffee consumers do not drink coffee often,

just once in a while. No one in the sample takes more than five cups a day.

Willingness to Pay for Organic Coffee

Based on the CVM, about 40% of the sample were willing to pay a high price

premium for organic coffee against 57% that were not willing to pay high price

premiums for organic coffee. About 3% were indifferent. (Refer to Figure 5-1). Based on

our results, the 40% were willing to pay an average price of MK 816.75 per 250 g of

organic coffee representing a price premium of MK164.75 per 250 g which indicates a

25% price premium over the average market price for conventional coffee of MK652 per

25g (Refer to Figure 5-2). On the other hand, the 40% reported that they actually paid

an average price of MK539.87 per 250 g of coffee (either conventional or organic) found

on the domestic market. This demonstrates that these consumers were willing to pay an

extra MK 276.89 for 250 g of organic coffee representing a 51% price increase for

organic coffee on top of the reported actual price for coffee. (Refer Table 5-2).










Frequency Distribution of WTP
o
















-I -



Figure 5-1. Frequency distribution of consumer WTP for organic coffee


Average WTP for Organic Vs Conventional Coffee
R1 R 7SS


C)


I mean of avgpric H mean of wtp


Figure 5-2.Average market price for conventional coffee and WTP for organic coffee









Table 5-2. Expressed price for organic versus actual price paid for coffee1
Description Average price % Increase over reported
(MK/250g) actual price paid for coffee
Reported actual price paid for 539.87 0%
coffee2
Expressed price for organic 816.75 51%
coffee
Note: 1The table is for the 40% of the sample
2Average price paid for coffee on domestic market as reported by survey respondents

However, taking into consideration every participant of the survey, our results

show that there was no willingness to pay for organic coffee. According to Table 5-3

below, which summarizes results for all respondents (not just the 40% of the sample

who were willing to pay high price premiums for organic coffee); individuals were willing

to pay an average price of MK599.65 per 250 g of organic coffee. This price is below

the average market price for conventional coffee by MK 52.34 per 250g. Regarding the

average price paid for coffee on the domestic market as reported by survey

respondents, participants actually paid an average price of MK483.56 per 250g of

coffee.

Taking into consideration the total sample, our results also indicated that

respondents were willing to pay an additional MK116.09 for 250 g of organic coffee on

top of the reported actual price paid for coffee representing an extra 24% price increase

over the reported price (MK 483.56 per 250 g). This is similar to the 40% of the sample

who were also willing to pay an extra price for organic coffee on top of the actual price

they pay for coffee signifying a relatively high value they attached to organic coffee.









Table 5-3. Expressed price for organic versus actual price paid for coffee
Description Average price % Increase over reported actual
(MK/250g) price paid for coffee
Reported actual price paid 488.56 0%
for coffee
Expressed price for organic 599.65 24%
coffee
Note: These estimates are for total sample (including the 40%)

Motivation for the WTP for Organic Coffee

The survey instrument used in the study also collected information that enabled us

to distinguish the specific reasons that prompted respective consumers to value organic

coffee with high price premiums as well as low price premium compared to the

conventional coffee. These questions were multiple in natures in the sense that

consumers were allowed to give out as many reasons as possible. According to Table

5-4 below, the major reason that motivated consumers to register high price premiums

for organic coffee might be the health issues associated with organic products. This is

most especially due to the fact that they are grown without the use of synthetic fertilisers

and so have low content of chemical substances that are hazardous to one's health.

About 33% of the respondents based their motivation on health related issues while

24% based it on the specific fact of low use of chemicals in organic production. These

findings are similar to those by Rodriguez et al. (2006), Wikstrom (2003), Loureiro and

Hine (2002), Didier and Lucie (2008), and Zanoli and Naspetti (2001). In these studies

WTP and preferences for organic products were influenced to a significant extent by

health related issues associated with organic products.

On the other side, the major reason for low WTP for organic coffee was mainly

due to the fact that participants believed that organic production ought to be cheaper









than conventional production since it does not use inorganic fertilizers that are relatively

expensive than organic manure. About 50% of the sample attributed their low

willingness to pay to this reason (Refer Table 5-5).

Table 5-4. Percentage of the sample per motivation factor for positive WTP
Reason for positive WTP Percentage of the
sample
Avoid possible chemical substances 24%
It gives value for money 3.9%
To support local farmer 16.28%
Organic coffee has purer taste 9.3%
To protect environment 12.4%
It makes me different 0%
I feel better 3.1%
Health related reasons 33.3%


Table 5-5. Percentage of the sample per motivation factor for negative WTP
Factor for negative WTP Percentage of the
sample
I can not afford organic 10.9%
Factor for negative WTP Percentage of the
Sample
I do not know how it taste 0.78%
I do not care what type 0%
No need to change coffee habits 1.6%
Its cheap to produce 49.6%

Support towards Organic Coffee Production

Information on some of the aspects that contribute to the high price premiums of

organic coffee was shared at the end of the interview. One of the aspects is the

certification process, which is relatively costly compared with other issues factored into

the cost of production. Participants were then asked to choose the best approach they









felt could contribute towards the certification fee associated with organic coffee.

According to Table 5-6 below, about 75% of the sample felt that the contribution should

be made through government subsidies in order to make it more accessible by many

people, 14% chose to contribute through consumer taxes while 11% thought the whole

cost should be left to the producers themselves. A significant proportion of the 11%

pointed out that eventually the cost will still be transferred to the consumer through high

price premiums so found the option of taxes to be similar to the one they chose and

secondly they felt government should be relieved as it is already subsidising a number

of industries (like the agricultural input subsidy programme).

Table 5-6. Support for certification of organic coffee production
Variable category No. of participants % of the sample
Cost to be left to producer 14 10.85%
alone
Consumer taxes 18 13.95%
Government subsidies 97 75.19%

According to Figure 5-3, the 14 % of the participants were willing to support

organic production through taxes of an average of 2%.

Empirical Analysis of Data from CVM

The regression model that was run to estimate consumer WTP for organic coffee

using the CVM methodology is Model 3-8 as specified in chapter 3. The process of how

we arrived at this final model will not be explained in this section as it has already been

well elaborated in chapter 3. However, it should be noted that not all variables described

under the descriptive summary were included in the model. It should also be highlighted

that some of the categories appearing in Table 5-1 above are not appearing in the

regression model as they were compressed into lesser categorical levels mainly due to









the fact that frequency distribution was scanty across the categories and also to avoid

having too many dummy variables in our model that could significantly reduce the

degrees of freedom of the regression model. For instance in Table 5-1, 'age' variable

has 11 categories that were compressed into 4 categories in OLS model; the 'education'

variable had also 11 categories but these were compressed into 4 categories as well in

the OLS model and lastly the 'income' variable had 10 categories and these were

divided into three income categories in the regression model.


Frequency Distribution of Tax

CO



,O













0 5 10 15 20
Tax (%)



Figure 5-3. Frequency distribution of tax

Prior to model estimation, all 23 regressors were tested for collinearity. The

variables were correlated with each other but no problem of multicollinearity existed.

Most of the coefficients of correlation between the variables were below 0.5 and a few

close to 0.7 such as dummy variables within the same group e.g. amongst the









education and denomination groups. In addition, the number of our variables is in such

a way that 'n> K +1'-'129>24.' This supports one of the Gauss Markov assumptions of

no perfect collinearity; hence our model can be estimated by OLS (Wooldridge J.M,

2009, pg. 86).

The data that were collected in the current study is cross-sectional; as such it is

likely to have the problem of heteroskedasticity. Since our sample size is large enough,

we did not conduct a special test of heteroskedasticity. Instead, we reported

heteroskedasticity robust standard errors to address the problem of heteroskedasticity.

This is a convenient approach of addressing heteroskedasticity that data may be

subjected to without even knowing its form (Wooldridge, 2009). Table 5-7 therefore

represents the results of our estimated model. Based on Table 5-7, our R2 is 0.233,

indicating that about 23.3% variation in the dependent variable can be explained by the

regressors in the model. The F-Statistic is quite large and significant at 5% level. As

such, it supports the R2for the purposes of both prediction of our model and explanation

of relationship between the independent variables (Marti, 2008). The significance of the

F-Statistic implies that at least one of the variables in the model was able to explain our

dependent variable (diffwtp).

As already stated in chapter 3, our model has 23 independent variables, these

were expected to explain the dependent variable in our model. Out of these 23 variables

which represent 13 original variables, four variables were statistically significant. These

are: actpric (actual price for coffee), Dage_d (60 years and older), Dincome_c (high

income of over MK322, 000 per month) and Orgnut (attitudinal value depicting as to










Table 5-7. Estimated OLS model
Name of Coefficient of Robust standard T-statistics P-values
variable variable error
Actpric 0.175 0.070 2.520 0.013**

Female -58.851 47.331 -1.240 0.216
Dage_b 37.719 27.417 1.380 0.172
Dage_c -2.765 45.123 -0.060 0.951
Dage_d -122.108 57.782 -2.110 0.037**
Deducb -20.287 28.598 -0.710 0.480
Deduc 4.113 22.260 0.180 0.854
Deducd -40.316 48.435 -0.830 0.407
Dincomeb -34.326 28.729 -1.190 0.235
Dincomec 132.082 53.063 2.490 0.014**
Ddeno_pres 6.443 30.013 0.210 0.830
Ddenocath 28.021 42.828 0.650 0.514
Ddeno_angl -129.288 87.474 -1.480 0.142
DdenoPent -1.026 38.221 0.030 0.979
DdenoSev -8.428 52.920 -0.160 0.874
Ddeno Isl 53.916 64.213 0.840 0.403
Orgfert 13.198 17.651 0.750 0.456
Orgchm 3.595 17.160 0.210 0.834
Orgnat 11.126 25.816 0.430 0.667
Orgsup -2.174 31.122 -0.070 0.944
Orgris -11.106 22.436 -0.490 0.622
Orgnut -32.309 18.633 -1.730 0.086*
Orgtas -0.744 20.411 -0.040 0.971


n = 129 F-Statistic = ** 5% 10% significant levels R2= 0.233
1.830

whether an individual thinks organic products offer more of some nutrients than their

conventional counterparts). This is in line with a number of studies that conclude that

consumer demographic variables are key determining factors of WTP for organic

products. These include studies by Peterson et al. (2008), Mabiso (2005) and others.









Our findings therefore support the second hypothesis of our study that consumer WTP

for organic coffee is influenced by consumer socio-demographic variables.

Price (actpric) was significant at 5% level and had a positive sign although its

coefficient was quite small. It means if price of conventional coffee paid by respondents

increased by 100 units the consumer WTP for organic coffee would increase by

MK17.51 per 250 g. A number of studies on WTP have also concluded that price is an

important determining factor of WTP for organic product (Rodriguez et al. (2007); Wang

and Sun (2003); Thomas (2009).

Age (60 years and older) was significant at 5% level of significance and it has a

negative coefficient which is relatively large compared to that of price. An individual of

65 years of age or older would be likely to pay a price premium of MK 122.11 per 250 g

of organic coffee less than an average individual. This could be attributed to the fact that

organic production is a relatively new concept in Malawi and so individuals of this age

group are not conversant with their benefits. It is so surprising because elder people are

expected to be sensitive with the type of foods they eat as most of people falling in this

age group are more prone to diseases like cancer than the average individual. Engel

(2008) found that age influenced WTP for organic fruits positively, the 'younger and

middle aged' were willing to pay high price premiums which is similar to our study.

Peterson et al. (2008) and Loureiro and Hine (2002) concluded that age had a

significant impact on WTP as well.

Dincome_c (high income of over MK322, 000 per month) was highly significant at

5% level just like price. Income is positively impacting our dependent variable and has a

high economic impact on the dependent variable evident by the largest coefficient









estimate as expected. If an individual belongs to the high income bracket, he is likely to

pay a price premium of MK132.08 per 250 g of organic coffee more than an average

individual. This supports the notion that organic products are normally demanded by

people with high affluence (Wilier and Yussefi, 2007). In a related development, a

number of studies have come up with a similar conclusion. Peterson et al. (2008),

Dong-Churl (2000), Werner et al. (2002) and Aulong et al. (2008) found that income had

a significant effect on WTP for respective products under valuation in their studies.

Nevertheless, Rodriguez et al (2007) reported that the relationship between income and

WTP is very controversial.

Orgnut (whether an individual thinks organic products offer more of some nutrients

than their conventional counterparts) is significant at 10% level assuming a negative

value. It therefore means that the more an individual agrees that organic products offer

more of some nutrients than their conventional counterparts, the more the individual is

willing to pay high price premiums for organic coffee. For instance, if an individual

increases his extent of agreement to this sentiment by one level (e.g. from just 'agree' to

the higher level of 'strongly agree' the individual is likely to increase his WTP for organic

coffee by MK 32.31 per 250 g of the coffee. This relationship was expected since out of

experience an individual would attach a relatively high value to a product which is more

nutritious than the one with low levels of nutrition. In his study of Rodriguez et al. (2006)

concluded that consumers' perceptions on organic products were the key and better

determinants of WTP for the products than the socio-demographic variables of the

consumers. Our study therefore confirmed this although the other six attitudinal

variables were insignificant.









Gender, Education, all denomination variables and six of the attitudinal variables

(orgfert, orgchm, orgnat, orgsup, orgris and orgtas) were insignificant in our study.

Likewise psychographic and socio-demographic variables of consumers did not have

any significant influence on WTP for beef (Umberger et al. 2002). Rodriguez et al.

(2007) even alluded to the fact that the relation between education and willingness to

pay is also controversial just as with income. However, in his study, there was a

significant positive relationship between lower levels of education and WTP for organic

products. In addition, Peterson et al. (2008), Engel (2008), Zepeda and Li (2007) and

others also reported that there were significant relationships between some of the

demographic variables including gender and education in their respective research.

Engel (2008) found that there was a significant positive relationship between being

Christian and WTP for organic fruits, and religious affiliation was also one of the

significant factors determining WTP in a study by Zepeda and Li (2007).

Empirical Analysis of Data from CE

As already highlighted above, the survey conducted had 129 observations. In the

Choice Experiments, an individual was asked to choose one option amongst the three

options provided to him/her. He/she was supposed to choose either organic coffee,

conventional coffee or the 'None' option. The experiment had 13 choice sets which

meant that the process of choosing the preferred coffee was done 13 times. In the

analysis conducted, each option in a set was turned into an independent observation

thereby resulting into '3 x 13 = 39' observations per each participant. Eventually we had

a total of 5028 observations in the choice experiment.

A test of collinearity was done amongst the independent variables (Organic, price

and None) and the results showed that no variable was perfect collinear to each other.









Descriptive Statistics

Based on Figure 5-4, about 54% of the participants expressed their preference for

organic coffee, about 40% chose conventional coffee while only 6% were indifferent.


Frequency of Choice of Coffee


- -


Figure 5-4. Frequency for choice of coffee

Results from Conditional Logit Model

A Conditional Logistic Regression Model was used to analyze data collected from

the CE. Below is a table of the estimated model:

Table 5-8. Estimates for conditional logistic model
Variable Coefficient Z-statistics P-value
Organic 3.129 33.770 0.000***
Price -0.004 -5.730 0.000***
None -3.520 -7.410 0.000***

n=5028 LR=2533.950 P-value = 0.000 ***1% significant level









The estimated model (Table 5-8) has a likelihood ratio of 2534, its P-value is very

low (0.000) suggesting that the overall model is highly significant. It therefore implies

that at least one of the independent variables used was able to influence our dependent

variable (probability of choosing a type of coffee). All of our variables were statistically

significant with very small P-Values (P<0.000).

Based on our model, the coefficient of 'Organic" was 3.13; this implies that holding

the other independent variables constant, the log of odds in favour of the 'choice of

organic coffee' against that for conventional coffee increased by 3.13. It therefore

means that the likelihood for the choice of organic coffee was higher than that for

conventional coffee. The variable had a coefficient which was larger than that of 'price.'

This indicates that consumers based their coffee preferences more on the method of

production of the coffee than the other attributes such as price. Similarly, in another

study a certain segment of participants were more sensitive to labels (organic and fair-

trade) than price and taste of the products bearing the labels (Didier and Lucie, 2008).

In addition, Thomas (2009) found that the method of production among other variables

was significant in influencing consumers' decision to buy organic oranges. To the

contrary, Boxall et al. (2007) concluded that price and not the method of production was

a significant determinant for increased probabilities that a consumer would purchase

organic bread. Contrary to our findings, both price and the method of production were

insignificant in influencing a participant's decision to buy organic carrots (Thomas,

2009).

On the other hand, the coefficient estimate for 'price' was about -0.004. Price is

therefore the regressor with the smallest coefficient estimate among the three









regressors used in our model. This means that with a unit increase in the price for

coffee, there was a decrease in the log of odds in favor of a consumer choice for a

particular type of coffee of only 0.004, holding other regressors constant. Based on this,

it is clear that 'price' did not have a large economic impact on consumer choice for

coffee. The participants were thus less sensitive to price when they were making their

respective choices for coffee. Our findings are both in support and in contradictory to a

number of previous studies. Thomas (2009) concluded that price was insignificant in

impacting consumers' preference over organic carrots, as already highlighted though it

was significant in impacting consumers' preference for organic oranges. In addition,

although price was significant, group 1 of participants were more sensitive to price and

least to labels of the organic chocolates under valuation in a study by Didier. and Lucie

(2008). Our findings point out that 'method of production' is more influential in affecting

consumer choice for coffee than the 'price.'

Lastly, the coefficient for 'none' was -3.52; this is the largest compared with that of

the other independent variables. Based on this, it means that the log of odds in favor of

the option 'None' (that a consumer would not choose either of the two types of coffee)

decreased by 3.52, holding other variables constant. This makes sense since only

about 6% of the individuals opted for the 'None' option. Most participants chose organic

while others chose conventional coffee.

Based on the analysis, an attempt was done to calculate the WTP for organic

coffee as below:


WTP = ....................... ........................ ........ (5-1)
b1

Where al is the coefficient for organic coffee, and









bl is the coefficient for price

WTP = 3.129
WTP =
0.004

WTP = 782.25

Based on Equation 5-1, the WTP for organic coffee was MK782.25 per 250g which

represents the price premium consumers were willing to pay for 250 g of organic coffee.

It therefore means that consumers were willing to pay an average price of MK 1,434.25

per 250 g of organic coffee (MK 652 + 782.25) which represents over 100% price

premium for organic coffee over the average market price for conventional coffee of MK

652 per 250 g of coffee.

Our results really show that a segment of consumers were indeed willing to pay

high price premiums for organic coffee compared to conventional coffee. The

preference for organic coffee was mainly made based on its type of production and to a

lesser extent its price. This could probably be due to the consumers' perceived benefits

related to organic production and its products. These findings therefore support our first

hypothesis which states that 'because of the perceived benefits associated with product

attributes such as method of production (e.g. organic production) consumers' WTP and

preference for organic coffee is likely to be higher than that of conventional coffee.'

According to past research, it is very clear that product attributes impact the

purchasing behavior of consumers differently. In our study, the method of production of

coffee was more influential than price in influencing consumers' preference for coffee.

Studies conducted by Didier and Lucie (2008) are in support of our results. Casadesus-

Masanell et al (2009) also concluded that WTP for organic garments was registered by

consumers regardless of other related costs (including price) associated with the









garments. However, consumer behaviour is not only influenced by product attributes;

other key factors that influence consumer choice of products are consumer

characteristics. In our study, these were only factored in the OLS regression that was

run prior to the Conditional Logistic Model. However, some studies used Generalized

Logit Models having both product attributes and consumer characteristics as factors

influencing consumer behavior in purchasing a number of products. This could give out

more realistic results as it would allow for interactions of the two types of regressors in

influencing consumer behavior among other things.

Comparison of Results from CVM and CE

Based on our findings, it is very clear that results from the two models used did not

converge (Refer to Table 5-9) below. Based on the CVM, the total sample of the survey

was not willing to pay high price premiums for organic coffee. Participants were willing

to pay an average price of MK599.65 per 250g of organic coffee which is lower than the

average market price for conventional coffee by MK52.35 per 250 g. However, the 40%

were willing to pay high price premiums for organic coffee with an average price of

MK816.75 per 250 g of organic coffee. This represented a price premium of MK164.75

for 250 g of organic coffee over the average market price of conventional coffee. On the

other hand, according to CE, consumers were willing to pay an average price of MK

1,434.25 per 250 g of organic coffee representing a price premium of MK 782.25 per

250g of organic coffee which represents over 100% price premium over the average

market price for conventional coffee.

The divergence of the results could be due to the obvious fact that the

methodologies are different yielding two different results. However, convergence of

results particularly for overall sample, would have given policy makers the confidence to









adopt the results and use them to make well informed decisions regarding organic

coffee. Adoption of the results from the respective methodologies should therefore be

treated with caution since at this stage it is tricky to conclude which ones could be close

to the reality. This therefore calls for need to conduct a similar study using 'Revealed

Preference Models' of eliciting WTP for products as these are commended for their

reliability as compared to the 'Stated Preference Models.'

Table 5-9. Comparisons of WTP price premiums between CVM and CE
Methodology Type of coffee VVTP (MK per 250g)
CVMa Organic (52.34)
CVMb Organic 164.75
CEc Organic 782.25
a Price premium for overall sample and is negative
b Price premium for 40% of sample who expressed high price premiums for organic coffee
c Price premium for overall sample









CHAPTER 6
CONCLUSION

Our study analyzed consumer WTP for Malawi Organic Coffee. The study used

two different methodologies of eliciting WTP which were both stated models. These

include CVM and CE. Information from participants was collected using a survey

targeting 129 participants. Data from CVM was analyzed using an OLS model while that

from CE was analyzed using a Conditional Logistic Model.

Our results showed that there is WTP for organic coffee in Malawi. Based on

CVM, about 40% of the participants were willing to pay a price premium of MK164.75

per 250 g of organic coffee. Nevertheless, when responses of all participants were

considered, participants were not willing to pay high price premiums for organic coffee.

They were willing to pay MK52.34 less per 250 g of organic coffee than the average

market price for conventional coffee. Based on CE, participants were willing to pay a

price premium of MK782.25 per 250 g of organic coffee. Many studies have indeed

drawn similar results that certain segments of consumers have a WTP for organic

products that is higher than their conventional counterparts. These include Rodriguez et

al. (2007), Engel (2008), Wikstrom (2003), Casadesus-Masanell et al. (2009), Dong-

Churl (2000), Werner et al. (2002) and Aulong et al. (2008), just to mention some.

However, some studies have contradicting results. For instance, WTP was higher for

fair trade products than organic products in studies by Didier and Lucie (2008), and

Loureiro and Lotade (2005). Similarly, WTP for organic potatoes was not higher than

that for local potatoes (Loureiro and Hine, 2002).

Results from the two models both confirm and contradict results from previous

studies. For instance, out of the 23 independent variables in the OLS model, only four









were statistically significant. These are: actpric (actual price for coffee), Dage_d (60

years and older), Dincome_c (high income of over MK322, 000 per month) and Orgnut

(attitudinal variable depicting whether participants think that organic products offer more

of some nutrients that their conventional counterparts). This indeed confirms studies by

Peterson et al. (2008) and Mabiso (2005) including many others that assert that

consumer demographic variables are key determining factors of WTP for organic

products although Rodriguez et al. (2006) concludes otherwise. In his study he reported

that consumers' perceptions on organic products are better determinants of WTP than

socio-demographic variables.

Gender, Education, all denomination variables and six of the attitudinal variables

(Orgfert, orgchm, orgnat, orgsup, orgris and orgtas) were insignificant in our study. This

is in line with Umberger et al. 2002 who concluded that psychographic and socio-

demographic variables of consumers did not have any significant influence on WTP.

Rodriguez et al. (2007) even alluded to the fact that relation between education and

willingness to pay is also controversial.

According to the Conditional Logistic Model, 'organic', 'price' and 'none' were

significant variables influencing the choice for coffee. Organic had a higher economic

impact than price over the choice of coffee which implies that the choice for coffee was

highly influenced by the type of production of coffee. This confirms previous studies by

(Didier and Lucie 2008) and Thomas (2009) that the method of production was more

important in influencing consumer decision to buy organic products than the price

attribute.









The major reason that motivated individuals to register higher price premiums for

organic coffee than conventional coffee was health related issues. Most people (about

33%) believed that organic coffee is a healthy drink since its production does not use

inorganic fertilizers.

In order to support organic production through contribution towards certification fee

of the product, about 75% of the individuals opted for government subsidies, 14% opted

to contribute through taxes of 2% on average while 11% chose to leave it all to the

producer.

Based on our results, consumer choices of coffee are based on the method of

production and to a lesser extent the price of coffee. Generally people feel that organic

coffee is healthy compared to conventional coffee. Therefore, it is very likely that there

exists a niche market for organic coffee in Malawi.

Study Limitations

Overall, the study had a number of limitations that were brought about due to

mainly resource constraints (budgetary and time). This resulted into challenges as

follows:

First and foremost, the use of Stated Preference Models allowed for some

hypothetical biases which could have been reduced if the Revealed Preference Models

were used. Initially the study wanted to use Experimental Auctions as a method of data

collection, which renders results that are more reliable than those from Stated

Preference Models (e.g. CVM and CE). However, since this is so expensive to

implement, we opted to use both the CVM and CE, which are relatively cheaper. The

two models were used for purposes of convergent validity check of results.









Secondly, the study wanted to target over 300 participants but we only interviewed

129 people due to budget constraints since we were using face-to-face household

surveys. A larger sample size is ideal in order to capture a number of participants with

divergent characteristics to represent the population and hence improve the efficiency of

the statistics used in making inferences.

Lastly, the study wanted to target the USA and EU markets since they form the

largest market for organic products in the world. However, due to time constraint it was

not possible to conduct the survey in the USA hence the current focus of targeting the

domestic as a step forward towards targeting the international market.

Recommendations and Further Research

The study recommends the following to be implemented in order to come up with

more reliable results on WTP for organic coffee in Malawi and thereafter to promote

organic production in Malawi.

A similar study should be conducted using Revealed Stated Preference Models

(preferably Experimental Auctions). It should also consider using over 300 observations

to make sure that all segments of consumers in Malawi are represented in the sample.

This will ensure that the results are more reliable.

A similar study should also be conducted targeting the international market,

preferable the USA and Europe since the two form the largest market for organic

products in the world. It would thus be feasible to produce organic coffee targeting the

international market in addition to the domestic market since the international market

fetches higher piece premiums for organic products than the domestic market due to the

fact that the international market has consumers that are affluent.









If the government decides to adopt production of organic coffee, there will be need

for a coherent policy for organic production in Malawi. Within this broad policy

framework, there will be need for a specific strategy for organic coffee production

whereby responsibilities of the main actors in the industry will be spelt out. This includes

the Government as a policy regulator, donors as financiers, producers, processors,

marketers, just to mention a few. Proper strategies on transferring the technology to

farmers need to be developed in the policy including proper strategies to support the

production to ensure that the type of production is sustainable and is contributing

significantly to economic growth of the agricultural sector in Malawi and the economy as

a whole. Since organic products are generally more expensive than conventional ones,

the Government should consider implementing subsidies in organic coffee production

among others, to make sure that the products are accessible by many people at

affordable price (Rodriquez et al. 2006).









APPENDIX A
SURVEY INSTRUMENT (VERSION A)

Assessing Consumer WTP for Malawi's Organic Coffee:
Evidence from a Consumer Survey



ID for Respondent ......... ............ ......... ........................... ..........

Nam e of Interview er..................................................... ................................

Supervisor .................................................... ........................................

Remarks by Supervisor .............................................. .............................

Date of Interview ..................................................... .................. ..........


PART A: Consumption Pattern for Coffee: This part attempts to assess
participants' consumption pattern of coffee.

1. Do you drink coffee?

(1) Yes
(2) No

(If no, skip to part b)

2. How often do you drink coffee?

(1) One cup a day
(2) Two cups a day
(3) Three to five cups a day
(4) Six to ten cups
(5) More than ten cups a day
(8) Other (Specify) ......... .. ................................... .........

3. When do you normally drink the coffee?
(Check all that apply)

(a) At Breakfast
(b) At Lunch
(c) At Dinner
(d) In between meals










Part B: Quiz on General Knowledge of Organic Coffee and Coffee in general

The quiz is aimed at assessing consumers' perception over organic coffee as well as
coffee in general. In order to assess the attitude, the options of the answers will be one
of (Strongly Agree, Agree, Uncertain, Disagree, and Strongly Disagree)

4. Coffee is the world's second most valuable "traded" commodity behind
only petroleum.



(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

5. Many coffee producing countries use highly toxic chemicals that have been
banned or restricted in many countries (e.g. DDT).

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree



6. Organic coffee is grown without the use of synthetic fertilizers.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree









(5) Strongly Disagree


7. Organic coffee is grown without the use of any pesticides or chemicals.



(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

8. By buying organic products you as a consumer are supporting the natural
and healthiest way to grow crops

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree



9. By buying organic coffee you are supporting the small holder farmer.



(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree









10. By drinking organic coffee there is lower risk of ingesting synthetics or
chemicals.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

11. Organically grown food may offer more of some nutrients than their
conventionally produced counterparts.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

12. Organically grown food have better taste than their conventionally
produced counterparts.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree


13. Europe and North America form the largest market for organic products.

(1) Strongly Agree

(2) Agree









(3) Uncertain


(4) Disagree

(5) Strongly Disagree


14. Malawi grows organic coffee which is also sold as an export crop.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree


Part C: Questions on WTP

Definition of Organic Coffee:

Organic coffee is coffee that has been certified as having been grown without the use of
inorganic fertilizers, synthetic pesticides, herbicides, or other chemicals. It can also
refer to farms which incorporate socially responsible activities such as recycling,
composting, soil health and environmental protections.

15. The average price for Malawi conventional coffee is MK 652 per 250 g.

How much are you willing to pay for organic coffee per 250 g?



16. If the WTP is positive, why would you be willing to pay more for it?
(Check all that apply)

(a) To avoid possible chemical substances in my coffee
(b) The organic coffee will give me the most value for the money
(c) To support local farmers
(d) Its got a purer taste
(e) To help protect environment
(f) It makes me different from people drinking conventional coffee
(g) I feel better to drink organic coffee
(h) Other, specify ................................................................ ....










17. If there's no WTP, what are the reasons:


(Check all that apply)


(a) I would wish to pay more for Organic coffee but I can't afford it
(b) I hesitate to choose organic coffee since I don't know how it tastes
(c) I don't care whether the coffee I buy is organic or not
(d) I see no reason to change my coffee habits
(e) Other, specify .................. ............... ............... ........ ....

18. In your last purchase, how much did you pay for a 250 g of coffee?


Part D: Choice Experiments (CE)

In this category, you are required to choose either 250 grams packet of organic coffee
or conventional coffee as you are shopping in the market, or choose None option if you
are not satisfied with both coffees. You will do this for 13 combinations.


Combination 1


Conventional
Coffee
MK795/ 250 g
O


Organic Coffee

MK699/250 g
O


Combination 2


Conventional
Coffee
MK699/ 250 g
O


Organic Coffee

MK699/250 g
O


Combination 3


Conventional
Coffee
MK720/ 250 g


Organic Coffee

MK660/250 g


Combination 4


None


None


None


Pricer


Pricer


Price
I Choose










Conventional
Coffee
MK485/250 g


Organic Coffee

MK485/250 g


Combination 5


Conventional
Coffee
MK475/ 250 g


Organic Coffee
None


MK485/250 g


Combination 6


Conventional
Coffee
MK729/ 250 g


Organic Coffee
None
MK475/250 g


Combination 7


Conventional
Coffee
MK475/ 250 g


Organic Coffee
None
MK475/250 g


Combination 8


Organic Coffee

MK 795/250 g
O


Conventional Coffee

MK729/250 g
O


Combination 9


Organic Coffee

MK720/ 250 g
O


Conventional Coffee

MK660/250 g
O


Combination 10


None


None


None


Price
I Choose


Price
I Choose


Price
I Choose


Price
I Choose


Pricem


Pricem
I Choose










Organic Coffee

MK729/ 250 g
O


Conventional Coffee

MK795/250 g
O


Combination 11


Organic Coffee

MK720/ 250 g
O


Conventional Coffee

MK699/250 g
O


Combination 12


Organic Coffee

MK729/ 250 g
O


Conventional Coffee

MK485/250 g
O


Combination 13


Organic Coffee

MK795/ 250 g
O


Part E: Questions on Socio-Demographics


31. Gender of the respondent:


(1) Male
(2) Female


32. What is your marital status?


Married
Divorced
Single


None


None


None


Conventional Coffee

MK720/250 g
O


None


Pricem
I Choose


Pricer


Price!
I Coos


Pricer









(8) Other, Specify ............... .........


33. Do you have children living in your household that fall into these age categories?
(Check all that apply)


(a) Under 2 years
(b) 2 to 5 years
(c) 6 to 12 years
(d) 13 to 18 years
(e) None



34. How old are you?


(1) 15 to 19 years
(2) 20 to 24 years
(3) 25 to 29 years
(4) 30 to 34 years
(5) 35 to 39 years
(6) 40 to 44 years
(7) 45 to 49 years
(8) 50 to 54 years
(9) 55 to 59 years
(10) 60 to 64 years
(11) 65 years and older


35. What's your highest level of education?


(1) Completed post-graduate degree (Masters or Ph.D)
(2) Completed University Undergraduate Degree
(3) Attended University Undergraduate
(4) Completed College Degree
(5) Completed College Diploma
(6) Attended Some College
(7) Some Post Secondary Technical School
(8) Completed High School Certificate/Secondary School (e.g. MSCE)
95









(9) Attended Some High School/Secondary School Certificate (e.g. MSCE)
(10) Completed Elementary/Primary School
(11) Attended Some Elementary/Primary school


36. What is your net monthly total household income?


(1) MK15, 000 and below
(2) MK16, 000 to MK66, 000
(3) MK67, 000 to MK117, 000
(4) MK118, 000 to MK168, 000
(5) MK169, 000 to MK219, 000
(6) MK220, 000 to MK270, 000
(7) MK271, 000 to MK321, 000
(8) MK372, 000 to MK423, 000
(9) MK474, 000 to MK525, 000
(10) MK526, 000 and above


37. Do you work with an organization that deals with issues related to Organic
farming, food safety and other environmental related issues?


(1) Yes
(2) No


(If No, go to 39)


38. W which Organization.................................. ...........................


39. You belong to which denomination?


(1) Presbyterian
(2) Catholic
(3) Anglican
(4) Pentecostal
(5) Seventh day Adventist
(6) Islam
(8) Other, specify ............ ............. ....................... .... ..
96











40. In which city do you reside in?


(1) Blantyre
(2) Lilongwe
(3) Mzuzu



41. Organic coffee may bring relatively high incomes to a farmer and the nation
as a whole and thus likely be one of the alternatives for tobacco as a major
foreign exchange earner for the country. However, the certification fee of
organic coffee may be high. Who do you think should pay for the higher
cost related to organic coffee certification?


(1) The producer as part of his/her cost of production
(2) The Consumer through taxes
(3) The Government through subsidies

42. If answer is (2) in 41, how much tax would you be willing to pay in support
of organic coffee certification?









APPENDIX B
SURVEY INSTRUMENT (VERSION B)

Assessing Consumer WTP for Malawi's Organic Coffee:
Evidence from a Consumer Survey



ID for Respondent ......... ............ ......... ........................... ..........


Nam e of Interview er..................................................... ................................


Supervisor .................................................... ........................................


Remarks by Supervisor .............................................. .............................


Date of Interview ..................................................... .................. ..........



PART A: Consumption Pattern for Coffee: This part attempts to assess
participants' consumption pattern of coffee.


1. Do you drink coffee?


(1) Yes
(2) No

(if no, skip to part b)


2. How often do you drink coffee?


(1) One cup a day
(2) Two cups a day
(3) Three to five cups a day
(4) Six to ten cups
(5) More than ten cups a day
(8) Other (Specify) .............. .......... .. ............................

98











3. When do you normally drink the coffee?
(Check all that apply)


(a) At Breakfast
(b) At Lunch
(c) At Dinner
(d) In between meals



Part B: Quiz on General Knowledge of Organic Coffee and Coffee in general


The quiz is aimed at assessing consumers' perception over organic coffee as well as
coffee in general. In order to assess the attitude, the options of the answers will be one
of (Strongly Agree, Agree, Uncertain, Disagree, and Strongly Disagree)

4. Coffee is the world's second most valuable "traded" commodity behind
only petroleum.




(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

5. Many coffee producing countries use highly toxic chemicals that have been
banned or restricted in many countries (e.g. DDT).

(1) Strongly Agree

(2) Agree









(3) Uncertain

(4) Disagree

(5) Strongly Disagree




6. Organic coffee is grown without the use of synthetic fertilizers.




(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree




7. Organic coffee is grown without the use of any pesticides or chemicals.




(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

8. By buying organic products you as a consumer are supporting the natural
and healthiest way to grow crops
100









(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree




9. By buying organic coffee you are supporting the small holder farmer.




(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

10. By drinking organic coffee there is lower risk of ingesting synthetics or
chemicals.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree


101









11. Organically grown food may offer more of some nutrients than their
conventionally produced counterparts.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree

12. Organically grown food have better taste than their conventionally
produced counterparts.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree



13. Europe and North America form the largest market for organic products.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree


102









14. Malawi grows organic coffee which is also sold as an export crop.

(1) Strongly Agree

(2) Agree

(3) Uncertain

(4) Disagree

(5) Strongly Disagree



Part C: Questions on WTP


Definition of Organic Coffee:


Organic coffee is coffee that has been certified as having been grown without the use of
inorganic fertilizers, synthetic pesticides, herbicides, or other chemicals. It can also
refer to farms which incorporate socially responsible activities such as recycling,
composting, soil health and environmental protections.


15. The average price for Malawi conventional coffee is MK 652 per 250 g.


How much are you willing to pay for organic coffee per 250 g?




16. If the WTP is positive, why would you be willing to pay more for it?
(Check all that apply)


(a) To avoid possible chemical substances in my coffee
(b) The organic coffee will give me the most value for the money
(c) To support local farmers
(d) Its got a purer taste
(e) To help protect environment
(f) It makes me different from people drinking conventional coffee


103









(g) I feel better to drink organic coffee
(h) Other, specify ......... ......... ............ .............


17. If there's no WTP, what are the reasons:


(Check all that apply)


(a) I would wish to pay more for Organic coffee but I can't afford it
(b) I hesitate to choose organic coffee since I don't know how it tastes
(c) I don't care whether the coffee I buy is organic or not
(d) I see no reason to change my coffee habits
(e) Other, specify .................. ............... ............... ........ ....
18. In your last purchase, how much did you pay for a 250 g of coffee?


Part D: Choice Experiments (CE)


In this category, you are required to choose either 250 grams packet of organic coffee
or conventional coffee as you are shopping in the market, or choose None option if you
are not satisfied with both coffees. You will do this for 13 combinations.


Combination 1


Organic Coffee

MK 795/250 g
O


Conventional Coffee

MK729/250 g
O


Combination 2


Organic Coffee

MK795/ 250 g
O


Conventional Coffee

MK720/250 g
O


Combination 3


Organic Coffee


Conventional Coffee


104


None

O


None

[


None


Pricer


Pricem
I Choose









MK729/ 250 g


Combination 4


Organic Coffee

MK729/ 250 g
O


Conventional Coffee

MK485/250 g
O


Combination 5


Organic Coffee

MK720/ 250 g
O


Conventional Coffee

MK699/250 g
O


Combination 6


Organic Coffee

MK720/ 250 g
O


Conventional Coffee

MK660/250 g
O


Combination 7


Conventional
Coffee
MK795/ 250 g


Organic Coffee
None
MK699/250 g


Combination 8


Conventional
Coffee
MK699/ 250 g


Organic Coffee
None
MK699/250 g


Combination 9


Conventional
Coffee


Organic Coffee


105


None


None


None


None


MK795/250 g


Pricer


Price!
I Coos


Price
I Choos


Price!
I Coos


Price
I Choos


Price!
I Coos










MK720/ 250 g
O


MK660/250 g
O O


Combination 10


Conventional
Coffee
MK485/250 g
O


Organic Coffee

MK485/250 g
O


Combination 11


Conventional
Coffee
MK475/ 250 g
O


Organic Coffee

MK485/250 g
O


Combination 12


Conventional
Coffee
MK729/ 250 g
O


Organic Coffee

MK475/250 g
O


Combination 13


Conventional
Coffee
MK475/ 250 g
O


Part E: Questions on Socio-Demographics


31. Gender of the respondent:


Male
Female


106


None


None


None


Organic Coffee

MK475/250 g
O


None


Price
I Choose


Pricer


Price
I Choos


Price
I Choos


Price
I Coos











32. What is your marital status?


(1) Married
(2) Divorced
(3) Single
(8) Other, Specify ............... .........


33. Do you have children living in your household that fall into these age categories?
(Check all that apply)


(a) Under 2 years
(b) 2 to 5 years
(c) 6 to 12 years
(d) 13 to 18 years
(e) None



34. How old are you?


(1) 15 to 19 years
(2) 20 to 24 years
(3) 25 to 29 years
(4) 30 to 34 years
(5) 35 to 39 years
(6) 40 to 44 years
(7) 45 to 49 years
(8) 50 to 54 years
(9) 55 to 59 years
(10) 60 to 64 years
(11) 65 years and older


35. What's your highest level of education?


(1) Completed post-graduate degree (Masters or Ph.D)
(2) Completed University Undergraduate Degree
107









(3) Attended University Undergraduate
(4) Completed College Degree
(5) Completed College Diploma
(6) Attended Some College
(7) Some Post Secondary Technical School
(8) Completed High School Certificate/Secondary School (e.g. MSCE)
(9) Attended Some High School/Secondary School Certificate (e.g. MSCE)
(10) Completed Elementary/Primary School
(11) Attended Some Elementary/Primary school


36. What is your net monthly total household income?


(1) MK15, 000 and below
(2) MK16, 000 to MK66, 000
(3) MK67, 000 to MK117, 000
(4) MK118, 000 to MK168, 000
(5) MK169, 000 to MK219, 000
(6) MK220, 000 to MK270, 000
(7) MK271, 000 to MK321, 000
(8) MK372, 000 to MK423, 000
(9) MK474, 000 to MK525, 000
(10) MK526, 000 and above


37. Do you work with an organization that deals with issues related to Organic
farming, food safety and other environmental related issues?


(1) Yes
(2) No


(If No, go to 39)


38. W which Organization.................................. ...........................


39. You belong to which denomination?


(1) Presbyterian
108









(2) Catholic
(3) Anglican
(4) Pentecostal
(5) Seventh day Adventist
(6) Islam
(8) O their, specify ........................................................... .... ....

40. In which city do you reside in?


(1) Blantyre
(2) Lilongwe
(3) Mzuzu



41. Organic coffee may bring relatively high incomes to a farmer and the nation
as a whole and thus likely be one of the alternatives for tobacco as a major
foreign exchange earner for the country. However, the certification fee of
organic coffee may be high. Who do you think should pay for the higher
cost related to organic coffee certification?


(1) The producer as part of his/her cost of production
(2) The Consumer through taxes
(3) The Government through subsidies

42. If answer is (2) in 41, how much tax would you be willing to pay in support
of organic coffee certification?


109









LIST OF REFERENCES


Agresti, A. and Finlay, B. (2009) Statistical Methods for the Social Sciences (Fourth
Edition). Prentice Hall, Inc. New Jersey.

Alpizar, F. et al. (2001) Using Choice Experiments for Non-Market Valuation.Working
Papers in Economics No. 52. Goteborg University, Sweden.

Annual Gross Margin Analysis Report, MCPCU, 2010.

Annual Sales Data, Tobacco Control Commission (TCC).http://www.tccmw.
com/Annual%20sales%20data.htm. (Accessed in August 2009).

Annual Production Data, Tobacco Control Commission (TCC). http://www.tccmw.
com/downloads.htm#production. (Accessed in January 2010).

Antle, J. (1999) The New Economics of Agriculture. American Journal of Agricultural
Economics, Proceedings, 81(5): 993-1010.

Aulong, S. and Rinaudo, J. (2008) Assessing the Benefits of Different Groundwater
Protection Level: Results and Lessons Learnt from a Contingent Valuation
Survey in the Upper Rhine Valley Aquifer, France.

Blend, J. and Van Ravenswaay, E. (1998) Consumer Demand for Ecolabelled Apples:
Survey Methods and Descriptive Results. Staff Paper 98-20. Department of
Agricultural Economics, Michigan State University.

Boxall, P. et al. (2007) The Role of Sensory Attributes and Information on the
Willingness to Pay for Organic White Bread. University of Alberta, USA.

Breslow, N.E. (1974) Covariance Analysis of Censored Survival Data. Biometrics 30:
89-95.

Bruin, J. (2006) Newtest: Command to Compute Newtest. UCLA: Academic
Technology Services, Statistical Consulting Group. http://www.ats.ucla.edu/
stat/stata/ado/analysis/

Buzby, J., Ready, R. and Skees, J. (1995) Contingent Valuation in Food Policy
Analysis: A Case Study of Pesticide-Residue Risk Reduction. Journal of
Agricultural and Applied Economics, 27(2):613-625.

Carson, R.T. et al. (1994) Contingent Valuation and Lost Passive Use: Damages from
the Exxon Valdez. Resources for the Future Discussion Paper. QE 94-18,
Washington D.C.


110









Casadesus-Masanell, et al. (2009) Households' Willingness to Pay for "Green" Goods:
Evidence from Patagonia's Introduction of Organic Cotton Sportswear. Journal of
Economics and Management Strategy, Vol. 18, Issue 1, pp. 203-233.

Chirwa E. et al. (2008) Smallholder Coffee Commercialization in Malawi, Policy Brief,
Future Agricultures.

CIA World Fact Book (ISSN 1553-8133) https://www.cia.gov/library/publications/the-
world-factbook/geos/xx.html. (Accessed on 28th July 2010).

Cochran, W.G. (1977) Sampling Techniques (3rd Edition) New York: John Wiley and
Sons.

Corsi, A. (2007) Ambiguity of Measured WTP for Quality Improvements When Quantity
is unconstrained: a Note. Oxford University Press and Foundation for the
European Review of Agricultural Economics Advance. (Access published online
on November 26, 2007).

Czaja, R. and Jonny, B. (1995) Designing Surveys: A Guide to Decisions and
Procedures, Pine Forge Press.

Darby, M.R. and Karni, E. (1973) Free Competition and the Optimal amount of Fraud.
Journal of Law and Economics, 16, 67-88.

Didier, T. and Lucie, S., (2008) Measuring Consumer's Willingness to Pay for Organic
and Fair Trade Products. International Journal of Consumer Studies, 32(2008)
479-490, Blackwell Publishing Ltd.

Dong-Churl, S. (2002) Consumers' Willingness to Pay for Pharmacy Services that
Reduce Risk of Medication-Related Problems. Journal of the American
Pharmacists Association.

Earth Trends Environmental Information. Agriculture and Food Searchable Database.
World Resource Institute, 2007.

Elzakker, B. et al. (2007) Organic Farming in Africa. The World of Organic Agriculture,
Statistics and Emerging Trends, Chpt.12 pg. 93-105.

Engel, W. (2008) Determinants of Consumer Willingness to Pay for Organic Food in
South Africa. A paper submitted in partial fulfillment of the requirement for the
degree Mlnst Agrar, University of Pretoria.

Ericsson, G., Kindberg, J. and Bostedt, G. (2007) Willingness to Pay (WTP) for
Wolverine Gulo Gulo Conservation. Wildlife Biology 13(sp2):2-13.


111









Fieser, E. (2010) Organic Coffee: Why Latin America's Farmers are Abandoning It.
Christian Science Monitor. http://seattletimes.nwsource.com/html/
businesstechnology/2011109993_organicoffee07.html. (Accessed on July 29,
2010).

Friedman, D., and Sunder, S. (1994) Data Analysis in Experimental: A Primer for
Economists. New York: Cambridge University Press.

Fox, J.A. (1995) Experimental Auctions to Measure Willingness to Pay for Food Safety.
Boulder, CO: West View Press.

Gao, Z. and Schroeder, T. (2007) Effect of Additional Quality Attributes on Consumer
Willingness to Pay for Food Labels, A Dissertation Submitted in Partial Fulfilment
of the Requirements for the Degree of Doctor of Philosophy, Kansas State
University.

Greene, W.H. (1997) Econometric Analysis. Upper Saddle River, NJ: Prentice-Hall.

Greene, W.H. (1998) LIMDEP Version 7.0 User's Manual (Revised Edition).Plainview
New York: Econometric Software, Inc.

Gwendolyn, H., John, C. and Bernard (2008) Consumer Willingness to pay for
Sustainable Apparel: The Influence of Labeling for Fibre Origin and Production
Methods. International Journal of Consumer Studies, Vol. 32, Issue 5, pages
491-498.

Hoffman, S.D. and Duncan, G.J. (1988) Multinomial and Conditional Logit Discrete-
Choice Models in Demography. Demography, Vol. 25, No.3: 415-427.

Hosmer, D.W. Jr. and Lemeshow, S. (2000) Applied Logistic Regression 2nd Edition.
New York: John Wiley & Sons.

Hustvedt, G. and Bernard, J.C. (2008) Consumer Willingness to pay for Sustainable
Apparel: The Influence of Labelling for Fibre Origin and Production Methods.
International Journal of Consumer Studies, 32 (2008): 491-498.

Introduction to SAS. UCLA: Academic Technology Services, Statistical Consulting
Group. http://www.ats.ucla.edu/stat/sas/notes2/ (Accessed November 24,2007).

Jaffee, S. (2003) "Malawi's Tobacco Sector-Standing One Strong Leg is Better Than on
None," Africa Region Working Paper Series, No. 55, World Bank.

Jauregui C.E. (2007) Consumers' Use of Food Labels: An Application of Ordered Probit
Models. A Doctoral of Philosophy Dissertation, University of Florida, FL-USA.


112









Jordan, J. and Elnagheeb, A. (1991) Public Perception of Food Safety. Journal of Food
Distribution Research, 22(3): 13-22.

Kim Minbo, Qualitative and Limited Dependent Variable Models Using the New QLIM
Procedure, Statistics and Data Analysis. SAS Institute Inc., Cary, NC. Paper 279-
25.

Krystallis, A. and Chryssohoidis, G. (2005) Consumers' Willingness to Pay for Organic
Food: Factors that affect it and Variation per Organic Product Type. British Food
Journal, 107, 320-343.

Lancaster, K. (1966) A New Approach to Consumer Theory. Journal of Political
Economy, LXXIV(2): 132-157.

Lin, Biing-Hwan et al. (2008) Demand for Organic and Conventional Fresh Fruits.
Selected Paper Prepared for Presentation at the American Agricultural
Economics Association Annual Meeting, Orlando.

Loureiro, M.L. and Hine, S. (2002) Discovering Niche Markets: A Comparison of
Consumer Willingness to Pay for Local (Colorado Grown), Organic and GMO-
Free Products. Journal of Agricultural and Applied Economics, 34, 3(December
2002):477-487. Southern Agricultural Economics Association.

Loureiro, M. and Lotade, J. (2005) Do Fair Trade and Eco-Labels in Coffee Wake up the
Consumer Conscience? Ecological Economics, 53, 129-138.

Loureiro, M.L. et al. (2001) Assessing Consumer Preferences for Organic, Eco-
Labelled, and Regular Apples. Journal of Agricultural and Resources Economics
26(2): 404-416, Western Agricultural Economics Association.

Lusk, J.L. and Schroeder, T.C. (2004) Are Choice Experiments Incentive Comparable?
A Test with Quality Differentiated Beef Steaks. American Agricultural Economics
Association. Amer J. Agr. Econ. 86(2): 467-482.

Mabiso A. (2005) Estimating Consumer's Willingness to Pay for Country of Origin
Labels in fresh Apples and Tomatoes: A Double-Hurdle Probit Analysis of US
Data, Master Thesis, University of Florida, Florida, USA.

Martin G.K. (2008) Assessing the Fit of Regression Model. Flashpoint
(http://www.selfgrowth.com).

MCCCI Monthly Bulletin. Published by the Malawi Confederation of Chambers of
Commerce and Industry, January 2009.

McCluskey, J.J. and Loureiro, M.L. (2003). The Consumer Response to Food Labelling.
http://www.farmfoundation.org


113










McFadden, D. (1974) Conditional Logit Analysis of Qualitative Choice Behaviour. In
Frontiers in Econometrics, ed. P. Zarembka, 105-142, New York: Academic
Press.

Measuring Consumer Interest in Mexican Shade-Grown Coffee: An Assessment of the
Canadian, Mexican and US Markets. Commission for Environmental
Cooperation. Published by the Communications and Public Outreach Department
of the CEC Secretariat, October 1999. http://www.cec.org.

Medium-Term Prospects for Agricultural Commodities, Projections to the Year 2010,
FAO, 2003.

Millock, K. et al. (2002) Willingness to Pay for Organic Foods: A Comparison between
Survey Data and Panel Data from Denmark.

Mitchell, R. and Carson, R. (1989). Using Surveys to Value Public Goods: The
Contingent Valuation Method. Resource for the Future, Washington Dc.

NOAA (1993) National Resource Damage Assessment Under the Oil Pollution Act of
1990, Federal Registers 58(10):4601-4614.

Nelson, P. (1970) Information and Consumer Behaviour. Journal of Political Economy,
78, 311-329.

Noussair, C., Robin, S. and Ruffieux, B. (2004) A Comparison of Hedonic Rating and
Demand -Revealing Auctions. Food Quality and Preference, 15, 393-402.

Pendergrast, M. (2006) Coffee: The Drink that Changed the World. GENERATE, CEA
Group, Germany.

Population and Housing Census Preliminary report. National Statistical office, GOM,
Malawi, 2008.

Peterson, et al. (2008) US Consumers' Willingness to Pay for Wool Product Attributes,
Selected Paper Prepared for Presentation at the Southern Agricultural
Economics Association, Annual Meeting, Dallas, TX.

Poelman, A., Mojet, J., Lyon, D. and Sefa-Dedeh, S. (2008) The Influence of
Information about Organic Production and Fair Trade on Preferences for and
Perception of Pineapple. Food Quality and Preference, 19, 114-121.

Rahmatian, M. (2005) Contingent Valuation Method, Session 14. California State
University. http://www.iwlearn.net/publications/misc/caspianer_rahmatian-
cvm.pdf. (Accessed on July 38, 2010).


114









Rodriguez, et al. (2006) Consumers' Perceptions about Food Quality Attributes and
their Incidence in Argentinean Organic Choices. Paper Presented at the
International Association of Agricultural Economics Conference, Gold Cost,
Australia.

Rodriguez E. et al. (2007) Willingness to Pay for Organic Food in Argentina: Evidence
from a Consumer Survey. Contributed Paper Presented at the 105th EAAE
Seminar 'International Marketing and International Trade and Quality Food
Products, Bologna, Italy.

Sahota, A. (2007) Overview of the Global Market for Organic Food and Drink. The
World of Organic Agriculture, Statistics and Emerging Trends, Chpt. 7 pg. 52-
55,IFOAM and FIBL, Switzerland.

Singleton, R. (1993) Approaches to Social Research, New York-Oxford University
Press.

The Agricultural Sector Wide Approach: Malawi's Prioritised and Harmonised
Agricultural Development Agenda. Ministry of Agriculture and Food Security
(MoAFS), (2010).

The Trade and Environmental Effects of Ecolabels: Assessment and Response, UNEP.

Thomas, D.S. (2009) Taste Test of Organic Versus Conventional Products and What
Influences the Consumers' Decision to Purchase. A Thesis for Masters Degree,
University of Florida, FL-USA.

Thurstone, L.L. (1927) A Law of Comparative Judgement. Psychology Review,34:
273-286.

Yussefi, M. and Wilier, H. (2007) Organic Farming Worldwide 2007: Overview and Main
Statistics. The World of Agricultural Production, Chpt. 3, pp. 9-16.

Umberger, W.J. (2002) U.S. Consumer Preference and Willingness to Pay for Domestic
Corn-Fed Beef Versus International Grass-Fed Beef measured through an
Experimental Auction. Agribusiness, Vol. 18 (4) (2002): 491-504. Published
online in Wiley InterScience. http://www.interscience.wiley.com.

Wang, Q. and Sun, J. (2003) Consumer Preference and Demand for Organic Food:
Evidence from a Vermont Survey. Paper Prepared for American Agricultural
Economics Association Annual Meeting.

Werner et al. (2002) Family Caregivers Willingness to Pay for Drugs Indicated for the
Treatment of Alzheimer's Disease, Dementia, 1, 59-74.Whittington, D. (1998)
Administering Contingent Valuation Surveys in Developing Countries. World
Development 26(1), 21-30.


115










Wikstrom D. (2003) Willingness to Pay for Sustainable Coffee: A Choice Experiment
Approach. Master's Thesis, University of Technology, Sweden.

Wilier, H. (2009) The World of Organic Agriculture 2009: Summary. IFOAM & FIBL,
Switzerland.

Wilier, H. and Yussefi, M. (2007) The World of Organic Agriculture, Statistics and
Emerging Trends. IFOAM & FIBL, Switzerland.

Wilier, H. and Yussefi, M. (2007) Organic Farming Worldwide 2007: Overview and Main
Statistics. The World of Organic Agriculture, Statistics and Emerging Trends,
Chpt. 8, pg. 9-16, IFOAM & FIBL, Switzerland.

Wooldridge, M.J. (2009) Introductory Econometrics: A Modern Approach (4th Edition),
Thomson, South-Western C.

Wuyang, H. (2006) Comparing Consumers' Preferences and Willingness to Pay for
Non-GM Oil Using a Contingent Valuation Approach. Empirical Economics,
Volume 31, Number 1, pp. 143-150(8).

Zanoli, R. and Naspetti, S. (2001) Consumer Motivations in the Purchase of Organic
Food: A Means-End Approach. British Food Journal, 8, 643-653.

Zepeda, L. and Li, J. (2007) Characteristics of Organic Food Shoppers, Journal of
Agricultural and Applied Economics, 39:17-28.


116









BIOGRAPHICAL SKETCH

Fiskani Esther Nkana is a young lady born and raised in Malawi, the warm heart of

Africa. She did her primary, secondary and undergraduate studies in Malawi. She holds

a Bachelor of Social Science with Economics as a major which was obtained from the

University of Malawi (UNIMA) Chancellor College in 2003. Based on her background,

she decided to pursue a Master of Science in Food and Resource Economics which is

very vital to the development of Agriculture in her country and Africa as a whole.


117





PAGE 1

1 ASSESSING CONSUMER WILLINGNESS TO PAY FOR MALAWI ORGANIC COFFEE: EVIDENCE FROM A CONSUMER SURVEY By FISKANI ESTHER NKANA A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF TH E REQUIREMENTS FOR MASTER OF SCIENCE DEGREE UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Fiskani Esther Nkana

PAGE 3

3 To my father and mother, I dedicate this to you

PAGE 4

4 ACKNOWLEDGMENTS I thank my parents for instilling discipline in me and making me who I am to day. You still remain my greatest role models. Many thanks to your prayers! I appreciate the support rendered by my sisters; Lydia, Irene, Tawonga and my brother Temwanani. My heartfelt appreciation goes to Herbert, my fianc; I appreciate your support tow ards my studies during the entire period. I thank m y Supervisor Dr. Zhifeng Gao, he had a lot of things to handle but he still set aside some time for this work He was quite a great resource. I also acknowledge Dr. James Sterns who was the member of my co mmittee. I thank the Ministry of Agriculture and Food Security for letting me participate in this program and for their support during my studies most especially during research work. I would like to thank USAID for awarding me this scholarship. This goes to both the DC and the Malawi Local Office. Specifically, my appreciation goes to Martin Banda, Jean Msosa Maganga and the entire team at USAID Malawi. In the same regard, I recognize the support by Martin Kanjadza of the American Embassy Education Departm ent. I also thank Dr. Walter Bowen, Dr. Burkhadt, Jessica Herman, Mart i and their entire team at the University of Florida for their support. My heartfelt appreciation goes to Mr. Peter Njikho of the Coffee Association of Malawi (CAMA L ) and all its subsidi aries for their assistance in sharing necessary data for this research In addition I thank all the households that I interviewed under this

PAGE 5

5 study for their cooperation and honesty in giving out the data. Lastly but not least, I thank all my friends for t heir moral support. Above all, I thank my God for his abundant grace during my studies, May his name be always praised!

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 General Problem ................................ ................................ ................................ ..... 14 Spec ific Problem ................................ ................................ ................................ ..... 18 Project Objectives ................................ ................................ ............................ 19 Testable Hypotheses ................................ ................................ ........................ 19 2 L ITERATURE REVIEW ................................ ................................ .......................... 20 Overview of Organic Agriculture ................................ ................................ ............. 20 Previous Studies on Consumer Preference and WTP for Organic Products .... 22 Previous Studies on Consumer Preference and WTP for Non Organic Products ................................ ................................ ................................ ........ 3 2 Methods of Eliciting Consumer WTP ................................ ................................ ...... 36 The Contingent Valuation Method (CVM) ................................ ......................... 37 The Choice Experiment (CE) ................................ ................................ ............ 39 Experimen tal Auction ................................ ................................ ....................... 40 3 RESEARCH METHODS AND DATA ................................ ................................ ...... 42 Theoretical Model ................................ ................................ ................................ ... 42 Model Estimation ................................ ................................ ................................ .... 43 Model Estimation under CVM ................................ ................................ ........... 43

PAGE 7

7 Model Estimation under CE ................................ ................................ .............. 48 4 DATA COLLECTION ................................ ................................ .............................. 49 Design of CVM ................................ ................................ ................................ ........ 53 Design of the CE ................................ ................................ ................................ ..... 55 5 RESULTS OF THE EMPIRICAL ANALYSIS ................................ .......................... 58 Descriptive Statistics ................................ ................................ ............................... 58 Willingness to Pay for Organic Coffee ................................ .............................. 64 Motivation for the WTP for Organic Coffee ................................ ....................... 67 Support towards Organic Coffee Production ................................ .................... 68 Empirical Analysis of Data from CVM ................................ ................................ ..... 69 Empirical Analysis of Data from CE ................................ ................................ ........ 75 Descriptive Statistics ................................ ................................ ........................ 76 Results from Conditional Logit Model ................................ ............................... 76 Comparison of Results from CVM and CE ................................ .............................. 80 6 CONCLUSION ................................ ................................ ................................ ........ 82 Study Limitations ................................ ................................ ................................ .... 84 Recommendations and Further Research ................................ .............................. 85 APPENDIX A SURVEY INSTRUMENT (VERSION A) ................................ ................................ .. 87 B SURVEY INSTRUMENT (VERSION B) ................................ ................................ .. 98 LIST OF REFERENCES ................................ ................................ ............................. 110 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 117

PAGE 8

8 LIST OF TABLES Table page 1 1 Total annual sales for tobacco in Malawi ................................ ............................ 15 4 1 Sample representativeness in terms of the demographic structure of the population of Malawi (gender, age and religion) ................................ ................. 51 4 2 Price levels for coffee in MK/250 grams (Version B) ................................ .......... 57 5 1 Summary for descriptive statistics ................................ ................................ ...... 59 5 2 Expressed price for organic versus actual price paid for coffee .......................... 66 5 3 Expressed price for organic versus actual price paid for coffee .......................... 67 5 4 P ercentage of the sample per motivation factor for positive WTP ...................... 68 5 5 Percentage of the sample per motivation factor for negative WTP ..................... 68 5 6 Support for certification of organic coffee production ................................ .......... 69 5 7 E stimated OLS model ................................ ................................ ......................... 72 5 8 Estimates for conditional logistic model ................................ .............................. 76 5 9 Comparisons of WTP price premiums between CVM and CE ............................ 81

PAGE 9

9 LIST OF FIGURES Figure page 4 1 Choice set in choice experiment ................................ ................................ ......... 57 5 1 Frequency distribution of consumer WTP for organic coffee .............................. 65 5 2 Average market price for conventional coffee and WTP for organic coffee ........ 65 5 3 Frequency distribution of tax ................................ ................................ .............. 70 5 4 Frequency f or choice of coffee ................................ ................................ ........... 76

PAGE 10

10 LIST OF ABBREVIATION S BDM Becker De Groot CAMAL Coffee Association of Malawi CE Choice Experiment CIA Central Intelligence Agency COOL Country of Origin Labelling DC District of Colombia DD T Dichlorodiphenyltrichloroethane EA Enumeration Area ETEI Emissions Trading Education Initiative GDP Gross Domestic Product GM Genetically Modified GMO Genetically Modified Organisms GOM Government of Malawi IFOAM International Federation of Organic Agric ulture MCCCI Malawi Confederation Chambers of Commerce and Industry MCPCU Mzuzu Coffee Planters Cooperation Union MK Malawi Kwacha MoAFS Ministry of Agriculture and Food Security MSCE Malawi School Certificate of Education MT Metric Tonnes NOAA National Oc eanic Aviation Administration NOP National Organic Program OLS Ordinary Least Squares

PAGE 11

11 SAS Statistical Analysis Software TCC Tobacco Control Commission UNCTAD United Nations Conference on Trade and Development UNIMA University of Malawi US$ United States Do llar USA United States of America WHO World Health Organization WTA Willingness to Accept WTP Willingness to Pay Exchange Rate = MK150/US$

PAGE 12

12 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfill ment of the Requirements for the Master of Science Degree ASSESSING CONSUMER WILLINGNESS TO PAY FOR MALAWI ORGANIC COFFEE: EVIDENCE FROM A CONSUMER SURVEY By Fiskani Esther Nkana August 2010 Chair: Zhifeng Gao Major: Food and Resource Economics Tobac co is a major cash crop for Malawi; however its pe rformance is currently dwindling mainly due to anti smoking lobby by WHO. One of the potential alternatives to tobacco is coffee the seventh largest export crop in Malawi and t h e second most commonly inter nationally traded commodity However consumers are becoming more sensitive to the type of coffee they c onsume. T aste characteristics, label of origin and other unobservable credence attributes (e.g. organic ) are becoming of great concern to consumers It is against this background that demand for organic products is continuously growing worldwide This study therefore aims at assessing Consumer WTP for organic coffee in Malawi and to determine important f actors that influence consumer preference and WTP fo r organic coffee. Data w ere collected from 129 participants through a household survey from three major cities of Malawi ( Blantyre, Lilongwe and Mzuzu ) using the CVM and CE CVM data was analyzed using an OLS model while CE data used a Conditional Logi s t ic Model.

PAGE 13

13 Based on CV M, about 40% of the sample was WTP an average price of MK816.75 per 250g of organic coffee translating into a price premium of MK164.75 per 250g of organic coffee representing about 25% price premium over the average market price for con ventional coffee of MK652 per 250g. T aking into consideration the whole sample, participants were WTP an average price of MK599.66 per 250g of organic coffee which is lower than the average market price of conventional coffee by about 8% Only four variabl es were significant in influencing WTP and these are: actual price paid for coffee, being of 60 years and older, being in the high income group of over MK322, 000 per month, and an attitudinal variable depicting whether an individual thinks that organic pr oducts may offer more of some nutrients than their conventional counterparts. Based on CE people were WTP an average price of MK1 444.38 per 250g of organic coffee translating to a price premium of MK 792.38 per 250 g of organic coffee which represent s o ver 100% price premium over the average market price of conventional coffee. The significant variables that influenced the pro bability of choice of coffee w ere represe nt ing th at individuals did not choose either organic or conventional preference for coffee was mainly based on the method of production rather than price High price pre miums were registered for organic coffee mainly due to health related issues. The majority of the sample opted for government subsidies in organic coffee for increased accessibility. Our results show that there exists a potential niche market for organic c offee in Malawi

PAGE 14

14 CHAPTER 1 INTRODUCTION General Problem Malawi is situated in the southeast of Africa. It is bordered by Zambia to the northwest, Tanzania to the northeast and Mozambique to the east, south and west of the country. The economy of Malawi i s agro based. The agricultural sector in the country employs about 80% of the labor force, contributes over 80% of foreign exc hange earnings and accounts for 39 % of the Gross Domestic Product (GDP) ( MoAFS, 2010). The agricultural sector in Malawi is dualis tic in nature. It has the smallholder sub sector, which contributes more than 70% to the agricultural GDP, and the estate sub sector, which contributes less than 30% to agricultural GDP (MoAFS, 20 10 ). The main food crops that are grown include maize, cassa va, and sweet potatoes, while the main cash crops grown are tobacco, sugar, tea, coffee, macadamia nuts, and cashew nuts (MoAFS, 2010). Tobacco is the major cash crop for Malawi and the major foreign exchange earner as well. Being a country without mineral resources, the crop is normally called the Limbe Leaf and Alliance One including others. The crop accounts for 60% of the (Jaffee, 2003). Therefore, the crop has been crucial for economic growth of the country. Nevertheless, tobacco production is currently dwindling due to a number of factors. For instance in 2007, according to the Tobacco Control Commission (TCC) of Malawi, total production was about 111 million tonnes against 155 million tonnes in

PAGE 15

15 2006. In terms of foreign exchange earnings, the country registered export earnings of about US$160 million in 2006 against US$162.1 million in 2005. Generally production and foreign exchange levels have been erratic for the past 10 years. (Refer to Table 1 1 for detailed estimates). Among others, this is highly attributed to antismoking campaigns led by public health activist s with support of th e World Health Organisation (WHO). To that effect, there are fears amongst stakeholders in the agricultural sector that this development is likely to result in a lot of producers abandoning the crop for other more lucrative ones. Table 1 1 Total annual sa les for t obacco in Malawi Year Volume (Tones) Realization (US$) 1995 130,181 201,562,572 1996 141,662 237,755,361 1997 158,113 248,406,791 1998 1 3 3 996 178,451,093 1999 134,386 186,784,038 2000 159,869 164,734,418 2001 124,669 1 4 3, 880 881 200 2 138,181 163,114,209 2003 13 4 ,326 144,061,678 2004 180,181 347,179,018 2005 145,267 162,061,730 2006 155,098 160,110,819 2007 110,715 195,547,819 2008 194,708 471,583,38 7 Total 2,020, 352 3,190,547,231 Source: TCC In this regard, the Govern ment of Malawi (GOM) has recently laid down a number of strategies aimed at coming up with alternative crops with potential to replace tobacco as a major exchange earner for the country.

PAGE 16

16 One of the policies is crop diversification which is principally impl emented as a risk management strategy. In complement, GOM devoted its efforts to revitalize production a nd marketing of crops with high potential for growth such as coffee, cotton and others whose performance ha ve bee n worsening in the recent past. For ins tance, Arabica coffee which is the major type of coffee grown in Malawi is the seventh largest export crop for the country and remains an essential source of income for farmers. On average, it has an estimated production of over 4,176 mt annually (MCCCI, 2 009). It is grown by both estate and smallholder farmers. The smallholder farmers are concentrated in the highland areas of the northern region of Malawi particularly in the districts of Chitipa, Rumphi, Mzimba and Nkhata Bay. In the Southern region, the i ndustry is dominated by large scale farmers most especially in Thy o lo and Chiradzulu districts. Contrary to its poor performance in the past, the coffee sub sector has seen great improvement in its performance in recent years. For example, in the northern region alone production increased from 2,250 mt in 2007 to 2,600 metric tonnes in 2008 and there are indications of continued improvements in the industry (Chirwa et al., 2008). C offee production areas in Malawi have favourable climatic conditions with a l t itudes ranging from 1,000 to 2,500 metres above sea level. It is for this reason that Malawi coffee has a very fine flavour with a balanced body and acidity In addition, Malawi coffee is gain ing popularity both domestically and globally. One of the most p opular brands sold is both international and local recognition. According to reports, the sale price of Mzuzu coffee has in general gone up register ing a price premium of up to 47%. Malawian coffee is exported to The

PAGE 17

17 Ne therlands, Germany, South Africa, Switzerland, Japan, Australia, United States of America (USA), Italy, among others (MCCCI, 2009) On the other hand, t he world consumption of coffee is projected to increase from 6.7 million tonnes in 1998 2000 to 6.9 mill ion tonnes in 2010 by 0.4% annually (FAO, 2003). In the meanwhile, consumers are becoming more sensitive to the type of coffee they consume. They are specifically conscious about search, experience and credence attributes of a product (as defined by Nelson (1970), and Darby and Karni (1973) ) An example of a search attribute is colour; that for experience attributes is taste while that for credence attributes could be the type of production used to produce a particular product (e.g. organic or fair trade pr oduction). According to a number of market studies, consumers pay much attention to these attributes mainly due to health concerns associated with the products as well as environmental and social justice concerns associated with their production or marketi ng methods. It is for this reason that the world industry for organic products has been growing to meet the growing demand for food with special attributes. Evidently, in 200 7 the global retail sales of organic products increased to US $41.6 billion agains t US$ 23 billion in 2002. Despite this alarming increase in demand, the organic industry still remains undersupplie d worldwide (Willer 200 9 ). Organic coffee has established a niche market within the market for organic foods. The primary markets for the pr oduct are North America and Europe. Currently, the two regions account for about 97% of the global sales for organic products ( Willer 2009 ). It is generally believed that because such countries have more affluent people with higher purchasing power consu mers in these countries are willing to pay high price premiums

PAGE 18

18 for expensive products. In general, consumers in developed countries are willing to pay an average of 15 % to 25% premium for orga nic coffee alone ( Willer and Yussefi 2007). With the current pe rformance of coffee production in Malawi coupled with adoption of Organic Agriculture in most countries in sub Saharan Africa e.g. South Africa, Kenya, Tanzania, Uganda and others, there is a high potential for production of organic coffee in Malawi. Like in the other countries, organic coffee is likely to find niche markets in both, the domestic and international markets th ereby potentially increas ing incomes and national incomes as well. Because coffee is the second most economically important c ommodity in the world after oil, ( Pendergrast 2006) there is a high possibility that the marketing of the crop will be sustainable and hence become one of the most suitable candidates replacing tobacco as the major foreign exchange earner for Malawi. Sp e cific P roblem The critical question for the Government of Malawi to answer before supporting production of organic coffee is whether markets exist for the crop. Particularly, can Malawi o rganic coffee attract higher price premiums from domestic consumers a s compared to conventional coffee? If so, what are the possible factors that would influence the consumer preferences that may determine the market segmentation of the organic coffee market? Th is study attempts to provide answers to these questions and thu s offer valuable information to Malawi an government policy makers seeking suitable substitutes for tobacco as major foreign exchange earners. If the study shows higher Willingness to Pay (WTP) for organic coffee it wi ll suggest that there exists a clear n iche market for organic coffee in the country. Policy makers will therefore be advised accordingly to promote production of organic coffee so

PAGE 19

19 as to meet the potential demand. The potential growth of the domestic market for the crop will therefore be seen a s a step towards targeting international markets that have relatively the highest level of demand for organic products in general. In complement to organic coffee, other crops with great potential for growth could also be promoted. Such a crop diversificat ion strategy is likely to replace tobacco as the major foreign exchange earner for the country. Project Objectives The overall objective of this study is to assess the potential demand for organic coffee in Malawi. Th is will be achieved through the followi ng specific objectives : 1. To assess consumer preference and WTP for organically produced coffee versus conventional coffee 2. To determine important factors that influence consumer preference and WTP for organic coffee. Testable Hypotheses The following hypoth ese s will be tested: 1. Because of the perceived benefits associated with product attributes such as method of production (e.g. organic production), c onsumer WTP and preference for organic coffee is likely to be higher than that of conventional coffee across consumers. 2. Heterogeneous p references exist among consumer and thus t he WTP for organic coffee socio demographic factors e.g. income, gender, age, level of education, including their perceptions.

PAGE 20

20 CHAPTER 2 LITERATURE REVIEW Ove rview of Organic Agriculture Organic Agriculture is increasingly being practised in more than 1 41 countries of the world. There are also strong assumptions that uncertified organic production is being practis ed by more countries (Yussefi et al, 2007). Curr ently, about 32.2 million hectares of land are being subjected under organic production representing 0.8% of the total land for a griculture world wide and the rate is estimated to be increasing ( Willer, 2009 ). The regions with the largest land under organi c agriculture are Oceania, Europe and Latin America The rate of increase is mainly a response to the global market for organic food which is rapidly increasing and mostly constitutes affluent countries (Sahota 2007). As already highlighted in Chapter 1, global sales for organic food increased from US$ 23 billion in 2002 to US$ 41.6 billion in 200 7 (Willer 200 9 ). Although this is the case, the market for organic products still remains undersupplied because of underproduction of organic food globally. To t hat effect, many consuming countries are relying on imports but the supply is still insufficient. The demand for organic products is concentrated in North America and Europe. In 2005, sales of organic products in North America were about US$14.9 billion r dominated by the USA mainly due to the National Organic Program (NOP) of 2002 which propelled growth in the production of organic food s in the country. In recent years, Europ e has overtaken North America as the largest consumer of organic food s and drinks mainly due to the appreciation of the Euro against the US dollar. Revenue for

PAGE 21

21 organic product s in beverages ( Sahota 2007 ). The major countries include Germany, United Kingdom, France and Italy. Other emerging markets are Denmark, Sweden and the Netherlands. It is thus clear that there is a great disparity between production and consumption of organic food s an d products in the world. sustainability is that consumption remains concentrated in Europe and America. There are thus fears that a slight change in consumption patterns in these two regions would likely cause a signifi cant impact on the world production and trade trends. For instance, if these countries decide to stop importing organic products, there is likely going to be oversupply of the products in the producing countries thereby depress ing their prices in those cou ntries and the world at large As such, organic producers are advised to develop their own domestic markets for their organic products rather than rel y on export markets only. Among others, this would be used as a risk management strategy in their business of organic pr oduction and so s ustain organic production world wide ( Sahota 2007) It is partly against this background that organic production is increasing in several Africa n countries where domestic markets are also opening up. Certified production is m ainly being practised in Uganda, Tanzania, Ghan a, Ethiopia, Kenya and Zambia. The main certified crops produced are: Fresh Vegetables, Bananas, Citrus Fruits, Coffee, Tea, Cocoa, Suga r, Cotton and others (Elzakker 2007). Although a large proportion of the production is geared for export markets e.g. North America, Europe and a little bit in Japan, regional markets have recently opened up. The major ones constitute the Republic of South Africa and the Gulf area.

PAGE 22

22 Many studies have concluded that people with higher disposable incomes are the largest spenders o n organic food (Sahota 2007). This conclusion is therefore a threat to the sustainability of domestic markets of organic food in countries with relatively less affluent people. However, a number of studi es have concluded otherwise. Among others, Rodriguez et al. (2007) ascertain that the relationship between income and income among some segments of the market while in other segments the relationship does not exist. Other demographic variables tend to influence WTP for organic products e.g. education, consumer perceptions, age, price, religion, gender (Rodr i guez et al, 2007; Zepeda and Li 2007 ; Peterson et al., 2008; Engel, 2 008). This current study also attempts to assess consumer preference for organic coffee in a domestic market, particularly to determine whether consumers in Malawi are willing to pay a higher premium for organic coffee than the conventional one Pr evious S tudies on Consumer P reference and W TP for O rganic P roducts A number of studies have been conducted on consumer preference and WTP for organic food, of these a few have been done on organic coffee. This section will highlight findings of studies on organic food s in general ; those for non organic products and services will be hi ghlighted in the next section. The non organic products include protection From the literature review, we expect to gain general knowledge on consumer preference and WTP for organic food s Rodriguez et al, (200 7 ) conducted a study on WTP for various organic foods in Argentina. nic products available in the Argentinean domestic market with the view of providing useful

PAGE 23

23 evidence to the government to support promotion of such crops, regulating processes and labelling programs. Data w ere collected from both organic and non organic fo od consumers using the Contingent Valuation Method (CVM). A Binomial Multiple Logistic Regression model was used to estimate the parameters of the targeted products (Regular Milk, Leafy Vegetables, Whole Wheat Flour, Fresh Chicken and Aromatic Herbs) by Ma ximum Likelihood. The dependent variable of the model was WTP for a i ncome r Based on the notion that quality has become a key concept in Demand Theory (Lancaster, 1966; Antle, 1999) among others, the results of the study confirmed this as it turned out that Argentinean consumers were willing to pay price premiums of 6% to 200% in order to acquire the bette r quality products (Rodriguez et al, 200 7 ). Based on the empirical analysis, there was a significant relation between consumer income and the WTP for the organic products in question. The other major factors determining the willing ness to a hindrance to consumer access to products thereby acting as a threat to expansion of the domestic market in Argentina. Another study on WTP for organic food was conducted by Mill ock et al (2002). In his study he used both panel and survey data. Organic food was identified as a product with The food attributes used in his study included environmental concerns, animal welfare and food safety/healthy conce rns. A comparison was made between results drawn from the use of CVM and those from observed WTP. Based on the results, avoidance of

PAGE 24

24 chemicals the highest However, order ing of the valued attributes did not differ at all across organic product types The study conducted both in store interviews and in store experiments on purchases of organic products. The questionnaire that was used had four sets of questions : on purchase habits and food culture (choice of store, impo rtant product characteristics, statements on risks from eating certain foods), questions on organic food production (identification of the Danish Organic label, statements on organic production and its effects), questions on habits and environmental attitu des (use of recycled toilet paper, aluminium foil, membership of environmental associations, WTP for four different products (milk, rye bread, potatoes and minced beef). Ac cording to the study, the majority of the sample was willing to pay more than the stated conventional market price for the products. About 59% of the sample was willing to pay more for the organic milk, 48% for potatoes, 51% for rye bread and 41% for mince d beef. Specific ally the price premium for organic milk was 32.1%, 40.2% for organic potatoes, 23% for rye bread and 18.5% for minced beef. A logistic maximum likelihood was estimated (defined as willingness to pay for all four organic products). It was f ound that about 32% of the sample was indeed willing to pay more for all products. In the same study using the actual purchase data to measure revealed WTP, about 55% of the sample were willing to pay more for organic milk, 35% were willing to pay more for organic rye bread, 14% of the sample were willing to pay more for organic potatoes while 6% were willing to pay more for minced beef. Results from the two methods found that elicited (stated) WTP is overestimated compared to the revealed (real) WTP and in this study the practise was dominant in milk. Surprisingly, for the

PAGE 25

25 other products (organic rye bread, organic potatoes and organic minced beef), consumers were actually paying more than their stated WTP for the products. For future studies, the team expr essed interest in modelling organic foods with demographical variables as independent factors such as income, geographic location, age, etc. Th is research focus ed on this area as one way of WTP for organic products, most especially the organic and fair trade chocolates. It was mainly aim ed at measuring consumer preferences and WTP for organic and/or Fair Trade Labels. Two methods of data collection were used; experimental auction and survey. An experiment was used in order to measure the actual consumer WTP as the method creates a real bi dding set up that reduces any social desirability bias (Noussair et al., 2004). Specifically, the Becker DeGroot On the other hand, the survey was used to collect information to measure elicited WTP Selection of the chocolates by consumers was made on two criteria: hedonic characteristics and the price level of the product. The bidding was done in 13 sessions each containing three stages. The first stage required that participants taste a bar of choc olate without any information of the chocolate; the participant thus attributed a hedonic rating for the chocolate and declared his/her WTP for the chocolate tasted. This was done in order to ing of the product (e.g. taste). On the second stage the consumers declared their WTP for each chocolate based on information on labels as a way of determining their WTP for the label(organic

PAGE 26

26 and fair trade) independent of liking. The last stage involved t asting the chocolates tasted from stage 1 but with comprehensive information. Two of the chocolates had organic and fair trade labels while the other two had neither organic nor fair trade labels. This was done in order to analyse the evolution of arbitrag es between the hedonic evaluation of chocolates and the information provided The results of the study specifically for stage 3 indicate that the organic and fair trade labelled chocolates received the highest bids compared to the standard ones confirming that the information positively. This study also conducted a consumer typology according to the valuation of the organic and fair trade labels. In this regard, the samp le was divided into three clusters based on their WTP for chocolates using comprehensive information provided for the chocolates Cluster 1 represented about 42% of the total sample; it had the least number of women (about 63% against 71% in the total samp le), it included students and people without an occupation with average age of 35 years. Generally, t he group constituted mainly non consumers and occasional consumers of organic products. The second cluster had 41% participants of the total sample; it had about 7 1 % women, had people with average age of 45 years with professions as Executives, commercials sector workers and retired officers Members of this group were regular consumers of organic and fair trade products. The last group had the least partici pants representing only 17% of the sample; it had about 88% women with people aged 32 years on average without specific professions This group constituted participants who consume organic products moderately but fair trade products more occasionally. Acco rding to the study, the first cluster registered the lowest WTP for the chocolates; the second cluster

PAGE 27

27 registered twice as much WTP as the previous cluster while the third cluster registered the highest WTP. or organic food occupation etc. chocolates across the 3 groups. The findings showed that the additional information of the chocolates provided to the participants had varying influences over the valuation process. It was also noted that consumers in cluster 1 were more sensitive to price tha n els but without any conditions while those in cluster 3 were sensitive to labels but conditioned on taste. Lastly, a comparison test was carried out to assess the factors that motivated the consumers in the valuation process. The results showed that group 1 were more influenced by taste and health issues associated with the labels and not the environmental and the social concerns associated with them, to the contrary consumers in group 2 were mainly influenced by the environmental and the social concerns as similar to those in group 2 apart from the fact that they considered taste as the main motivation factor as well. In conclusion, according to the results nearly half of the s ample was insensitive to the organic and fair trade labels This proportion of the sample was mainly motivated by price in its choice of the chocolates, then taste and health related issues and lastly the environmental and social factors associated with the chocolates. The remaining proportion was mainly motivated by the environmental and social concerns associated with the labels as they were able to valu e the products to 20% to 30% of the product price Although the environmental and social concerns wer e

PAGE 28

28 some of the motivational factors, some consumers based their WTP on the liking of the products. The labels just enhanced the valuation process and varied a lot across consumers. The study therefore concluded that consumers were not ready to pay more for organic and fair trade products and so the market for such should not be overestimated. Wikstr m (2003) measured WTP for sustainable coffee (organic and fair trade certified) in Sweden. He also made an attempt to determine the underlying factors for the ch oice of sustainable coffee. In his study he used choice experiments as a method of collecting data and conducted data analysis using a binary probit model. The analysis was based on the neoclassical demand theory. The study targeted 100 respondents who wer e required to make a choice between a number of alternatives of coffee provided to them in the choice experiments In the end, the analysis had a total of 900 observations since each participant was required to make 9 choices. Results of the study showed t hat there was higher WTP for the organic certified coffee as compared to the fair trade coffee although the monetary attribute of the coffee had a significant impact on the consumer utility. The implication of this conclusion is that consumers were willin g to pay higher price premium for the two types of coffee; the choice was made at a minimum cost. On the other hand, social demographic factors and the attitude factors were significantly influential in the consumer choice of the coffee. The regular consum ers of coffee were less likely to buy organic and fair trade coffee due to their high premiums, and consumers who we re aware of environmental concerns, were more likely to buy the coffee than those without the knowledge. The results therefore concluded that there

PAGE 29

29 was an existence of a market for both certified and fair trade certified coffee in Sweden as consumers were willing to pay high price premium s for the products under the study In this ligh t, recommendations were made to organisations in the industry to consider lowering the prices of these products as one way of expand ing the market shares for their brands. In addition, organisations were encouraged to incorporate health benefits of the cof fee in their marketing campaigns as about 20% of the respondents based t heir choice of the two types of coffee on the aspect of reduced levels of chemical substance s in the products The Commission for Environmental Cooperation in 1999 conducted a consumer demand study on Mexican shade grown coffee. Shade grown coffee is most of the times referred to as organic because it is generally grown natural ly, it does not use heavy chemicals. The main purpose of the study was to assess the potential market for Mexi can Shade grown coffee in USA, Canada and Mexico. Data collection was done through personal surveys and focus group discussions where individuals even conducted taste tests of Mexican Shade grown coffee compared with other brands. Based on the results, 22% 42% and 50% of the people interviewed in USA, Canada and Mexico respectively were willing to pay $1 more per pound for the coffee. Consumers were motivated to register WTP for the coffee because of its environmental friendliness. Q uality and taste were a lso key factor s in determining consumer choices As such the study recommended that emphasis should be done on quality and taste of the Mexican shade grown coffee in marketing campaign s Loureiro and Hine (2002) also assessed consumer preference and WTP f or local (Colorado grown) ; organic and GMO free potatoes as one way of discovering their

PAGE 30

30 potential niche markets. The study also focused on determining factors ( socio demographic factors and quality characteristics ) of consumer response to a particular att ribute of the products in question Revealed consumer preference data w ere collected in a survey and the analysis was done using a multiple bounded probit model. T he results show ed that the locally grown potatoes had the highest WTP estimate of 9.37 cents while the organic and GMO free potatoes had 6.64 cents and 5.55 cents respectively. The social demographic variables and quality characteristics had different effects on WTP for the three attributes. For organic and GMO free potatoes, consumers who were se nsitive to freshness and nutrition registered higher premiums; WTP was negatively related to age. In addition, there was also a negative relationship between WTP and the number of children per household. Even though the WTP was the highest for locally grow th potatoes, the only variable that had statistically significant positive d that there wa s a potential niche market for the loca lly grown potatoes in Colorado. Engel (200 8) calculated consumer WTP for major organic products (wine and fruit juice) in South Africa. He used the CVM in data collection. The study used the binary logit model to analy z e consumer decision to purchase organic food or not and the ordered logit model to analy z e the determinants of WTP for organic wine and fruit juice. Based on the results, the significant socio demographic variables influencing the decision to purchase organic food were age, marital status (being married) and level of education. Age p ositively influenced the decision to purchase organic food; marital status had a negative influence over the decision to buy and level of education had a positive influ ence over the decision to buy. The results from ordered logit model

PAGE 31

31 demonstrated that ag e, language (Afrikaans), head of household and citizenship significantly affect ed consumer WTP for organic product s. Specifically language was negatively related to WTP as the majority of Afrikaans speakers have low disposable incomes. Overall, South Africans were willing to pay bid values of $0.25, $0.37 and $0.49 more for organic fruit juice compared to conventional fruit juice For organic wine, the significant ind ependent variables were age (younger and older), Afrikaans, English or language other than Afrikaans, English or Xhosa as the home language and Christian faith. Age was positively related to WTP, Afrikaans and home English speakers were negatively related to WTP and being Christian was positively related to WTP. Peterson et al (2008) made a contribution to the study of demand for non food organic goods. The study products (gloves) made in USA and Austr alia. The attributes used for the research included country of origin, environment focused (organic and pro environment), animal focused and price. In their study, choice experiments were used to collect data and data analysis was based on the conditional logit model. According to results; consumers were willing to pay $1.20 more for a pair of USA wool glove compared to a pair of acrylic ones and WTP for Au stralian wool glove was $0.25. It was noted that the pro environmental label was valued more than the organic label by 14 cents. This could be attributed to the relatively low recognition of organic clothing than food by the participating consumers In addition consumer preferences for

PAGE 32

32 the gloves varied by socio economic and psychographic characteristics e.g. gender, age, income, education attainment, location and beliefs of animal rights. sportswear), Casadesus Masanell et al (2009) used sales data to elicit consumer reveale d preferences and WTP. The study showed that consumers were willing to pay significant premiums for organic cotton clothes regardless of the related costs associated with the apparel. Recently, Hustvedt and Bernard (200 8 ) examined consumer WTP for three cr edence attributes of organic socks made from cotton and corn. The attributes assessed were: origin (imported, US and Texas), type, and production method [conventional, organic and non genetically modified (GM)]. Data were collected through e xperimental auc tions and w ere analyz ed using a Tobit regression model. Bidding in the first round was conducted without information about the credence attributes while in the second round respondents were provided with various attribute information. The model included de mographic variables as possible factors determining WTP for the attributes. According to the results consumers were willing to pay the highest premium of $1.86 for organic socks which w as slightly higher than the premium for non GM socks. Regarding the e ffect of demographics, females were less willing to pay for the U S fibe r s than men, and Hispanics were less willing to pay for organic or non GM fiber. Among others, the study concluded that there is a potential market for organic garments in USA which is in line with the results of Casadesus Masanell (2009) Previous S tudies on Consumer P reference and W TP for N on O r ganic P roducts M ethodologies used in consumer preference studies are vital to the analyses of the current study In this regard this chapte r highlights previous studies on consumer

PAGE 33

33 WTP for non organic products. The major aim wa s to explore the methodologies used in pursuing these studies which later assist ed in the selecti on of appropriate data collection and analysis method s that are most s uitable to the research obj ective s Umberger et al (2002) assessed the consumer preference and WTP for domestic corn fed beef against international grass fed beef. The study targeted two locations in the USA, Chicago and San Francisco. Data w ere collected through panel taste testing and experimental auctions (fourth price Vickrey Auction). The taste test was conducted WTP f or t he preferred steak. The collected data w ere analy z ed using two types of models. Data on preference w ere analy z ed using a multinomial logit model based on random utility theory The dependent variable was n fed beef over grass fed, 1 for those indifferent and 2 for those preferring grass fed beef over the corn fed beef), the independent variables included: location, age, gender, ethnic, income, education, family size and other factors representing character istics of the consumer. In the analysis of dependent variables were the same as that used in t he logit model. The study found that on average, consumers preferred the domestic steak on all sensory qualities and they were willing to pay a 30.6% premium for corn fed beef. To be specific, about 62% of the participants were willing to pay an average pr emium of $1.61 more per pound for the corn fed beef, 23% were willing to pay a premium of $1. 36 more per pound for the grass fed beef and only 15% of the consumers were indifferent. These results show

PAGE 34

34 that there exist respective niche markets for the two t ypes (corn fed and grass fed) of beef as well as beef with country of origin labelling. The demographic factors such as age, ethnic ity beef knowledge and quality grade were seen to have some influence over the flavour preference. However these factors di d not have any influence over the bid difference. It was thus difficult to predict the type of consumers willing to pay for the product they prefer. Mabiso (2005) estimated the WTP for Country of Origin Labelling (COOL) for American fresh apples and tomato es and established the major determining factors for the WTP. In the study, experimental auctions (Vickrey fifth bi d sealed price) were used for data collection. The study used the double hurdle probit model for analysis of the data. The findings indicated that 99% of the consumers were willing to pay $0.49/lb the labels. The demographic and psychographic variables such as food quality Numerous studies on valuation of goods and services have been done in the health sector as well. Dong Churl (2000) measured WTP for pha rmacists' services directed toward reducing the risk of medication related problems. The study also attempted to determine factors that have a significant influence on WTP. Like most of the studies highlighted above, Dong collected data using the CVM. Data analysis was based on logistic regression and semi log regression models. Overall, there was WTP For instance, the mean WTP for pharmacy services ranged from $4.02 to $5.48 per prescription. Of the factors used in the regres sions,

PAGE 35

35 magnitude of risk reduction had an influence on WTP; income was positively related to WTP although it was not statistically significance. Werner et al Data w ere collec ted using two methods. The first one was through experimental auctions and the second approach used a questionnaire with open ended questions. The data collected w ere analyzed using an econometric model with WTP as the dependent variable and psychological factors, social demographic factors and other characteristic variables as independent variables. According to the results, the mean WTP for the treatment was $188.45 (using open ended questions) and some independent variables such as income, age, cognitive status and periods of caring for the sick had a significant impact the WTP Aulong and Rinaudo (2008) assessed population WTP for ground water protection in the Upper Rhine Valley. The valuation was elicited based on two scenarios (restoring drinking wate r quality and eliminating all traces of polluting substances ) They used the standard contingent valuation method and analyzed the data using three models. The logit model was used to assess whether or not participants were willing to pay for proposed scen ario; a linear regression which exclude d some of the variables (protest answers) was used to elicit stated WTP while the Tobit model was used to capture the same but included the variables capturing protest answers. Based on the results, 62% of the respond ents were willing to pay for the first scenario at a mean WTP of $59.6 per household while 52% were willing to pay for the second scenario at a mean WTP of $107.72 per household. It was also noted that some of the independent variables used in the three mo dels were statistically significant. For the linear logic model such

PAGE 36

36 variables include realism of the scenario, number of children in the household, income and the number of known polluting substances. the linear regression a nd the Tobit model ), the statistically significant variables were knowledge of the water bill, income, concerns about groundwater pollution, leisure and use and non use values of groundwater Gao and Schroeder (2009) investigated the effects of a dditional beef steak attributes on consumer WTP in two different US markets. They used Choice Experiment (CE) and analyzed the data using Random parameter logit models. The survey had four questionnaires; two of the questionnaires were aimed at collecting data to test the effect of additional attributes when cue attributes exists while the others tested the effect of additional attributes when no cue attributes are available. Results from both set s of questionnaires showed consistent results of the effect o f additional attribute information on consumer WTP. Based on the results, response of additional attribute information was twofold. In some instances, WTP for the most important attributes decreased when consumers were provided with additional attribute in formation whilst in certain instances it was the opposite, WTP was positively related with additional information of the most important attributes of the study. It was thus concluded that the varying WTP for the attributes was conditioned on the relationsh ip existing between the attributes and the additional ones. Methods of Eliciting Consumer WTP In reviewing literature, there exist a number of techniques that are used in estimation of WTP for products. This includes the CVM, Experimental Auctions, Hedonic Pricing models Conjoint Analysis, and others. Amongst these the CVM has proved to be the most widely used method in many market research studies However,

PAGE 37

37 the Experimental Auctions are more reliable and are currently being used in market research the mo st. Below is a detailed summary of some of the methods. The Contingent Valuation M ethod (CVM) preferences and WTP for products. It is used to attach monetary value to products most es pecially when their markets do n o t exist. The valuation is based on the change in attributes of a particular product such as prices and quality Consumer preference for the product is therefore assessed based on the monetary value attached to it. The valua tion process is also extended to services. The method thus creates a hypothetical market situation for those goods. Through the valuation process, the data collected forms what consumers are willing to pay for a particular product. In CVM, the valuation of the products is done using a questionnaire which is administered through mail, telephone and face t o face interviews. The survey instrument used offers the respondents an opportunity to make an economic decision on the non or market goods (Rahmatian, 2005 ). The valuation process is therefore contingent upon the simulated market presented to the respondents. Product valuation is done through bidding. The bidding takes different formats e.g. open ended questions, bidding game, payment card, dichotomous choic e questions and randomized card sorting In open ended questions, participants disclose their WTP without the use of a starting bid level T he bidding game uses a number of discrete choice questions but one open ended WTP question and also provides a start ing bid value In payment cards, visual aids bearing product monetary values for attribute changes are used while the in dichotomous choice questions researchers use yes or no questions on whether

PAGE 38

38 consumers are willing to pay for a particular product at a certain price In addition, this format uses additional follow up questi on specifying lower bid levels. Carson et al. (1994) doc umented the advantages of CVM. According to him, CVM is a flexible tool for product valuation, it i s easy to apply and cost effe ctive Aulong et al (2008), Werner et al (2002), Dong Churl (2000),Rodriguez et al. (200 7 ), Engel (2008), Millock et al (2002) and others used the method in their respective studies as highlighted in the literature review above. However, being hypothetic the CVM has got a number of flaws. The major one is of response bias which mainly emanates from the use of open ended questions. Mitchell et al. (1989) reported biases on the use of open ended questio ns mainly due to high non response rates. In addition, part i cipants overstate their preferences which most of the times is different from their actual purchase behaviour. Many consumers state high WTP but are less willing to pay the exact amount during ac tual purchases. Nevertheless, based on Rahmatian (2005), CVM are more reliable when one is using test retest (conducting CVM on a different sample of the same population overtime) or when c onvergent validity checks are employed. This compares results obtai ned from CVM with other methods e.g. CE, travel cost or hedonic. These precautionary measures ensure that results based on CVM are more reliable even with the presence of hypothetical biases. After assessing the reliability of the CVM, the National Oceanic Aviation Administration (NOAA) (1993) recommended that the CVM should incorporate the following:

PAGE 39

39 1. The use of face to face interviews 2. The use of WTP as opposed to WTA, 3 Provision of comprehensive information about a product to be valued, 4 The need to remind consumers of the budget constraints they are subjected to in the course of the valuation process, 5 Inform the participants the possible substitutes of the product under valuation, 6 Need to use probing to ensure that respondents unders tand issues being Th is study adopted these recommendations during its implementation The Choice E xperiment (CE) Like the CVM, CE is also a stated preference method based upon the utility model of consumer economics (Lancaster, 1966). In this method, individuals are given a hypothetical setting and asked to choose their preferred alternative among several alternatives in a choice set and they are usually asked to perform a sequence of such choices. Each alternative is described by a nu mber of attributes or characteristics. A monetary value is included as one of the attributes, along with other attributes of importance, when describing the profile of the alternative. Thus, when individuals make their choice, they implicitly make trade of fs between the levels of the attributes in the different alternatives presented in a choice set. Accor ding to Alpizar et al. (2001), there are four major steps that need to be followed when designing Ch oice Experiments and these are: 1. Definition of attr ibutes, attribute levels and customisation 2. Experimental design 3. Experimental context and questionnaire development, and 4. Choice of sample and sampling strategy Within the stated preference models, the CE is currently being widely used. Among ot hers this is highly due to the fact that their use reduces some of the potential

PAGE 40

40 biases created by the use of CVM M ore information is elicited from each respondent compared to CVM and it allows for the possibility of testing for internal consistency. Base d on our literature review, Gao and Schroder (2009), Peterson et al (2008) and Wikstrom (2003) used CE in their respective studies. Experimental A uction Unlike the other two models, Experimental Auction is a revealed preference model. It is used to captur e revealed (actual) WTP for a particular attribute of a product as it creates a real market auction bidding environment The Vickrey sealed bid, second price auction is the most commonly used experimental auction model. This requires parti cipants to submit written bids of a particular product in a real auction environment (Friedman et al., 1994). In a sealed bid, second price auction, bids are ranked from highest to lowest. The highest bidder wins the bid and purchases the product at the second highest pric e. Unlike the CVM and the CE, it is advantageous in the sense that it is designed to reveal true preferences; the use of real money for bidding in additional to other factors like repetitive bidding ensures reliability of results from this methodology. The method also reports less bias by non responses (Fox et al., 1995). However, the major flaw of the Experimental Auction is that it is very expensive to implement. Werner et al (2002), Mabiso (2005), Umberger et al (2002), Hustvedt and Bernard (200 8 ) Didi er and Lucie (2008) and others used Experimental Auctions in order to estimate WTP in their studies (details in literature review above). It is very clear according to published studies that revealed preference models are preferred to stated preference mod els because of their reliability. However, due to the target (sample) of this study and resource constraints the CVM and the CE w ere used as method s of data collection. CVM is flexible and easy to use and most importantly

PAGE 41

41 they are easier to implement in de veloping countries than industrialised countries (Whittington, 1998). The main purpose of multiple valuation technique s is for convergent validity of the estimates. If the results from t he two models (CVM and CE) converge, this is likely to give policy mak ers the confidence to reliably base their decisions on the results. The analysis of data collected from CVM will be based on an Ordinary Least Squares Regression Model while that of CE will be based on a Conditional Logistic Model.

PAGE 42

42 CHAPTER 3 RESEARCH MET HODS AND DATA Theoretical Model consumer cognitive demand theories and random utility models. The r andom utility is based on the assumption that individual utility is a function of observable product attributes individual characteristics and an unobservable random component. Such as ( 3 1) Where X ij is a ro w v ector of independent variables. These variables could represent characteristics specific to the individual and also attributes of the choices ij is a ij is an error term. It is assumed t hat the error term is independently and identically distributed with certain distribution (Greene, 1998). Based on this framework, it is also assumed that a consumer chooses the attribute combination or a product that gives him or her the maximum utility. The CVM and the CE that the study employed are in tandem with According to Thurstone (1927), Random Utility Theory specifically explains the way a consumer makes his choices out of a set of choices provided to him. On the other hand, the Lancaster Theory asserts to disaggregate utilities of products into utilities derived from respective attributes (Lancaster, 1966). In the CVM, the error term is usually assumed to be normally distributed while in the CE, it is normally assumed to follow an extreme maxima value distribution which wi ll result in C onditional L ogit M odels (Hoffman and Duncan, 1988).

PAGE 43

43 Model Estimation As already mentioned above, the current study used two models based on the ty pe of data collection procedure. Ordinary Least Squares Model was used to estimate consumer WTP using data from CVM while the Conditional Logistic Model was used to analyze data collected through the CE method. Model E stimation under CVM Within the framewo rk of Random Utility Theory, WTP is estimated using Ordinary independence and normal distribution of the error terms is thus in line with one of the Gauss Markov assumptions of the OLS which also asserts that the error term be normally distributed. In the estimation of WTP for beef from USA and Argentina, Umberger et al. (2002) used an OLS model She also investigated the impact of socio demographic variables on consumer taste pr eferences and WTP. The current study use d a similar model as below: 2) Where: Y = the bid price premium (the difference between the maximum price an individual is willing to pay for Malawi Organic Coffee a nd the average price of Malawi Conventional coffee ) X i represents the regressors as follows: Actprice = the actual price consumers pay for coffee at domestic market Demographic Variables: Female= gender (1= Female)

PAGE 44

44 Age= (1= 15 24 years, 2=25 39 years, 3=40 59 and 4=60 years and older) Education = (1=at least degree education, 2= at least college certification but no degree, 3= Secondary School qualification (M.S.C.E) and 4=Elementary School certificate) Income= (1=MK66, 000 and below, 2= MK67,000 MK321,000 and 3=MK322,000 and above) Denomination=(1=Presbyterian, 2= Catholics, 3=Anglican, 4=Pentecostals, 5= Sevethday Adventist, 6=Muslims and 7= Other denominations) Attitudinal V ariables : OrgFertr = if one believes that organic coffee is grown without the u se of fertilizers (1=strongly agree, 2=agree, 3=uncertain 4 =disagree,5=strongly disagree) Orgchm = if one believes that organic coffee is grown without the use of pesticides or chemicals (1=strongly agree, 2=agree 3 =uncertain, 4=disagree 5 =strongly dis agree) Orgnat = if one believes that by buying organic products you are supporting natural and healthiest way to grow crops ( 1=strongly agree, 2=agree, 3=uncertain 4 =disagree,5=strongly disagree ) Orgsup = if one believes that by buying organic coffee you are supporting farmer s ( 1=strongly agree, 2=agree, 3=uncertain 4 =disagree,5=strongly disagree ) Orgris = if one believes that by drinking organic coffee there is lower risk of ingesting chemicals ( 1=strongly agree, 2=agree, 3=uncertain,4=disagree,5=stro ngly disagree ) Orgnut = if one believes that organically grown food may offer more of some

PAGE 45

45 nutrients than conventional counterparts ( 1=strongly agree, 2=agree, 3=uncertain, 4=disagree, 5=strongly disagree ) Orgtas = if one believes that organically grown food have better taste than c onventional counterparts ( 1=strongly agree, 2=agree, 3=uncertain, 4=disagree, 5=strongly disagree ) Model 3 2 is the original model with 1 3 independent variables. Out of the 1 3 variables, 1 2 O O O O O O form would create problems most especially in interpretation of the estimates of the coefficients of the polytomous variables (those taking more than two levels) This is so because the intervals of the levels may not be standard except for the attitudinal variables whose intervals are assumed to be standard in our study Dummy coding of these variables wa s therefore essential in order to address this problem. In dummy coding, e ach of the polytomous variable s apart from the attitudinal variables was made binary, thus taking the value of either 1 or 0. Instead of having 1 3 independent variables this modification resulted in h aving a total of 2 7 independent variables in the model Nineteen (1 9 ) of these were dummy variables, seven were discrete (polytomous) while only one was continuous. T radition all y the model with 2 7 regressors as explained above is normally run by dropping one of the dummy variables per each group of the polytomous variables in order to set it as a base group. Among others, this addresses the problem of dummy variable trap and so makes the model easy to estimate. In this case, the intercept of the

PAGE 46

46 estimated model assumes the value of all five of the base groups Interpretation of the estimates of the coefficients of the dummy variables in the model is thus done in comparison to the intercept. However, it becomes so complicated to make such interpretations in a situation whereby the base groups are more than one as one is laboured to remember all of them when making the comparisons among other thing s. According to Jauregui (2007), t he best approach is therefore to perform an effect coding of the dummy variables. With the effect coding the intercept takes the value of the average household of the sample instead of a particular base group hence interpretation of results becomes easier For variable, which has three levels according to model 3 2 In effect coding the first stage require s that the variable be decomposed into three dummy variables representing each of the three levels and thi s is represented in equation 3 3 below. 3) Where y = dependent variable, D i are the dummy variables representing the three income categories i are estimates of the coefficients of the dummy variables. The process way that the sum of their respective coefficients is equal to zero at the mean of the dummy variables as in E quation 3 4 : 4) where represent the mean of the respective income dummy variables This implies that / ) / 5) If Equation 3 5 is inserted in Equation 3 3 the following is yielded:

PAGE 47

47 / + / ) 6 ) This eventually is transformed into: / + / ... ( 3 7 ) let ( / ) = DIncome 1 representing the restricted variable f or t and ( / ) = DIncome 2 representing the restricted variable Model 3 7 is what is eventually run as the model with restricted variables. The dummy variable of the third income category ( D 3 ) is what has been dropped from the original model to avoid dummy variable trap When all the discrete variables are at their means 1 2 equal zero (0 ). This is what transforms the intercept to represent the average household of the sample. The process was replicated to the other polytomous variables and we finally run a model containing all variables as outlined in Model 3 2 Instead of having thir t een ( 1 3 ) variables as in Model 3 2 we eventually had 2 3 (27 4 = 23 variables) in our final model. The variables add up to 23 because we factored out 4 variables each from the set of the polytomous variables to avoid dummy variable trap. It should therefor e be noted that (Female) was not restricted because it is not a polytomous variable; interpretation of its coefficient will therefore be in comparison to its base group (Male) and not the average household. The attitudinal variables were not restricted as well. I nterpretation of the se variables will therefore be done similarly to continuous variables

PAGE 48

48 based on the assumption that the ir respective intervals 5 for strongly dis agree) are constant as explained above Model E stimation und er C E The second model that was estimated in our study is the Conditional Logi s t ic Model. The model is based on the Lancaster ation. Based on this theory, a consumer chooses the product that maximizes his or her utility. The prob ability that an alternative j is chosen among J alternatives is : ) for all other (Greene, 199 7 ). Assuming T ij is independent and identically distributed following an extreme maxima value distribution. In this case, the probability of an alternative j can be chosen as below: Prob [Y i =choice j] = Where Y i is a random variable that indicates the choice made by the i th individual and is e qua preference; V ij is the utility an individual obtained and is determined by individual specific characteristics and product attributes In our current study, consumer utility is assumed to be a function of price and the m ethod of p roduction of a product such as: Where X j1 is the price of coffee, X j2 is the dummy of organic coffee.

PAGE 49

49 CHAPTER 4 DATA COLLECTION Data collection was conducted using two stated preference methods, namely: CVM and the CE. The multiple valuation techniques were used mainly for convergent validity of the estimates from the two models. In order to conform to the recommendations of NOAA (1993), face to face household interviews were conducted in data collection with an aim of reducing hypothetical bias when one is using CVM. According to Cochran ( 1977), a formula for determining a sample size expressed as a percentage is; 1) where t 2 = the standard deviation sco re that represents the probability level of a variable of falling within a confidence interval when the variable is normally distributed (p)(q) = Variance j 2 = confidence interval The following are the results after incorporating our data variables into the formula: 384 The probability level and confidence interval of 1.96 and 0.05 respectively were used as these are the commonly used estimates and normally accord estimation process efficiency. The variables making up the variance represent the proportion of consumers and non consumers of coffee according to our study. Since it was difficult to source the

PAGE 50

50 specific estimates for these proportions, Czaja and Jonny (1995) recommends that a 50% proportion for each is id eal. We thus needed to collect a sample size of about 384 to represent the target population of our study. However, due to budgetary constraints, time and other factors, the study managed to collect a sample size of 129. This is still a significant figure considering that it i s still a large sample and it was randomly collected. The survey therefore targeted 129 participants in the three main cities of Malawi. These are: Blantyre, Lilongwe and Mzuzu in the Southern, Central and Northern regions respectively The sample was randomly selected using the Systematic Random Sample Method in order to reduce response biases. This was based on a sampling frame collected from the Malawi National Statistics Office (NSO) of the 2008 Census Household list. According to t he methodology used by NSO in conducting surveys, the country is divided into clusters and then further broken into Enumeration Areas (EAs). Maps of these clusters including their respective EAs were used to locate the households that were selected in the sample. Selection of households was done in such a way that diverse income categories of the Malawi population be represented in our sample. This was done by first dividing our target population into three major clusters representing the low, middle and hi gh income groups. For instance, households of people with high income levels were selected from the low density clusters while those of people with low incomes were selected from the high density clusters. Households of participants of the middle class wer e selected from the clusters in the middle of the two. It should however be noted that collection of data on household incomes was not limited to the three main income groups, ten (10) categories of household income data were collected guided by the three main categories (Refer A ppendices A and B). This

PAGE 51

51 information was used for descriptive statistics. On the other hand, the ten categories were compressed into three income brackets that were eventually used for estimation of the OLS. The survey targeted peop le of at least eighteen ( 18 ) years of age although the minimum age captured in the survey instrument used as per A ppendices A and B is fifteen ( 15 ). This was done so because the study adopted the age categories normally used in surveys conducted in Malawi It should be emphasized that only those with a minimum age of 18 were targeted in the survey. Apart from the age, there were no other restrictions in terms of the characteristics of an individual so long as they were able to speak either the local langua ge or English. The survey targeted both consumers and non consumers of coffee. The non consumers were interviewed most especially to get their perceptions on organic coffee. Table 4 1 presents some of the demographic structure of our sample in relation to the population terms of gender, age and religious affiliation. Table 4 1. Sample r epresentat i veness in terms of the demographic structure of the population of Malawi (gend er, age and religion) Demographic Variable Category Sample (%) Population (%) Gender Male 32% 49% Female 68% 51% Denomination Christians 98% 80% Muslims 2% 13% Age 15 64 years 98% 52% 65 years and older 2% 3% Source: For Population figu res Gen der Census Report 2008 and Denomination and Age CIA World Fact book For Sample figures

PAGE 52

52 Our sample estimates do not converge with the population estimates although they both portray a similar pattern. The distribution of gender is sk ewed towards females for both the survey sample and the population. Sixty eight percent (68%) of our sample are women while 51% of the population are females. This could be highly attributed to the fact that the interviews were mostly conducted during work ing h ours when most men were on duty. In addition, most men insisted that their wives be interviewed in situations were both were available to make sure that there was a good rapport since all interviewers were also females. In terms of denomination, both structures show that the country is dominated by Christians. About 98% of our sample were Christians while 80% of the Malawi population are Christians. Among others, this could be due to the fact that the Muslim population is not very significant in the ci ties (Blantyre, Lilongwe and Mzuzu) that the survey targeted. The dominant denominations were Presbyterians and Catholics representing 36% and 20% of the sample respectively. This is in line with CIA World Fact book which states that 80% of Christians in M alawi constitute the major denominations of Presbyterians and Catholics with the former taking a higher percentage. According to age distribution, our target population were people of at least 18 years S ince the comparison is done with categories provided by the CIA World Fact book; this leaves greater propor tion of our sample in the age category (15 64 years). However, on the elderly population (65 years and older), the estimates are close to each other. About 2% of our sample were participants belonging to this age group which is close to 3% of the population of Malawi. Based on this comparison, our sample qualifies to be representative of the Malawi population.

PAGE 53

53 Design of CVM The CVM is a survey instrument used to obtain preferences of respondents in mone tary values for changes in the price or quality of a particular good or services (Engel 2008). The current study used a well structured questionnaire to collect information from participating consumers. In order to save on time both CVM questions and CE q uestions were captured on the same instrument (R efer A ppendices A and B for detailed survey instruments). The survey instrument s had five main parts in all parts discrete questions were asked except for the one that captured the WTP for organic coffee Th e first part had questions aimed at collecting information related to consumer consumption pattern of coffee Participants were asked whether they drink coffee or not, in what quantities they take the coffee (how many cups per day) and when they normally t ake the coffee (breakfast, lunch, supper or in between meals). T he second part contained questions that were aimed at information on their perception of o rganic products or organic a griculture in general In this section a number o f sentiments related to organic products and production were read out and the consumer was required to either strongly agree or just agree or were not sure. T he third part was an open ended question that asked consumers to provide their WTP for organic coffee. Before giving out the WTP estimate, a definition of organic products/production was read out to participants to make sure they have the knowledge of the coffee to be valued. Two packets of coffee were then shown to the participants ; one was conventional while the other was organic by assumption (this was done because Malawi does not grow organic coffee yet) The packaging was done i n such a

PAGE 54

54 way that all coffee attribut es were held constant i.e. brand, quantity, taste, country of origin, etc., apart from type of production and price. Participants were then asked to assume they were in a grocery shop to make a coffee purchase. They were then asked to bid for the organic c of the CVM that the survey adopted. The bench mark price that was used in the bidding process was the average market price for conventional coffee in Malawi which was MK 652 per 250g The average price was calculated using retail coffee prices collected from the Consumer Association of Malawi. The survey did not use the actual price paid f or coffee on the domestic market as reported by the survey participants, a s the bench mark price because there was a possibility th at some of the coffee purchased could be organic though imported. The WTP question that was asked by participants during the bidding session was; how willing are you organic coffee was thus revealed through the calculated price premium for organic coffee (Expresse d WTP for organic coffee minus a verage market price for conventional coffee (MK 652 per 250 g). If the price premium was positive (if a participant was willing to pay more than MK 652 for organic coffee ) the implication was that he preferred organic to conventional and if the price premium for a participant was negative ( he /she was willing to pay less than MK 652 for organic coffee ) the implication was he preferred conventional coffee to organic. Consumers were also required to give out the reasons to back up their willingness to pay for the organic coffee. These reasons acted as their motivation for their preference f or coffee. After giv ing out the reasons, the participating consumers

PAGE 55

55 were asked to provide information of the actual price they pay for coffee on the domestic market. T he fourth category of the questionnaire had the CE questions whose details will be elaborated in the subsequ ent section of CE design. The last part had discrete questions aimed at captur ing socio demographic information of the participants. These included age, marital status, income, level of education, number of children under 18 years staying in their househol d, occupation (whether they work with an environmental related organisation) and denomination. Lastly consumers were asked to choose an approach that would contribute towards the production of organic coffee in Malawi from the three approaches given ( leavi ng every cost of production to producers through government subsidies or through consumer taxes), if one chose through taxes he/she was required to estimate the rate of tax he would desire to contribute. Design of the CE As already highlighted, the fourt h part of the questionnaire had questions of the CE. The survey had two types of questionnaires (Versions A and B), in which the order of the product s in a choice set, and the order of the choice sets were different to avoid certain types of order effects. Each participant was required to answer one version of the questionnaires. Design of t he CE was based on the fractional factorial design in SAS to maximize the D efficiency. In the choice experiment, each respondent w as asked to choose between and at corresponding price level. In preferred any type of coffee. Due to the fact that organic products are relatively new to Malawi an

PAGE 56

56 consumers, Orga nic Coffee with lower price was not treated as one dominant choice. Therefore, the choice experiment include d some sets of choice options with Organic Coffee having prices lower than C onventional Coffee counterpart. The final choice experiment composed of 13 sets of choice options with the D efficiency of 63%. The lower D efficiency resulted from the use of 7 level real coffee prices in the CE design. The prices used in the CE were prices for the major brands of coffee for the past four months (May t o Augus t 2009) in Malawi. Less price level may increase the D efficiency of the CE design, but the real prices we used make the choice scenario facing respondents more realistic. In addition, the learning efforts of the respondents were reduced which may be more important than small improvement in the statistical efficiency. In order to eliminate the potential order effect, the order of the Organic and Conventional Coffee were changed after reaching a certain choice set in the choice experiment. For instance, the order was changed after reaching the 7 th choice set under Version A and it changed after the 6 th choice set under Version B. Refer Figure 4 1 for a n example of a choice set in the CE A n example of the combination and ordering of prices is as in Table 4 2

PAGE 57

57 Please choose a 250 grams packet of coffee as you are shopping in the market, or choose None option if you are not satisfied with both coffees. Figure 4 1. C ho ice set in choice e xperiment Table 4 2. Price l evels for c offee in MK/250 grams (Version B) Organic c offee Conventional c offee 795 729 795 720 729 795 729 485 720 699 720 660 Conventional c offee Organic c offee 795 699 699 699 720 660 485 485 4 75 485 729 475 475 475 Organic Coffee Conventional Coffee None Price MK720/ 250 g MK660/250 g I Choose

PAGE 58

58 CHAPTER 5 RESULTS OF THE EMPIRICAL ANA LYSIS This chapter presents empirical analy z es of our data. The presentation will be twofold; the first section will highlight results based on CVM and the second one will include those from the CE methodology. In each respective section descriptive statistics will first be presented before results from the models. It should be noted that amongst the 129 individuals that were interviewed in the survey, nobody refused to divulge any informati on that was required. As such, there were no incomplete questionnaires hence 100% response rate. This could be due to the use of face to face interviews that gave enough room for probing, clarification of questions, and others One refusal was encountered but this was addressed by selecting the appropriate alternative household using the random sampling method that was employed under the study. As already explained under the section of data collection, the survey collected five categories of information fro m the respondents the next section outlines descriptive statistics of our sample. Descriptive Statistics This section gives out the major summary of descriptive statistics of the sample (R efer to T able 5 1 for details). The survey interviewed a total of 1 29 p articipants in the three major cities of the country. About 31.78% of the respondents were from the city of Blantyre while Lilongwe and Mzuzu cities had the same percentage of about 34.11% of the sample. Out of the 129 respondents, 31.78% were male wh ile the majority about 68.22 %, were female. This could be attributed highly to the fact that the population distribution of Malawi is skewed towards women and secondly the interviews were conducted during

PAGE 59

59 working hours when most men were at work. This is b ased on the fact that most women in developing countries are unemployed due to low levels of literacy when compared to men among other things. Table 5 1. Summary for descriptive s tatistics No. Name of v ariable Variable c ategory No. of p articipants Percenta ge of s ample 1. Gender Male 41 31.78% Female 88 68.22% 2. Marital Status Married 68 52.71% Divorced 11 8.53% Single 43 33.33% Other (divorced and w idowed) 7 5.43% 3. Children u nder 18 years old (multiple answer) Under 2 years 28 21.71% 2 to 5 years 42 32.56% 6 to 12 years 56 43.41% 13 to 18 years 82 63.57% None 24 18.6% 4. Age 15 to 19 years 23 17.83 20 to 24 years 24 18.6% 25 to 29 years 20 15.5% 30 to 34 years 19 14.73% 35 to 39 years 10 7.75% 40 to 44 years 10 7 .75% 45 to 49 years 4 3.1% 50 to 54 years 7 5.43% 55 to 59 years 6 4.65% 60 to 64 years 4 3.1% 65 years and older 2 1.55%

PAGE 60

60 Table 5 1. Continued No. Name of v ariable Variable c ategory No. of p articipants Percentage of s ample 5. Education C ompleted p ost graduate degree 2 1.55% Comp leted university undergraduate d egree 3 2.33% Attended university u ndergraduate 3 2.33% Completed college d egree 4 3.1% Completed college d iploma 19 14.73% Attended some c ollege 15 11.63% Some pos t secondary technical s chool 7 5.43% Completed h igh school c ertificate (e.g. MSCE) 29 22.48% Attended some high s chool (e.g. MSCE) 29 22.48% Completed elementary/p rimary s chool 11 8.53% Attended some elementary/primary s chool 7 5.43%

PAGE 61

61 Table 5 1. Continued No. Name of v ariable Variable c ategory No. of p articipants Percentage of s ample 6. Income MK 15,000 and below 20 15.5% MK16,000 to MK 66,000 44 34.11% MK67,000 to MK117,000 19 14.73% MK118,000 to MK168,000 16 12.4% MK169,000 to M K219,000 6 4.65% MK220,000 to MK270,000 2 1.55% MK271,000 to MK 321,000 2 1.55% MK322,000 to MK423,000 2 1.55% MK424,000 to MK525,000 7 5.43% MK526,000 and above 11 8.53% 7. Occupation(i f one works with an environmental related organization ) Yes 16 12.4% No 113 87.6% 8. Denomination Presbyterians 47 36.43% Catholics 26 20.16% Anglicans 2 1.55% Pentecostals 23 17.83% Seventh Day Adventist 14 10.85%

PAGE 62

62 Table 5 1. Continued No. Name of v ariable Variable c ategory No. of p articipa nts Percentage of s ample Muslims 3 2.33% Other denominations 14 10.85% 9. City Blantyre 41 31.78% Lilongwe 44 34.11% Mzuzu 44 34.11% 10. Drink coffee Yes 110 85.27% No 19 14.73% 11. Frequency of coffee c onsumption One cup a day 41 37.27% Two cups a day 49 44.55% Three to five cups a day 6 5.45% Six to ten cups 0 0% More than ten cups a day 0 0% Other 14 12.73% 12. Drinking time (multiple) Breakfast 100 90.9% Lunch 6 5.45% Dinner 9 8.18% In between meals 62 56.36% As already alluded to, t he survey targeted people of at least 18 years of age. According to our sample the age group that registered the largest number of

PAGE 63

63 respectively of our sample. The 65 years and older age category registered the least number of respond ents at about 1.55% of the sample. In terms of education, our target group include d individuals with at least an Elementary or Primary School qualification. According to our sample, the majority were individuals with the highest Secondary School qualification (M.S.C.E) and those who attended Secondary School but have Junior Certificates of Education qualification (J.C.E) both representing 22.48% of the total respondents. About 14.73% of the sample completed their College Diploma while 11.63% have College Certificates. The category of individuals with post graduate school qualification ha d the least number of respondents representing just 1.6% of th e total sample. We had 1 0 group of monthly net earnings. The income category with the highest number of 34.11% of the sample followed by the minimum income group ( MK117,000 ( ( ( ( ( ( US$2,146.67) to MK 423,000 ( representi ng 1.55% of the total sample. Malawi has a variety of denominations and quite a number of them were represented in our sample. The proportion of Presbyterians were t he highest in all the

PAGE 64

64 cities representing about 36.43% of the total sample followed by Catholics who represented 20.16% of the sample. The least were Anglicans with a representation of only 1.55% of the sample. Based on the sample only 12.4% of the total r espondents work with organisations dealing with organic farming, food safety and environmental related against 87.6% who are not associated wi th such type of organisations. In terms of coffee consumption pattern which is very crucial in this study, the maj ority of the respondents drink coffee representing 85.27% of the total sample against only 14.73% who do n o t drink coffee at all. Of the majority, most people take two cups of coffee a day representing 44.55% of the coffee consumers and mostly at breakfast and in the evening. About 12.73% of the coffee consumers do n o t drink coffee often, just once in a while. No one in the sample takes more than five cups a day. Willingness to Pay for Organic C offee Based on the CVM, about 40% of the sample were willing to pay a high price premium for organic coffee against 57% that were not willing to pay high price premiums for organic coffee. About 3% were indifferent. (Refer to Figure 5 1). Based on our results, the 40% were willing to pay an average price of MK 816.75 per 250 g of organic coffee representing a price premium of MK164.75 per 250 g which indicates a 25% price premium over the average market price for conventional coffee of MK652 per 25g ( R efer to Figure 5 2). On the other hand, the 40% reported that they a ctually paid an average price of MK539.87 per 250 g of coffee (either conventional or organic) found on the domestic market. This demonstrates that these consumers were willing to pay an extra MK 276.89 for 250 g of organic coffee representing a 51% price increase for organic coffee on top of the reported actual price for coffee. (Refer Table 5 2).

PAGE 65

65 Figure 5 1. Frequency di stribution of consumer WTP for organic c offee Figure 5 2.Average m arket p rice for c onventional c offee and WTP for organic c offee

PAGE 66

66 Tab le 5 2. Expressed p rice for o rganic versus actual p rice p aid for c offee 1 Description Average p rice (MK/250g) % Increase o ver r eported actual p rice p aid f o r c offee Reported a ctual p rice paid for c offee 2 539.87 0% Expressed price for organic c offee 816.75 51% Note : 1 The table is for the 40% of the sample 2 Average price paid for coffee on domestic market as reported by survey respondents However, taking into consideration every p articipant of the survey, our results show that there was no willingness to p ay for organic coffee. According to Table 5 3 below which summarizes results for all respondents (not just the 40% of the sample who were willing to pay high price premium s for organic coffee); individuals were willing to pay an average price of MK599.65 per 250 g of organic coffee. This price is below the average market price for conventional coffee by MK 52.34 per 250g. Regarding the average price paid for coffee on the domestic market as reported by survey respondents, participants actually paid an aver age price of MK483.56 per 250g of coffee Taking into consideration the total sample our results also indicated that respondents were willing to pay an additional MK116.09 for 250 g of organic coffee on top of the reported actual price paid for coffee rep resent ing an extra 24% price increase over the reported price (MK 483.56 per 250 g) This is similar to the 40% of the sample who were also willing to pay an extra price for organic coffee on top of the actual price they pay for coffee signifying a relativ ely high value they attached to organic coffee.

PAGE 67

67 Table 5 3. Expressed price for organic versus a ctual price paid for c offee Description Average p rice (MK/250g) % Increase o ver r eported a ctual p rice p aid f o r c offee Reported a ctual p rice paid for c offee 488. 56 0% Expressed price for organic c offee 599.65 24 % Note : These estimates are for total sample (including the 40%) Motivation for the WTP for Organic C offee The survey instrument used in the study also collected information that enabled us to distinguis h the specific reasons that prompted respective consumers to value organic coffee with high price premiums as well as low price premium compared to the conventional coffee. These questions were multiple in natures in the sense that consumers were allowed t o give out as many reasons as possible. According to Table 5 4 below, the major reason that motivated consumers to register high price premiums for organic coffee might be the health issues associated with organic products. This is most especially due to t he fact that they are grown without the use of synthetic fertilisers About 33% of the respondents based their motivation on health related issues while 24% based it on the s pecific fact of low use of chemicals in organic production. These findings are similar to those by Rodriguez et al. (200 6 ), Wikstrom (2003), Loureiro and Hine (2002), Didier and Lucie (2008), and Zan oli and Naspetti (2001). In these studies WTP and prefere nces for organic products w ere influenced to a significant extent by health related issues associated with organic products. On the other side, the major reason for low WTP for organic coffee was mainly due to the fact that participants believed that organ ic production ought to be cheaper

PAGE 68

68 than conventional production since it does not use inorganic fertilizers that are relatively expensive than organic manure. About 50% of the sample attributed their low will ingness to pay to this reason (R efer Table 5 5). Table 5 4 Percentage of the sample per motivation factor for p ositive WTP Reason for p ositive WTP Percentage of the s ample Avoid possible chemical substances 24% It give s value for money 3.9% To support local farmer 16.28% Organic coffee has purer ta ste 9.3% To protect environment 12.4% It makes me different 0% I feel better 3.1% Health related r easons 33.3% Table 5 5 Percentage of the sample per motivation factor for n egative WTP Factor for n egative WTP Percentage of the s ample I can no t affo rd o rganic 10.9% Factor for n egative WTP Percentage of the Sample I do n o t know how it taste 0.78% I do n o t care what type 0% No need to change coffee habits 1.6% Its cheap to produce 49.6% S upport towards Organic C offee P roduction Information on so me of the aspects that contribute to the high price premiums of organic coffee was shared at the end of the interview. One of the aspects is the certification process which is relatively costly compared with other issues factored into the cost of producti on. Participants were then asked to choose the best approach they

PAGE 69

69 felt could contribute towards the certification fee associated with organic coffee. According to Table 5 6 below about 75% of the sample felt that the contribution should be made through go vernment subsidies in order to make it more accessible by many people, 14% chose to contribute through consumer taxes while 11% thought the whole cost should be left to the producers themselves. A significant proportion of the 11% pointed out that eventual ly the cost will still be transferred to the consumer through high price premiums so found the option of taxes to be similar to the one they chose and secondly they felt government should be relieved as it is already subsidising a number of industries (lik e the agricultural input subsidy programme). Table 5 6 Support for c ertification of organic coffee p roduction Variable c ategory No. of p articipants % of the sample Cost to be left to p roduc er a lone 14 10.85% Consumer t axes 18 13.95% Government s ubsidie s 97 75.19% According to Figure 5 3, the 14 % of the participants were willing to support organic production through taxes of an average of 2%. Empirical Analysis of Data from CVM The regression model that was run to estimate consumer WTP for organic cof fe e using the CVM methodology is Model 3 8 as specified in chapter 3. The process of how we arrived at this final model will not be explained in this section as it has already been well elaborated in chapter 3. However, it should be noted that not all vari ables described under the descriptive summary were included in the model. It should also be highlighted that some of the categories appearing in Table 5 1 above are not appearing in the regression model as they were compressed into lesser categorical level s mainly due to

PAGE 70

70 the fact that frequency distribution was scanty across the categories and also to avoid having too many dummy variables in our model that could significantly reduce the degrees of freedom of the regression model. For instance in Table 5 1, variable had also 11 categories but these were compressed into 4 categories as well in d these were divided into three income categories in the regression model. Figure 5 3. Frequency distribution of t ax Prior to model estimation, all 23 regressors were tested for collinearity. The variables were correlated with each other but no problem o f multicollinearity existed Most of the coefficients of correlation between the variables were below 0.5 and a few close to 0.7 such as dummy variables within the same group e.g. amongst the

PAGE 71

71 education and denomination groups In addition, the number of ou r variables is in such 129>2 4 This supports one of the Gauss Markov assumptions of no perfect collinearity; hence our model can be estimated by OLS (Wooldridge J.M, 2009 pg. 86 ). The data that w ere collected in the current study is cross sectional; as such it is likely to h ave the problem of heteroskedasticity. Since our sample size is large enough, we did not conduct a special test of heteroskedasticity. Instead, we reported heteroskedasticity robust standard errors to address the problem of heteroskedasticity. This is a co nvenient approach of addressing heteroskedasticity that data may be subjected to without even knowing its form (Wooldridge 2009). Table 5 7 therefore represents the results of our estimated model. Based on Table 5 7 our R 2 is 0.233, indicating that about 2 3 .3% variation in the dependent variable can be explained by the regressors in the model. The F Statistic is quite large and significant at 5% level. As such, it supports the R 2 for the purposes of both prediction of our model and explanation of relation ship between the in dependent variables (Marti, 2008). The significance of the F Statistic implies that at least one of the variables in the model was able to explain our dependent variable (diffwtp). As already stated in chapter 3, our model has 23 indepen dent variables, these were expected to explain the dependent variable in our model. Out of these 23 variables which represent 13 original variables, four variables were statistically significant. These are: actpric (actual price for coffee), Dage_d (60 yea rs and older), Dincome_c (high income of over MK322, 000 per month) and Orgnut (attitudinal value depicting as to

PAGE 72

72 Table 5 7 Estimated OLS m odel Name of v ariable Coefficient of v ariable Robust standard error T s tatistics P v alues Actpric 0.175 0.0 70 2.52 0 0.013 ** Female 58.85 1 47.331 1.24 0 0.216 Dage_b 37.71 9 27.41 7 1.38 0 0.172 Dage_c 2.76 5 45.12 3 0.06 0 0.951 Dage_d 122.10 8 57.78 2 2.11 0 0.037 ** Deduc_b 20.287 28.598 0.71 0 0.480 Dedu_c 4.11 3 22.260 0.18 0 0.854 Deduc_d 40.31 6 48.43 5 0.83 0 0.407 Dincome_b 34.32 6 28.72 9 1.19 0 0.235 Dincome_c 132.082 53.06 3 2.49 0 0.014 ** Ddeno_pres 6.44 3 30.01 3 0.21 0 0.83 0 Ddeno_cath 28.02 1 42.82 8 0.65 0 0.514 Ddeno_angl 129.28 8 87.47 4 1.48 0 0.142 Ddeno_Pent 1.02 6 38.221 0.03 0 0.979 Ddeno_Sev 8.42 8 52.92 0 0.16 0 0.874 Ddeno_Isl 53.916 64.213 0.84 0 0.403 Orgfert 13.198 17.651 0.75 0 0.456 Orgchm 3.595 17.160 0.21 0 0.834 Orgnat 11.12 6 25.816 0.43 0 0.667 Orgsup 2.174 31.122 0.07 0 0.944 Orgris 11.10 6 22.436 0.49 0 0.622 Orgnut 32.30 9 18.633 1.73 0 0.086 Orgtas 0.744 20.411 0.04 0 0.971 n = 129 F Statistic = 1.83 0 ** 5% 10% significant levels R 2 = 0.233 whether an individual thinks organic products offer more of some nutrients than their conventional counterparts). This is in lin e with a number of studies that conclude that consumer demographic variables are key determining factors of WTP for organic products. These include studies by Peterson et al. (2008), Mabiso (2005) and others.

PAGE 73

73 Our findings therefore support the second hypot hesis of our study that consumer WTP for organic coffee is influenced by consumer socio demographic variables. Price (ac t pric) was significant at 5 % level and had a positive sign although its coefficient was quite small. It means if price of conventional c offee paid by respondents increased by 100 units the consumer WTP for organic coffee would increase by MK1 7 5 1 per 250 g. A number of studies on WTP have also concluded that price is an important determining factor of WTP for organic product (Rodrigue z et al. (200 7 ); Wan g and Sun (2003); Thomas (2009). Age (60 years and older) was significant at 5 % level of significance and it has a negative coefficient which is relatively large compared to that of price. An individual of 65 years of age or older would be l ikely to pay a price premium of MK 1 22.11 per 250 g of organic coffee less than an average individual. This could be attributed to the fact that organic production is a relatively new concept in Malawi and so individuals of this age group are not conversan t with their benefits. It is so surprising because elder people are expected to be sensitive with the type of foods they eat as most of people falling in this age group are more prone to diseases like cancer than the average individual. Engel (2008) found Peterson et al. (2008) and Loureiro and Hine (2002) concluded that age had a significant impact on W TP as well. Dincome_c (high income of over MK322, 000 per month) was highly significant at 5 % level just like price. Income is positively impacting our dependent variable and has a high economic impact on the dependent variable evident by the large st coeff icient

PAGE 74

74 estimate as expected. If an individual belongs to the high income bracket, he is likely to pay a price premium of MK1 32.08 per 250 g of organic coffee more than an average individual. This supports the notion that organic products are normally deman ded by people with high affluence ( Willer and Yussefi, 2007). In a related development, a number of studies have come up with a similar conclusion. Peterson et al. (2008), Dong Churl (2000), Werner et al. (2002) and Aulong et al. (2008) found that income h ad a significant effect on WTP for respective products under valuation in their studies. Nevertheless, Rodriguez et al (2007) reported that the relationship between income and WTP is very controversial. Orgnut ( whether an individual thinks organic products offer more of some nutrients than their conventional counterparts) is significant at 10% level assuming a negative value. It therefore means that the more an individual agrees that organic products offer more of some nutrients than their conventional coun terparts, the more the individual is willing to pay high price premiums for organic coffee. For instance if an individual increases his extent of agreement to this sentiment by one level (e.g. from just agree to the higher level of strongly agree the individual is likely to increase his WTP for organic coffee by MK 32.31 per 250 g of the coffee. This relationship was expected since out of experience an individual would attach a relatively high value to a product which is more nutritious than the one wi th low levels of nutrition. In h is study of Rodriguez et al. (2006) were the key and better determinants of WTP for the products than the socio demographic variables of the consumers. Our study ther efore confirmed this although the other six attitudinal variables were insignificant.

PAGE 75

75 Gender, Education all denomination variables and six of the attitudinal variables (o rgfert, orgchm, orgnat, orgsup, orgris and orgtas) were insignificant in our study. L ikewise psychographic and socio demographic variables of consumers did not have any significan t influence on WTP for beef (Umberger et al. 2002). Rodriguez et al. (2007) even alluded to the fact that th e relation between education and willingness to pay is also controversial just as with income. However, in his study, there was a significant positive relationship between lower levels of education and WTP for organic products. In addition, Peterson et al (2008), Engel (2008), Zepeda and Li (2007) and others also reported that there were significant relationships between some of the demographic variables including gender and education in their respective research. Engel (2008) found that there was a significant positive relationship between being Christian an d WTP for organic fruits, and religious affiliation was also one of the significant factors determining WTP in a study by Zepeda and Li (2007). Empirical Analysis of Data from CE As already highlighted above, the survey conducted had 129 observations In t he Choice Experiments, an individual was asked to choose one option amongst the three o ptions provided to him/her. He/s he was supposed to choose either organic coffee, conventional coffee or the N option. The experiment had 13 choice sets which meant that the process of choosing the preferred coffee was done 13 times. In the analysis conducted, each option in a set was turned into an independent observation a tota l of 5028 observations in the choice experiment. A test of collinearity was done amongst the independent variables (Organic, price and None) and the results showed that no variable was p erfect collinear to each other.

PAGE 76

76 Descriptive S tatistics Based on Figure 5 4, about 54% of the p articipants expressed their preference for organic coffee, about 40% chose conventional coffee while only 6% were indifferent. Figure 5 4. Frequency for choice of c offee Results from Conditional Logit M o del A Conditional Logistic Regression Model was used to analy z e data collected from the CE Below is a table of the estimated model: Table 5 8 Estimate s for conditional l ogi s t ic m odel Variable Coefficient Z s tatistics P v alue Organic 3.129 33.77 0 0.000 ** Price 0. 00 4 5.73 0 0.000 *** None 3.520 7.41 0 0.000 *** n=5028 LR=2533.95 0 P v alue = 0.000 ***1% significant level 40% 54% 6% Organic Conventional None Frequency of Choice of Coffee

PAGE 77

77 The estimated model (Table 5 8 ) has a likelihood ratio of 2534, its P value is very low (0.000) suggesting that the overall mo del is highly significant. It therefore implies that at least one of the independent variables used was able to influence our dependent variable (probability of choosing a type of coffee). All of our variables were statistically significant with very small P Values (P<0.000). Based on our model, the coe 3 ; this implies that holding the other independent variables constant, the ional coffee increased b y 3.1 3 It therefore means that the likelihood for the choice of organic coffee was higher than that for conventional coffee. The variable ha d a coefficient which was larger than This indicates that consumers based their coffee preferences m ore on the method of production of the coffee than the other attributes such as price Similarly, in another study a certain segment of participants were more sensitive to labels (organic and fair trade) than price and taste of the product s bearing the l abels (Didier and Lucie, 2008). In addition, Thomas (2009) found that the method of production among other variables contrary, Boxall et al. (2007) concluded that price and n ot the method of production was a significant determinant for increased probabilities that a consumer would purchase organic bread. C ontrary to our findings, both price and the method of production were t o buy organic carrots (Thomas 2009). On the other hand, the 0.004. Price is therefore the regressor with the smallest coefficient estimate among the three

PAGE 78

78 regressors used in our model This means that with a u nit increase in the price for coffee, there wa s a decrease in the log of odds in favor of a consumer choice f or a particular type of coffee of only 0.004 holding other regressors constant Based on this, price did not have a large econo mic impact on consumer choice for coffee. The participants were thus less sensitive to price when they were making their respective choices for coffee. Our findings are both in support and in contradictory to a number of previous studies. Thomas (2009) con cluded that price was insignificant in though it was s although price was significant, group 1 of pa rticipants were more sensitive to price and least to labels of the organic chocolates under valuation in a study by Didier. and Lucie ( 2008). O ur findings point out that method is more influential in affecting consumer choice for coffee tha Lastly, the 3.52; this is the largest compared with that of the other independent variables. Based on this, it means t hat the log of odds in favor of the option N ( that a consumer would not choose either of t he two types of coffee ) decreased by 3.52 holding other variables constant This makes sense since only about 6% of organic while others chose conventional coffee. Based on the analysis, an attempt was done to calculate the WTP for organic coffee as below: 1) Where a 1 is the coefficient for organic coffee, and

PAGE 79

79 b 1 is the coefficient for price Based on E quation 5 1 the WTP for organic coffee was MK782.25 per 250g which represents the price premium consumers were willing to pay for 250 g of organic coffee. It therefore means that consumers were willing to pay an average price of MK 1,434.25 per 250 g of organic coffee (MK 652 + 782.25) which represent s over 1 00% price premium for organic coffee over the average market price f or conventional coffee of MK 652 per 250 g of coffee Our results really show that a segment of consumers were indeed willi ng to pay high price premiums for organic coffee compared to conventional coffee The preference for organic coffee was mainly made based o n its type of production and to a lesser extent efits related to organic production and its products. Th ese findings therefore support our first hypothesis which states that attributes such as method of production (e.g. organic production) consu According to past research, it is very clear that product attributes impact the purchasing behavior of consumers differently. In our study, the method of production of coffee Studies conducted by Didier and Lucie (2008) are in support of our results. Casadesus Masanell et al (2009) also concluded that WTP for organic garments w as registered by consumers regardless of other related costs (including price) associated with the

PAGE 80

80 garments. However, consumer behaviour is not only influenced by product attributes ; o ther key factors that influence consumer choice of products are consumer characteristics In our study, these were only factored in the OLS regression that was run prior to the Conditional Logistic Model. H owever, some studies used Generalized Logit Models having both product attributes and consumer characteristics as factors influencing consumer behavior in purchasing a number of products. This could give out more realistic results as it would allow for interactions of the two types of regressors in influencing consumer behavior among other things. C ompari son of Results from C VM and CE Based on our findings, it is very clear that results f rom the two models used did not converge (Refer to Table 5 9) below B ased on the CVM, the total sample of the survey was not willing to pay high price premiums for organic coffee. Participant s were willing to pay an average price of MK599.65 per 250g of organic coffee which is lower than the average market price for conventional coffee by MK52.35 per 250 g. However, the 40% were willing to pay high price premiums for organic coffee with a n ave rage price of MK816.75 per 250 g of organic coffee This represented a price premium of MK16 4 .75 for 250 g of organic coffee over the average market price of conventional coffee On the other hand, according to CE, consumers were willing to pay an average price of MK 1,434.25 per 250 g of organic coffee representing a price premium of MK 782.25 per 250g of organic coffee which represent s over 100% price premium over the average market price for conventional coffee. The divergence of the results could be due to the obvious fact that the methodologies are different yielding two different results. However, convergence of results particularly for overall sample, would have given policy makers the confidence to

PAGE 81

81 adopt the results and use them to make well informed decisions regarding organic coffee. Adoption of the results from the respective methodologies should therefore be treated with caution since at this stage it is tricky to conclude which ones could be close to the reality. This therefore calls for need to Table 5 9 Comparisons of WTP price p remiums between CVM and CE Methodology Type of c offee WTP (MK per 250g) CVM a Organic (52.34) CVM b Organic 16 4 .75 CE c Organic 782.25 Price premium for overall sample and is negative Price premium for 40% of sample who expressed high price premiums fo r organic coffee Price premium for overall sample

PAGE 82

82 CHAPTER 6 CONCLUSION Our study analyzed consumer WTP for Malawi Organic Coffee. The study used two different methodologies of eliciting WTP which were both stated models. These inclu de CVM and CE. Information from participants was collected using a survey targeting 129 participants. Data from CVM was analyzed using an OLS model while that from CE was analyzed using a Conditional Logistic Model. O ur results showed that there is WTP for organic coffee in Malawi Based on CVM, about 40 % of the participants were willing to pay a price premium of MK16 4 .75 per 250 g of organic coffee Nevertheless, when responses of all participants were considered participants were not willing to pay high price premiums for organic coffee. They were willing to pay MK52.34 less per 250 g of organic coffee than the average market price for conventional coffee. Based on CE, participants were willing to pay a price premium of MK 782.25 per 250 g of organic coffe e. Many studies have indeed drawn similar results that certain segments of consumers have a WTP for organic products that is higher than their conventional counterparts. These include Rodriguez et al. (2007), Engel (2008), Wikstrom (2003), Casadesus Masane ll et al. (2009), Dong Churl (2000), Werner et al. (2002) and Aulong et al. (2008), just to mention s ome However, some studies have contradicting results. For instance, WTP was higher for fair trade products than organic p roducts in studies by Didier and Lucie (2008), and Loureiro and Lotade (2005). Similarly, WTP for organic potatoes was not higher than that for local potatoes (Loureiro and Hine, 2002). Results from the two models both confirm and contradict results from previous studies. For instance, ou t of the 2 3 independent variables in the OLS model, only four

PAGE 83

83 were statistically significant. These are: actpric (actual price for coffee), Dage_d (6 0 years and older), Dincome_c (high income of over MK322, 000 per month) and Orgnut (attitudinal variable d epicting whether participants think that organic products offer more of some nutrients that their conventional counterparts ) This indeed confirms studies by Peterson et al. (2008) and Mabiso (2005) including many others that assert that consumer demograph ic variables are key determining factors of WTP for organic products although Rodriguez et al. (2006) concludes otherwise. In his study he reported socio demographic variab les Gender, Education, all denomination variables and six of the attitudinal variables (Orgfert, orgchm, orgnat, orgsup, orgris and orgtas) were insignificant in our study. This is in line with Umberger et al. 2002 who concluded that psychographic and so cio demographic variables of consumers did not have any significant influence on WTP. Rodriguez et al. (2007) even alluded to the fact that relation between education and willingness to pay is also controversial According to the Conditional Logistic Model organic price were significant variables influencing the choice for coffee. Organic had a higher economic impact than price over the choice of coffee which impl ies that the choice for coffee was highly influenced by the type of production of coffee. This confirms previous studies by (Didier and Lucie 2008) and Thomas (2009) that the method of production was more important in influencing consumer decision to buy organic products than the price attribute.

PAGE 84

84 The major reason that motivated in dividuals to register higher price premiums for organic coffee than conventional coffee was health related issues. Most people (about 33%) believed that organic coffee is a healthy drink since its production does not use inorganic fertiliz ers. In order to support organic production through contribution towards certification fee of the product, about 75% of the individuals opted for government subsidies, 14% opted to contribute through taxes of 2% on average while 11% chose to leave it all to the producer. B ased on our results consumer choices of coffee are based on the method of production and to a lesser extent the price of coffee. Generally people feel that organic coffee is healthy compared to conventional coffee. Therefore, it is very likely that there exists a niche market for organic coffee in Malawi. Study Limitations Overall, the study had a number of limitations that were brought about due to mainly resource constraints (budgetary and time). This resulted into challenges as follows: First and foremo st, the use of Stated Preference Models allowed for some hypothetical biases which could have been reduced if the Revealed Preference Models were used. Initially the study wanted to use Experimental Auctions as a method of data collection which renders re sults that are more reliable than those from Stated Preference Models (e.g. CVM and CE). However, since this is so expensive to implement, we opted to use both the CVM and CE which are relatively cheaper. The two models were used for purposes of convergen t validity check of results.

PAGE 85

85 Secondly, the study wanted to target over 300 participants but we only interviewed 129 people due to budget constraints since we were using face to face household surveys. A larger sample size is ideal in order to capture a num ber of participants with divergent characteristics to represent the population and hence improve the efficiency of the statistics used in making inferences Lastly, the study wanted to target the USA and EU market s since they form the largest market for or ganic products in the world. However, due to time constraint it was not possible to conduct the survey in the USA hence the current focus of targeting the domestic as a step forward towards targeting the international market. Recommendations and Further Re search The study recommends the following to be implemented in order to come up with more reliable results on WTP for organic coffee in Malawi and thereafter to promote organic production in Malawi. A similar study should be conducted using Revealed Stated Preference Models (preferably Experimental Auctions). It should also consider using over 300 observations to make sure that all segments of consumers in Malawi are represented in the sample. This will ensure that the results are more reliable. A similar s tudy should also be conducted targeting the international market, preferable the USA and Europe since the two form the largest market for organic products in the world. It would thus be feasible to produce organic coffee targeting the international market in addition to the domestic market since the international market fetches higher piece premiums for organic products than the domestic market due to the fact that the international market has consumers that are affluent.

PAGE 86

86 If the government decides to adopt production of organic coffee there will be need for a coherent policy for organic production in Malawi. Within this broad policy framework, there will be need for a specific strategy for organic coffee production whereby responsibilities of the main actor s in the industry will be spelt out. This includes the Government as a policy regulator, donors as financiers, producers, processors, marketers, just to mention a few. Proper strategies on transferring the technology to farmers need to be develop ed in the policy including proper strategies to support the production to ensure that the type of production is sustainable and is contributing significantly to economic growth of the agricultural sector i n Malawi and the economy as a whole. Since organic products a re generally more expensive tha n conventional ones, the Government should consider implementing subsidies in organic coffee production among others, to make sure that the products are accessible by many people at affordable price (Rodriquez et al. 2006).

PAGE 87

87 APPENDIX A S URVEY INSTRUMENT (VE RSION A) Evidence from a Consumer Survey ID for Respondent Supervisor Remarks by Supervisor PART A: Consumption Pattern for Coffee: This part attempts to assess 1. Do you drink coffee? (1) Yes (2) No (If no, skip to part b) 2. How often do you drink coffee? (1) One cup a day (2) Two cups a day (3) Three to five cups a day (4) Six to ten cups (5) More than ten cups a day (8) 3. When do you normally drink the coffee? (Check all that apply) (a) At Breakfast (b) At Lunch (c) At Dinner (d) In between meals

PAGE 88

88 Part B: Quiz on General Knowledge of Organic Coffee and Coffee in general over organic coffee as well as coffee in general. In order to assess the attitude, the options of the answers will be one of (Strongly Agree, Agree, Uncertain, Disagree, and Strongly Disagree) 4. ty behind only petroleum. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 5. Many coffee producing countries use highly toxic chemicals that have been banned or restricted in many countries (e.g. DDT). (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 6. Organic coffee is grown without the use of synthetic fertilizers. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree

PAGE 89

89 (5) Strongly Disagree 7. Organic coffee is grown without the use of any p esticides or chemicals. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 8. By buying organic products you as a consumer are supporting the natural and healthiest way to grow crops (1) Strongly Agree (2) Agree (3) Uncertain (4 ) Disagree (5) Strongly Disagree 9. By buying organic coffee you are supporting the small holder farmer. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree

PAGE 90

90 10. By drinking organic coffee there is lower risk of ingesting synthet ics or chemicals. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 11. O rganically grown food may offer more of some nutrients than their conventionally produced counterparts. (1) Strongly Agree (2) Agree (3) Uncertain (4) D isagree (5) Strongly Disagree 12. O rganically grown food have better taste than their conventionally produced counterparts. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 13. Europe and North America form the largest mark et for organic products. (1) Strongly Agree (2) Agree

PAGE 91

91 (3) Uncertain (4) Disagree (5) Strongly Disagree 14. Malawi grows organic coffee which is also sold as an export crop. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree Par t C: Questions on WTP Definition of Organic Coffee: Organic coffee is coffee that has been certified as having been grown without the use of inorganic fertilizers, synthetic pesticides, herbicides, or other chemicals. It can also refer to farms which in corporate socially responsible activities such as recycling composting soil health and environmental protections. 15. The average price for Malawi conventional coffee is MK 652 per 250 g. How much are you willing to pay for organic coffee per 250 g? 16. If the WTP is positive, why would you be willing to pay more for it? (Check all that apply) (a) To avoid possible chemical substances in my coffee (b) The organic c offee will give me the most value for the money (c) To support local farmers (d) Its got a purer taste (e) To help protect environment (f) It makes me different from people drinking conventional coffee (g) I feel better to drink organic coffee (h)

PAGE 92

92 17. (Check all that apply) (a) (b) how it tastes (c) (d) I see no reason to change my coffee habits (e) 18. In your last purchase, how much did you pay for a 250 g of coffee? Part D: Choice Experiments (CE) In this category, you are required to choose either 250 grams packet of organic coffee or conventional coffee as you are shopping in the market, or choose None option if you are not satisfied with both coffees. You will do this for 13 combinations. Combination 1 Conventional Coffee Organic Coffee None Price MK795/ 250 g MK699/250 g I Choose Combination 2 Conventional Coffee Organic Coffee None Price MK699/ 250 g MK699/250 g I Choose Combination 3 Conventional Coffee Organic Coffee None Price MK720/ 250 g MK660/250 g I Choose Combination 4

PAGE 93

93 Conventional Coffee Organic Coffee None Price MK485/ 250 g MK485/250 g I Choose Combination 5 Conventional Coffee Organic Coffee None Price MK475/ 250 g MK485/250 g I Choose Combination 6 Conventional Coffee Organic Coffee None Price MK729/ 250 g MK475/250 g I Choose Combination 7 Conventional Coffee Organic Coffee None Price MK475/ 250 g MK475/250 g I Choose Combination 8 Organic Coffee Conventional Coffee None Price MK 795/ 250 g MK729/250 g I Choose Combination 9 Organic Coffee Conventional Coffee None Price MK720/ 250 g MK660/250 g I Choose Combinatio n 10

PAGE 94

94 Organic Coffee Conventional Coffee None Price MK729/ 250 g MK795/250 g I Choose Combination 11 Organic Coffee Conventional Coffee None Price MK720/ 250 g MK699/250 g I Choose Combination 12 Organic Coffee Conventional Coffee N one Price MK729/ 250 g MK485/250 g I Choose Combination 13 Organic Coffee Conventional Coffee None Price MK795/ 250 g MK720/250 g I Choose Part E: Questions on Socio Demographics 31. Gender of the respondent: (1) Male (2) Female 32. What is your marital status? (1) Married (2) Divorced (3) Single

PAGE 95

95 (8) 33. Do you have children living in your household that fall into these age categories? (Check all that apply) (a) Under 2 years (b) 2 to 5 years (c) 6 to 12 years (d) 13 to 18 years (e) None 34. How old are you? (1) 15 to 19 years (2) 20 to 24 years (3) 25 to 29 years (4) 30 to 34 years (5) 35 to 39 years (6) 40 to 44 years (7) 45 to 49 years (8) 50 to 54 years (9) 55 to 59 years (10) 60 to 64 years (11) 65 years and older 35. (1) Completed post graduate degree (Masters or Ph.D) (2) Completed University Undergraduate Degree (3) Attended University Undergraduate (4) Completed College Degree (5) Completed Coll ege Diploma (6) Attended Some College (7) Some Post Secondary Technical School (8) Completed High School Certificate/Secondary School (e.g. MSCE)

PAGE 96

96 (9) Attended Some High School/Secondary School Certificate (e.g. MSCE) (10) Completed Elementary/Primary Schoo l (11) Attended Some Elementary/Primary school 36. What is your net monthly total household income? (1) MK15, 000 and below (2) MK16, 000 to MK66, 000 (3) MK67, 000 to MK117, 000 (4) MK118, 000 to MK168, 000 (5) MK169, 000 to MK219, 000 (6) MK220, 000 to MK270, 000 (7) MK271, 000 to MK321, 000 (8) MK372, 000 to MK423, 000 (9) MK474, 000 to MK525, 000 (10) MK526, 000 and above 37. Do you work with an organization that deals with issues related to Organic farming, food safety and other environmental relate d issues? (1) Yes (2) No (If No, go to 39) 38. 39. You belong to which denomination? (1) Presbyterian (2) Catholic (3) Anglican (4) Pentecostal (5) Seventh day Adventist (6) Islam (8)

PAGE 97

97 40. In which city do you reside in? (1) Blantyre (2) Lilongwe (3) Mzuzu 41. Organic coffee may bring relatively high incomes to a farmer and the nation as a whole and thus likely be one of the alternatives for tobacco as a major fore ign exchange earner for the country. However, the certification fee of organic coffee may be high. Who do you think should pay for the higher cost related to organic coffee certification? (1) The producer as part of his/her cost of production (2) The Consumer through taxes (3) The Government through subsidies 42. If answer is (2) in 41, how much tax would you be willing to pay in support of organic coffee certification?

PAGE 98

98 APPENDIX B S URVEY INSTRUMENT (VE RSION B) Evidence from a Consumer Survey ID for Respondent Supervisor Remarks by Superviso r PART A: Consumption Pattern for Coffee: This part attempts to assess 1. Do you drink coffee? (1) Yes (2) No (If no, skip to pa rt b) 2. How often do you drink coffee? (1) One cup a day (2) Two cups a day (3) Three to five cups a day (4) Six to ten cups (5) More than ten cups a day (8)

PAGE 99

99 3. When do you normally drink the coffee? (Check all that apply) (a) At Breakfast (b) At Lunch (c) At Dinner (d) In between meals Part B: Quiz on General Knowledge of Organic Coffee and Coffee in general coffee in ge neral. In order to assess the attitude, the options of the answers will be one of (Strongly Agree, Agree, Uncertain, Disagree, and Strongly Disagree) 4. only petroleum. (1) Strongly Ag ree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 5. Many coffee producing countries use highly toxic chemicals that have been banned or restricted in many countries (e.g. DDT). (1) Strongly Agree (2) Agree

PAGE 100

100 (3) Uncertain (4) Disagree (5) Stron gly Disagree 6. Organic coffee is grown without the use of synthetic fertilizers. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 7. Organic coffee is grown without the use of any pesticides or chemicals. (1) Str ongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 8. By buying organic products you as a consumer are supporting the natural and healthiest way to grow crops

PAGE 101

101 (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 9. By buying organic coffee you are supporting the small holder farmer. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 10. By drinking organic coffee there is lower risk of ingesting synthetics or chemicals. (1) Strongly A gree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree

PAGE 102

102 11. O rganically grown food may offer more of some nutrients than their conventionally produced counterparts. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 12 O rganically grown food have better taste than their conventionally produced counterparts. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree 13. Europe and North America form the largest market for organic products. (1) Str ongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree

PAGE 103

103 14. Malawi grows organic coffee which is also sold as an export crop. (1) Strongly Agree (2) Agree (3) Uncertain (4) Disagree (5) Strongly Disagree Part C: Questions on WTP Definitio n of Organic Coffee: Organic coffee is coffee that has been certified as having been grown without the use of inorganic fertilizers, synthetic pesticides, herbicides, or other chemicals. It can also refer to farms which incorporate socially responsible activities such as recycling composting soil health and environmental protections. 15. The average price for Malawi conventional coffee is MK 652 per 250 g. How much are you willing to pay for organic c offee per 250 g? 16. If the WTP is positive, why would you be willing to pay more for it? (Check all that apply) (a) To avoid possible chemical substances in my coffee (b) The organic coffee will give me the most valu e for the money (c) To support local farmers (d) Its got a purer taste (e) To help protect environment (f) It makes me different from people drinking conventional coffee

PAGE 104

104 (g) I feel better to drink organic coffee (h) 17. (Check all that apply) (a) (b) (c) whether the coffee I buy is organic or not (d) I see no reason to change my coffee habits (e) 18. In your last purchase, how much did you pay for a 250 g of coffee? Part D: Cho ice Experiments (CE) In this category, you are required to choose either 250 grams packet of organic coffee or conventional coffee as you are shopping in the market, or choose None option if you are not satisfied with both coffees. You will do this for 1 3 combinations. Combination 1 Organic Coffee Conventional Coffee None Price MK 795/ 250 g MK729/250 g I Choose Combination 2 Organic Coffee Conventional Coffee None Price MK795/ 250 g MK720/250 g I Choose Combination 3 Organic C offee Conventional Coffee None

PAGE 105

105 Price MK729/ 250 g MK795/250 g I Choose Combination 4 Organic Coffee Conventional Coffee None Price MK729/ 250 g MK485/250 g I Choose Combination 5 Organic Coffee Conventional Coffee None Price MK720/ 250 g MK699/250 g I Choose Combination 6 Organic Coffee Conventional Coffee None Price MK720/ 250 g MK660/250 g I Choose Combination 7 Conventional Coffee Organic Coffee None Price MK795/ 250 g MK699/250 g I Choose Comb ination 8 Conventional Coffee Organic Coffee None Price MK699/ 250 g MK699/250 g I Choose Combination 9 Conventional Coffee Organic Coffee None

PAGE 106

106 Price MK720/ 250 g MK660/250 g I Choose Combination 10 Conventional Coffee Organic C offee None Price MK485/ 250 g MK485/250 g I Choose Combination 11 Conventional Coffee Organic Coffee None Price MK475/ 250 g MK485/250 g I Choose Combination 12 Conventional Coffee Organic Coffee None Price MK729/ 250 g MK475/250 g I Choose Combination 13 Conventional Coffee Organic Coffee None Price MK475/ 250 g MK475/250 g I Choose Part E: Questions on Socio Demographics 31. Gender of the respondent: (1) Male (2) Female

PAGE 107

107 32. What is your marital statu s? (1) Married (2) Divorced (3) Single (8) 33. Do you have children living in your household that fall into these age categories? (Check all that apply) (a) Under 2 years (b) 2 to 5 years (c) 6 to 12 years (d) 13 to 18 years (e) None 34. How old are you? (1) 15 to 19 years (2) 20 to 24 years (3) 25 to 29 years (4) 30 to 34 years (5) 35 to 39 years (6) 40 to 44 years (7) 45 to 49 years (8) 50 to 54 years (9) 55 to 59 years (10) 60 to 64 years (11) 65 years and older 35. Wh (1) Completed post graduate degree (Masters or Ph.D) (2) Completed University Undergraduate Degree

PAGE 108

108 (3) Attended University Undergraduate (4) Completed College Degree (5) Completed College Diploma (6) Attended Some Co llege (7) Some Post Secondary Technical School (8) Completed High School Certificate/Secondary School (e.g. MSCE) (9) Attended Some High School/Secondary School Certificate (e.g. MSCE) (10) Completed Elementary/Primary School (11) Attended Some Elementary/ Primary school 36. What is your net monthly total household income? (1) MK15, 000 and below (2) MK16, 000 to MK66, 000 (3) MK67, 000 to MK117, 000 (4) MK118, 000 to MK168, 000 (5) MK169, 000 to MK219, 000 (6) MK220, 000 to MK270, 000 (7) MK271, 000 to MK 321, 000 (8) MK372, 000 to MK423, 000 (9) MK474, 000 to MK525, 000 (10) MK526, 000 and above 37. Do you work with an organization that deals with issues related to Organic farming, food safety and other environmental related issues? (1) Yes (2) No (If N o, go to 39) 38. 39. You belong to which denomination? (1) Presbyterian

PAGE 109

109 (2) Catholic (3) Anglican (4) Pentecostal (5) Seventh day Adventist (6) Islam (8) 40. In which city do you reside in? (1) Blantyre (2) Lilongwe (3) Mzuzu 41. Organic coffee may bring relatively high incomes to a farmer and the nation as a whole and thus likely be one of the alternatives for tobacco as a major foreign exchange earner for the coun try. However, the certification fee of organic coffee may be high. Who do you think should pay for the higher cost related to organic coffee certification? (1) The producer as part of his/her cost of production (2) The Consumer through taxes (3) The G overnment through subsidies 42. If answer is (2) in 41, how much tax would you be willing to pay in support of organic coffee certification?

PAGE 110

110 LIST OF REFERENCES Agresti, A. and Finlay, B. (2009) Statistical Methods f or the Social Sciences (Fourth Edition). Prentice Hall, Inc. New Jersey. Alpizar F. et al. (2001) Using Choice Experiments for No n Market Valuation. Working Papers in Economics No. 52. Goteborg University, Sweden. Annual Gross Margin Analysis Report, MCP CU, 2010. Annual Sales Data, Tobacco Control Commission (TCC). http://www.tccmw. com/Annual%20sales%20data.htm ( Accessed in August 2009 ). Annual Production Data, Tobacco Control Commission ( TCC). http://www.tccmw. com/downloads.htm#production ( Accessed in January 2010 ) Antle, J. (1999) The New Economics of Agriculture. American Journal of Agricultural Economics, Proceedings, 81 (5): 993 1010. Aulong S. and Rinaudo, J. (2008) Assessing the Benefits of Different Groundwater Protection Level: Results and Lessons Learnt from a Contingent Valuation Survey in the Upper Rhine Valley Aquifer, France. Blend, J. and Van Ravenswaay, E. ( 1998) Consumer Demand for Ecolabelled Apples: Survey Methods and Descriptive Results. Staff Paper 98 20. Department of Agricultural Economics, Michigan State University. Boxall, P. et al. (2007) The Role of Sensory Attributes and Information on the Willin gness to Pay for Organic White Bread. University of Alberta, USA. Breslow, N.E. (1974) Covariance Analysis of Censored Survival Data. Biometrics 30: 89 95. Bruin, J. (2006) Newtest: Command to Compute Newtest. UCLA: Academic Technology Services, Statist ical Consulting Group. http://www.ats.ucla.edu/ stat/stata/ado/analysis/ Buzby, J., Ready, R. and Skees, J. (1995) Contingent Valuation in Food Policy Analysis: A Case Study of Pesticide Re sidue Risk Reduction. Journal of Agricultural and Applied Economics, 27(2):613 625. Carson, R.T. et al. (1994) Contingent Valuation and Lost Passive Use: Damages from the Exxon Valdez. Resources for the Future Discussion Paper. QE 94 18, Washington D.C.

PAGE 111

111 Casadesus Masanell Economics and Management Strategy, Vol. 18, Issue 1, pp. 203 233. Chirwa E. et al. (2008) S mallholder Coffee Commercialization in Malawi, Policy Brief, Future Agricultures. CIA World Fact Book (ISSN 1553 8133) https://www.cia.gov/library/publications/the wor ld factbook/geos/xx.html ( Accessed on 28th July 2010 ). Cochran, W.G. (1977) Sampling Techniques (3 rd Edition) New York: John Wiley and Sons. Corsi A (2007) Ambiguity of Measured WTP for Quality Improvements When Quantity is unconstrained: a Note. Oxf ord University Press and Foundation for the European Review of Agricultural Economics Advance ( Access published online on November 26, 2007 ) Czaja, R. and Jonny, B. (1995) Designing Surveys: A Guide to Decisions and Procedures, Pine Forge Press. Darby, M.R. and Karni, E. (1973) Free Competition and the Optimal amount of Fraud. Journal of Law and Economics, 16, 67 88. Didier T. and Lucie S., and Fair Trade Products. International Journal of Co nsumer Studies, 32(2008) 479 490, Blackwell Publishing Ltd. Dong Churl S. (2002) Reduce Risk of Medication Related Problems. Journal of the American Pharmacists Association Earth Trends Environme ntal Information. Agriculture and Food Searchable Database. World Resource In stitute, 2007. Elzakker, B. et al. (2007) Organic Farming in Africa. The World of Organic Agriculture, Statistics and Emerging Trends, Chpt.12 pg. 93 105. Engel, W. (2008) Deter minants of Consumer Willingness to Pay for Organic Food in South Africa. A paper submitted in partial fulfillment of the requirement for the degree MInst Agrar, University of Pretoria. Ericsson, G., Kindberg, J. and Bostedt, G. (2007) Willingness to Pay ( WTP) for Wolverine Gulo Gulo Conservation. Wildlife Biology 13(sp2):2 13.

PAGE 112

112 Christian Science Monitor. http://seattletimes.nwsource.com/html/ businesstechnology/2011109993_organicoffee07.html. (Accessed on July 29 2010 ) Friedman, D., and Sunder, S. (1994) Data Analysis in Experimental: A Primer for Ec onomists. New York: Cambridge University Press. Fox, J.A. (1995) Experimental Auctions to Measure Willingness to Pay for Food Safety. Boulder, CO: West View Press. Gao, Z. and Schroeder, T. (2007) Effect of Additional Quality Attributes on Consumer Willi ngness to Pay for Food Labels, A Dissertation Submitted in Partial Fulfilment of the Requirements for the Degree of Doctor of Philosophy, Kansas State University. Greene, W.H. ( 1997 ) Econometric Analysis Upper Saddle River, NJ: Prentice Hall. Greene, W. New York: Econometric Software, Inc. Gwendolyn H ., John C. and Bernard (2008) Consumer Willingness to pay for Sustainable Apparel: The Influence of Labeling for Fibre Origin and Prod uction Methods. International Journal of Consumer Studies, Vol. 32, Issue 5, pages 491 498. Hoffman S.D. and Duncan, G.J. ( 1988 ) Multinomial and Conditional Logit Discrete Choice Models in Demography. Demography, Vol. 25, No.3: 415 427. Hosmer, D.W. Jr and Lemeshow, S. (2000) Applied Logistic Regression 2 nd Edition. New York: John Wiley & Sons. Hustvedt, G. and Bernard, J.C. (2008) Consumer Willingness to pay for Sustainable Apparel: The Influence of Labelling for Fibre Origin and Production Methods. International Journal of Consumer Studies, 32 (2008): 491 498. Introduction to SAS. UCLA: Academic Technology Services, Statistical Consulting Group. http://www.ats.ucla.edu/stat/sas/notes2/ (A ccesse d November 24,2007). Standing One Strong Leg is Better Than on Models. A Doctoral of Philosophy Dissertation, University of Florida, FL USA.

PAGE 113

113 Jordan, J. and Elnagheeb, A. (1991) Public Perception of Food Safety. Journal of Food Distribution Research, 22(3): 13 22. Kim Minbo, Qualitative and Limited Dependent Variable Models Using the New QLIM Procedure, Statistics and Data Analysis. SAS Institute Inc., Cary, NC. Paper 279 25. Food: Factors that affect it and Variation per Organic Pr oduct Type. British Food Journal, 107, 320 343. Lancaster, K. (1966) A New Approach to Consumer Theory. Journal of Political Economy, LXXIV(2): 132 157. Lin, Biing Hwan et al. (2008) Demand for Organic and Conventional Fresh Fruits. Selected Paper Prepar ed for Presentation at the American Agricultural Economics Association Annual Meeting, Orlando. Loureiro M L. and Hine S (2002) Discovering Niche Markets: A Comparison of Consumer Willingness to Pay for Local (Colorado Grown), Organic and GMO Free Prod ucts. Journal of Agricultural and Applied Economics, 34, 3(December 2002):477 487. Southern Agricultural Economics Association Loureiro, M. and Lotade, J. (2005) Do Fair Trade and Eco Labels in Coffee Wake up the Consumer Conscience? Ecological Economics 53, 129 138. Loureiro, M.L. et al. (2001) Assessing Consumer Preferences for Organic, Eco Labelled, and Regular Apples. Journal of Agricultural and Resources Economics 26(2): 404 416, Western Agricultural Economics Association. Lusk, J.L. and Schroeder T.C. (2004) Are Choice Experiments Incentive Comparable? A Test with Quality Differentiated Beef Steaks. American Agricultural Economics Association. Amer J. Agr. Econ. 86(2): 467 482. Mabiso y of Origin Labels in fresh Apples and Tomatoes: A Double Hurdle Probit Analysis of US Data Master Thesis, University of Florida, Florida, USA. Marti n G.K. (2008) Assessing the Fit of Regression Model. Flashpoint (http://www.selfgrowth.com). MCCCI Month ly Bulletin. Published by the Malawi Confederation of Chambers of Comme rce and Industry, January 2009. McCluskey J.J. and Loureiro M L. (2003). The Consumer Response to Food Labelling http:// www.farmfoundatio n.org

PAGE 114

114 McFadden, D. (1974) Conditional Logit Analysis of Qualitative Choice Behaviour. In Frontiers in Econometrics, ed. P. Zarembka, 105 142, New York: Academic Press. Measuring Consumer Interest in Mexican Shade Grown Coffee: An Assessment of the Canad ian, Mexican and US Markets. Commission for Environmental Cooperation. Published by the Communications and Public Outreach Department of the CEC Secretariat, October 1999. http://www.cec.org Medium Term Prospects for Agr icultural Commodities, Projections to the Year 2010, FAO, 2003. Millock, K. et al. (2002) Willingness to Pay for Organic Foods: A Comparison between Survey Data and Panel Data from Denmark. Mitchell, R. and Carson, R. (1989). Using Surveys to Value Publi c Goods: The Contingent Valuation Method. Resource for the Future, Washington Dc. NOAA (1993) National Resource Damage Assessment Under the Oil Pollution Act of 1990, Federal Registers 58(10):4601 4614. Nelson, P. (1970) Information and Consumer Behaviou r. Journal of Political Economy, 78, 311 329. Noussair, C., Robin, S. and Ruffieux, B. (2004) A Comparison of Hedonic Rating and Demand Revealing Auctions. Food Quality and Preference, 15, 393 402. Pendergrast, M. (2006) Coffee: The Drink that Changed t he World. GENERATE, CEA Group, Germany. Population and Housing Census Preliminary report. National Statistical office, GOM, Malawi, 2008. Selected Paper Prepared for Pr esentation at the Southern Agricultural Economics Association, Annual Meeting, Dallas, TX. Poelman, A., Mojet, J., Lyon, D. and Sefa Dedeh, S. (2008) The Influence of Information about Organic Production and Fair Trade on Preferences for and Perception of Pineapple. Food Quality and Preference, 19, 114 121. Rahmatian, M. ( 2005 ) Contingent Valuation Method, Session 14. California State University. htt p://www.iwlearn.net/publications/misc/caspianer_rahmatian cvm.pdf. (Accessed on July 38 2010 )

PAGE 115

115 their Incidence in Argentinean Organic Choices. Pape r Presented at the International Association of Agricultural Economics Conference, Gold Cost, Australia. Rodriguez E. et a l. (2007) Willingness to Pay for Organic Food in Argentina: Evidence from a Consumer Survey. Contributed Paper Presented at the 105 th EAAE Products, Bologna, Italy. Sahota, A. (2007 ) Overview of the Global Market for Organic Food and Drink. The World of Organic Agriculture, Statistics and Emerging Trends, Chpt. 7 pg. 52 55,IFOAM and FIBL, Switzerland. Singleton, R. (1993) Approaches to Social Research, New York Oxford University Pre ss. The Agricultural Development Agenda Ministry of Agriculture and Food Security ( MoAFS ), (2010) The Trade and Environmental Effects of Ecolabels: Assessment and Response, UNEP. T homas, D.S. (2009) Taste Test of Organic Versus Conventional Products and What University of Florida, FL USA. Thurstone, L.L. (1927) A Law of Comparative Judgement. Psychology Re view,34: 273 286. Yussefi, M. and Willer, H. (2007) Organic Farming Worldwide 2007: Overview and Main Statistics. The World of Agricultural Production, Chpt. 3, pp. 9 16. Umberger W J. (2002) U.S. Consumer Preference and Willingness to Pay for Domestic Corn Fed Beef Versus International Grass Fed Beef measured through an Experimental Auction. Agribusiness, Vol. 18 (4) (2002) : 491 504. Published online in Wiley InterScience http:// www.interscience.wiley.com Wang, Q. and Sun, J. (2003) Consumer Preference and Demand for Organic Food: Evidence from a Vermont Survey. Paper Prepared for American Agricultural Economics Association Annual Meeting. Werner et al. (2002) Family Caregiver s Willingness to Pay for Drugs Indicated for the 74. Whittington, D. ( 1998 ) Administering Contingent Valuation Surveys in Developing Countries. World Development 26(1), 21 30.

PAGE 116

116 Wikstrom D. (2003) Willingness to Pay for Sustainable Coffee: A Choice Experiment Willer H. ( 2009 ) The World of Organic Agriculture 2009: Summary. IFOAM & FIBL, Switzerland. Willer, H. and Yussefi, M. (2007) The World of O rganic Agriculture, Statistics and Emerging Trends. IFOAM & FIBL, Switzerland Willer, H. and Yussefi, M. (2007) Organic Farming Worldwide 2007: Overview and Main Statistics. The World of Organic Agriculture, Statistics and Emerging Trends, Chpt. 8, pg. 9 16, IFOAM & FIBL, Switzerland. Wooldridge, M.J. (200 9 ) Introductory Econometrics: A Modern Approach (4 th Edition), Thomson, South Western C. Wuyang H. (2006) Comparing C onsumers' P references and W illingness to P ay for N on GM O il U sing a C ontingent V alu ation A pproach. Empirical Economics Volume 31 Number 1, pp. 143 150(8) Zanoli, R. and Naspetti, S. (2001) Consumer Motivations in the P urchase of Organic Food: A Means End Approach. British Food Journal, 8, 643 653. Ze peda L. and Li, J. (2007) Characteristics of Organic Food Shoppers, Journal o f Agricultural and Applied Economics 39:17 28.

PAGE 117

117 BIOGRAPHICAL SKETCH Fiskani Esther Nkana is a young lady born and raised in Malawi, the warm heart of Africa. She did her primary, secondary and undergraduate studies in Malawi. S he h old s a B achelor of Social Science with Economics as a major which was obtained from the University of Malawi (UNIMA) Chancellor College in 2003. B ased on her background, she decided to pursue a Master of Science in Food and Resource Economics which is very vital to the development of Agriculture in her country and Africa as a whole.