PATTERNS OF PARTICIPATION IN WEST AFRICAN PERI-URBAN DAIRY MARKETS: 
ANALYSES OF CONSUMERS, RETAILERS, PROCESSORS, AND PRODUCERS IN MALI 

 

 
By  

Ryan Vroegindewey 

 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

A DISSERTATION 

Submitted to  

Michigan State University  

in partial fulfillment of the requirements  

for the degree of 

Community Sustainability – Doctor of Philosophy  

  

 

2019 

ABSTRACT 

 
By  

Ryan Vroegindewey 

 

 

PATTERNS OF PARTICIPATION IN WEST AFRICAN PERI-URBAN DAIRY MARKETS: 
ANALYSES OF CONSUMERS, RETAILERS, PROCESSORS, AND PRODUCERS IN MALI 

 

The Malian dairy sector has the potential to play an important role in food security and 

economic development. However, less than 5% of Malian milk is marketed, dairy processing is 

largely dependent on powdered milk, and imports account for over 70% of household dairy 

consumption in Mali’s capital, Bamako.  

The objective of this research is to investigate the market failures behind this pattern, in order 

to identify opportunities for strengthening the competitiveness of the Malian dairy value chain. 

Three studies focus on key segments of this value chain, with a focus on peri-urban Bamako. The 

first study uses random parameters logit and latent class analysis of choice experiment data, to 

examine Malian consumer and retailer preferences for fresh milk as an ingredient in pasteurized 

milk, as well as for a number of quality-signaling attributes. The second study analyzes nine 

representative cases of artisanal, semi-industrial, and industrial dairy processors in Bamako, in 

order to understand the factors that drive a firm’s choice to procure and use fresh milk, powdered 

milk, or some blend of both. The third study applies a double hurdle market participation model 

to a nationally-representative household dataset, to identify the barriers that constrain Malian 

milk producers from entering and supplying markets.  

Among the key results, I find that most Malian consumers have a significant willingness-to-

pay for dairy products that are manufactured from fresh milk, as opposed to powdered or blended 

milk. I also identify quality-signaling mechanisms that can help to address information 

asymmetry in consumer and retailer markets, and policy measures that could encourage fresh 

milk supply in markets. Together, these measures should help the Malian dairy value chain to 

capture a greater share of growing dairy demand. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 
 
Copyright by  
RYAN VROEGINDEWEY 
2019

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

who is beautiful in every way,  

not least in her indelible love and support 

To my wife, Ruth,  

 

v 

 
 
 
 

 

ACKNOWLEDGEMENTS 

 
 

This dissertation research owes its completion to many extraordinary people.  

First of all, I would like to thank my advisor, Robert Richardson, for his faith in my potential, 

his ongoing generosity of time, and his unwavering support as I have navigated the many twists 

and turns of doctoral work. I am also deeply grateful to Véronique Thériault, who has mentored 

me as a researcher for the past six years, who first suggested and committed funding to this 

dissertation topic four years ago, and with whom I am proud to share a deep passion for Mali and 

a hope for its future. I am also very grateful to my other two committee members for their 

kindness and expertise. Specifically, David Ortega has taught me a great deal about the design, 

implementation, and analysis of choice experiments. Kimberly Chung has deepened my 

understanding and appreciation for good qualitative research. To all of my committee members: 

thank you. Any strengths of this dissertation are direct reflections of your skills as researchers 

and dedication as teachers. 

Second, I want to thank my amazing family. Ruth, you have been steadfast in your faith and 

support, even though you, of all people, have made the most and costliest sacrifices for this PhD. 

Hadassah, you came and brought joy, grounding, and perspective when I needed these 

interruptions the most. Also, you are the only one-year-old I know who can consume her weight 

in cheese, milk, yoghurt, and butter – whether this is some externality of her father doing nothing 

but dairy research the first year of her life will remain an empirical question. My Vroegindewey 

and Abaya parents and siblings have also been a wonderful base of support, as they have 

encouraged and pushed and cheered me these past years, all along the way patiently enduring 

many conversations about doctoral drama and Malian dairy value chains.  

 

vi 

Third, I want to express sincere gratitude to the U.S. Agency for International Development, 

for funding much of this research through its Feed the Future Innovation Lab for Food Security 

Policy grant to the MSU Department of Agricultural, Food, and Resource Economics. I am also 

grateful to the Department of Community Sustainability, and to centers at MSU, which have also 

generously financed various aspects of data collection and conference presentations.    

Fourth, I want to thank the many Malians who contributed to this research project through 

various means. My three field visits to Mali never could have been so successful or efficient if I 

hadn’t enjoyed the excellent support, guidance, and facilitation of the MSU/PREPOSAM staff, 

namely Yenizie Koné, Abdramane Traoré, Naman Keita, and Inah Niaré. Nor could I have been 

successful in the field without the assistance, resourcefulness, and daily perseverance of Assitan 

Traoré, whom I am confident will go on to do great and impactful things in the Malian 

agricultural sector. Over the past two years, Diop Fatoumata Thiam has also provided invaluable 

expertise on Malian dairy and, more importantly, has become a good friend to my family.  

I am also deeply grateful for my friends and colleagues Bino Témé and Yacouba Santara, 

who had supported me as a young professional at various points over the past ten years. Your 

lives and work—including your effusive passion for the Malian livestock sector—have had a 

great impact on me and so many others. May your souls rest in peace.  

To the hundreds of dairy farmers, processors, retailers, consumers, and sector champions 

whom I have had the great privilege to work with: from the bottom of my heart, thank you so 

much for your time and trust. Ultimately, this research is for you. It is my deepest hope that all of 

this effort will make some meaningful and material difference in your lives and work.  

Not least of all, I want to thank God, for his boundless love and faithful provision, and for 

establishing the work of our hands.  

 

vii 

TABLE OF CONTENTS 

 
 
LIST OF TABLES .......................................................................................................................... x 
 
LIST OF FIGURES ....................................................................................................................... xi 
 
CHAPTER ONE: INTRODUCTION ............................................................................................. 1 
REFERENCES ............................................................................................................................... 4 
 
CHAPTER TWO: ANALYSIS OF CONSUMER AND RETAILER PREFERENCES 
TOWARDS LOCAL MILK AND OTHER DAIRY PRODUCT ATTRIBUTES ........................ 6 
2.1 Introduction ........................................................................................................................... 6 
2.2 Theoretical framework and econometric modeling ............................................................ 10 
2.3 Data and Survey .................................................................................................................. 13 
2.3.1 Sampling Strategy ........................................................................................................ 13 
2.3.2 Socio-demographic survey and sample characteristics ................................................ 16 
2.3.3 Choice experiment and dairy product attributes .......................................................... 21 
2.4 Results and discussion ........................................................................................................ 27 
2.4.1 Results from the consumer random parameters logit models ...................................... 27 
2.4.2 Comparison of consumer and retailer preferences and WTP ...................................... 33 
2.4.3 Results from the latent class models ............................................................................ 37 
2.5 Conclusion .......................................................................................................................... 45 
REFERENCES ............................................................................................................................. 50 
 
CHAPTER THREE: ANALYSIS OF THE FACTORS MOTIVATING DAIRY PROCESSORS 
TO USE LOCAL OR IMPORTED MILK INPUTS IN THEIR OPERATIONS ........................ 55 
3.1 Introduction ......................................................................................................................... 55 
3.2 Conceptual Framework ....................................................................................................... 57 
3.2.1 Firm competitive advantage ......................................................................................... 57 
3.2.2 Cost .............................................................................................................................. 59 
3.2.3 Differentiation .............................................................................................................. 60 
3.3 Methods............................................................................................................................... 62 
3.3.1 Case study methods and case selection ........................................................................ 62 
3.3.2 Data collection and analysis ......................................................................................... 63 
3.4 How do firms use fresh and powdered milk in their operations? ....................................... 65 
3.4.1 Fresh milk firms ........................................................................................................... 67 
3.4.2 Powdered milk firms .................................................................................................... 69 
3.5  How does each milk input influence cost? ........................................................................ 71 
3.5.1 Purchase Price .............................................................................................................. 71 
3.5.2 Coordination costs ....................................................................................................... 73 
3.5.3 Other transaction costs ................................................................................................. 75 
3.5.4 Other idiosyncratic costs .............................................................................................. 78 
3.5.5 Summary of the relative costs ...................................................................................... 79 
3.6 How does milk input choice influence differentiation? ...................................................... 80 
3.6.1 Product types and sub-types ......................................................................................... 81 

 

viii 

3.6.2 Enhancing or preserving product quality ..................................................................... 83 
3.6.3 Quality-signaling mechanisms ..................................................................................... 85 
3.6.4 Packaging ..................................................................................................................... 91 
3.7 Conclusion .......................................................................................................................... 94 
REFERENCES ............................................................................................................................. 97 
 
CHAPTER FOUR: ANALYSIS OF THE DETERMINANTS OF MILK PRODUCERS’ 
PARTICIPATION IN OUTPUT MARKETS ............................................................................ 101 
4.1 Introduction ....................................................................................................................... 101 
4.2 Conceptual framework ...................................................................................................... 104 
4.3 Materials and methods ...................................................................................................... 106 
4.3.1 Data sources ............................................................................................................... 106 
4.3.2 Econometric model and estimation ............................................................................ 107 
4.3.3 Variable definitions .................................................................................................... 109 
4.4 Results and discussion ...................................................................................................... 115 
4.4.1 Descriptive statistics .................................................................................................. 115 
4.4.2 Econometric results .................................................................................................... 117 
4.5 Conclusion ........................................................................................................................ 123 
APPENDIX ................................................................................................................................. 128 
REFERENCES ........................................................................................................................... 126      
 
CHAPTER FIVE: CONCLUSIONS .......................................................................................... 128 

 

 

 

 

ix 

LIST OF TABLES  

 
 
Table 1: Summary statistics of consumer sample………………………………………….…...19 
 
Table 2: Summary statistics of retailer sample…………………………………………………..……20  
 
Table 3: Pasteurized milk product attributes and levels in choice experiments……………..……21 
 
Table 4: Results from the consumers random parameters logit model……………………….……29  
 
Table 5: Results from the retailers random parameters logit models…………..…………..………35  
 
Table 6: Comparison of consumer and retailer WTP……………………………..….………………36  
 
Table 7: Results from latent class analysis of consumers……………………………..………….…39  
 
Table 8: Profiles of consumers from each latent class…………………………………………….…40  
 
Table 9: Results from latent class analysis of retailers………………………………………...….…41  
 
Table 10: Profiles of retailers from each latent class………………………………………..…….…42 
 
Table 11: Overview of firms and dairy products……………………………………………...………66 
 
Table 12: Costs of fresh and powdered milk in Bamako……………………….……………………80 
 
Table 13: Variable definition and summary statistics…………………………….…………………110  
 
Table 14: Distribution of milk sales across quintiles, weighted…………………………...………115 
 
Table 15: Summary statistics……………………………………………………………..…………….116  
 
Table 16: Cragg model results and average partial effects………………………………………….119 
 
Table 17: Robustness checks: alternative Tobit and Cragg model specifications……………..…131 

 

x 

LIST OF FIGURES 

 
 
Figure 1: Summary of dissertation research……………………...……………………….………2 
 
Figure 2: Map of areas sampled in Bamako……………………...……………………………...15 
 
Figure 3: Example choice task for a consumer and retailer……………………………………..……27 
 
Figure 4: Research framework and questions, based in Porter’s comparative advantage……….59 

Figure 5: Monthly volumes collected by peri-urban FENALAIT dairy cooperatives,  
                2012-16..........………………………………………………...………………………………..77 
 
Figure 6: Per capita supply of domestic milk and dairy imports in Mali (L/per person) …….…104 
 
 

 
 
 
 
 
 
 
 
 
 

 

xi 

 

CHAPTER ONE: INTRODUCTION 

The Malian dairy sector would appear to hold enormous growth potential. First, dairy 

consumption has been increasing over the past decades. Recent estimates of the income elasticity 

of dairy demand-1.1 and 1.3 in rural and urban areas, respectively- indicate that, as incomes rise, 

the consumption of milk is set to increase as a share of total food consumption (Zhou and Staatz, 

2016). Second, Mali’s livestock sector, which provides employment to about 85% of the 

population and has about six cows for every ten people in the population, has the potential to 

meet this demand (FAO, 2019; MSU, 2011). Additionally, there is some evidence suggesting 

that Sahelian West African consumers have a strong preference for local milk (or fresh milk), 

compared to imported dairy products (Lefèvre, 2014).  

Although one might see these characteristics as a recipe for the rapid growth of Mali’s 

domestic dairy sector over imports, what one observes is the opposite. Per capita availability of 

imports—especially powdered milk—is growing at the national level, while per capita 

availability of domestic milk is falling (authors’ calculations from FAO, 2019). In Mali’s capital, 

Bamako, imports have already captured the majority share of household dairy consumption 

(authors’ calculations from World Bank, 2015). 

This puzzling pattern is not unique to Mali. Two decades ago, Staal et al. (1997) documented 

similar features in East Africa, and identified them as the “tale-tell signs” of “dairying under 

large transaction costs.” The objective of this dissertation research is to identify the key market 

failures that constrain the development of the Malian dairy value chain, and to propose policy 

options and value chain investments that can redress these challenges. To do this, I draw heavily 

from New Institutional Economics, which has been developed to explain and address market 

failures in their many forms (Kirsten et al., 2009). The classic definition of market failure refers 

 

1 

to a situation in which a free market fails to bring about a Pareto-efficient allocation of a good 

(e.g., raw milk or packaged pasteurized milk), such as when there exist unexploited trades 

between two parties that could make at least one party better off without making the other worse 

off (Besley, 1998).  

This research focuses on market failures at key transactional points in the dairy value chain: 

between farmers and buyers in milk output markets, between processors and their value chain 

partners, and in the retailing of dairy products to consumers. Figure 1 visualizes these foci. I 

begin the analysis with the consumption and retailer segments, where demand is the primary 

driver for dairy value chain development (Reardon, 2015), then progressively move upstream.  

 
Figure 1: Summary of dissertation research 

Study #3

Econometric analysis 

of household data

Study #2
Case studies 
of processors

Study #1

Choice experiments with 
consumers and retailers

Milk

Production

 

Processing

Retailing

Consumption

 

In the first chapter, I use choice experiments to examine Malian retailer and consumer 

preferences for fresh milk as an ingredient in pasteurized milk, the most widely-consumed fluid 

dairy product in Mali, as well as for other novel product attributes. In doing so, I investigate 

information asymmetry as a source of market failure and also identify quality-signaling 

mechanisms that can help redress this problem. The second chapter turns to the processing 

segment of the Malian dairy value chain. I use case study methods to understand the factors that 

drive a dairy firm’s choice to use fresh milk, powdered milk, or some blend of both, as an input 

 

2 

in the manufacturing of consumer dairy products. In the final analysis, I conduct econometric 

analysis of household data in order to identify the barriers that constrain Malian milk producers 

from entering and supplying output markets. In the final chapter, I synthesize the main findings 

from these three studies, and highlight the broad implications for policy and research.  

Among the key results, I find that most Malian consumers have a significant willingness-to-

pay for dairy products that are manufactured from fresh milk, as opposed to powdered or blended 

milk. I also identify quality-signaling mechanisms that can help to address information 

asymmetry in consumer and retailer markets, and policy measures that should encourage fresh 

milk supply in output markets. Together, these measures should help the Malian dairy value 

chain to capture a greater share of growing dairy demand. 

 

 
 
 
 
 
 
 
 
 

 

 

3 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

REFERENCES 

 

4 

 

 

 

REFERENCES 

Besley, T. (1998). “How Do Market Failures Justify Interventions in Rural Credit Markets?" In  
 

Eicher, C. and J. Staatz (eds). International Agricultural Development. Baltimore, MD: 
Johns  Hopkins Univ. Press, pp. 370-389.  

 
Food and Agriculture Organization (2019). FAOSTAT. http://fao.org/faostat/en/#data 
 
Kirsten, Johann F., Karaan, Mohammad, A.S., and Dorward, R. Andrew. (2009). Introduction 
to the Economics of Institutions. In Kirsten, Johann F., Dorward, R. Andrew, and Poulton, 
Colin (Eds.), Institutional economics perspectives on African agricultural development 
(35-74). IFPRI, Washington, D.C. 

 
Lefèvre, M. (2014). Do Consumers Pay More for What They Value More? The Case of Local 

Milk-based Dairy Products in Senegal. Agricultural and Resource Economics Review, 
43(01), 158–177.  

 
Reardon, T. (2015). The hidden middle: the quiet revolution in the midstream of agrifood 

value chains in developing countries. Oxford Review of Economic Policy, 31(1), 45–63. 

 
Staal, S., Delgado, C., and Nicholson, C. (1997). Smallholder dairying under transactions costs 

in East Africa. World Development, 25(5), 779–794. 

 
World Bank. (2015). Living Standards Measurement Survey – Integrated Survey on Agriculture 

2014.  

 
Zhou, Y., and Staatz, J. (2016). Projected Demand and Supply for Various Foods in West Africa: 

Implications for Investments and Food Policy. Food Policy 61:198–212.   

 

 

 

 

5 

CHAPTER TWO: ANALYSIS OF CONSUMER AND RETAILER PREFERENCES 

TOWARDS LOCAL MILK AND OTHER DAIRY PRODUCT ATTRIBUTES 

 
2.1 Introduction 

 

Consumer demand for dairy in West Africa has been growing for several decades and is 

positively correlated with both income growth and urbanization (Hollinger and Staatz, 2015). For 

instance, from the late 1980s to mid-2000s, the share of household food expenditure on dairy 

increased by 50% and by 30% in urban and rural areas of Mali, respectively (Ibid.). The Malian 

capital, Bamako, consumes over one-fifth of the national dairy production that is marketed (83 of 

376 million liters (L) per year), which is the equivalent of a per capita dairy consumption of 34 L 

per person per year (author’s calculations from World Bank, 2015).1  

The potential to meet this growing demand with domestic milk supply is high. Mali ranks 6th 

in total livestock holdings among African countries (FAO, 2017). The country’s cattle holding is 

estimated at eleven million heads and the potential production of cow’s milk is nearly 600,000 

tons (Government of Mali, 2017). There is some evidence, although limited, that West African 

consumers have a preference for dairy products that are manufactured from local fresh milk 

rather than from imported powdered milk (2019; Hollinger and Staatz, 2015; Lefèvre, 2014). 

Powdered milk accounts for about one third of dairy consumption in Mali as a whole 

(author’s calculations from World Bank, 2015). This dependence is accented in urban areas, due 

to higher population densities and greater distances to agricultural production zones, combined 

with weak overall commercial infrastructure. In Bamako, household purchases of powdered milk 

account for 70% of volumes consumed in liquid milk equivalents (Ibid.). Fluid milk and 

yoghurts account for another 18% of consumption (author’s calculations from World Bank, 

1 In this paper, I present dried milk products in Liquid Milk Equivalents (LME) using a conversion factor 
recommended by Meyer and Duteurtre (1998). 

 

 

6 

2015). Although Bamako-based companies manufacture most of these products, the most 

established local brands are mostly or entirely manufactured from powdered milk. Most fresh 

milk-based dairy products that are available in Bamako are manufactured by peri-urban dairy 

cooperatives or small artisanal processors. This dependence on powdered milk in urban 

processing and consumption—despite an overall preference for fresh milk—is a common pattern 

that has been documented elsewhere in Senegal (Lefèvre, 2014) and more generally across West 

Africa (Hollinger and Staatz, 2015). 

One reason for this apparent market failure may be information asymmetry between the 

manufacturers and buyers of fresh milk-based dairy products. That is, it could be that consumers 

and retailers of dairy products demand certain credence or experience attributes that may or may 

not be available on the market, but are in any case difficult or costly for them to verify ex ante 

(Caswell and Mojduszka, 1996; Rosenman and Wilson, 1991). It could also be the case that 

certain value actors, such as processors or retailers, are unaware of consumer preferences for 

certain attributes and, therefore, do not offer them.  

Two issues are relevant in developing contexts such as Mali, where food quality and labeling 

regulations are limited and/or weakly enforced. First, the ingredient composition of a dairy 

product is a quality attribute that is usually unobservable to consumers prior to the purchase or 

consumption. For example, Lefèvre (2014) estimates that Senegalese consumers are willing to 

pay 80% above average prices for a fresh milk-based dairy product; however, in urban markets 

this preference is not transmitted into an actual price premium, because many consumers do not 

know the ingredient composition of different brands and are also misled by ambiguous or 

deceitful labeling practices. Second, recent qualitative research conducted in several West 

African capitals found that consumers value local foods, but concerns over product safety and 

 

7 

other quality problems lead them to consume foreign brands instead (Hollinger and Staatz, 

2015). Other studies conducted in Africa have showed that consumer dairy purchase decisions 

are influenced by perceptions of aflatoxin levels, unhygienic handling, color and smell, and fat 

content (Mtimet et al., 2015; Fadiga and Makokha, 2014).  

The objectives of this paper are threefold. First, I examine Malian preferences for fresh milk 

as an ingredient in dairy products. Second, I investigate information asymmetry as a contributing 

factor to the market failure described above. Third, I analyze three quality-signaling 

mechanisms—ingredients listing, third-party certification, and enhanced packaging—that 

manufacturers of fresh milk-based products could be adopted to address information asymmetry. 

To accomplish this, I use discrete choice experiments to elicit buyer preferences, since quality 

certification, certain types of packaging, and the clear labeling of ingredients are dairy product 

attributes that are not yet offered in the Malian market. The study focuses on Mali’s largest city 

and capital, Bamako, given its high dependence on powdered milk. Additionally, large urban 

markets such as Bamako account for a growing share of food expenditure in developing 

countries, and also represent the frontier of changing tastes and preferences in these contexts 

(Hollinger and Staatz, 2015).    

A major contribution of this study is that I investigate the preferences of both consumers and 

retailers, the latter providing a critical link between the consumers and producers of local foods 

(Trivette, 2018). To do this, I conducted parallel choice experiments with both consumers and 

retailers, in the vein of the “stacked survey method” which allows for the statistical analysis of 

differences across value chain segments (Reardon et al., 2012, p. 32). To my knowledge, this is 

the first study to take this approach in a developing country context. Fernández-Polanco et al. 

(2013) used a similar approach to study retailer and consumer preferences for seafood products 

 

8 

in Europe. This study concludes that retailers understand well the preferences of consumers with 

respect to several product attributes, but that retailers overestimate the price-sensitivity of 

consumers. The study design builds on Fernández-Polanco et al. (2013), by allowing retailer 

respondents to make choices as firms (rather than as consumers), and by using a larger sample of 

retailers.  

The approach allows me to probe more deeply into issues of information asymmetry and to 

provide more useful insights for the marketing of dairy products. Retailer value and procurement 

decision-making are partly driven by a retailer’s knowledge of consumer preferences for product 

attributes (Skytte and Bove, 2004). If I find evidence that retailer preferences are aligned with 

those of consumers, then it suggests that retailers have adequate information about consumer 

preferences. Further, if retailer preferences manifest in positive willingness-to-pay (WTP) for 

product attributes, then it implies that other determinants of retailer value, such as considerations 

of cost and competition, are also satisfied for retailers to purchase that product. In this case, dairy 

processors can have greater confidence that those product attributes will be commercially 

successful, compared to only having evidence of consumer demand. Further, estimates of retailer 

WTP for product attributes—which factors in the marketing margins that retailers would expect 

to capture—should provide manufacturers with a more relevant price premium target, compared 

to consumer WTP.  

Finally, because buyer preferences for dairy products are assumed to vary across individuals 

(Olynk and Ortega, 2013; Wolf et al., 2011; Olynk et al., 2010), I analyze the choice experiment 

data using models that account for preference heterogeneity. This analysis allows me to identify 

and characterize distinct consumer and retailer segments. This approach is useful for business 

 

9 

strategy and policy, because it enables them to target, and to tailor their approaches to, 

appropriate consumer groups along with the retailers that these consumers are likely to patronize. 

In the next section, I present the theoretical and empirical framework that underlies the 

analysis of retailer and consumer preferences. In section three, I describe the data and choice 

experiment design. I discuss the results from the data analysis in section four, then conclude by 

highlighting implications for agribusiness strategy and government policy in section five. In this 

paper, “fluid milk” refers to milk in its liquid form, and includes raw, pasteurized, or sterilized 

(i.e. ultra-high temperature-processed) milk products that are made from fresh milk, reconstituted 

powdered milk, or some combination of the two. Fresh milk refers to fluid milk that is locally-

sourced, i.e. produced in Mali. Mali does not produce any powdered or sterilized milk. 

Pasteurized milk refers specifically to fluid milk that has been pasteurized for sale as a consumer 

product. Dairy refers to any of these, plus any other product that is manufactured from a milk 

input. 

 

2.2 Theoretical framework and econometric modeling 

The theoretical framework of this research is grounded in random utility theory 

(Manski,1977; McFadden, 1974) and Lancaster (1966)’s consumer theory, which described 

product quality as a bundle of attributes (related to food safety, nutrition, packaging, processes, 

etc.) that together determine product performance (Caswell and Mojduszka, 1996).  

As a flexible empirical method that conforms to these theories (Adamowicz et al., 1998), 

choice experiments closely simulate real-world purchasing decisions in which an individual must 

select one product alternative from a set of options. In most applications, the buyer is the final 

consumer of the product (e.g., Ortega et al., 2011). The literature has given little attention to 

 

10 

retailer preferences  (one exception is Fernández-Polanco et al., 2013), although there is 

precedence for such an application given that researchers have adapted the approach to study 

other actors along livestock value chains, including farmers (e.g., Otieno et al., 2011) and traders 

(e.g., Ruto et al., 2008). 

unobservable random variable:  

of alternatives), where each alternative k represents a bundle of product attributes x. Subject to 

Consider a situation in which individual i faces K alternatives contained in ! (the global set 
the choice set in situation t, the individual will select alternative "∈!	to maximize her utility 
%&'(. Because of incomplete information and other errors, I model utility as a latent and 
 %&'(=*+,&'(-+	/&'(,                                                                                                                                    (1) 
where *+,&'(- is an observable deterministic utility component generated from the selected 
alternative, and /&'( is the random component of utility assumed to be independently and 
alternatives; formally %&'(≥	%&1(	∀	"≠4. Thus the probability of her choosing product j is 
 5678&'(=5678(*&'(+	/&'(	≥*&1(+	/&1(;		∀4∈!,∀4≠")                                                       (2) 
Assuming that the deterministic component of %&'( is linear in parameters, I can specify 

product j if her utility derived from j is greater than or equal to her utility derived from the 

identically distributed across all individuals and choice situations. Individual i will choose 

 

 

given by: 

individual i’s utility function as:  

 

11 

 *&'(=	<′&'(>+/&'(                                                                                                                                 (3) 
where <′&'( is a vector of attributes for the jth alternative, > is a vector of taste parameters, and 
/&'( is the stochastic component which follows a Gumbel (extreme value Type I) distribution.  

 

In a simple conditional logit (CL) models, individuals are assumed to have homogenous 

preferences. However, in reality, consumers preferences for food quality differ from one another. 

Therefore, I investigate preferences of dairy consumers using two approaches that account for 

preference heterogeneity.  

The first approach is the random parameters logit (RPL) model, which relaxes limitations in 

the CL model by allowing preferences to vary randomly within a sample according to a specified 

distribution (McFadden and Train, 2000). The probability that individual i selects alternative j 

H(>)I>                                                                                                               (4)                

from the choice set ! in situation t is: 
 5&'(= ?@A+BCDE-
∑ ?@A(BCDE)
D
where > is a vector of random parameters with distribution H(∙).  
2002). In the LC model, H(>) is discrete and takes on C distinct values. The probability that 
individual i selects alternative j from the choice set ! in situation t is: 

 

 The second approach is the latent class (LC) model. This model segments individuals into a 

number of C latent classes across which preferences vary discreetly (Boxall and Adamowicz, 

 

 

12 

POQR

?@A+KCDELM-
∑ ?@A(KCDELM)
D

N&O

                                                                                                                   (5) 

5&'(=∑
where >M is the specific parameter for class c, and N&O is the probability that individual i is a 
 N&S= ?@A(TUVW)
∑ ?@A(TUVX)
X
where YZ is a set of observable characteristics that affect individual i’s class membership, and [S 

member of class c. I can model this probability of class membership as:  

                                                                                                                                     (6) 

 

 

is the parameter vector for consumers in class c. 

I use the RPL and LC models to analyze data collected through a choice experiment (see next 

section for a detailed discussion). For both models, the choice experiment data allows me to 

estimate parameters in preference-space. These estimated coefficients represent consumers’ 

marginal utilities for different product attributes, while the ratio of any two coefficients 

represents the marginal rate of substitution of one attribute over the other; use of the price 

coefficient in the denominator yields willingness to pay.  The specification of the RPL model can 

be reparametrized in WTP-space so that the model coefficients are themselves WTP estimates.2    

 

2.3 Data and Survey 

2.3.1 Sampling Strategy  

I collected the data for this study in March 2018, by surveying 218 dairy consumers and 193 

retailers located across Bamako, which has a population of approximately 2.5 million (World 

2 For more details on models specified in WTP-space, see Train and Weeks (2005). 

 

 

13 

Bank, 2018). In order to obtain a sample that is representative of the Bamako population in terms 

of wealth, I used a stratified random sampling approach. In the first stage, I stratified the city by 

neighborhood wealth levels. Given that no official information exists on average household 

incomes by neighborhood, I classified each of Bamako’s 61 neighborhoods into one of three 

wealth levels—low, medium, and high—based on research and field partners’ knowledge of the 

neighborhoods in terms of housing conditions (e.g., access to electricity), infrastructure (e.g., 

whether roads are paved), and other socio-economic indicators (e.g., where expatriates reside). 

According to this classification, twenty-five neighborhoods (41%) are low-income, twenty-four 

(39%) are medium-income, and twelve (20%) are high-income. I randomly selected eight, seven, 

and four neighborhoods from each class, respectively, in order to have the same neighborhood 

weight in the sample as in the population. In the second stage, survey teams randomly sampled 

households within each selected neighborhood. Because censuses of these populations were not 

available, I developed a geographic random sampling approach aimed at capturing an 

appropriate cross-section of the neighborhood.  

In each neighborhood, random and purposive sampling approaches to select retail outlets 

were used. Two types of small retail formats make up the vast majority of shops selling dairy and 

other food products in Bamako (Theriault et al., 2018). First, boutiques are small traditional 

shops that sell a limited selection of food items and non-food items (e.g., soaps, batteries) 

displayed behind a counter. Many are equipped with one or two refrigerators, which vary widely 

in quality and make available a limited selection of cold beverages and dairy products. Second, 

alimentations are small self-service grocery stores that mostly carry processed foods, including 

cold products, and are equipped with one cash register. Although most neighborhoods have at 

least a few alimentation shops, these formats are much less numerous than boutiques and are 

 

14 

typically located on main paved roads, rather than on the smaller residential dirt roads. One 

retailer survey team followed one of the two residential roads taken by the household survey 

teams, selecting every three retail shops that had at least one working refrigerator, a prerequisite 

for marketing pasteurized milk. The second retailer survey team purposively sampled 

alimentations located along the main road. Although Bamako has about half a dozen 

supermarkets, these were not sampled because of their small number and their unbalanced 

distribution throughout the city. Figure 2 presents the location of sampled households using their 

GIS coordinates. The clusters of map markers represent the sampled households in different 

neighborhoods. As this map shows, the randomly-selected neighborhoods are fairly well-

distributed throughout the city.   

 
Figure 2: Map of areas sampled in Bamako  

 

15 

  

  
2.3.2 Socio-demographic survey and sample characteristics  

The surveys data had two parts: a questionnaire and a choice experiment, each tailored for 

either a consumer or retailer respondent. The survey teams conducted each consumer survey with 

the household member who was responsible for food purchasing decisions. In cases where this 

person was unavailable, the teams conducted the survey with the individual who was second-in-

charge.3 The teams conducted each retailer survey with the owner or manager of each shop (57% 

of the retailer sample) or, when he was unavailable, the agent on duty.4 All questionnaires 

collected data on basic socio-demographic characteristics of the respondent and on knowledge, 

attitudes, and preferences regarding multiple dairy product quality issues. Additionally, the 

household survey included variables that might influence dairy preferences, including household 

composition and income; access to refrigeration, electricity, and a motor-vehicle; and dairy 

product transactions and consumption in the previous week. The retailer survey aimed at 

capturing basic information and variables influencing retailer value with respect to the 

procurement of dairy products, including: how retailers obtain information on consumer 

preferences and product attributes, customer characteristics, competitive positioning, and drivers 

of cost (Skytte and Bove, 2004).  

Table 1 summarizes relevant characteristics of the consumer sample, disaggregated by 

neighborhood wealth level. Most respondents were females who had received at least some 

elementary school education. Almost 60% of the households are poor, based on the estimates of 

 

3 In about 75% of the households where the primary decision-maker was interviewed, this was a male or female 
household head. In about 70% of the households where the secondary decision-maker was interviewed, this was the 
female spouse of the household head. Thus, in most households, the household head was the primary decision-
maker. 
4 Whenever neither individuals were available in households or shops, or else when they refused to participate, the 
survey team continued sampling using the same predesignated interval. In total, there were about seventy-five and 
sixty such cases for households and retailers, respectively.  

 

16 

per capita household income and the 2017 national poverty line of 178,000 FCFA (338 USD) per 

year (Government of Mali, 2018).5 The data suggests that household assets, access to electricity, 

and expenditures for fresh milk and all dairy products are statistically different across 

neighborhood wealth levels, confirming the stratified sampling approach. Just over half of the 

households have access to both electricity and a refrigerator, with the worst and best access in 

poor and wealthy neighborhoods, respectively. Depending on the neighborhood type, households 

had spent an average total of 4,125 FCFA (7.84 USD) to 6,872 FCFA (13.06 USD) on dairy 

products during the previous week. Households allocated about a quarter of this expenditure to 

fluid milk, and about one-third to powdered milk. Approximately 70% of households had 

consumed fluid milk at least once in the previous week. Of these households, about half had 

consumed fresh milk in the previous week. 

Table 2 presents characteristics of the retail shop sample, disaggregated by retailer type. On 

average, all shops are quite small, each with about two employees and one or two refrigerators. 

However, as expected, boutiques are smaller than alimentations in terms of number of 

employees and refrigerators, product variety, weekly volumes of fluid milk, and margins.  

On average, boutiques carry 2.7 distinct fluid milk products (i.e. different brands and/or sizes of 

pasteurized or sterilized milk) and procure 57 liters of fluid milk per week. In comparison, 

alimentations carry 4.6 different products and procure about 118 liters per week. The average 

marketing margin earned on fluid dairy products (i.e., the difference between the unit price at 

which retailers purchase and resell a product) is 16% for boutiques and 18% for alimentations. I 

did not find any statistically significant differences in the mean values of these variables when 

testing across neighborhood wealth levels. However, I do find that the average shop located in 

5 At the time of field work, the FCFA-to-USD conversion rate was .0019. 

 

 

17 

neighborhoods of intermediate wealth level experience power outages more frequently and of 

longer duration, compared to shops in poor or wealthy neighborhoods. Overall, retailers reported 

an average of five power outages in the previous week, with the average duration of the longest 

outage being almost two hours. 

 
 
 

 

18 

  

Total (N=218) 
SD 
Mean 
 

Poor (N=92) 
SD 
 

 

 

 

  

0.31 
0.78 

Table 1: Summary statistics of consumer sample  
  
 
 
Respondent characteristics 
Primary HH food purchases decisionmaker (yes/no) 
Female (yes/no) 
Education level attained (yes/no) 
   None 
   Elementary school or more 
   Middle school or more 
   Highschool or more 
   University 
Household characteristics 
Below the poverty line  (yes/no) 
Has access to electricity and refridgerator (yes/no) 
Household dairy consumption in past week 
Total dairy expenditure (FCFA) 
Expenditure share - all fluid milk (%) 
Expenditure share - powdered milk  (%) 
Consumed any fluid milk (yes/no) 
Consumed fluid milk made from local milk (yes/no) 

0.26 
0.37 
0.70 
0.37 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively

0.16 
0.84 
0.72 
0.44 
0.34 

0.58 
0.54 

5,128.11 

 

 

0.46 
0.41 

 

0.36 
0.37 
0.45 
0.50 
0.47 

 

0.49 
0.50 

 

5,869.56 

0.27 
0.31 
0.46 
0.48 

 

0.35 
0.76 

0.16 
0.84 
0.72 
0.42 
0.30 

  
 
   Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

0.25 
0.33 
0.61 
0.26 

0.63 
0.43 

 

 

4,124.73 

0.48 
0.43 

 

0.37 
0.37 
0.45 
0.50 
0.46 

 

0.49 
0.50 

 

4,427.14 

0.30 
0.30 
0.49 
0.44 

Wealthy (N=45) 
SD 
 

 

 

 

 

 

 

SD 
 

0.46 
0.41 

0.31 
0.79 

0.22 
0.81 

Intermediate (N=81) 

0.37 
0.38 
0.46 
0.50 
0.49 

0.16 
0.83 
0.69 
0.47 
0.40 

Neighborhood wealth level 
 
 
   Mean 
   Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
  

0.13 
0.87 
0.78 
0.44 
0.31 

0.27 
0.38 
0.71 
0.47 

0.26 
0.41 
0.79 
0.43 

0.24 
0.32 
0.41 
0.50 

0.55 
0.64 

0.54 
0.60 

0.50 
0.49 

5,298.69 

4,734.80 

 

 

 

 

 

6,872.40 

0.42 
0.40 

 

0.34 
0.34 
0.42 
0.50 
0.47 

 

0.50 
0.48 

 

9,129.26 

0.26 
0.31 
0.46 
0.50 

∆ p-value 

  
 

 

 

0.49 
0.89 

 

0.96 
0.93 
0.72 
0.88 
0.55 

 

0.58 
0.06 

 

0.03 
0.90 
0.25 
0.12 
0.07 

 
 
 
 
 
 
 
 
 
 
 

 

 
 
 

* 

* 

* 

  

  
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 

19 

Table 2: Summary statistics of retailer sample  
  

  

  

  

Retail format type 

Total  
(N=193) 

Boutique  
(N=144) 

  

∆ p-value 

 

  

 

 

* 

 
** 
 
 
 
 
 
*** 
*** 

 
 
 
*** 
*** 
*** 
*** 

 
 

Respondent characteristics 
Manager or owner of shop (yes/no) 
Education level attained (yes/no) 
   None 
   Elementary school or more 
   Middle school or more 
   Highschool or more 
   University 
Shop characteristics and cold chain 
No. employees (#) 
No. functioning refridgerators (#) 
No. power outages in previous week (#) 
Duration of longest power outage (minutes) 
Product availability and transactions 
No. distinct fluid milk products available - fluid milk (#) 
No. other distinct products available - other dairy (#) 
Volumes fluid milk procured in previous week (L) 
Mean marketing margin on fluid milk (% of consumer price) 

Mean 
 

0.56 

0.13 
0.85 
0.76 
0.61 
0.51 

 

 

2.02 
1.43 
5.05 
106.35 
 

3.18 
5.17 
70.47 
0.16 

SD 
 
0.50 
 
0.34 
0.35 
0.43 
0.49 
0.50 
 
1.13 
0.63 
4.09 
161.98 

 
1.65 
3.66 
71.21 
0.03 

 
0.60 
 
0.16 
0.83 
0.75 
0.60 
0.53 
 
1.77 
1.30 
4.78 
110.49 

 
   Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
2.69 
3.76 
56.70 
0.16 

SD 
 
0.49 
 
0.37 
0.37 
0.43 
0.49 
0.50 
 
0.87 
0.49 
3.63 
175.81 

 
1.40 
2.25 
57.30 
0.02 

 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

Alimentation  (N=49) 

Mean 

 
0.45 
 
0.04 
0.92 
0.80 
0.63 
0.45 
 
2.75 
1.82 
5.85 
94.18 
 
4.61 
9.31 
118.34 
0.18 

SD 
 
0.50 
 
0.20 
0.28 
0.41 
0.49 
0.50 
 
1.45 
0.83 
5.18 
112.57 

 
1.50 
3.87 
92.20 
0.04 

 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
0.06 
 
0.03 
0.15 
0.52 
0.72 
0.30 
 

0.00E+00 
0.00E+00 

0.13 
0.54 
 

0.00E+00 
0.00E+00 
0.00E+00 
0.00E+00 

Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively.

 

20 

2.3.3 Choice experiment and dairy product attributes  

In addition to the questionnaire, each survey included a choice experiment that I tailored for 

either a consumer or retailer respondent. However, both experiments aimed at eliciting 

preferences regarding four pasteurized milk product attributes: 1) ingredients labeling, 2) third-

party quality certification, 3) packaging, and 4) price. Table 3 summarizes the four dairy product 

attributes and the four levels for each. 

 

 
Table 3: Pasteurized milk product attributes and levels in choice experiments 
Attribute 
Ingredient composition 
claim 

Attribute levels 
No claim (status quo) 
100% reconstituted powdered milk  
Fresh milk mixed with powdered milk  
100% fresh milk 

 

 
 

 

Food safety 
certification   
 

 

Packaging  

 

Price (FCFA/.5L) 
 

No certification (status quo) 
Certified by the Malian government  
Certified by a private third-party  
Double-certification by government and private third-party  

 
 

Tied transparent plastic sack (status quo for fresh milk) 
Sealed opaque plastic pouch 
Sealed transparent plastic bottle 
Sealed cardboard carton 
     
Consumers:        280, 370, 460, 550 
Retailers:            240, 310, 380, 450 
 

 

Note: The size of the hypothetical product was held constant at .5 L, which is a common size for packaged milk.  
 
 

2.3.3.1 Ingredients Composition  

The first product attribute is labeling that indicates the product’s composition in terms of 

milk ingredients. Many Bamako retailers state that inquiries about the purity and origin of dairy 

products are the most frequent questions that they receive from customers. In this market, 

pasteurized milk comes in two forms (Ibid.). First, there are a few semi-industrial and industrial 

 

21 

processors that pasteurize milk that they reconstitute entirely from powdered milk or that is some 

blend of reconstituted and fresh milk. According to some of these processors, blending milk 

ingredients is a strategy that reduces input and transaction costs while preserving some of the 

taste of local milk that many consumers desire.  

Second, there is a number of milk producer cooperatives and small artisan processors that 

pasteurize fresh milk that is sourced from peri-urban farms on the belief that many Bamako 

consumers prefer such purity to blended or powdered milk-based products. However, this is 

difficult to confirm from actual product data. Although most pasteurized milk products do 

provide some indication of ingredients, the location of this information on packaging varies 

across products. Terminology (i.e. indicating fresh or powdered milk) also varies across 

products, and in any case is always in French. This is a key observation, given high illiteracy 

rates in Mali and the presence of other packaging elements (e.g., icons of cows or the use of 

Bambara words) on powder milk-based products that could misinform buyers about the type and 

origin of the milk ingredients. For example, in the sample only 20% of retailers could correctly 

identify the ingredient composition of the most common brand of pasteurized milk. Lefèvre 

(2014) documented a similar situation in neighboring Senegal. She used stated and revealed 

choice data to confirm that urban consumers have strong preferences for fresh milk, while 

showing that misinformation regarding ingredient composition prevents them from actually 

allocating higher prices to these products.   

To obtain a detailed assessment of consumers’ and retailers’ preferences for milk ingredient 

composition, and how they value information on composition, I include an ingredients label 

attribute in the experiment, with four levels: no ingredients label provided (status quo); a label 

indicating that the product is composed of 100% fresh milk; a label indicating that the product is 

 

22 

composed of 100% reconstituted powdered milk; or label indicating that the product is composed 

of a blend of fresh milk and reconstituted powdered milk. I expect findings consistent with 

Lefèvre (2014) – that consumers have the strongest preferences for products labeled with 100% 

fresh milk but that no labeling (which can give the false impression that a product is made from 

fresh milk) is preferred to labeling that indicates any powdered milk. 

 

2.3.3.2 Quality certification  

The second product attribute is a third-party quality certification that can assure retailers and 

consumers regarding the quality and safety of a product. Experts cite two categories of dairy 

food safety hazards—biological and chemical—that pose significant public health risks. 

Biological hazards include food-borne pathogens (e.g., Salmonella, brucellosis, and tuberculosis) 

and aflatoxins, as well as spoilage organisms and hygienic contamination that can affect 

reconstituted powdered milk as well as fresh milk (Kenny, 2013; Bonfoh et al. 2003; Hetzel et 

al., 2005). Chemical hazards include residues from antibiotics and other veterinary drugs, which 

enter milk when herds are improperly treated, farmers do not respect withholding periods 

between treatment and milking, or when handlers intentionally adulterate milk to forestall 

spoiling (Kenny, 2013).  

The Malian government has already set up a system in which all food products must be 

certified by l’Agence Nationale de la Sécurité Sanitaire des Aliments (ANSSA) (Government of 

Mali, 2006). The ANSSA seal, which should be printed on product packaging, indicates to 

consumers that the product has met product-specific norms and standards that the Government 

has adapted from CODEX. However, few, if any, imported or domestic dairy products carried 

the seal at the time of fieldwork, due to high certification and compliance costs, combined with 

 

23 

historically weak enforcement by the government and limited recognition of the certification by 

Malian consumers.  

In order to provide useful information on how consumers would value ANSSA certification 

if they were sensitized to its significance, I provided a short explanation to respondents, and 

included the ANSSA seal as one level of the quality certification attribute. The other attribute 

levels were: no certification (status quo), quality certification backed by a hypothetical private 

third party (e.g., a non-government organization or a private company), or double-certification 

(i.e. ANSSA seal plus another private certification).  

The expectations regarding preferences for these attributes fall in line with similar research 

that has used experimental methods to investigate food quality certification. Focusing on Malian 

preferences for infant food quality, Masters and Sanogo (2002) found that Bamako consumers 

have a positive WTP for products that are backed by third-party quality certification. Other 

studies outside of Mali have used choice experiments to investigate the welfare gains associated 

with the certification of dairy products in particular. In Kenya, Mtimet et al. (2015) found large 

welfare effects for milk that is certified aflatoxin-free. Several studies considering different 

certification schemes in the U.S. context find welfare gains associated with most certification 

options (as opposed to none at all) but show that consumers prefer government to other industry 

of private certifying agencies (Olynk and Ortega, 2013; Wolf et al., 2011; Olynk et al., 2010).  

 

2.3.3.3 Packaging  

The third product attribute is packaging, which can create value for retailers and consumers 

by protecting product contents and improving convenience. Packaging quality can also 

potentially convey information about unobservable product attributes to buyers. Hollinger and 

 

24 

Staatz (2015) observed that improved product presentation, labeling, and packaging are key 

strategies for earning consumer acceptance of locally-processed products in West Africa. Recent 

research conducted in Bamako found that dairy processors use enhanced packaging features to 

differentiate their brands from those of rivals. For example, hedonic price analysis of retailer 

inventory data revealed that sterilized milk packaged in imported materials (e.g., cartons) earned 

a premium over locally-available materials (plastic pouches and bottles) (Theriault et al., 2018). 

In Senegal, analysis of choice-based conjoint data indicated that yoghurt consumers are willing 

to pay a price premium for individually-sized pouch packaging (possibly for its convenience 

value), and hedonic price analysis of product data revealed price premiums for cup packaging 

(versus simple pouch packaging) (Lefèvre, 2014). In Kenya, Mtimet et al. (2015) estimated 

positive WTP associated with plastic bottles and TetraPak packaging of fluid milk.  

However, research also cautions that the welfare effects of different packaging types can be 

sensitive to consumer demographics and other factors. For example, Fadiga and Makokha (2014) 

found that fluid milk consumers living in Nairobi or belonging to middle or wealthy classes 

preferred sealed to unsealed packaging, while consumers who were poor or living in another city 

were indifferent. In Bamako, the value of packaging may also depend on the ingredient 

composition of the product. Many consumers identify fresh milk products by a particular form of 

packaging: hand-tied transparent plastic sacks. However, this packaging poorly protects product 

contents while its low cost potentially undercuts its value as an effective quality-signaling 

mechanism. Although factory-sealed plastic pouches and bottles are a common type of enhanced 

packaging for dairy products, many consumers associate this packaging with powdered milk-

based products. Meanwhile, there are other types of packaging that local processors have not yet 

introduced. For example, cardboard cartons are recyclable and offer superior product protection 

 

25 

compared to plastic, but their use in Mali is currently limited to imported sterilized milk. In the 

choice experiment, I include four packaging attribute levels to explore preferences towards each 

of these four packaging types.  

 

2.3.3.4 Price  

Finally, I included price as a fourth product attribute to allow for the estimation of money-

metric measures of WTP and in order to make welfare comparisons. For each experiment, I 

specified four price levels that were consistent with actual market prices and margins. 

Specifically, I selected consumer price levels that covered the full range of actual consumer 

prices for pasteurized milk in Bamako retail shops. The retailer price levels reflect a 14-18% 

margin on the consumer price levels. As a reference point, the second levels of consumer and 

retailer prices correspond roughly to the retail and wholesale unit prices of the most established 

and widely-distributed pasteurized milk product in Bamako, Mali Lait milk. This industrially-

manufactured product has opaque plastic pouch packaging without certification and clear 

labeling of its ingredients; thus, it represents two the three status quo product attribute levels. 

The choice tasks were designed following Street, Burgess, and Louviere (2005). The product 

attributes and corresponding levels were first used to develop an orthogonal fractional factorial 

design reducing the original attribute level combinations to 16. Following, the generators 

described by Street and Burgess (2007), were used to generate 16 choice tasks composed of two 

product alternatives and no purchase option (design D-efficiency of 94.49%). I designed the 

experiments using the Ngene software. Each choice experiment included sixteen choice-sets, 

each of these consisting of two hypothetical half-liter product alternatives with the above 

attributes and one opt-out option. To avoid respondent fatigue, the design was blocked in two so 

 

26 

that each respondent evaluated eight choice tasks.6 To facilitate comprehension among choice 

experiment participants with varying levels of literacy, the survey teams presented illustrations 

of each choice to the participants. Figure 3 displays a sample choice task for a consumer and 

retailer, respectively. 

 
Figure 3: Example choice task for a consumer and retailer 
 

B 1 / S 8

B 1 / S 7

B 2 / S 7

Option A

Block 1 / Scenario 15

Lait Frais 
Pasteurisé

+

Option B

Block 1 / Scenario 14

Lait Frais 
Pasteurisé

Option A

Lait Frais 
Pasteurisé

Block 2 / Scenario 13

B 2 / S 7

Option B

Block 2 / Scenario 13

Lait Frais 
Pasteurisé

 

380 FCFA     .5L

 

450 FCFA     .5L

 

280 FCFA     .5L

 

370 FCFA     .5L

 

To reduce hypothetical bias in the choice experiment results, the survey teams delivered a 

cheap talk script to all participants (Lusk, 2003). The teams instructed consumer respondents to 

imagine themselves in an actual retail shopping scenario, while they instructed retailer 

respondents to imagine themselves purchasing merchandise from a supplier which they would 

resell in their shops.  

 

2.4 Results and discussion  

2.4.1 Results from the consumer random parameters logit models 

I first fitted the consumer choice data with an RPL model, in order to analyze preferences 

and to test for heterogeneity within the sample.7 I also estimated results in WTP-space. Both 

 

6 This yielded a statistical sample size of 1,744 observations for consumers and 1,640 observations for retailers, after 
taking into account a few respondents who did not complete all eight choice sets.  
7 In the RPL model, I keep price and opt-out as fixed, while assuming the other parameters to be random (Ubilava 
and Foster, 2009). 

 

27 

models Table 4 presents results from the RPL model. On average, consumers prefer some 

certification compared to no certification and any ingredient label compared to no label. Of the 

packaging attributes, they only have a preference for plastic bottle packaging in comparison to 

plastic sacks. The results in WTP-space are similar to those in preference-space, except that there 

is no significant WTP for any packaging attribute. These results, which represent mean effects 

for the entire consumer sample, are useful for providing an overall picture. However, the 

statistical significance of the standard deviation of every random variable in the RPL model 

supports the hypothesis of preference heterogeneity among consumers. In section 4.3 I will 

examine this heterogeneity more closely. Below, I explore the RPL results for each product 

attribute more closely.  

 
 

 

 

28 

Coeff. 

Std. 
Error 
 

 
 
0.31  *** 
3.47 
0.30  *** 
0.92 
0.33  *** 
3.95 
0.21 
-0.18 
 
* 
0.22 
0.38 
0.20 
-0.13 
 
0.28  *** 
1.13 
0.27  *** 
2.27 
0.29  *** 
1.11 
0.47 
-0.53 
 
-0.01  1.05E-03  *** 

 

1.63 
1.99 
1.54 
0.83 
1.17 
0.73 
2.42 
1.84 
2.33 

 

 

 
0.29  *** 
0.33  *** 
0.29  *** 
0.36 
** 
0.27  *** 
0.35 
** 
0.36  *** 
0.30  *** 
0.40  *** 
 
 
 
 
  

 
 
 
 
  

 

 

Coeff. 
 
431.96 
103.53 
505.04 
-30.91 
29.39 
-10.34 
138.44 
280.21 
146.73 
-676.23 
--- 

 
178.68 
237.62 
200.51 
101.26 
159.97 
128.38 
314.73 
225.14 
268.34 
 
1720 
20 
1,209.35 
1.43 

  

Std. 
Error 
 

 
 
45.72  *** 
37.94  *** 
51.02  *** 
25.51 
 
27.17 
 
29.64 
 
36.88  *** 
35.83  *** 
38.35  *** 
371.80 
* 
--- 
 

 

 

 
37.07  *** 
45.30  *** 
38.53  *** 
51.81 
* 
27.89  *** 
44.13  *** 
39.96  *** 
41.13  *** 
40.35  *** 
 
 
 
 
  

 
 
 
 
  

Table 4: Results from the consumers random parameters logit model  

 

preference space 

  

 

WTP space 

 

 

 

 
 

  

Parameter means 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 
Opt-out 
Price 

Random parameter standard deviations 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 

 

 

N 
d.f 
Log-Likelihood 
AIC/n 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. 

1720 
20 
1,213.14 
1.43 

 

29 

 
 

The results in Table 4 confirm that, on average, consumers value the ingredient composition 

of 100% fresh milk more than either 100% reconstituted powdered milk or a blend thereof. This 

confirms other stated preference data that I collected: 92% of consumers in the sample said that 

they prefer pasteurized milk that is made purely from fresh milk while less than 1% prefer 

blended milk ingredients. The estimated WTP for fresh milk labeling versus no labeling is 280 

FCFA (.53 USD) per half-liter, which is about 140 FCFA (.27 USD) more than the WTP for any 

labeling indicating powdered milk. This fresh milk-powdered milk price difference represents 

about one-third of the average price of the hypothetical products, and almost twenty percent of 

the price of the most popular pasteurized milk product sold on the Bamako market. In 

comparison, Lefèvre (2014) estimated a fresh milk-powdered milk price differential that 

amounted to about 80% of the average price in her choice-based conjoint experiment in the 

Dakar context.8 In the results, the marginal utility and WTP for blended ingredients are almost 

the same as for 100% powdered milk. This challenges the belief that consumers view the practice 

of blending milk ingredients as a value-addition to using only powdered milk. On the contrary, 

the results indicate that the average Malian—when clearly informed about the ingredient 

composition of pasteurized milk—has a strong preference for pure fresh milk-based products.  

Additionally, the results suggest that Malians are willing to pay for clear labeling in and of 

itself, regardless of the actual ingredient composition of a product. Because current packaging 

often does not provide clear labeling of ingredients, this result provides evidence of information 

asymmetry in the pasteurized milk consumer market. Yet, it is also somewhat surprising: in 

 

8 In Lefèvre (2014)’s analysis, 100% powdered milk was the sole alternative to 100% fresh milk and ranged between 
225 FCFA and 325 FCFA for an unspecified package size. In this analysis, the lower bound of the confidence 
intervals for the fresh milk-powdered milk price difference amounted to 36% of the average price.  

 

30 

Lefèvre (2014)’s hedonic price analysis of market data, she found that “ambiguous” ingredients 

labeling actually earns a small price premium in the Dakar dairy market, possibly because it 

sometimes leads consumers to falsely believe that a given product is manufactured from fresh 

milk. In the absence of clear labeling, consumers should impose their own assumptions about a 

given product’s ingredient composition. Even if this assumption reflects the least-desired 

composition of milk ingredients (e.g., 100% powdered milk), I should find no significant 

difference between preferences for no labeling and preferences for labeling indicating that 

particular ingredient composition. One possible explanation is that consumers are also concerned 

about the presence of non-dairy ingredients (e.g., preservatives) in pasteurized milk, and that 

participants in the choice experiment were mindful of this possibility when considering the 

hypothetical products with no labeling. It is also possible that the value of clear ingredients 

labeling goes beyond what it says about the product contents themselves, and that it also instills 

consumer trust in the brand. 

Table 4 also shows that each certification variable has a positive and statistically significant 

influence on consumer utility, confirming that Malians are concerned about safety and demand 

more information on these attributes. Stated preference data from the survey indicates that 

consumers and retailers appear to be most concerned about quality issues originating in the 

processing and distribution stages of the value chain (i.e., unhygienic handling, inadequate 

pasteurization, adulteration of product with water spoilage due to a weak cold chain) and less 

concerned about issues originating at the farm-level (i.e., feed quality, aflatoxins, pesticide and 

antibodies residues).  

The significant WTP coefficient for government certification suggests that consumers would 

pay a sizeable price premium—equal to about 100% of the average product price—for 

 

31 

pasteurized milk products that carried the ANSSA certification (compared to the status quo of no 

certification), if they were better-informed about the ANSSA seal. While the magnitude of this 

premium is striking, it is smaller in comparison to results from Mtimet et al. (2015)’s Kenya 

study, which estimated a WTP of almost 140% of the average product price for aflatoxin-free 

certification of fluid milk. I emphasize that better awareness of the ANSSA seal among Malian 

consumers is a critical condition for this WTP to be realized in actual markets. In the sample, 

only about 20% of consumers and retailers in the sample recognize the ANSSA seal on 

packaging, and just over 10% of the sample had some idea of the meaning.  

Significant differences in WTP across certification types confirms that part of the value of 

certification is derived from who is verifying the claim (Olynk et al., 2013). In the results, 

consumer WTP for government certification is four times the WTP for private certification. 

Double certification provides only a little additive value compared to government certification 

alone. This pattern of greater consumer confidence in government certification (compared to 

private certification) resembles the preferences of U.S. dairy consumers as reported in several 

other choice experiment studies in this context (Olynk and Ortega, 2013; Wolf et al., 2011; 

Olynk et al., 2010). The consumer survey probed further into this question by asking respondents 

to rate their level of confidence in different verifiers on a four-point Likert scale ranging from 

one (no confidence at all) to four (complete confidence). Consumers gave the highest average 

rating (3.5) to government, an intermediate rating to a Malian consumer association (2.3), and 

the lowest rating (less than two) to a Malian industry association, or to a local or international 

company or non-government association. Overall, the relatively weak WTP for private 

certification may also explain why there is no alternative third-party certification option in Mali, 

especially when one considers the large costs of setting up and maintaining such systems. 

 

32 

Turning to packaging, the mean estimates of WTP suggest that consumers do not value any 

of the three types of enhanced packaging any more than they do plastic sacks. On one hand, this 

seems surprising, given that other choice experiments examining dairy preferences in Africa 

have found positive and statistically significant WTP for enhanced packaging features (Mtimet et 

al., 2015; Fadiga and Makokha, 2014; Lefèvre 2014). Additionally, when I asked consumer 

respondents to state their ideal type of packaging for pasteurized milk, the three most popular 

responses were plastic bottle (49% of respondents), carton (22%), and plastic pouch (8%). Only 

one respondent stated that plastic sack packaging was ideal. On the other hand, this result does 

seem to be consistent with qualitative accounts that many consumers in Bamako associate such 

traditional packaging with local fresh milk. It also seems consistent with consumer responses, 

when asked to rate the level of importance they accord to packaging as a source of information 

about a product. The average rating was only 1.9, on the basis of a four-point Likert scale 

ranging from one (not important) to four (very important). In the next sub-section, I will examine 

the variation of consumer preferences with respect to packaging, which will help shed some light 

on this puzzle.  

 
2.4.2 Comparison of consumer and retailer preferences and WTP 

In order to compare the preferences and marginal rates of substitution of consumers and 

retailers, I fitted the retailer choice data with the same RPL model used for consumers. Table 5 

presents these results. Overall, the preferences and WTPs of the two groups are well-aligned, in 

terms of sign, significance, and ordering of the estimated coefficients. One exception is that for 

retailers the estimated coefficient on bottle packaging in preference space is not significant, 

while the corresponding estimate for consumers is statistically significant and positive at the 

 

33 

10% level of significance. This overall pattern of alignment suggests that retailers have good 

knowledge of consumer preferences with respect to the dairy product attributes in question. In 

their study of the Spanish seafood market, Fernández-Polanco et al. (2013) also found that 

retailer preferences towards several experience and credence attributes were well-aligned with 

customer preferences in terms of sign and significance, although the rank-order of marginal 

utility varied across actors.  

To investigate this further, Table 6 compares the WTP estimates from the consumer and 

retailer RPL models. Paired t-test results show that, for almost every attribute, there are no 

statistically significant price differences between consumer and retailer WTP.9 The one 

exception is the WTP coefficients for double certification, which have a difference that is 

statistically significant at the 10% level. At the same time, I note that for every attribute that has 

statistically significant coefficients in both datasets (i.e., all certification and ingredients 

composition variables except for private certification), the average retailer WTP is less than the 

average consumer WTP, depending on the attribute.10 This is unsurprising, because I would 

expect a share of consumer price to be captured by retailers’ marketing margins.11 I conjecture 

that retailer WTP can be interpreted as the maximum procurement price that retailers are willing 

to pay for each attribute, which theoretically takes into account retailers’ expectations about 

marketing costs, consumers’ WTP, and competition-related considerations.  

 

9 In contrast, although Polanco et al. do not conduct a similar test between consumer and retailer WTP estimates, 
they conclude that retailer and consumer WTP diverge on the basis of statistically significant differences in some 
marginal utilities including that of price (on which the estimated WTP is based). They also acknowledge that one 
weakness of their analysis is large standards errors for retailers, due to a small sample size.  
10 Fernández-Polanco et al. (2013) report a similar pattern, although in their results retailer WTP was as much as 
194% less than consumer WTP. 
11 However, I note that the percentage point differences in price shares that I examine here are not calculations of 
market shares. 

 

34 

Taken together, these results suggest that any information asymmetry in the pasteurized milk 

value chain regarding these attributes is not due to retailers’ limited understanding of consumer 

preferences. On the contrary, Bamako retailers of dairy products appear to understand well what 

their customers desire and what they are willing to pay.  

 
Table 5: Results from the retailers random parameters logit models  

RPL 

(preference space) 

  

  

Parameter means 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 
Opt-out 
Price 

Random parameter standard deviations 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 

 

 

 

 
 

Coeff. 

2.21 
0.70 
2.45 
-0.04 
0.08 
-0.06 
0.54 
1.44 
0.80 
-1.38 
-0.01 

1.33 
1.36 
1.16 
1.11 
0.98 
1.02 
1.83 
1.57 
1.40 

 

 

1544 
N 
21 
d.f. 
-1,178.80 
Log-Likelihood 
AIC/n 
1.55 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively.

Std. Error 
 

 
 
0.23  *** 
0.22  *** 
0.24  *** 
0.18 
 
0.18 
 
0.18 
 
0.22 
** 
0.23  *** 
0.23  *** 
0.40  *** 
1.07E-03  *** 
 
 
0.28  *** 
0.28  *** 
0.29  *** 
0.27  *** 
0.28  *** 
0.30  *** 
0.28  *** 
0.24  *** 
0.32  *** 
 
 
 
 
  

 
 
 
 
  

 
 

RPL 

(WTP space) 
Coeff. 
 
330.54 
101.92 
360.80 
-8.00 
4.64 
-8.28 
90.83 
234.11 
139.23 
-1,519.58 
--- 
 
 
178.38 
203.92 
153.31 
162.22 
153.49 
169.52 
271.59 
266.72 
197.66 
 
 
 
 
  

Std. Error 
 

  
 
 
52.03  *** 
35.75  *** 
57.73  *** 
28.64 
 
28.11 
 
29.21 
 
35.22  *** 
42.06  *** 
38.94  *** 
376.04  *** 

 

--- 
 
 

 
 
47.33  *** 
46.36  *** 
52.03  *** 
45.30  *** 
34.78  *** 
48.61  *** 
46.07  *** 
51.90  *** 
50.18  *** 
 
 
 
 
  

 
1544 
21 
-1,261.28 
1.55 

 

35 

Table 6: Comparison of consumer and retailer WTP  
  

Consumers (N=218) 

  

Attribute 

Mean WTP 
(FCFA/.5L) 

Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively.

431.96 
103.53 
505.04 
-30.91 
29.39 
-10.34 
138.44 
280.21 
146.73 

 
 
 
 
 
 
 
 
 
  

SE 
45.72 
37.94 
51.02 
25.51 
27.17 
29.64 
36.88 
35.83 
38.35 

  

  
*** 
*** 
*** 

 
 
 

*** 
*** 
*** 

  

 
 
 
 
 
 
 
 
 
  

Retailers (N=193) 

  

Mean WTP 
(FCFA/.5L) 

330.54 
101.92 
360.80 
-8.00 
4.64 
-8.28 
90.83 
234.11 
139.23 

SE 
52.03 
35.75 
57.73 
28.64 
28.11 
29.21 
35.22 
42.06 
38.94 

  
*** 
*** 
*** 

 
 
 

*** 
*** 
*** 

  

 
 
 
 
 
 
 
 
 
  

Difference  

101.42 
1.61 
144.24 
-22.91 
24.75 
-2.06 
47.61 
46.10 
7.50 

  

∆ p-value 

0.14 
0.98 

     0.06   * 

0.55 
0.53 
0.96 
0.35 
0.40 
0.89 

 

36 

2.4.3 Results from the latent class models  

Given the statistical significance of the standard deviation estimates in the RPL models, I 

fitted the consumer and retailer datasets with LC models in order to further investigate 

preference heterogeneity among these actors. The comparative advantage of the LC model is that 

it facilitates the identification of market segments (i.e., different groups of consumers and 

retailers with similar underlying preference) via covariates in the class membership function. 

Because formal statistical tests are not available for identifying an optimal number of segments 

in a population, in empirical applications it is common to make this choice based on information 

criteria, such as the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), 

and the Consistent Akaike Information Criterion (CAIC) (Pacifico and Yoo, 2013).12 For the 

consumer and retailer models I identified three and two classes as optimal, respectively, based on 

these information criteria, while also ensuring that each class was large enough for statistical 

analysis and that each had a statistically significant price coefficient. 

I also tried multiple combinations of potential class membership variables and ultimately 

selected a combination of variables that improved model fit and resulted in negative statistically 

significant price coefficients (as is consistent with conventional demand theory). For the 

consumer model, I included three covariates in the latent class membership model. The first two 

covariates are a dummy variable estimating whether or not the household is below the national 

poverty line and a continuous variable indicating the share of household members that under the 

age of fourteen. I expect households with higher incomes and more children to have stronger 

 

12 AIC, BIC, and CAIC are each estimators of the relative quality of a model. Each takes into account the model 
goodness of fit (measured by -2lnL where L is the maximized sample log likelihood statistic) and the simplicity of 
the model (measure by some function of the number of parameters). The criteria differ only in how each measures 
the latter aspect, with BIC and CAIC penalizing models with extra parameters more heavily (i.e., by using penalty 
functions that increase in the number of choice makers) compared to AIC (Pacifico and Yoo, 2013).   

 

37 

preferences for product quality, while greater income should also positively influence WTP by 

reducing price sensitivity. The third covariate is household ownership of a motor vehicle, 

reflecting the hypothesis that access to a means of transportation affects fluid milk preference by 

improving household access to fresh milk, and perhaps. Additionally, consumers transporting 

milk by a motor-vehicle (e.g., by motorcycle, the most common means of transportation in 

Bamako) may have different preferences for packaging compared to others.  

For retailers, I included one covariate, a dummy indicating whether the retail shop is a 

boutique. On one hand, because these shops have limited refrigerated shelf-space and tend to 

offer less variety than alimentations, I might expect these retailers to have weaker preferences 

for novel product attributes. However, Theriault et al. (2018) found that boutiques in Mali tend to 

offer a larger share of local dairy products compared to more modern formats, including 

alimentations and supermarkets, which tend to carry a higher share of imported products.  

I present results from the consumer and retailer LC models in Tables 7 and 9, respectively, 

along with derived WTP estimates. Each probability reported at the top of these tables refers to 

the probability that a randomly-selected household or retailer belongs to that class. The estimated 

thetas, when statistically significant, indicate whether the class membership variable is positively 

or negatively correlated with membership in that class, compared to membership in class three 

(the base class in each model). In Tables 8 and 10, I also present descriptive statistics for each 

consumer and retailer class to allow for additional class profiling.  

 

 

38 

 

Table 7: Results from latent class analysis of consumers  
Table 7: Results from latent class analysis of consumers  

  

 

 
Parameter means 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 
Opt-out 
Price 

 

Class probability 
 
Thetas in class probability model 
HH owns motor-vehicle 

Coeff. 

2.94 
-2.36 
3.35 
-0.07 
-0.29 
-0.68 
1.56 
2.89 
3.25 
1.89 
-0.01 

0.17 

 

 

 

 

0.57 

0.83 

 

1.40 
0.48 

-2.09 
0.10 

5175 
42 

Share of HH members who are <15 
years 
 
HH below international poverty line 
 
 
 
N 
 
No. of parameters 
 
Log-Likelihood 
 
AIC 
 
BIC 
  
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. 
 
 
 
 

-1182.01 
2446.01 
2714.63 

 
 
 
 
 
 
 
  

 

 

Class 1: Price-sensitive  

Class 2: Fresh milk-focused  

Class 3: Quality-conscious  

Preference-space 

WTP-space 

Preference-space 

WTP-space 

Preference-space 

SE 
 

 
 

2.36E-03 

*** 

Coeff. 

 

362.49 
-291.43 
413.80 
-8.30 
-36.05 
-83.34 
192.38 
356.88 
401.51 
233.37 

 
 
 
 
 
 

 

*** 

 
 
 
** 
*** 
*** 

 

*** 

 
 
 
 
 

0.79 
3.95 
0.78 
0.74 
0.65 
0.62 
0.85 
0.75 
0.81 
1.90 

 
 
 

 
 
 
 
 
  

95% Confidence 

Intervals 

 

[60.08, 664.91] 

[-1,241.59, 658.74] 

[77.81, 749.78] 
[-185.18, 168.58] 
[-182.18, 110.08] 
[-220.00, 53.33] 
[-79.88, 464.64] 
[68.39, 645.36] 
[60.83, 742.18] 
[-338.18, 804.92] 

Coeff. 

 

-0.30 
0.49 
-0.30 
0.43 
0.32 
0.25 
0.48 
1.54 
0.45 
-0.51 

SE 
 

0.32 
0.29 
0.35 
0.29 
0.31 
0.27 
0.30 
0.28 
0.32 
0.56 

 
 
 
 
 
 
 
 
 

*** 

 
 

Coeff. 

 

-107.26 
175.16 
-107.13 
154.73 
114.95 
90.03 
171.63 
551.11 
159.37 
-181.55 

95% Confidence 

Intervals 

 
[-346.04, 131.53] 
[-65.66, 415.98] 
[-376.14, 161.88] 
[-79.62, 389.07] 
[-115.91, 345.80] 
[-114.66, 294.72] 
[-64.27, 407.52] 
[76.66, 1,025.56] 
[-105.40, 424.15] 
[-478.08, 114.98] 

Coeff. 

 

3.11 
1.70 
3.60 
-0.06 
0.58 
0.09 
0.83 
1.38 
0.84 
0.52 

 
 
 
 
 
 

 
 
 
 
 
 
 
  

-2.80E-03 

1.11E-03 

*** 

 

0.21 

 
 

 
 
 
 

 
 
 
 

-1.93 

0.58 

*** 

3.42 
-1.14 

1.40 
0.57 

 
 
 
 
 
  

 
 
 
 
 
  

** 
** 
 
 
 
 
 
  

 
 
 
 
 

 
 
 
 
 

 

 
 

  

 
 
 
 
 

 
 
 
 
 
 
 

 

-3.17E-03 

 

0.63 

 
 
 

 
 
 
 
 
 
 

 
 
 

 
 
 
 
 
 
 
  

  

  

 

SE 

0.28 
0.30 
0.33 
0.20 
0.25 
0.21 
0.21 
0.23 
0.25 
0.64 

9.00E-04 
 

 

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
  

 

 
*** 
*** 
*** 

 
** 
 

*** 
*** 
*** 

 

*** 
 
 
 
 
 

 
 
 
 
 
 
 
  

WTP-space 

Coeff. 

 

980.78 
534.82 
1136.91 
-18.58 
183.28 
29.95 
261.99 
435.68 
266.10 
164.90 

95% Confidence 

Intervals 

 
[400.06, 1,561.50] 
[167.90, 901.74] 
[452.77, 1,821.05] 
[-138.14, 100.97] 
[-24.75, 391.30] 
[-108.60, 168.49] 
[48.07, 475.90] 
[107.92, 763.44] 
[40.70, 491.50] 
[305.62, 635.43] 

 

 

 
 
 
 
 

 
 
 
 
 
 
 
  

 
 
 
 
 

 
 
 
 
 
 
 
  

 

 

 

39 

39 

 
0.58 
0.16 
0.45 
0.15 
0.64 
0.53 
 

  
  Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

11.22 
0.37 
0.42 
0.44 
0.06 
0.06 
 
1.14 
2.51 
1.39 
2.61 
2.89 
 
1.97 
1.63 

SD 
 
0.50 
0.25 
0.37 
0.27 
0.49 
0.51 
 
6.61 
0.17 
0.50 
0.50 
0.08 
0.08 
 
0.49 
1.63 
0.84 
1.29 
1.39 
 
1.54 
0.49 

 
0.72 
0.17 
0.58 
0.14 
0.89 
0.54 
 

  
  Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

13.47 
0.32 
0.56 
0.62 
0.07 
0.07 
 
1.43 
2.86 
1.58 
3.35 
3.13 
 
2.46 
1.75 

SD 
 
0.45 
0.23 
0.36 
0.24 
0.31 
0.50 
 

10.17 
0.17 
0.50 
0.49 
0.07 
0.07 
 
0.93 
1.66 
1.02 
0.96 
1.25 
 
1.42 
0.44 

∆ p-value 

  

 
0.40 
0.24 
0.01 
0.76 
0.04 
0.99 
 
0.12 
0.02 
0.21 
0.42 
0.19 
0.49 
 
0.26 
0.19 
0.71 
0.02 
0.26 
 
0.24 
0.07 

  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

  

 
 
 
 

*** 

 
** 
 
 
 
** 
 
 
 
 
 
 
 
 
** 
 
 
 

* 

Class 1: 
Price-
sensitive 
(N=32) 

Class 2: 
Fresh milk-
focused 
(N=36) 

Class 3: 
Quality-
conscious 
(N=148) 

 
0.75 
0.24 
0.38 
0.20 
0.94 
0.53 
 

  
  Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

10.09 
0.26 
0.41 
0.50 
0.05 
0.05 
 
1.59 
3.28 
1.63 
3.25 
3.56 
 
2.25 
1.50 

SD 
 
0.44 
0.25 
0.37 
0.33 
0.25 
0.51 
 
7.68 
0.16 
0.50 
0.51 
0.07 
0.07 
 
1.04 
1.55 
1.07 
1.08 
0.84 
 
1.46 
0.51 

Table 8: Profiles of consumers from each latent class  

Variable 

 
Househld access to, storage of, and consumption of milk (yes/no) 
Consumed any fluid milk in past week 
Purchased any dairy product from a traditional outlet in past week  
Purchased any dairy product from a boutique in past week 
Purchased any dairy product from a modern outlet in past week 
Owns a vehicle 
Owns a functioning refridgerator and has access to electricity 
 
Household composition 
Total household members+ 
Share of household members that are under 14 years of age 
Household includes an infant (< 2 years) (yes/no) 
Share of household members who are infants  an infant (< 2 years)  
Household includes a pregnant or lactating mother (yes/no) 
Share of household members who are mothers 
 
Knowledge and preferences regarding dairy products (1/disagree - 5/strongly agree) 
Recognize and know meaning of ANSSA seal 
Food safety of imports is superior to Malian dairy products 
Environmental issues are key element in packaging preferences  
Closeability is key element in packaging preferences  
Transparence is key element in packaging preferences  
 
Respondent characteristics 
Educational attainment of respondent (1/little none - 4/university) 
Respondent is primary household decisionmaker (vs secondary) purchases (yes/no) 
Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. 
 

 

Total 
(N=218) 

Mean 
 

0.70 
0.18 
0.53 
0.15 
0.86 
0.54 

 
12.59 
0.32 
0.52 
0.06 
0.58 
0.06 

 

 

1.40 
2.85 
1.55 
3.20 
3.13 

2.35 
1.69 

SD 
 
0.46 
0.24 
0.37 
0.26 
0.35 
0.50 
 
9.39 
0.17 
0.50 
0.07 
0.50 
0.07 
 
0.89 
1.65 
0.99 
1.08 
1.25 
 
1.45 
0.46 

 

40 

Table 9: Results from latent class analysis of retailers  

Class 1: Fresh milk-focused 

Class 2: Higher variety  

Model results 

Derived WTP 

Model results 

Derived WTP 

Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. 
 
 

 

 

Coeff. 

1.56 
0.62 
1.61 
0.15 
0.31 
0.31 
0.23 
0.76 
0.31 
-2.38 

SE 
 

0.16 
0.16 
0.16 
0.14 
0.16 
0.15 
0.15 
0.15 
0.17 
0.50 

-1.77E-03 

7.98E-04 

  

 

 
Parameter means 
Government certification 
Private certification 
Double-certification 
Pouch packaging 
Bottle packaging 
Carton packaging 
100% powdered milk claim 
100% fresh milk claim 
Blended ingredients claim 
Opt-out 
Price 

Class probability 

 

 

Thetas in class probability model 
Retail outlet is a boutique 

 

 

 

 
 

 

N 
No. of parameters 
Log-Likelihood 
AIC/n 
BIC/n 

0.69 

0.68 

4,632 
39 

-1162.19 

0.51 
0.55 

 
 
 
 

0.37 

 
 
 
 
 
  

SE 
 

0.31 
0.29 
0.29 
0.23 
0.26 
0.24 
0.38 
0.36 
0.29 
0.68 

1.26E-03 

 
 
 
 
 
 
 
 
 
 
 
  

 
 

*** 

 

*** 

* 

 

 

* 
*** 
*** 

 

*** 

 
 
 
 
 
 
 
 
 
 
  

Coeff 

95% Confidence Intervals 

174.20 
54.96 
224.48 
-14.90 
-89.89 
-60.90 
127.94 
251.61 
249.40 
-54.73 

--- 

 

 
 
 

 
 
 
 
 
 
 
  

 

[29.08, 319.32] 
[-61.45, 171.37] 
[65.98, 382.98] 
[-98.18, 68.37] 
[-175.86, -3.92] 
[-146.34, 24.53] 
[-20.65, 276.53] 
[71.24, 431.98] 
[87.97, 410.82] 
[-285.97, 176.50] 

--- 
 
 
 
 
 
 
 
 
 
 
  

 

 
 
 

 

 
 

*** 
*** 
*** 

 

* 
** 
 

*** 
* 
*** 
** 
 
 
 
 

* 

 
 
 
 
 
  

Coeff 

95% Confidence Intervals 

Coeff. 

877.34 
348.60 
909.62 
85.68 
173.18 
173.35 
131.97 
429.86 
176.44 
-1343.26 

--- 

 

 
 
 
 
 
  

 

0.93 
0.29 
1.20 
-0.08 
-0.48 
-0.33 
0.68 
1.35 
1.34 
-0.29 
-0.01 

 

0.32 

 
 
 
 
 
 
 
 
  

 

[94.79, 1,659.89] 
[-10.99, 708.19] 
[78.09, 1741.15] 
[-92.47, 263.84] 
[-66.01, 412.36] 
[-56.03, 402.73] 
[-85.50, 349.44] 
[-3.91, 863.64] 
[-58.49, 411.38] 

[-2,346.45, -340.06] 

--- 
 
 
 
 
 
 
 
 
 
 
  

41 

Table 10: Profiles of retailers from each latent class  

Variable 

 
Retailer type, size, product availability 
Retailer is boutique format (yes/no) 
Store is alimentation format (yes/no) 
Total number of employees 
Total number of unique fluid milk products available 
Total number of other unique dairy products available 
Volumes (L) of fluid milk purchased per week 
Concerned about milk quality issues (1/disagree - 5/strongly agree) 
Feed quality of dairy herd  
Aflatoxins  
Pesticide residues  
Pathogens associated with unvaccinated herds  
Antibiotic residues   
Pathogens associated with unhygienic milk handling  
Inadequate pasteurization 
Quality deterioration due to weak cold chain  
Adulteration of product with water 
Respondent characteristics 
Whether respondent is owner or manager 
Education level attained 

Note: ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively.

Class 1: 
Fresh milk-
focused 
(N=135) 

Class 2: 

Higher variety  

(N=58) 

Total (N=193) 
SD 
Mean 
 
 
0.44 
0.44 
1.13 
1.65 
3.66 
71.21 

0.75 
0.25 
2.02 
3.18 
5.17 
70.47 
 

 
0.43 
0.34 
0.44 
0.48 
0.42 
0.48 
0.50 
0.50 
0.48 
 
0.50 
1.48 

0.24 
0.13 
0.27 
0.35 
0.23 
0.65 
0.56 
0.55 
0.66 

0.56 
2.78 

 

  
  Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
0.79 
0.21 
2.02 
3.01 
4.90 
68.01 
 
0.24 
0.15 
0.30 
0.32 
0.22 
0.64 
0.52 
0.58 
0.61 
 
0.60 
2.84 

 
0.64 
0.36 
2.00 
3.57 
5.79 
76.33 

  
  Mean 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
0.22 
0.10 
0.21 
0.43 
0.26 
0.67 
0.66 
0.48 
0.76 
 
0.48 
2.66 

SD 
 
0.41 
0.41 
1.12 
1.54 
3.68 
68.50 
 
0.43 
0.36 
0.46 
0.47 
0.42 
0.48 
0.50 
0.50 
0.49 
 
0.49 
1.50 

 

∆ p-value 

 
 
0.02 
0.02 
0.90 
0.03 
0.12 
0.48 
 
0.76 
0.41 
0.20 
0.13 
0.58 
0.71 
0.08 
0.23 
0.05 
 
0.13 
0.28 

 
 
** 
** 

** 

 

 

* 

* 

  

 

 
 

 
 
 
 
 
 

 

 

SD 
 
0.48 
0.48 
1.15 
1.84 
3.55 
77.66 
 
0.42 
0.31 
0.41 
0.50 
0.44 
0.47 
0.48 
0.50 
0.43 
 
0.50 
1.46 

 

42 

The largest “quality-conscious” consumer class (class three, representing 63% of consumers) 

appears to be the most conscious of dairy quality. It has the highest WTP for government 

certification and is the only class to significantly value private certification. It is also the only 

class with a statistically significant WTP for an enhanced type of packaging, i.e. factory-sealed 

plastic bottles. The class profiles (Table 9) highlight a factor that may be driving this distinct 

preference. I asked respondents to rate the ability to re-close product packaging as an important 

consideration when purchasing dairy products, on a Likert scale ranging from one (not at all 

important) to four (very important). Respondents in class three had the highest average rating 

(3.4), compared to classes one (3.2) and two (2.6). Finally, consumers in this class have positive 

WTP for any ingredients label, but value 100% fresh milk the most. They value blended 

ingredients about the same as 100% powdered milk.  

Compared to class two (i.e., based on the class two membership covariates), households 

owning a motor vehicle are more likely to belong to this class, as are households that are below 

the poverty line. Although the pairing of these household characteristics may seem 

counterintuitive, they may be explained by the fact that poor households have a much larger 

number of household members (sixteen on average) compared to non-poor households (eight on 

average), which overall should increase the likelihood that the household includes at least one 

person who owns a vehicle. Households with larger shares of adult members are also more likely 

to belong to this class, compared to class two. Table 8 shows that these consumers are also the 

most-likely to have purchased a dairy product from a boutique in the previous week.  

The second largest “fresh milk-focused” class (class two, representing 21% of consumers) is 

exclusively focused on the purity and freshness of product ingredients. Of all classes, it has the 

highest WTP for 100% fresh milk, and does not value any other ingredient composition nor any 

 

43 

other product attribute. Compared to class three, above the poverty line, households with larger 

shares of children, and households without a motor-vehicle are most likely to be in this consumer 

class. These households are also the least-likely to have purchased a dairy product from a 

boutique in the previous week.  

The smallest “price-sensitive” consumer class (class one, representing 17% of consumers) is 

the most price-sensitive, as indicated by its relatively low price coefficient in preference-space. 

This class demands more quality and product information than the fresh milk-focused 

consumers. However, it values fewer attributes than quality-conscious consumers, i.e. it does not 

value certification that is not at least backed by the government and does not value any enhanced 

packaging. Additionally, WTP values are also generally smaller than those of the quality-

conscious consumers. One exception is that, of all consumers, this class has the highest WTP for 

blended milk products, and even values this attribute more than 100% fresh milk products. Table 

8 shows that these consumers have the lowest average share of children under fourteen in their 

households. These households also have the highest share of motor vehicle ownership and were 

the least likely to purchase a dairy product from a boutique in the previous week (but may have 

purchased from another outlet).  

Turning now to the LC results for retailers (Tables 9 and 10), I identified two segments of 

retailers: “fresh milk-focused” retailers (representing 69% of retailers) and “higher variety” 

retailers (representing 32% of retailers. As its name suggests, the fresh milk-focused segment 

only has statistically significant preferences for labeling that indicates some inclusion of fresh 

milk (pure or blended), and has the largest WTP (at the 10% level) for pure fresh milk labeling. 

The negative and statistically significant marginal utility of opting out suggests that these 

retailers value the ability to offer a pasteurized milk product, as opposed to offering none. In 

 

44 

contrast, higher variety retailers have positive marginal utilities associated with any type of milk 

ingredients labeling and are willing to pay a smaller price differential for fresh milk labeling. 

Consistent with Theriault et al., (2018), results from the class membership model show that 

boutique shops are most likely to fall into the fresh milk focused segment. According to Table 

10, about 80% of shops in this segment are boutiques, compared to 64% of higher variety 

retailers. Although this latter segment tends to offer more variety of fluid milk products, these 

are not necessarily local fresh milk-based products. The descriptive statistics in Table 10 suggest 

that retailer attitudes towards a couple quality issues may partially explain the differentiated 

preferences towards fresh milk. On average, higher variety retailers expressed greater concern 

over inadequate pasteurization and the illicit addition of water, two quality control problems that 

are commonly associated with informal fresh milk supply chains.  

However, both segments have significant WTP for any certification entirely or partially 

backed by the government. Fresh milk-focused retailers have a much larger WTP for these 

attribute levels, and are also the only segment with a positive WTP for private certification. 

Fresh milk focused retailers have positive marginal utilities associated with bottle and carton 

packaging while higher variety retailers have negative marginal utility associated with bottle 

packaging; however, in each case the WTP is not statistically significant at the 10% level.  

 

2.5 Conclusion 

This study investigates information asymmetry as a potential problem that can help to 

explain the limited consumption of fresh milk in urban Mali, despite accounts that Malians 

largely prefer fresh milk to imported powdered milk. In a novel approach to using choice 

experiments to study an issue that has value chain-wide implications, I conducted parallel 

 

45 

discrete choice experiments on random samples of consumers and retailers in Bamako. Due to 

preference heterogeneity within these populations, I analyzed the data using random parameters 

logit and latent-class models and compared the results across samples. Five key findings emerge 

from this analysis, which have several implications for agribusiness strategy and government 

policy.  

First, I find a positive and significant WTP for attributes that provide information on product 

ingredients, safety, and other dimensions of quality. Together, these results provide evidence of 

information asymmetry between the manufacturers and buyers of pasteurized milk products. This 

resonates broadly with Lefèvre (2014)’s study of the Dakar dairy market, and I echo that paper’s 

recommendation for measures that improve information flows amongst consumers and other 

value chain actors.  

One policy option is to require dairy manufactures to clearly indicate milk ingredients on 

product packaging using standardized wording or icons (to accommodate illiterate consumers). 

Manufacturers’ compliance with such a regulation should cost them little; furthermore, the fact 

that consumers have a positive WTP for clearer ingredients labeling in and of itself (i.e., 

regardless of what milk ingredients the product actually contains) should operate as a positive 

incentive for compliance. The Malian government could develop other labeling regulations that 

more specifically aim at improving the competitiveness of fresh milk-based dairy products, 

although passing and enforcing such policies are likely to be more challenging. For example, the 

government could limit the use of certain words (e.g., local terms for milk), phrases (e.g., “made 

in Mali”), and imagery (e.g., images of cows in pastures) on powdered milk-based product 

packaging, which might mislead Malian consumers to believe that they are consuming locally-

sourced fresh milk-based products (Lefèvre, 2014).  

 

46 

The second finding identifies another example of a policy enhancement that could improve 

information flows and reduce information asymmetry. I find that consumers have a specific 

preference for government certification of product quality, compared to private certification, and 

are willing to pay an important price premium for this assurance. This result suggests that dairy 

manufacturers can create value for their brands by improving product quality and adopting the 

existing government ANSSA certification, which is also required by law. However, several other 

conditions must be met in order for the ANSSA system to be effective. Among these, the Malian 

government must improve consumers’ awareness of ANSSA certification, e.g., through public 

information campaigns and requiring more prominent placement of the ANSSA seal on 

packaging. Simultaneously, it must strengthen enforcement of corresponding regulations in the 

dairy market, e.g., through better monitoring and sanctioning for counterfeit certifications.  

Third, I find that Bamako consumers are willing to pay a significant price premium for 

pasteurized milk that is made purely from fresh milk. Further, LC analysis indicates that nearly 

85% of consumers prefer fresh milk most, compared to pure or blended powdered milk which 

these consumers value similarly. Given the current limited availability of fresh milk-based dairy 

products in Bamako, this finding points to an important market opportunity for fresh milk 

producers and processors. However, in order to successfully compete against imported powdered 

milk, the Malian fresh milk value chain must identify upgrades that reduce production and 

transaction costs while better differentiating their brands from those that are manufactured from 

powdered milk. The ingredients labeling and certification mechanisms that I analyze in this study 

are two options for improving differentiation.   

The fourth finding pertains to another possible upgrade – enhanced packaging. However, the 

analysis showed that only one type of enhanced packaging (bottle packaging) is valued by just 

 

47 

one consumer segment (representing about 65% of consumers) who may especially appreciate 

their ability to reclose this packaging. Overall, consumers do not have strong preferences for 

upgrades from transparent plastic sack packaging. On one hand, this finding is surprising, given 

that other dairy consumer studies show that consumers derive value from higher-quality 

packaging. On the other hand, other research in Bamako shows that many consumers may view 

traditional packaging as a signaling mechanism for fresh milk-based products. I suggest that 

packaging preferences are especially dependent on particular packaging features and can vary 

widely across consumers, and I this as an important area for further research.  

Fifth, the analysis of retailer data demonstrates that retailer preferences and WTP are well-

aligned with those of consumers. These results suggest that retailers can be a useful source of 

information and, hence, a strategic distribution partner for upstream value chain actors wishing 

to better understand consumer preferences and demand. However, the LC analysis reveals 

significant segmentation among retailers, implying that pasteurized milk manufacturers should 

carefully select their distributors. For example, fresh milk products may obtain the greatest price 

premium among high-volume retailers that prioritize being able to offer a variety of dairy 

products. In contrast, many low-volume, low-variety retailers (representing about 30% of 

retailers) have zero WTP for fresh milk and any other product attributes. A third segment 

(representing another 40% of retailers) has positive WTP for each attribute, including packaging. 

Overall, these insights demonstrate the usefulness of complementing information on consumers 

with retailer analysis in a food market study. The paper illustrates the use of stacked choice 

experiments as one promising tool for this endeavor. 

As with any stated preference method, there are some limitations that need to be considered.  

The hypothetical nature of choice experiments can potentially introduce a negative bias in 

 

48 

estimates of the marginal utility of price, and thus potentially overinflate WTP. Ideally, 

comparable future research should evaluate the results against results generated from other 

methods that mitigate hypothetical bias, i.e. analysis of actual market data (once available) or the 

use of incentive-compatible valuation methods. Additionally, one potential limitation of adapting 

choice experiments for retailers is that the design requires them to select a single product 

alternative, while in reality retailers may purchase multiple products at once in order to offer 

variety to their clients or to test out demand for new products. This could possibly introduce an 

upward bias on the marginal utilities estimated for retailers, which would lead to greater 

overestimates of retailer WTP. Taking these words of caution together, value chain managers 

and policy makers may wish to focus on the lower bound values of the WTP confidence 

intervals. Nonetheless, because the present study is the first of its kind to estimate WTP in the 

Malian dairy sector—as well as the first to compare retailer and consumers preferences using 

choice experiment methods in a developing market context—I expect that it will help to inform 

and benchmark future research in these areas.  

 
 

 

 

 

 

49 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

REFERENCES 

 

50 

 

 

 

REFERENCES 

Adamowicz, W., Boxall, P., Williams, M., and Louviere, J. (1998). Stated Preference 

Approaches for Measuring Passive Use Values: Choice Experiments and Contingent 
Valuation. American Journal of Agricultural Economics, 80(1), 64–75.  

 
Bonfoh, B., Wasem, A., Traoré, A. N., Fané, A., Spillmann, H., Simbé, C. F., Alfaroukh, I.O., 

Nicolet, J., Zinsstag, J. (2003). Microbiological quality of cows’ milk taken at different 
intervals from the udder to the selling point in Bamako (Mali). Food Control, 14(7), 495–
500. 

 
Boxall, P.C., and Adamowicz, W.L., 2002. Understanding heterogeneous preferences in random 
utility models: a latent class approach. Environmental and Resource Economics 23 (4), 421–
446.  

 
Caswell, J. A., and Mojduszka, E. M. (1996). Using informational labeling to influence the 
market for quality in food products. American Journal of Agricultural Economics, 78(5), 
1248-1253. 

 
Fadiga, M. L. and Makokha, S. (2014). Consumer valuations of the quality and safety attributes 
of milk and meat in Kenya. African Journal of Agricultural and Resource Economics, 9(2): 
91–105. 

 
Fernández-Polanco, J., Loose, S. M., and Luna, L. (2013). Are Retailers’ Preferences for 

Seafood Attributes Predictive for Consumer Wants? Results from a Choice Experiment for 
Seabream (Sparus aurata). Aquaculture Economics and Management, 17(2), 103–122.  

 
Food and Agriculture Organization (2017). FAOSTAT. Retrieved from 

http://www.fao.org/faostat/en/#data 

 
Government of Mali. (2006). DECRET N°06-259/P-RM DU 23 JUIN 2006 INSTITUANT 
L’AUTORISATION DE MISE SUR LE MARCHE DES DENREES ALIMENTAIRES, 
DES ALIMENTS POUR ANIMAUX ET DES ADDITIFS ALIMENTAIRES.  

 
Government of Mali. (2017). MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 

NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. 2017. 
RAPPORT ANNUEL 2017. 

 
Government of Mali. (2018). INSTITUT NATIONAL DE LA STATISTIQUE DU MALI. 

CONSOMMATION, PAUVRETE, BIEN-ETRE DES MENAGES: AVRIL 2017-MARS 
2018.  

 
Hetzel, M. W., Bonfoh, B., Farah, Z., Simbe, C. F., Alfaroukh, O. I., and Zinsstag, J. (2005). 

Milk consumption pattern in an area with traditional milk production: data from a case 

 

51 

control study in peri-urban Bamako, Mali. Revue Africaine de Santé et de Productions 
Animales, 3, 174-7. 

 
Hollinger, F., and Staatz, J. M. (2015). Agricultural Growth in West Africa. Market and policy 

drivers. FAO, African Development Bank, ECOWAS.  

 
Kenny, Mary. (2013). “Chapter 6: Safety and Quality.” In Muehlhoff, E. et al., ed. Milk and 

Dairy Products in Human Nutrition. Food and Agriculture Organization of the United 
Nations (FAO). Pp. 243-273.  

 
Lancaster, K. J. (1966). A new approach to consumer theory. Journal of Political 

Economy, 74(2), 132-157. 

 
Lefèvre, M. (2014). Do Consumers Pay More for What They Value More? The Case of Local 

Milk-based Dairy Products in Senegal. Agricultural and Resource Economics Review, 
43(01), 158–177.  

 
Lusk, J. L. (2003). Effects of cheap talk on consumer willingness-to-pay for golden 

rice. American Journal of Agricultural Economics, 85(4), 840-856 

 
Manski, C. F. (1977). The structure of random utility models. Theory and Decision, 8(3), 229-

254. 

 
Masters, W. A., and Sanogo, D. (2002). Welfare Gains from Quality Certification of Infant 

Foods: Results from a Market Experiment in Mali. American Journal of Agricultural 
Economics, 84(4): 974–989.  

 
McFadden, D. (1974). The measurement of urban travel demand. Journal of Public 

Economics, 3(4), 303-328. 

 
McFadden, D., and Train, K. (2000). Mixed MNL models for discrete response. Journal of 

Applied Econometrics, 15(5), 447–470.  

 
Meyer, C., and Duteurtre, G. (1998). “Equivalents lait et rendements en produits laitiers: modes 

de calculs et utilisation.” Revue d’élevage et de médecine vétérinaire des pays 
tropicaux, 51(3), 247-257. 

 
Mtimet, N., Walke, M., Baker, D., Lindahl, J. F., Hartmann, M., and Grace, D. (2015). Kenyan 

awareness of aflatoxin: An analysis of processed milk consumers. 

 
Olynk, N. J., and Ortega, D. L. (2013). Consumer preferences for verified dairy cattle 
management practices in processed dairy products. Food Control, 30(1), 298–305  

 
Olynk, N. J., Tonsor, G. T., and Wolf, C. A. (2010). Consumer Willingness to Pay for Livestock 

Credence Attribute Claim Verification. Journal of Agricultural and Resource Economics, 
35(2), 261–280. 

 

52 

 
Ortega, D. L., Wang, H. H., Wu, L., and Olynk, N. J. (2011). Modeling heterogeneity in 

consumer preferences for select food safety attributes in China. Food Policy, 36(2), 318–324. 

 
Otieno, D. J., Ruto, E., and Hubbard, L. (2011). Cattle Farmers’ Preferences for Disease-Free 
Zones in Kenya: An application of the Choice Experiment Method. Journal of Agricultural 
Economics, 62(1), 207–224. 

 
Pacifico, D., and Yoo, H. I. (2013). lclogit: A Stata command for fitting latent-class conditional 
logit models via the expectation-maximization algorithm. The Stata Journal, 13(3), 625-639. 

 
Reardon, T. A. (Ed.). (2012). The quiet revolution in staple food value chains: enter the dragon, 

the elephant, and the tiger. Metro Manila, Philippines: Asian Development Bank ; 
International Food Policy Research Institute.  

 
Rosenman, R. E., and Wilson, W. W. (1991). Quality differentials and prices: are cherries 

lemons?. The Journal of Industrial Economics, 649-658. 

 
Ruto, E., Garrod, G., and Scarpa, R. (2008). Valuing animal genetic resources: a choice 

modeling application to indigenous cattle in Kenya. Agricultural Economics, 38(1), 89–98. 

 
Scarpa, R., Thiene, M., and Train, K. (2008). Utility in Willingness to Pay Space: A Tool to 
Address Confounding Random Scale Effects in Destination Choice to the Alps. American 
Journal of Agricultural Economics, 90(4), 994–1010.  

 
Skytte, H., and Bove, K. (2004). The concept of retailer value: A means-end chain analysis. 

Agribusiness, 20(3), 323–345.  

 
Street, D. J., Burgess, L., and Louviere, J. J. (2005). Quick and easy choice sets: constructing 

optimal and nearly optimal stated choice experiments. International Journal of Research in 
Marketing, 22(4), 459-470. 

 
Street, D. J., and Burgess, L. (2007). The construction of optimal stated choice experiments: 

Theory and methods (Vol. 647). John Wiley and Sons. 

 
Theriault, V., Vroegindewey, R., Assima, A., and Keita, N. (2018). Retailing of Processed Dairy 

and Grain Products in Mali: Evidence from a City Retail Outlet Inventory. Urban Science, 
2(1), 24.  

 
Train, K. (2003). Discrete choice methods with simulation. New York: Cambridge University 

 
Train, K., and Weeks, M. (2005). Discrete choice models in preference space and willingness-to-

pay space. In Applications of Simulation Methods in Environmental and Resource 
Economics (pp. 1-16). Springer, Dordrecht. 

Press. 

 

 

53 

Trivette, S. A. (2018). The importance of food retailers: applying network analysis techniques to 

the study of local food systems. Agriculture and Human Values.  

 
Ubilava, D., and Foster, K. (2009). Quality certification vs. product traceability: Consumer 

preferences for informational attributes of pork in Georgia. Food Policy, 34(3), 305–310. 

 
Wolf, C., G. Tonsor, and N. Olynk. (2011). Understanding US Consumer Demand for Milk 

Production Attributes. Journal of Agricultural and Resource Economics 36: 326-342. 

 
World Bank. (2015). Living Standards Measurement Survey – Integrated Survey on Agriculture 

2014. Retrieved from 
http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EXTLSMS/0,
,contentMDK:23635506~pagePK:64168445~piPK:64168309~theSitePK:3358997,00.html   

 
World Bank (2018a). “Data: Mali.” Accessed 2018. 

https://data.worldbank.org/country/mali?view=chart   

 
World Bank (October, 2018b). Poverty and equity brief: Sub-Saharan Africa: Mali. 

 

 

 

54 

CHAPTER THREE: ANALYSIS OF THE FACTORS MOTIVATING DAIRY PROCESSORS 

TO USE LOCAL OR IMPORTED MILK INPUTS IN THEIR OPERATIONS  

 

 
 
3.1 Introduction 
 

In Bamako, nearly all dairy that is consumed by households has been, to some extent, 

processed and packaged. It follows that, if consumption of Malian milk is to increase, local 

processors must choose to use it more as an input in their products. However, most dairy 

processors in Mali—and especially the largest firms—favor the use of imported powdered milk 

over fresh milk.13 Recent analyses have pointed to global trends that may further expand the 

processing and consumption of powdered milk in West Africa (Choplin, 2016; Orasmaa et al., 

2016).14 Clearly, competition from powdered milk is an important feature of West African dairy 

markets. From the perspective of the fresh milk stakeholders in these countries, the growth of 

these imports represent an existential threat to the local dairy sector (e.g., Baché, 2018,  March 2; 

Livingstone, 2018, April 4).  

Although there is a substantial economic literature focusing on African dairy development, 

studies have been surprisingly silent on the issue of import competition and, more importantly, 

how fresh milk value chains can strengthen their competitiveness against this threat. The 

objective of this study is to investigate this under-researched issue, by answering the following 

research question: What factors drive a dairy processing firm’s choice to use local fresh milk, 

 

13 Because Mali does not manufacture powdered milk, all supply of powdered milk is imported. Also, in this paper I 
use the following nomenclature. “Fresh milk” refers to any fluid milk that is produced in Mali. “Dairy product” 
refers to any consumer product made from powdered and/or fresh milk inputs. “Local dairy product” refers to any 
dairy product that is manufactured in Mali, regardless of the type of milk input used. I also use the terms milk 
“inputs” and “ingredients” interchangeably. “Firm” refers to any dairy processor.   
14 Among these trends, European dairies have been making large investments in processing and distribution capacity 
in West Africa, including in Bamako, following the end of the European Union milk production quotas in 2015 and 
due to other factors in global milk markets (Choplin, 2016; Orasmaa et al., 2016). 

 

55 

powdered milk, or some blend of both, as an input in the manufacturing of consumer dairy 

products? I focus on competition between fresh and powdered milk as an input in dairy 

processing, because of the determining role that processor strategies play in the 

commercialization of these milk commodities.  

To understand how and why processors select fresh or powdered milk, I draw on concepts 

from strategic management, industrial organization, and organizational economics. Specifically, 

I analyze how each of the two milk input types contributes to or detracts from a processor firm’s 

competitive advantage, based on its influence on cost and on product differentiation (Porter, 

1985). I employ case study methods and examine nine Bamako firms that use fresh milk and 

powdered milk to varying degrees in their operations, and which (together) are representative of 

the Bamako dairy processing industry.   

From this analysis, I am able to identify several strategic factors on which fresh milk value 

chains and government policy can focus to promote the use of fresh milk by processor firms. 

This goal is of great importance to economic development, because increased competitiveness of 

domestic agricultural products should contribute to job and income growth while enhancing food 

and nutritional security in urban areas where access to dairy products can be limited (Theriault et 

al., 2018; Tschirley et al., 2015).  

Additionally, this study responds to a longstanding need to apply industrial organization and 

agribusiness concepts to address challenges in developing agricultural markets (e.g., Bellemare 

and Bloem, 2018; Cook and Chaddad, 2000), particularly in the “hidden middle” segments of 

value chains (Reardon, 2015). These middle segments include wholesalers, transporters, and 

processors, who together account for a large share of added value and costs. For example, in 

dairy value chains, processors play the critical role of preserving liquid milk through 

 

56 

pasteurization of fermentation and converting it into numerous other products (Jaffee, 1995). 

Broadening analyses beyond the farming segment is also important because, in increasingly 

globalized food markets, value chains compete against other value chains (Boelhje, 1999).  

In the next section, I define competitive advantage and its two domains—cost and 

differentiation—and relate these concepts to African dairy value chains. In Sections 3, I describe 

the case study methods that I use in this research. In Section 4, I describe the nine dairy 

processors representing the nine cases. In Section 5, I analyze the cases to answer two underlying 

questions: (1) How does each milk input influence cost? and (2) How does each input influence a 

firm’s ability to differentiate? The answers to these questions, together, allow me to address the 

broader question: Why do dairy processor firms choose to use fresh or powdered milk, or some 

blend of both, in their operations? In the final section, I present implications for agribusiness 

strategy and policy, with the aim of improving the competitiveness of local dairy value chains in 

Mali.  

 
3.2 Conceptual Framework  
 

There are several techno-economic features of milk, milk production, and dairy 

processing which generate risks as well as opportunities for dairy producers and 

marketing enterprises and which might be expected to influence the industrial 

organization of dairy commodity systems in Africa (Jaffee, 1995, p. 200). 

 
3.2.1 Firm competitive advantage    

In economic theory, a firm makes decisions in order to maximize its expected profit—or, 

under the principle of duality, to minimize costs—subject to constraints. However, the 

conventional neoclassical production model makes several strong assumptions, including the 

 

57 

existence of perfect information, a complete set of markets, product homogeneity, competition 

between a large number of firms and buyers with ease of market entry and exit, and the absence 

of economies of scale (Kirsten et al., 2009). By relaxing these assumptions, organizational 

economics provide a framework for explaining the existence and behavior of different firms, as 

they seek to maximize profits and minimize costs in response to various market failures 

(Mahoney and Pandian, 1992).15  

Drawing from this framework, as well as from concepts in industrial organization, strategic 

management theory redefines the firm’s objective as the creation of competitive advantage, 

which is the ability of a firm to establish a profitable position in an industry, by differentiating its 

products from those of competitors in order to create value for its customers, while ensuring that 

firm costs do not exceed this value (Porter, 1985). Embedded within this definition are two 

domains that determine competitive advantage— cost and differentiation (Ibid.). The premise of 

this paper is that, by analyzing how a firm’s choice of inputs influences these two domains, one 

can understand how different inputs contribute or detract from the firm’s competitive advantage  

and, thus, identify the critical factors that shape firm preferences for the different inputs.  

Figure two visualizes Porter (1985)’s conceptual framework and depicts its 

operationalization for investigating the strategic choices of dairy processors with respect to 

different milk inputs. I explain the concepts of cost and differentiation below, and further discuss 

the research questions in the following section.  

 

 

 

15 Organizational economics is the branch of economics that includes, among other theoretical frameworks, 
transaction cost theory and information economics. The shared premise in these frameworks—established by Coase 
(1937)—is that market failures explain not only the existence of firms, but also the strategic characteristics of firms 
including their size, scope, and internal organization (Mahoney and Pandian, 1992). A market failure occurs 
whenever a free market fails to bring about a Pareto-efficient allocation of a good or service, i.e. when there exist 
additional unexploited trades between two parties that could make at least one party better off without making the 
other worse off (Besley, 1998). 

 

58 

Figure 4: Research framework and questions, based in Porter’s comparative advantage 

CONCEPTUAL FRAMEWORK:

Goal of Firm Strategy: Competitive Advantage 
Ability to establish a profitable position

1st Domain: Cost reduction
Ability to operate at a 
lower per-unit average cost

2nd Domain: Differentiation
Ability to distinguish brand on 
basis of consumer preferences

Operationalize to investigate 
why dairy processors choose 
to use fresh or powdered milk 
or both in their operations

RESEARCH QUESTIONS:

Overarching Research Question: How does each milk 
input contribute to firm competitive advantage?

1st Underlying Research Question: 
How does each milk input 
contribute to cost reduction? 

2nd Underlying Research Question:
How does each milk input 
contribute to differentiation?

 

3.2.2 Cost  

The first domain of competitiveness is cost. For a dairy firm, costs are composed of the 

purchase price of milk and other inputs, the production costs associated with processing inputs 

into a consumer product (e.g., reconstitution, pasteurization and cooling, fermentation, 

packaging), the distribution costs for getting the product to the buyer (e.g., storage, transport, and 

promotional costs), and any associated transaction costs in these processes (e.g., gathering 

information on the buyer, assessing the product, contracting).16  

Cost reduction contributes most to competitive advantage in the form of cost advantage, 

which is a firm’s capability to operate at a lower per-unit average cost than its rivals, 

independent of scale (Waldman and Jensen, 2013; Porter, 1985). Cost advantage is rooted in 

superior (i.e., less costly) access to key resources such as a critical input, enhanced technology, 

financial capital, specialized human resources, or desirable production or distribution locations. 

It originates from imperfections in various capital, input, output, or labor markets, and arises 

when firms incur differently the transaction costs resulting from these market failures. Thus, 

identifying the sources of transaction costs are especially important to understanding competitive 

advantage.  

 

16 Transaction costs are the costs incurred by two parties in using a spot market prices to transaction a good (Coase, 
1937). They include the ex ante costs of searching for and screening trade partners and negotiating an agreement; 
and the ex post costs of monitoring and enforcing the agreement, transferring ownership of the product, and any 
costs associated with adapting to disturbances (Staal et al, 1997). 

 

59 

Such imperfections are a common feature of developing agricultural markets, and dairy value 

chains are no exception (Jaffee, 1995). In fact, as the section epigraph suggests, milk has several 

fundamental attributes that generate significant transaction costs within the value chain. These 

include its high perishability and bulkiness, wide quality variability, the lumpy specialized assets 

required for production, and the seasonality of output (Ibid.). There is a broad literature 

describing the constraining presence of transaction costs in the supply and procurement of fresh 

milk in East Africa (e.g., Burke et al., 2015; Holloway et al., 2005; Holloway, 2004; Holloway et 

al., 2000), as well as in its processing (Jaffee, 1995). The major insight from these studies is that 

high transaction costs impede participation, efficiency, and supply in these markets. On the other 

hand, there is little documentation on the procurement and processing of powdered milk, and to 

what extent the costs associated with this milk input differ from those generated by fresh milk.  

 

3.2.3 Differentiation 

Differentiation refers to a firm’s ability to distinguish its brand on the basis of consumer 

preferences for some perceived product attribute, services associated with products, or costs 

involved in switching from one product to another (Waldman and Jensen, 2013; Porter, 1985). 

As a domain of competitive advantage, differentiation allows a firm to sell more products at a 

given price, or to raise its price relative to the prices of its rivals while still retaining customers 

who prefer its products (Porter, 1985; Bain, 1956). Differentiation must be valuable to buyers 

and relatively rare on the market in order for it to lead to competitive advantage (Porter, 1985).  

In addition to generating transaction costs, the techno-economic features of milk also create 

opportunities for differentiation (Jaffee, 1995). One potential basis for differentiation is the input 

composition of a dairy product. Some consumers may have a particular preference for dairy 

 

60 

products processed from fresh milk or, alternatively, from powdered milk, on the basis of taste, 

consistency, place of origin, or some other attribute that is associated with one milk type but not 

the other. For example, dairy firms might differentiate their products on the basis of consumers’ 

preference for locally produced fresh milk instead of imported powdered milk as a product 

ingredient. Using choice experiment data collected in Bamako, Vroegindewey et al. (2019) 

showed that 85% of consumers have a distinct preference for 100% local fresh milk as the 

ingredient in pasteurized milk. Additional stated preference data suggests that similar preferences 

hold for other dairy products as well.  

The choice of milk input might further affect competitive advantage through its influence on 

several other bases for differentiation. First, in general milk is amenable to processing into a very 

wide range of products and sub-product types. However, a firm’s options for dairy product 

differentiation may depend on the input that is used.   

Second, the taste, and safety of a dairy product depends on its chemical and biological 

quality, which can be highly variable and depends on an array of factors related to herd 

management as well as elements in processing, transport, and storage. Given the considerable 

differential in conditions between dairy farming in Mali and in other exporting countries, I would 

expect that dairy product quality is affected by the origin of its inputs.  

Third, it is costly or impossible for consumers to directly observe many of these variations in 

quality, which creates information asymmetry between consumers and the other actors in the 

value chain. Consequently, to avoid problems of adverse selection, dairy processor firms must 

successfully differentiate their own products from other lower-quality products through quality-

 

61 

signaling mechanisms, such as advertising, packaging, third-party certification, or an effective 

sales force (Rosenman and Wilson, 1991).17  

3.3 Methods  

3.3.1 Case study methods and case selection  

A qualitative research methodology is appropriate for this study, because it seeks to answer 

several process-type questions about firm strategy that are outlined in Figure 4 (Maxwell, 2013). 

The objective of the research is to address the following question: Why do dairy processor firms 

choose to use fresh or powdered milk, or some blend of both, in their operations? Utilizing the 

concept of competitive advantage focuses the investigation on two underlying research 

questions. How does the use of each input influence cost? How does the use of each input 

influence differentiation? Qualitative methods are an effective tool for investigating such why 

and how-type questions, because they allow for thick description of processes, through the use of 

various types of data and methods (Yin, 2014).    

More specifically, I adopt a qualitative case study approach because the phenomenon of 

interest—the milk input decision—is embedded within a context of multiple integrated systems – 

the procurement, production, distribution, and marketing activities of the firm – which, together, 

manifest and contribute to a firm’s cost position and its basis for differentiation (Porter, 1985, p. 

33). Thus, analysis of the milk input decision (i.e., its relationship with competitive advantage) 

cannot be abstracted from the variables of these broader systems. A case study approach is 

uniquely suited to cope with such complexity, in which the boundaries between the 

 

17 Information asymmetry exists in a transaction when one party has more information than the other. Adverse 
selection is a situation of information asymmetry in which buyers must rely on market prices alone as an imperfect 
indicator of quality. Akerlof (1970) showed that there will be a reduction in the average quality of marketed dairy 
products and buyers with a preference for higher quality will eventually exit the market.  

 

62 

“phenomenon and context are may not be clearly evident,” and there are “many more variables 

of interest than data points” (Yin, 2014; pp. 16-17).  

In the first step of case selection, I developed a preliminary list of processor firms operating 

in Bamako, based on secondary sources, informant interviews, and dairy product packaging that 

was gathered during preliminary fieldwork. I then refined this list of case candidates through 

initial interviews with the firms. In order to examine the relationship between the domains of 

competitive advantage and milk input decisions, I adopted a multiple-case study design in which 

the primary case selection criteria was to obtain a sample of firms that varied in the extent to 

which they use fresh versus powdered milk in their operations (Yin, 2014).  

To ensure that the research is relevant to the entire Bamako dairy processing sector, the 

second case selection criteria was to obtain a sample that represented the range of dairy firms 

operating in Bamako, in terms of three variables. The first variable is product mix, or the variety 

of different types of dairy products that a firm produces. The second is scale, as indicated by a 

firm’s annual dairy output, its number of employees, and its level of capitalization. The third 

variable is production technology, which I break down into three groups. I define an artisanal 

firm as one which completes all production and packaging steps by hand, a semi-industrial firm 

as one which has mechanized at one part of its production operations, and an industrial firm as 

one which has mechanized most of its operations.   

 

3.3.2 Data collection and analysis  

I collected the case study data during three waves from July 2017 to April 2018, to obtain a 

seasonal understanding of dairy production and marketing activities, to build rapport with each 

firm, and to facilitate a participatory process of data collection and validation. Field work 

 

63 

focused on several types and sources of data in order to obtain in-depth coverage of each case, 

and to facilitate data triangulation (Yin, 2014).  

As mentioned above, one objective of the first wave of data collection was case selection. 

During this fieldwork, I collected preliminary information on most cases through relatively short 

interviews with the owner of each respective firm.18 These interviews focused on obtaining a 

broad overview of firm history and activities, including information on the shares of milk 

ingredients that each firm used and each firm’s product mix, production technology, and scale. 

The case information that I gathered from the initial interviews also provided a basis for deeper 

inquiry in the second wave of data collection.  

During the second wave, I conducted in-depth semi-structured interviews with the owner and 

key personnel of each firm, as well as a short quantitative survey. These instruments focused on 

obtaining a more detailed understanding of how each firm used different types of milk inputs, 

and to explore how each input type influenced costs and differentiation in every area of firm 

activity (i.e., in procurement, production, distribution, and marketing). I also conducted semi-

structured interviews with each firm’s primary input supplier in order to gather additional data on 

procurement, and with a selection of buyers (i.e., the sellers, independent retailers, or 

wholesalers with whom the firm transacts directly) in order to gather additional data on 

marketing and distribution. I recorded interview data with audio recording interviews and hand-

written notes, and also collected samples or photographs of packaging and other marketing 

artifacts (e.g., billboards, storefronts, electronic media) for each firm.  

After the second wave of field work, I developed narratives of each case. These described in 

detail the strategic choices and processes associated with each area of firm activity and, as an 

18 In the case of the largest processor, the primary firm respondent was the assistant director of operations. For 
convenience, I will also refer to him as “owner.” 

 

 

64 

analytical technique, began to organize the complexity of each case and set the stage for 

answering the research questions (Yin, 2014). I also summarized each case narrative into a short 

brief and shared it with the owner of each firm. During the final wave of field work, I invited 

each owner to correct any errors in the narratives and to provide any additional data or 

comments. In this way, the case narratives also served as a data validation tool and helped to 

generate a final layer of information.  

Following the final wave of field work, I analyzed the case study data in three steps. First, I 

grouped each firm by its use of fresh milk versus powdered milk to understand how each input is 

used among the firms. Second, I identified the instances of cost reduction and differentiation 

within each group to understand the factors by which firms create competitive advantage. 

Finally, I analyzed the relationship between these factors and milk input choice.  The results of 

each analysis are presented and discussed in the next three sections. 

 

3.4 How do firms use fresh and powdered milk in their operations? 

Table 11 presents summary information on each firm that I selected for the research, 

including the dairy products that the firms commercialize and whether each is made from fresh 

milk, powdered milk, or some combination thereof. The production technologies, scales, and 

different product types in the case studies (which I describe in the following sub-section) 

represent the entire range of domestically-manufactured packaged dairy products that are 

available in Bamako.19  

 
 

 

19 Ghee and butter are also manufactured in Bamako by some households and artisanal processors in Bamako. 
However, the level of commercialization of these products is very low, and no packaged version exists.  

 

 

65 

Table 11: Overview of firms and dairy products 

Milk inputs  

Dairy  

Production 
Technology 

Firm 
 
100% VOLUMES FROM FRESH MILK    
FM1 

Artisanal 

Product Mix 

 

 

Pasteurized milk 
Féné 

95% VOLUMES FROM FRESH MILK 
FM2 

Artisanal 

FM3 

Artisanal 

Pasteurized milk 
Drinking yoghurt 
Pasteurized milk  
Féné   
Drinking yoghurt 
Dégué 

Semi-

PM8  

Artisanal 

Industrial 

1% – 20% VOLUMES FROM FRESH MILK    
PM4 

Dégué  
Pasteurized milk 
Féné 
Drinking yoghurt  
Strained yoghurt 
Dégué  
Pasteurized milk 
Pasteurized milk 
Drinking yoghurt 
Strained yoghurt 
Soft cheese 
Fresh cream 
0% VOLUMES FROM FRESH MILK    
PM5 

Industrial 

Semi-

PM9 

Pasteurized milk 
Drinking yoghurt 

PM6 

PM7 

Industrial 

Semi-

Industrial 

Semi-

Industrial 

Primary 
input 
used 
(1st, 2nd) 

FM 
FM 

FM 
PM 
FM 
FM 

PM 

PM, FM 

PM 
FM 
FM 

PM 
PM 

PM, FM 

FM, PM 
PM, FM 
PM, FM 

PM 

FM, PM 
FM, PM 

Share 
of total 

commercial 

volumes 

70% 
30% 

95% 
5% 
70% 
25% 
< 5% 
< 5% 

85% 
15% 
< 5% 
65% 
15% 
15% 
< 5% 
50% 
40% 
10% 
< 5% 
< 5% 

PM 
PM 

> 50% 
< 50% 

Total 

commercial 

volumes 
(L/year) 

30,000 

92,000 

60,000 

225,000 

1,047,000 

Scale 

Number of 
employees 

 

4 

5 

7 

16 

52 

 
 
 

 

Year 

Founded 

Capitalization 

(FCFA) 

 

3 million 

2010 

3 million 

4 million 

2002 

1995 

6 million 

2002 

284 million 

1996 

12,000,000 

125 

1,500 million 

1969 

33,000 

26000 

230,000 

70 

20 

19 

57 million 

72 million 

2004  
(2017) 

 

1999 

125 million 

1993 

Other 
products 

 
 

 
 

 

Juice 
drinks 

Juice 
drinks 

Packaged 

Water 

 

 

Drinking yoghurt 
Dégué 
Drinking yoghurt 
Strained yoghurt 

PM 
PM 
PM 
PM 

65% 
35% 
90% 
10% 

Source: Authors, based on firm interviews and surveys. Dairy product mix only includes product types that the firm regularly manufactured at the time of the study.  
Volumes and shares are calculated from 2016 figures, except for PM4, PM5, and PM9 which are based on 2017 figures.  
Capitalization refers to the reported value of all fixed investments.  Year founded refers to the year in which the firm began dairy operations. 

 

66 

Based on the share of each firm’s commercial volumes that is constituted by products 

manufactured from fresh milk, I classed the firms into one of four groups. One firm exclusively 

uses fresh milk for its volumes (100%), two firms mostly use fresh milk (for 95% of their 

volumes), three firms mostly use reconstituted powdered milk (for between 80% and 99% of 

their volumes), and two firms use 100% powdered milk. Interestingly, although many firms use 

some combination of fresh and powdered milk in their operations, all firms tend to depend 

heavily on one input or another with the dominant input accounting for 80% or more of volumes. 

In this paper, I will refer to the three firms in the first two groups as “fresh milk firms,” 

because they mostly or exclusively use fresh milk in their operations. I refer to the six firms in 

the last two groups as “powdered milk firms” for the same reason. In order to respect the 

anonymity of the case study participants, I also refer to each individual firm by a unique 

identifier composed of (i) “FM” or “PM” (to indicate whether it is a fresh milk firm or powdered 

milk firm, respectively), followed by (ii) a number that indicates its ordinal size relative to the 

other firms in the study.20 I measure firm size by its level of capitalization, in terms of fixed 

investments. Among the case studies, firm capitalization tends to increase with the age of the 

firm, the number of employees, and the dairy volumes manufactured. I also note that the largest 

dairy firms in these cases (and in Bamako generally) mostly or exclusively use powdered milk 

for their operations.   

 

3.4.1 Fresh milk firms  

The three fresh milk firms (FM1-FM3) are the smallest among the case studies, each with four 

million FCFA or less in physical capital. FM1, which exclusively uses fresh milk, is the smallest 

20 For example, FM1 refers to the smallest firm among the case studies, which is a FM fresh milk, while PM9 refers 
to the largest firm, which is a powdered milk firm.  

 

 

67 

and sells about 30,000 L per year. The other two process up to about 90,000 L per year. All three 

firms are artisanal in their production technology, meaning that they complete all production and 

packaging steps by hand, utilizing rudimentary equipment (e.g., basins, cooking pots, and wood 

or charcoal stoves). Production takes place within the walled courtyard of the owner’s residence 

or inside her house.  

These firms use fresh milk for one or two products that make up the majority of their 

volumes. The primary product (making up 70% or more of commercial volumes for each firm) is 

pasteurized milk (lait frais pasteurisé), which firms manufacture by heating raw fresh milk to 

about 90 degrees Celsius. The second-most important product is féné, a traditional full-fat 

fermented milk that firms manufacture by leaving pasteurized milk to ferment at the ambient 

temperature, sometimes with a local culture.  

The two larger firms in this group also have one or two additional yoghurt-based product that 

incudes powdered milk. Each of these yoghurt-based products makes up 5% or less of each 

firm’s total volumes. Both firms manufacture drinking yoghurt (yaourt à boire or yaourt 

brassé).21 FM2 makes this product entirely from powdered milk, while FM3 uses fresh milk as 

the base ingredient, and blends in some additional powdered milk. FM3 additionally 

manufactures a powdered milk-based dégué, a traditional product consisting of millet couscous 

mixed with a thick yoghurt. FM1 currently only manufactures pasteurized milk and féné; 

however, the owner has experimented with yoghurt in the past and wishes to begin regularly 

make this product in the near future. 

Each of the fresh milk firms is a member of the Bamako Cooperative of Local Milk Resellers 

(CRLLB). In the 2000s, several artisanal firms (including the owners of FM2 and FM3) 

21 In Mali, the French word lait caillé (“soured milk”) is sometimes also used colloquially to refer to either féné or 
drinking yoghurt. 

 

 

68 

collaborated with a local non-governmental organization (NGO) to create CRLLB. Its purpose 

was to build the capacity of processors of fresh milk-based products in Bamako, and to help 

coordinate their procurement of fresh milk. A broader goal was to develop a viable market 

channel in Bamako for eighteen peri-urban farmer cooperatives that are members of the National 

Federation of Malian Milk Producers (FENALAIT). In 2018, CRLLB had about 90 individual 

members (90% of whom are women), and delivered milk to about 160 artisanal firms throughout 

Bamako. Field work confirmed that the number of small firms selling fresh milk in Bamako 

greatly exceeds this amount, although CRLLB is the largest network of these actors.  

 

3.4.2 Powdered milk firms  

The other six firms in the case studies (PM4-PM8) use powdered milk for products that make 

up the majority, or all, of their volumes. Among these, one firm (PM4) is an artisanal firm. Like 

its fresh milk firm counterparts, this firm is also a member of CRLLB and manufactures a 

pasteurized milk and féné product using fresh milk. However, dégué accounts for 85% of its total 

commercial volumes (225,000 L per year), which greatly surpasses the output of the other 

artisanal firms. 

Another four powdered milk firms (PM5-PM8) are semi-industrial: each has a dedicated 

factory and has mechanized at least the packaging process of its operations (and sometimes 

additional steps as well). These firms have a capitalization of about 60 million to 280 million 

FCFA, and commercialize up to one million L of dairy products per year. Informant interviews 

and secondary sources suggest that these four firms are among the largest semi-industrial dairy 

firms in Mali.  

 

69 

Among these semi-industrial firms, the product accounting for the most volumes is drinking 

yoghurt. Additionally, two firms (PM6 and PM8) manufacture dégué and two firms (PM7 and 

PM8) manufacture strained yoghurt (yaourt étuvé), which is less-viscous than drinking yoghurt. 

The smallest and largest semi-industrial firm (PM5 and PM8) also manufacture a pasteurized 

milk product. All of these products are made from powdered milk, with two exceptions. For its 

occasional production of pasteurized milk, PM8 uses fresh milk, powdered milk, or some blend 

of the two, depending on the availability of fresh milk (the preferred input) at that moment. 

When there is leftover fresh milk from pasteurized milk production, PM8 also blends this into 

the drinking yoghurt formula. Finally, firms PM5 and PM8 also commercialize a product line of 

packaged water and fruit juices, respectively. 

The industrial firm PM9 has mostly mechanized facilities and is the oldest and largest dairy 

processor firm in Mali. It has a capitalization of 1,500 million FCFA and an output of about 

twelve million L per year. This firm manufactures its best-selling products, pasteurized milk and 

drinking yoghurt (together accounting for at least 75% of volumes) using a blend of reconstituted 

powdered milk and fresh milk. While the ratio of inputs varies, it is typically dominated by 

powdered milk. PM9’s second most popular product by volume is strained yoghurt made entirely 

from powdered milk. Finally, the industrial firm also manufactures a fresh milk-based soft 

cheese (fromage frais) and fresh cream (crème fraiche) product manufactured from a blend of 

fresh milk (the base input) and powdered milk.  

 

 

70 

3.5  How does each milk input influence cost?  

The case study data indicates that the costs of using one milk input instead of another can 

differ according to four elements: the purchase price of the input itself, the costs of coordinating 

input procurement in order to minimize transaction costs, any transaction costs that still remain 

after such coordination, and any idiosyncratic costs associated with processing and distribution. I 

assess the relative costs of using fresh or powdered milk by analyzing and comparing the cost 

elements of each input. I find that, for every element, firms face higher costs using fresh milk, 

implying that firms can reduce costs significantly by using powdered milk instead of fresh milk.  

 

3.5.1 Purchase price 

Each of the powdered milk firms cited the relatively high per-liter purchase price of fresh 

milk, compared to powdered milk, as a primary reason for its dependence on powdered milk. 

According to one dairy industry informant, among all of the factors influencing the use of fresh 

milk, its purchase price was the “brake” that constrained firms the most. Even fresh milk firms 

complained that the purchase price, in the context of intensifying competition, are increasingly 

squeezing their margins. PM9 stated that the purchase price negotiated with its suppliers is the 

primary factor determining fresh milk purchase volumes. This negotiated price has also been a 

perennial source of tension between the PM9 firm and its suppliers since its privatization in 

1996. The real price has risen over time—from 200 FCFA per liter in 1996 to 400 FCFA in 

2018—and a few facts suggest that it is a determinant of the fresh milk price throughout 

Bamako. First, PM9 is the single largest market channel for the largest peri-urban dairy 

 

71 

cooperatives, accounting for at least one quarter of the commercial volumes of these suppliers.22 

Second, PM9’s purchasing terms were cited by all four artisanal firms, as a variable affecting 

their own transaction terms with suppliers. These firms reported purchasing fresh milk (usually 

pasteurized) at prices ranging between 400 and 425 FCFA/L in 2018, depending on the seller. 

Interviews suggest that this purchase price fluctuates little throughout the year, despite 

significant intra-seasonal variations in supply (discussed below). 

Of issue, of course, is not merely the price of fresh milk but its relative position with respect 

to the price of powdered milk. In Mali, most powdered milk is imported from Europe, followed 

by New Zealand (FAOSTAT, 2018). According to Bamako powdered milk firms, and in contrast 

to fresh milk, the local powdered milk market is well integrated with international markets, due 

to the relatively low trade barriers in place for powdered milk and the high volumes that are 

imported. The per liter purchase price of these imports are cheaper in Bamako, compared to that 

of fresh milk, due to a combination of higher farm efficiencies and substantial policy support 

through the years (Choplin, 2016). Most powdered milk firms reported purchasing 25-kg sacks 

of whole powdered milk or filled powdered milk.23  

An additional factor has been the market entry of filled powdered milk, in which dairy fat is 

substituted for a cheaper vegetable fat. Filled powdered milk imports began appearing on the 

Malian market around the year 2000 and surpassed the imports of whole powdered milk 

beginning around 2013 (FAOSTAT, 2018), due to the relative affordability of this high-fat milk 

ingredient option. Firms reported purchasing filled powdered milk in late 2017 at prices ranging 

 

22 According to the three largest dairy cooperatives supplying PM9, these cooperatives delivered 1,800 L to 3,500 L 
of milk daily to PM9 in 2016. It is difficult to pin down PM9’s total daily procurement of fresh milk. According to 
company sources, it was procuring 10,000 to 15,000 per day during the 2017-2018 period; however, Corniaux 
(2014) reports that the daily average ranged between 2,000 L and 4,000 L.  
23 Whole powdered milk contains 100% milk fat. In contrast, manufacturers of filled powdered milk replace the milk 
fat with vegeß fats to maintain high fat contents ranging between 25% and 30%. 

 

72 

between 1,960 and 2,200 FCFA/kg, or 260 to 290 FCFA/L in liquid milk equivalents (LME).24 

Two artisanal firms, which purchase much smaller amounts of filled powdered milk packaged in 

consumer-sized packaging, reported slightly higher prices: 2,200 to 2,500 FCFA/kg (290 to 330 

FCFA/L in LME) during the same period. 

 

3.5.2 Coordination costs  

In addition to the costs that are captured through purchase price, the case study data 

demonstrates that the procurement of fresh milk generates additional transaction costs associated 

with such challenges as finding suppliers, procuring the right quantities at the right time, and 

quality. Firms attempt to minimize these transaction costs through better coordination with 

suppliers. According to the two most experienced fresh milk firms, in the past two decades this 

has been greatly facilitated by several structural changes, including the expansion of cell phone 

coverage, increases in urban demand, and the diffusion of higher-yielding cow breeds alongside 

the development of dairy cooperatives. 

However, coordination itself is also costly, both in terms of the fixed costs of setting up 

procurement arrangements and the daily costs of coordinating each individual transaction. 

Broadly speaking, fresh milk firms and suppliers use several types of supply arrangements. 

First, each of the fresh milk firms procure inputs through relational contracting with individual 

dairy farmers, farmer cooperatives, or intermediary traders who work closely with farmers. PM8 

uses a similar model for its occasional fresh milk purchases. Within these arrangements, firms 

coordinate carefully with suppliers on a daily basis, due to the perishability and bulk of fresh 

milk, limited cold storage capacity, and the tight production and distribution schedules that firms 

24 This price range is calculated from data from five of the six powdered milk firms, excluding the largest industrial 
processor for which this data was not accessible.  

 

 

73 

must keep. At minimum each day, fresh milk firms must call their suppliers to plan the next fresh 

milk delivery and must ensure that a capable agent is physically present to receive and inspect 

deliveries, to begin processing the milk, and to make any necessary adjustments for late 

deliveries or quantity or quality problems.  

Second, all fresh milk firms also purchase a share of their volumes through the CRLLB 

supply cooperative, which collects 25 FCFA on every liter of fresh milk (or about 6% of the per-

unit purchase price) to sustain its operations. This fee only represents a share of the daily 

coordination costs that firms incur for using this arrangement, given that firms must still 

coordinate with CRLLB (instead of directly with other suppliers) on a daily basis. Additionally, 

some CRLLB costs (e.g., salaries) are subsidized by external partners. 

Third, the industrial firm (PM9) purchases fresh milk that is delivered to a collection center 

at its Bamako factory. Each week-day, it procures up to 10,0000 L from a select group of forty to 

sixty suppliers mostly made up of cooperatives and large farmers. Suppliers must deliver 

unpasteurized milk before one p.m., and are paid once a month. In 2018, the second-largest firm 

(PM8) was also in the final stages of constructing a collection center complex next to its factory, 

with the aim of increasing its fresh milk procurement capacity. Funded through a public-private 

partnership and set to begin operations in late 2018, the center will have 8,000 L in cooling tank 

capacity, a training center, and a retail point for animal feed and veterinary products. 

In contrast to fresh milk, for which spot markets are not well developed in Bamako, 

powdered milk is retailed from virtually every food shop in the city. The relatively well-

developed state of the powdered milk market reduces the ex ante transaction costs involved with 

searching out suppliers and negotiating prices, while mitigating some uncertainty around 

 

74 

availability. For this reason, the smallest firms (FM1-FM3, PM4-PM6) depend on the spot 

market to make weekly or biweekly purchases from a retailer or wholesaler.  

There are several challenges, however, associated with sourcing powdered milk from spot 

markets. Powdered milk firms cited price volatility, quality uncertainty, and supply uncertainty.25 

In order to mitigate these problems associated with the spot market, the largest powdered milk 

firms (PM7-PM9) have developed two special procurement arrangements to better coordinate 

transactions. First, two semi-industrial firms (PM7, PM8) use ongoing relationships and repeated 

monthly transactions with reputed importers of high-quality powdered milk brands. These 

informal agreements provide some guarantees or incentives with respect to availability and/or 

price. However, these firms still maintain a second supplier as a backup. Second, the industrial 

firm (PM9) has partnerships with at least three foreign dairy companies, through which the firm 

imports and markets powdered milk and other products. This access, combined with the firm’s 

strong financial and storage capacity, enables it to purchase large amounts of powdered when it 

is relatively favorable to do so and to maintain robust stocks. The company additionally 

purchases other brands of high-quality powdered milk.  

 

3.5.3 Other transaction costs  

Even with these procurement arrangements in place to coordinate daily fresh milk 

transactions, fresh milk firms identify two enduring problems that they must frequently manage, 

 

25 Several firms and informants reported that powdered milk quality varies significantly across brands, does not 
always reflect the technical specifications printed on product packaging, and can even vary within the same brand. 
Informants report several potential reasons for this quality variability, including smuggling and informal trade, illicit 
repackaging, and counterfeit branding of powdered milk, combined with the fact that phytosanitary and food safety 
regulations are weakly enforced. A third problem is supply uncertainty. Most firms prefer a few select European 
brands that they trust, but which are not always available on the Malian market. According to firms, these problems 
can adversely affect the quality of their own manufactured products), result in stock-outs, or generate search costs. 
 

 

75 

which drive up the transaction costs of procuring fresh milk. The first problem is the uncertainty 

of milk supply interannually, across seasons, and even from day to day. Four of the powdered 

milk firms cited supply uncertainty, especially during the hot dry season of March to May, as a 

key challenge that drove and/or continually constrains them from using fresh milk. PM5 (the 

most recent entrant in the case studies) began its dairy operations in 2017 by sourcing fresh milk, 

but began during the hot dry season. Within a couple months, the firm made a complete switch to 

powdered milk. The owner cited the limited availability of fresh milk, in addition to its price and 

sensitivities in processing (see below), as the deciding factor. PM8 also stated that its choice of 

how much of each input to use in a given batch of pasteurized milk depends on the availability of 

fresh milk at that moment.  

Fresh milk processors also underlined the problem of supply uncertainty. For example, the 

year-on-year changes in total volumes procured by CRLLB from 2012 to 2016 range between – 

29% and 24%. As an indication of seasonal variability, Figure 5 displays the volumes collected 

by the FENALAIT cooperatives in the years 2012 to 2016. While the mean trendline suggests 

that supply does tend to peak during the rainy seasons due to the availability of pasture for 

grazing (June-September,), this pattern is less pronounced in certain years (e.g., 2012, 2015, and 

2016) and can fluctuate significantly from month to month. These broader dynamics translate 

into supply uncertainty for the CRLLB members and PM9 who source from the cooperatives. 

For example, CRLLB records in 2016 indicate that PM4 and FM1 received all of their ordered 

volumes only 35% and 55% of the time, respectively. To mitigate such uncertainty associated 

with any given supplier, FM3, FM2, and FM1, each maintained at least two additional suppliers 

from whom they regularly purchased throughout the week, as backup in the event that one of 

their orders was not fulfilled, or to supplement their input needs when they received large special 

 

76 

orders (e.g., for cultural ceremonies). In the words of the owner of FM2, “You cannot get what 

you want with only one or two suppliers.” 

 
Figure 5: Monthly volumes collected by peri-urban FENALAIT dairy cooperatives, 2012-16 

 

 

h
t
n
o
m
y
b
d
e
t
c
e

 

l
l

)
L
(

o
c
 
k
l
i

 

m
h
s
e
r
f
 
l

a
t
o
T

 320,000

 240,000

 160,000

 80,000

Janvier

Février

Mars

Avril Mai

Juin

Juillet

Septembre
Octobre
Août
Novembre
Décembre

Source: Author, based on FENALAIT (2017) 
 
 

2016

2015

2014

2013

2012

5-year average

2 per. Mov. Avg. (5-year
average)

 

A second enduring problem in the procurement of fresh milk is quality uncertainty, 

particularly spoilage that leads to losses. The three largest powdered milk firms (PM7, PM8, and 

PM9) cited the quality of fresh milk as a challenge. Supply cooperatives and firms usually 

observe some minimal form of quality control, depending on their access to different testing 

instruments. PM9 and many supply cooperatives utilize a lactodensimeter (a tool for measuring 

milk density) and an alcohol test to assess deliveries of raw milk regarding density (e.g., to gauge 

whether it has been mixed with water or cream has been removed) and acidity (e.g., bacterial 

growth and spoilage), respectively. However, artisanal firms do not have access to these 

resources. Instead, they depend on any quality control procedures taken by their supplier, if any, 

and otherwise only apply organoleptic tests (based on sight, smell, and visual tests) when they 

receive deliveries (Ibid.). Under the above-described coordination mechanisms, a buyer typically 

 

77 

holds the right to reject milk ex ante if it is observably bad; however, the buyer must absorb the 

costs of losses or quality problems that are discovered ex post. For example, the owner of FM1 

complained that, when deliveries are slow on hot days, milk can sour before it arrives at her 

door. She estimated that she rejects milk deliveries about twice per week. However, sometimes 

she is forced to work with milk that does not pass her test, because the alternative is no product 

until the following day. At other times (i.e., two or three times per month), it is only once she 

begins working with a batch of milk that she realizes that it was souring. Processing such milk 

into commercial products carries risks to her customers’ health and, perhaps also, to her 

reputation. For example, one study found that Bamako schoolchildren who regularly consume 

pasteurized fresh milk were four times more likely to experience diarrhea and/or vomiting in a 

previous two-week period (Hetzel et al., 2004). 

 

3.5.4 Other idiosyncratic costs  

Even after procurement, the bulkiness and perishability of fresh milk make it more costly to 

store and process. Firms can store powdered milk in its dried form and reconstitute it into fluid 

milk only when they are ready to process. In contrast, firms must either process fresh milk 

immediately or else refrigerate it. For the semi-industrial and industrial firms—which handle 

large volumes—acquiring this capacity requires large lumpy investments. Indeed, four of these 

firms stated that one of the advantages of powdered milk is that it is “less delicate,” “less 

difficult,” or “easier to work with” than fresh milk. PM5 and PM6, both of which have a couple 

large refrigerators and use unrefrigerated delivery vehicles, stated that these resources were 

insufficient to process fresh milk. PM6 stated that switching from powdered to fresh milk would 

require new investments in a cold room, a gas pasteurizer (to heat large amounts of fresh milk 

 

78 

quickly), and refrigerated trucks. Maintaining this expanded cold chain capacity would generate 

additional electricity and fuel costs as well. The only powdered milk firms that do use some fresh 

milk (PM8 and PM9) already have such cold chain capacity.  

 

3.5.5 Summary of the relative costs  

Table 12 summarizes the relative costs for fresh and powdered milk, including the purchase 

prices, transaction costs in procurement, and other idiosyncratic challenges in processing and 

distribution for each commodity. This comparison indicates that, for every element, firms face 

significantly higher costs using fresh milk. Even the highest-reported price for powdered milk 

(purchased in consumer-sized packaging from retailers) is about 20% cheaper than fresh milk in 

LME terms. Although firms face some transaction costs due to the price, quality, and supply of 

powdered milk, the largest firms are able to successfully manage these issues through more 

coordinated relationships with milk importers. On the other hand, fresh milk firms face 

significant transaction costs related to uncertainty of quality and supply, even after putting in 

place costly procurement arrangements with suppliers. Additionally, the bulkiness and 

perishability of fresh milk make it difficult to handle in storage and distribution. Taken together, 

this analysis suggests that firms can reduce costs significantly by using powdered milk instead of 

fresh milk.  

 
 
 

 

 

79 

Table 12: Costs of fresh and powdered milk in Bamako 

Milk input 

Purchase price 

Other costs 

 

Coordination and transaction costs in procurement: 
(cid:1)Frequent (daily) transactions 
(cid:1)Costly procurement arrangements used in every case (relational  
contracting, supply cooperative, maintaining collection center) 

(cid:1)High supply uncertainty (across years and seasons, and day-to-day) 
(cid:1)High quality uncertainty (spoilage, variations in composition and taste)  
 
Costs in processing: 
(cid:1)Requires large cold chain capacity and uncertainty prior to processing 
 
Coordination and transaction costs in procurement: 
(cid:1)Presence of spot markets, which some firms use. Others use 
procurement  

arrangements (relational contracting, partnerships). 

(cid:1)Price volatility 
(cid:1)Some related quality and supply uncertainty  
 

  

400 FCFA/L 

400 - 425 FCFA/L 

 

1,960-2,200 FCFA/kg 

260-290 FCFA/L 

(LME)  

2,200-2,500 FCFA/kg  

290-330 FCFA/L 

(LME)  

Fresh milk  
Raw milk, delivered by 
various suppliers to 
industrial firms 

Pasteurized milk, 
delivered by various 
suppliers to artisanal and 
semi-industrial firms 

Powdered milk  
Filled or whole 
powdered milk, 
purchased from 
importers or other 
wholesalers by industrial 
and semi-industrial firms 
 
Filled or whole 
powdered milk, 
purchased from retailers 
by artisanal firms 
Source: Authors.  
 
 

3.6 How does milk input choice influence differentiation?     

The case study analysis revealed four general sources of differentiation among dairy firms 

operating in Bamako. First, firms physically differentiate their products by offering new product 

types, or more variety within a given product type. Second, they enhance or preserve some 

product quality attribute through key ingredient choices. Third, they use quality-signaling 

mechanisms to provide information on product attributes that are otherwise difficult or costly to 

observe. Fourth, firms offer unique packaging features. However, I find that fresh milk firms are 

less likely to exploit each of these sources of differentiation.  

 

 

80 

3.6.1 Product types and sub-types  

One basic source of differentiation for all firms is to offer product types that are unique to others 

on the Bamako market. In Section 5, I presented the different product types that are currently 

available in Bamako. In recounting their beginnings, most firm owners stated that they launched 

their business on the back of one of these dairy products at a time that it was relatively rare on 

the market. For instance, PM9 was one of the first domestic firm to introduce packaged forms of 

several dairy products, including pasteurized milk and strained yoghurt. When all three of the 

largest semi-industrial firms (PM6-PM8) were launched in the mid-1990s nineties, they began 

with a similar flagship product, a drinking yoghurt, which at the time was uncommon in 

Bamako. Similarly, two artisanal firms (FM2, FM3) were among the first to begin regularly 

trading fresh milk in Bamako beginning in in the late 1990s. Almost simultaneously, these firms 

also diversified into féné, because of its simple conversion from fresh milk, and in order to offer 

some variety to customers.  

Over time, new entrants increased the competition in these product markets. Many even 

closely mimicked the packaging, labeling, or brand-names of the market leaders. For example, 

the owner of PM8 recounted an incident in which a consumer became gravely ill from such an 

illicit product that mimicked her own brand. Increased competition drove some incumbent firms 

to offer entirely new products in order to continue differentiating their brands. PM9 introduced 

the first locally-manufactured soft cheese and fresh cream in the early 2000s. Around 2010, both 

PM8 and PM6 diversified into another product that was rare on the Bamako market—packaged 

dégué. PM5, the newest entrant among the cases, manufacturers what are now very common 

products (pasteurized milk and drinking yoghurt) but has invested in equipment and special 

packaging to offer a novel frozen dairy product based on powdered milk. In most cases, these 

 

81 

firms have continued to offer the older product types as they introduce newer ones, in order to 

offer variety to buyers and maintain market share.  

Instead of introducing new product types, PM7 differentiated by creating variety within an 

existing product type. This firm, like PM8 and PM6, faced growing competition in the drinking 

yoghurt product category. However, it responded by increasing the flavor options from the 

standard three (strawberry, vanilla, and natural) to eleven, which represents the most variety 

within any dairy product type in Bamako. The firm selected this strategy based on a consumer 

study that it had conducted, which showed that Malians “have a curiosity for flavors.” The owner 

explained that the objective was to allow each consumer to find the “flavor that was right for him 

or her.” 

The fact that a few product types (i.e., soft cheese, and fresh cream, pasteurized milk) are 

made mostly from fresh milk, and one product type (i.e., féné) is made exclusively from fresh 

milk, suggests that this input has some advantage over powdered milk in the production of 

certain product types. For example, PM9 stated that the unique taste of its soft cheese and fresh 

cream products “depend on fresh milk.” The cultural familiarity that Malians have with 

pasteurized milk and féné may also play a role in preferences for fresh milk in these products. 

However, there is an implicit market standard that these products do not contain added sugar, 

flavoring, or other additive; thus, there may also be less scope for further differentiating these 

products, compared to powdered milk-based products.  

Also, the physical characteristics of Malian fresh milk may limit its use in certain dairy 

products—namely, yoghurt products (i.e., drinking yoghurt, strained yoghurt, or dégué). One 

firm specializing in yoghurt (PM7) explained that the fresh milk that is available in Bamako has 

a lower fat and protein content than powdered milk. To compensate and obtain the same amount 

 

82 

of yoghurt product, this firm would have to use larger amounts of fresh milk which would 

effectively double milk input costs. Thus, almost all of the eight firms that manufacture any 

yoghurt product do so from powdered milk. One exception is FM3, which reinforces fresh milk 

with some powdered milk when making drinking yoghurt. Also, after the production of their 

fresh milk-based products, PM8 and PM9 mix any remaining fresh milk into their drinking 

yoghurt recipe in order to enhance the taste and consistency of this product. On one hand, there 

is some evidence that Bamako consumers prefer yoghurt products made from fresh milk. 

However, yoghurt products typically include other less costly ingredients (i.e., sugar, flavoring, 

starch, or millet) intended to enhance taste or consistency, which may compensate for any 

marginal utility that is lost from the exclusion of fresh milk.  

 

3.6.2 Enhancing or preserving product quality   

Within a given product type, all firms also seek to differentiate on the basis of some 

experiential or credence quality attribute, such as taste and consistency, shelf life, nutrition, or 

purity. For powdered milk-based drinks, success in quality differentiation partly depends on the 

careful selection of an appropriate powdered milk brand. This is aligned with Theriault et al. 

(2018), who found that “price premiums are paid for powdered milk manufactured in Europe and 

by a well-reputed multinational company, Nestlé.” For example, although PM4 was not one of 

the first firms to offer dégué on the Malian market, according to the owner it has come to 

distinguish itself through careful attention to quality. The owner stated that she constantly seeks 

to improve her product formula by tasting and studying rival products, experimenting with her 

own product formula and presentation, and by upgrading to higher (and more expensive) grades 

of powdered milk. According to the owner, this strategy—combined with the owner’s high 

 

83 

attentiveness to customer service—has grown the firm: it now employees seventeen people, and 

commercialized over 200,000 L per year. It has also driven its transition from processing mostly 

fresh milk to powdered milk, with dégué accounting for 80% of these volumes. As another 

example, PM7 uses a powdered milk brand that is enriched with vitamins and minerals, which is 

highlighted on product labeling and plays into the firm’s mission to provide a nutritious food to 

children and elderly.  

Other powdered milk firms seek quality differentiation based on the addition or omission of 

certain non-milk ingredients. In order to extend product shelf life, PM5 stated that it often adds 

the preservative sorbate to its pasteurized milk and drinking yoghurt products, adjusting the 

amount depending on whether the powdered milk brand already contains this additive. PM7 also 

stated that it is considering adding a preservative to its products. In contrast, PM9 claims to use a 

particular ferment that offers better product resilience than others. Two other semi-industrial 

firms, PM5 and PM6, have introduced a corn or soy starch to thicken their products. Although 

these imported starches are expensive and sometimes unavailable, the firms report that the 

powders thicken the consistency of yoghurt in a way that is appealing to consumers and which 

allows them to reduce the amount of powdered milk by 10% to 20%, thus reducing milk input 

costs.  PM8 also considered using a starch, but ultimately decided against it because the owner 

was uncomfortable using such an additive, highlighting that the omission of an ingredient can 

also be a way to differentiate so long as consumers value product purity.  

As noted above, fresh milk products rarely contain additional non-milk ingredients, thus 

limiting this differentiation opportunity. Instead, quality differentiation depends to a larger 

degree on the quality of the procured milk ingredient. However, again, fresh milk is vulnerable to 

several quality issues. The three largest powdered milk firms (PM7-PM9)—together representing 

 

84 

the most established domestic dairy brands in Bamako—all cited the quality of fresh milk as a 

key challenge. Overall, quality control and traceability of fresh milk is still at a very early stage 

in Mali. At best, firms are aware of which cooperative the milk is sourced from, but not which 

farm or animal. Further, firms do not have detailed or verified knowledge of herd conditions, 

including herd health, feed, and veterinary care.  

For instance, all three of the fresh milk firms complained that CRLLB-sourced milk varied 

greatly in fat and other taste attributes, in comparison to milk from other supply channels. These 

firms and CRLLB suspected a number of possible causes, including the higher prevalence of 

mixed-breed dairy cows in among some cooperatives (which are popularly believed to provide 

milk that is less rich than the traditional breeds), feed quality, or a problem in the cooperative-

level pasteurization equipment. Milk quality in Mali also varies by season.  Although CRLLB 

and its milk cooperative partners have co-investigated quality complaints, including through a 

collection center inspection, they have been unable to identify the source of the problem. Other 

quality issues that are difficult to verify are even more deleterious to long term human health. 

For example, interviews with government technicians and value chain actors suggest that the 

proper use, documentation, and monitoring of veterinary drugs such as antibiotics and vaccines 

are not widely respected in the Bamako peri-urban production basin.  

 

3.6.3 Quality-signaling mechanisms 

Differentiation on the basis of experiential or credence quality attributes, on its own, is 

difficult for consumers to observe prior to the purchase and consumption of a product. Retailers, 

too, must be able to verify quality attributes that are important to the consumers they serve. In 

order to address this information asymmetry between processors and buyers, and to avoid a 

 

85 

situation of adverse selection, firms must couple such investments with attributes that provide 

information to retailers and consumers on quality attributes. In the Malian dairy sector, firms use 

several such quality-signaling mechanisms: sales representatives, product warranties, brand 

recognition, and third-party certification.  

 

3.6.3.1 Sales representatives 

All firms use sales representatives to provide product information to existing and potential 

clients. In the most direct model, all four artisanal firms interface directly with consumers 

through forward integration into retailing. According to fresh milk vendors, the two most 

frequent inquiries that they receive from customers are, “Is it real?” and “Where does it come 

from?” By maintaining their own retail points, firm owners or employees can provide personal 

assurance on these questions. Additionally, the owner of FM3 explained that Malians “have the 

mentality to go to the houses of sellers when they wish to purchase milk,” especially when the 

sellers are reputed for having good quality, and because dairy products are traditionally 

consumed at night when markets are closed. Indeed, all four of the artisanal firms (FM1-FM3, 

PM4) make some sales from their homes, and two have sales windows that are connected to their 

household concession.26   

As these artisanal firms have grown, they have acquired (through ownership or rental) retail 

points which they staff with family members or trusted employees. They are typically located in 

markets, near busy roads, or next to mosques in order to access high-traffic areas while offering 

convenience to customers who are commuting, shopping, or returning home from Friday prayers.  

 

26 Another advantage of maintaining a residential point of sale is that the owner can make sales throughout the day 
and evening without depending on an employee, while at the same time being present at home to receive milk 
deliveries, make and package products, and oversee any other domestic activities. 

 

86 

One owner stated that “it is not easy to have a retail point in Bamako that works.” She explained 

that it had taken her years of searching and trying out new locations before she secured her 

external points of sale. CRLLB has provided some assistance to its members while attempting to 

standardize and scale up the direct retailing model. Since 2010, the cooperative has provided 

locally-fabricated retail kiosks to about 45 of its members. The kiosks are locally-constructed out 

of metal, wood, and glass, and include a sales window and signage. CRLLB has offered the 

kiosks (valued at about 800,000 FCFA) and new refrigerators (valued at about 450,000 FCFA) to 

its members on zero-interest credit.  

In contrast, all semi-industrial and industrial firms outsource product distribution to external 

partners. Two artisanal firms (FM2, PM4) also make direct sales to retailers, which has become 

their primary market channel.  In these cases, maintaining direct contact with consumers 

becomes more difficult and the relationship with the distributor becomes most important. 

Consequently, these firms employ commercial agents who make deliveries to their distribution 

partners, provide them with information, and monitor how products are stored and displayed. 

Individuals on motorcycles deliver products for the artisanal firms, while semi-industrial firms 

deploy two-person teams (a driver and sales agent) in delivery vehicles.  

 

3.6.3.2 Warranties 

As firm scale and distance to the end consumers increases, product warranties are another 

mechanism by which a firm provides product information and signals quality. All firms selling to 

retailers or other intermediary clients offer product warranties, in which they guarantee quality 

by replacing any defective products ex post. The conditions of warranties are usually vaguely 

defined, but at minimum include cases in which a client discovers a defect in the product that is 

 

87 

clearly attributable to the firm, such as faulty packaging or spoilage that the firm can trace to 

faults in the product process. Several firms extend their warranties to cover some share of losses 

that are clearly not their own fault, such as due to power cuts or even when the causes are 

unknown.  

 

3.6.3.3 Brand recognition 

Building brand recognition is another signaling mechanism used by the largest firms (PM7-

PM9). These firms have developed brand recognition and reputation through time, scale, and/or 

mass communications. First, the largest firms are also the most experienced among the case 

studies, with the oldest having a market presence in Mali of over twenty years. Second, each firm 

has a substantial distribution network consisting of 500 retailers or more, and which reaches 

towns outside of Bamako. Third, each firm has invested in different forms of consumer 

advertising, which include branded trucks, websites, social media messaging, television and 

radio advertising, and billboard communications. As additional measures to develop brand 

recognition, PM7 has opened a yoghurt parlor in the national park and PM8 had used celebrity 

sponsorships and social media accounts to build brand recognition among youth.  

In partnership with CRLLB, fresh milk firms have also attempted to build a collective brand 

to signal that products are made from high-quality fresh milk. Interviews and marketing 

materials suggest that the CRLLB brand represents at least three principles that depend largely 

on CRLLB’s partnership with its FENALAIT suppliers and on a system of support, standards, 

and quality control set up within these organizations. The first principle is purity, i.e. that all 

CRLLB dairy products are made 100% from local milk. The second principle is food safety. 

CRLLB food safety standards (e.g., proper refrigeration and a clean retailing environment) are 

 

88 

checked weekly by marketing advisors who visit kiosks once a week. A third and related 

principle is traceability. Local milk is sourced from local farmers and through close 

relationships. Furthermore, firms have some degree of confidence that the production, collection, 

pasteurization, and transport of CRLLB milk is supported by an infrastructure of FENALAIT 

veterinary care, standards, and quality control. The cooperative’s forty-five retail kiosks, placed 

throughout the city, play a role in building recognition for this brand. These kiosks are 

identically painted blue and white, with the cooperative name and tagline (“milk from our 

cows”) written in Bambara in bold letters across the tops. On the sides are images of dairy 

farming and the phrase, “I love my local milk.” CRLLB has also made available to its members 

branded product labels with similar colors, images and messages.  

However, the effectiveness of the CRLLB brand faces several challenges. None of the 

interviewed members cited this common brand as a primary advantage to cooperative 

membership. Not all members have access to a kiosk, and others have additional unbranded retail 

points. Many members do not use branded labels because they are expensive for them to print 

and possibly also vulnerable to counterfeiting. CRLLB does not have the resources to advertise 

outside the kiosk signs. Feeble use of the CRLLB brand may also be due to weak 

implementation of its standards. First, many members sell products that are based on powdered 

milk alongside local milk products, sometimes in the CRLLB-branded kiosk. Second, 

traceability is weakened by the fact that firms source milk from outside CRLLB and the 

FENALAIT network. Firms often are not aware of the source of this milk, let alone the standards 

by which it was produced and handled. 

 

 

89 

3.6.3.4 Third-party certification 

A final potential mechanism to signal product quality is through third-party quality 

certification. According to government regulation, all food firms must obtain for each of their 

products a certification that is issued by l’Agence Nationale de la Sécurité Sanitaire des Aliments 

(ANSSA) (Government of Mali, 2006). The ANSSA seal, which should be printed on product 

packaging, indicates to consumers that the product has met product-specific norms and standards 

that the Government has adapted from CODEX. However, few, if any, imported or domestic 

dairy products carried the seal at the time of fieldwork. Among the cases, only PM5 had obtained 

an ANSSA certification, and in this case for its water product.  

One reason for this low certification compliance may be weak recognition of the seal by 

consumers and even retails. In a 2018 survey of Bamako consumers, only 20% of respondents 

recognized the ANSSA seal and most of these stated that they did not have a good idea of its 

meaning (Vroegindewey, 2019). Another reason for low compliance may be the complexity and 

costs of obtaining ANSSA certification. Among the required steps, a company must provide 

proof of necessary business registrations and tax payments, provide satisfactory results from lab 

tests of their products, and pass a factory inspection. PM5 estimated that obtaining certification 

for its water products took two years and cost about two million FCFA. Finally, low compliance 

might also be explained by the historically weak government enforcement of noncompliance. 

However, some of these trends could be changing: before its water carried the ANSSA seal, PM5 

had been fined 250,000 FCFA by the Ministry of Commerce. To avoid similar sanctions and 

disruptions to its dairy and juice business, the firm was in the process of certifying these other 

product lines as well. Among the firm cases, at least two other powdered milk firms (PM7, PM9) 

had also initiated the certification application process.  

 

90 

 

3.6.4 Packaging  

The use of distinct packaging features is the fourth source of differentiation in the Bamako 

dairy market. In the same context, Theriault et al. (2018) found price premiums to be associated 

with powdered milk products that were packaged in metal tins or cartons, as opposed to plastic 

pouches which could potentially be manufactured in Bamako. Packaging can potentially create 

value for retailers and consumers in three ways. First, packaging protects product contents from 

the external environmental, which can extend shelf-life and facilitate wider distribution. The 

fresh milk firms that primarily distribute through their own retail points (FM1, FM3) package all 

or most of their products by double-bagging them in thin transparent plastic bags that they hand-

tie to close. The two larger artisanal firms (FM2, PM4), all of which distribute most of their 

volumes through independent retailers, have upgraded to use a thicker plastic bags that they heat-

seal with a hot iron and, occasionally, locally-purchased plastic bottles. The powdered milk 

firms, all of which outsource their distribution to external partners, use even sturdier packaging: 

opaque plastic pouches (for drinking yoghurt and dégué), plastic bottles (for drinking yoghurt), 

or durable plastic cups (for strained yoghurt and dégué).  

Second, packaging is another means of communicating information about quality. In addition 

to exposing products to more shocks and longer lead times between production and consumption, 

a longer marketing chain also increased the potential for information asymmetry between a firm 

and consumers. To mitigate this and avoid problems of adverse selection, expensive packaging 

can signal to consumers that quality contents are superior to products with less sophisticate 

packaging. Among the cases, five of the seven firms (all powdered milk firms) with longer 

marketing channels use packaging that they either custom-manufacture in their factories or 

 

91 

import directly. Package labeling can also explicitly convey information on product attributes. 

For example, PM7 packaging indicates that yoghurt is made with “enriched milk,” a reference to 

the enrichment of the powdered milk they use.  

Despite its low cost and the minimal protection that it provides, informants claimed that 

many consumers draw a strong cultural connection between the very familiar appearance of 

hand-tied plastic bag packaging and quality fresh milk. The bags themselves are locally 

manufactured and ubiquitously used by Malian vendors in traditional markets, and their 

transparence allows Malians to visually examine the milk contents (e.g., color and consistency, 

and any debris). The owner of FM3 explained that, traditionally, milk was sold out of calabashes, 

which allowed buyers to verify the quality ex ante by sight, smell, or taste.  

Even when fresh milk firms upgrade to thicker plastic bags that are better suited for longer 

marketing chains, they maintain the transparency of their packaging. As distance between the 

firm and retail point of a product increases, labeling also becomes more important. One fresh 

milk firm does not even use labels for products sold from her home, because of the personal 

contact that she has with these consumers. Similar to their choice of packaging materials, fresh 

milk firms use simple labeling that evokes associations with traditional milk marketing. Most 

labels are locally-printed in monochromatic colors; feature icons of cows or traditional milk 

marketing; include a brand name that utilizes words from a local language and/or the personal 

name of the firm; and provide contact information on the firm.  

For instance, FM2’s label features a simple drawing of a Fulani woman selling milk from a 

bowl on her head, under the company name in Fulani, which translates as “milk from our cows.”  

The owner, who is herself Fulani, explained that “everyone knows that it is the Peuhl who sell 

and know milk.” To further underline (and help to protect) the product quality, “Fresh 

 

92 

pasteurized milk 90 degrees” and “conserve at 7 degrees” are printed vertically in large font up 

either side of the package. However, the owner of FM3 believed that there were limits on how 

much she should innovate on traditional packaging, because sophisticated packages have a 

“pharmaceutical” appearance that many Malians associate with the industrial processing of 

powdered milk. Indeed, all nicely-packaged yoghurts that are sold in Bamako are made from 

powdered milk. 

Finally, packaging form and size can improve the convenience and affordability of a product, 

especially for consumers with limited resources for purchasing, storing, and consuming dairy 

products. Across the cases, the firms indicated/stated that their most popular packaging formats 

are those that are smallest. The per-liter consumer prices of PM9’s pasteurized milk and drinking 

yoghurt are among the highest on the Bamako market. However, the firm packages these 

products in 200 ml pouches in order to offer a relatively low price-point to consumers. 

Consequently, this packaging has become the format preferred by wholesale distributors in 

Bamako and is an essential element for maximizing a brand’s breadth of distribution. PM5, PM8, 

and the two other industrial dairies in Bamako have each introduced a similar packaging for their 

milk and drinking yoghurt products, and PM7 has purchased new packaging equipment to do the 

same. Additionally, pouches and plastic bottles are convenient for more viscous products 

(pasteurized milk, drinking yoghurt, féné) because consumers can drink from them directly, 

while plastic cups are preferred for products that are consumed by a spoon (e.g., strained 

yoghurt). Some firms also offer bottles and cups with re-closable tops as an additional 

enhancement.  

 

 

93 

3.7 Conclusion  

This paper examines the competitiveness of local fresh milk—versus imported powdered 

milk—as an input for dairy processor firms in Bamako. To do this, I conduct case studies of nine 

dairy firms, analyzing the relationship between their use of each input and the creation of 
competitive advantage through cost advantage and/or product differentiation.  

Regarding cost, I find that the use of fresh milk is more costly than use of powdered milk. 

The purchase price of fresh milk is at least 20% higher than the liquid equivalent of powdered 

milk. Additionally, in contrast to powdered milk for which spot markets already exist in 

Bamako, all firms that procure fresh milk must carefully coordinate their transactions with 

suppliers. Yet, even with special procurement arrangements in place, these firms still face 
transaction cost problems related to quality and supply uncertainty.  

I also find four sources of differentiation that dairy firms use to distinguish their brands from 

competitors. First, firms offer unique products, such as a new type of yoghurt. Later, as rivals 

adopt similar products and competition increases, firms enhance the product or create variety by 

introducing new ingredients, such as flavors or additives. Second, within each product type firms 

seek to preserve or enhance quality. Third, firms adopt quality-signaling mechanisms to convince 

consumers that unobservable product attributes of their brand are superior to those of 

competitors. Mechanisms include the use of sales representatives, enhanced packaging and 

labeling, warranties, investments to improve brand recognition, and third-party certification. 

Finally, firms offer unique packaging features (e.g., different forms or materials) that create 

customer value by protecting contents, signaling quality, improving convenience, or increasing 
affordability.  

 

94 

However, I find that fresh milk firms face unique difficulties exploiting these differentiation 

opportunities, compared to firms of powdered milk. For example, these firms have adopted 

variations of three signaling mechanisms to communicate product information to consumers that 

have a preference for fresh milk, but each has significant disadvantages. One mechanism is to 

sell through small specialized retail points that are costly to scale up. Another is to use traditional 

packaging that poorly protects product contents and has limited appeal to many consumers. 

These firms also attempt to use collective branding that requires more coordination in order to 
work effectively.   

In order to improve the competitiveness of local dairy value chains, milk farmers and 

processors should work with the Malian government to bring down the costs of procuring and 

processing fresh milk. Measures that improve milk producers’ access to productive breeds, feed, 

and veterinary care should help to reduce the price of fresh milk, while improving quality and 

stabilizing supply. The government should also address problems in the business environment 

that drive up transaction costs in dairy procurement and marketing, especially the high costs of 
electricity, poor roads, and limited agricultural financing.  

Even with these measures, it is unlikely that the costs of fresh milk will drop below those of 

powdered milk in the near term. The competitiveness of local dairy value chains, therefore, will 

also depend on their ability to stimulate and exploit demand for products made from fresh milk 

through differentiation. One possibility is the development and commercialization of products 

that are well-suited for local fresh milk and difficult to imitate using powdered milk. Because 

milk ingredient type is a product attribute that is difficult for consumers to observe, investments 

in more effective quality-signaling mechanisms will also be critical. One option is the 

development of innovative packaging that maintains traditional features while offering better 

 

95 

protection and convenience. Government can also strengthen the enforcement of an existing 

certification system—which, technically, has a special designation for local fresh milk—as a tool 
to help local value chains differentiate their products.  
 
 

 

 

96 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

REFERENCES 

 

97 

 

 

 

REFERENCES 

Baché, David. (2018, March 2). Ces importations qui font tourner le lait malien. RFI. Grande 
Raportage. Retrieved from: http://www.rfi.fr/emission/20180302-mali-lait-importations-
elevage-viande-europe 

 
Bain, J. S. (1956). Barriers to new competition (Vol. 3, p. 55). Cambridge, MA: Harvard 

University Press. 

 
Bellemare, M. F., and Bloem, J. R. (2018). Does contract farming improve welfare? A review. 

World Development, 112, 259–271. 

 
Boehlje, M. (1999). Structural Changes in the Agricultural Industries: How Do We Measure, 

Analyze and Understand Them? American Journal of Agricultural Economics, 81(5), 1028–
1041.  

 
Burke, W. J., Myers, R. J., and Jayne, T. S. (2015). A Triple-Hurdle Model of Production and 

Market Participation in Kenya’s Dairy Market. American Journal of Agricultural Economics, 
97(4), 1227–1246.  

 
Choplin, Gerard. (2016). SOS Faim. Oxfam Solidarite. “Europe’s Dairy Sector has its eyes on 

Africa.” Report for Oxfam Solidarité and SOS Faim-Belgique.  

Cook, M. L., and Chaddad, F. R. (2000). Agroindustrialization of the global agrifood economy: 
bridging development economics and agribusiness research. Agricultural Economics, 23(3), 
207–218.  

 
Corniaux, C., Duteurtre, G., and Broutin, C. (2014). Filières laitières et développement de 

l'élevage en Afrique de l'Ouest: l'essor des minilaiteries. Paris: Editions Karthala. 

 
Food and Agricultural Organization. FAOSTAT. Accessed 2018.  
 
Government of Mali. (2006). DECRET N°06-259/P-RM DU 23 JUIN 2006 INSTITUANT 
L’AUTORISATION DE MISE SUR LE MARCHE DES DENREES ALIMENTAIRES, 
DES ALIMENTS POUR ANIMAUX ET DES ADDITIFS ALIMENTAIRES. 

 
Hollinger, F., and Staatz, J. M. (2015). Agricultural Growth in West Africa: Market and policy 

drivers, FAO, African Development Bank, ECOWAS.  

 
Holloway, G., Nicholson, C., Delgado, C., Staal, S., and Ehui, S. (2000). Agroindustrialization 

through institutional innovation Transaction costs, cooperatives and milk-market 
development in the east-African highlands. Agricultural Economics, 23(3), 279–288. 

 

 

 

98 

Holloway, G. (2004). A revised Tobit procedure for mitigating bias in the presence of non-zero 

censoring with an application to milk-market participation in the Ethiopian highlands. 
Agricultural Economics, 31(1), 97–106.  

 
Holloway, G. J., Barrett, C. B., and Ehui, S. K. (2005). Bayesian Estimation of the Double 
Hurdle Model in the Presence of Fixed Costs (SSRN Scholarly Paper No. ID 2633551). 
Rochester, NY: Social Science Research Network.  

 
Jaffee, S., (1995). “Perishable Profits: Private Sector Dairy Processing and Marketing in Kenya.” 

In Jaffee, S., and Morton, J. F., ed. Marketing Africa’s high-value foods: comparative 
experiences of an emergent private sector. Dubuque, IA: Kendall/Hunt Pub. milk – Sci-
Afrique. (n.d.). 

 
Kirsten, Johann F., Karaan, Mohammad, A.S., and Dorward, R. Andrew. (2009). Introduction to 

the Economics of Institutions. In Kirsten, Johann F., Dorward, R. Andrew, and Poulton, 
Colin (Eds.), Institutional economics perspectives on African agricultural development (35-
74). IFPRI, Washington, D.C. 

 
Livingstone, Emmet. (2018, April 4). How EU milk is sinking Africa’s farmers. Politico. 

Retrieved from https://www.politico.eu/article/eus-milk-scramble-for-africa/ 

 
Mahoney, J. T., and Pandian, J. R. (1992). The Resource-Based View Within the Conversation 

of Strategic Management. Strategic Management Journal, 13(5), 363–380. 

 
North, D. C. (1987). Institutions, transaction costs and economic growth. Economic 

 
Maxwell, J.A. (2013). Qualitative Research Design (3rd ed). Sage Publications, Inc. Thousand 

inquiry, 25(3), 419-428. 

Oaks, CA.  

 
Meyer, C., and Duteurtre, G. (1998). “Equivalents lait et rendements en produits laitiers: modes 

de calculs et utilisation.” Revue d’élevage et de médecine vétérinaire des pays 
tropicaux, 51(3), 247-257. 

 
Porter, M. E. (1985). Competitive advantage: creating and sustaining superior performance. 

1985. New York: FreePress, 43, 214. 

 
Reardon, T. (2015). The hidden middle: the quiet revolution in the midstream of agrifood value 

chains in developing countries. Oxford Review of Economic Policy, 31(1), 45–63.  

 
Siegfried, J. J., and Evans, L. B. (1994). Empirical studies of entry and exit: a survey of the 

evidence. Review of Industrial Organization, 9(2), 121-155. 

 
Sterns, J. A., Schweikhardt, D. B., and Peterson, H. C. (1998). Using case studies as an approach 
for conducting agribusiness research. The International Food and Agribusiness Management 
Review, 1(3), 311–327. 

 

99 

 
Tschirley, D. L., Snyder, J., Dolislager, M., Reardon, T., Haggblade, S., Goeb, J., … Meyer, F. 

(2015). Africa’s unfolding diet transformation: implications for agrifood system 
employment. Journal of Agribusiness in Developing and Emerging Economies, 5(2), 102–
136. 

 
Theriault, V., Vroegindewey, R., Assima, A., and Keita, N. (2018). Retailing of Processed Dairy 

and Grain Products in Mali: Evidence from a City Retail Outlet Inventory. Urban 
Science, 2(1), 24. 

 
Vroegindewey, R., Richardson, R., David Ortega, and Theriault, V. (2019). Consumer and 

Retailer Preferences for Local Dairy Products: Evidence from Stacked Choice Experiments 
in Urban Mali. East Lansing: Michigan State University. Research Paper # 134. 
https://www.canr.msu.edu/resources/consumer-and-retailer-preferences-for-local-dairy-
products-evidence-from-stacked-choice-experiments-in-urban-mali 

 
Waldman, D., and Jensen, E. (2013). Industrial organization: theory and practice. Routledge. 
 
World Bank. (2015). “Mali Living Standards Measurement Study.” 
 
Yin., R. (2014). Case Study Research Design and Methods (5th ed.). Thousand Oaks, CA: Sage.  
 

 

 

 
 

 

 

 

100 

CHAPTER FOUR: ANALYSIS OF THE DETERMINANTS OF MILK PRODUCERS’ 

PARTICIPATION IN OUTPUT MARKETS 

 

4.1 Introduction 

Substantial growth in the market supply of Malian milk will be necessary to meet the rising 

demand for dairy products, while also improving the livelihoods of milk producers and 

strengthening the competitiveness of the Malian dairy sector against imports. The aim of this 

paper is to understand the barriers that constrain Malian milk producers from entering and 

supplying markets, and to identify policy measures that can boost such market participation.  

The question of smallholder farm commercialization in Africa is an issue that has 

increasingly occupied the attention of policymakers, especially in recent years as liberalization of 

markets, globalization, and transformations in retailing and consumption are magnifying 

challenges and opportunities (Reardon and Timmer, 2012). As Barrett (2008, p. 300) 

summarizes, the theoretical benefits of market-oriented production and trade, relative to 

subsistence production for own consumption, are important, and include not only “the one-off, 

static welfare effects of trade according to comparative advantage” but also more rapid total 

factor productivity growth due to opportunities for larger-scale production and the increased 

interflow of ideas. Additionally, in the Malian context, a greater and more stable market supply 

of milk should have broad economic benefits, insofar as it strengthens the competitiveness of the 

local dairy value chain and contributes to greater availability and accessibility of diverse foods, 

perhaps especially in urban areas (Theriault et al., 2018).  

However, a body of evidence has showed that transaction costs are a significant impediment 

to the participation of African farmers in various agricultural markets (Barrett, 2008). 

Transaction costs are ex ante costs that a household faces in searching for a market and 

 

101 

negotiating a transaction, and the ex post costs of monitoring and enforcing the terms of the 

transaction (North, 1990). Numerous market participation studies have focused on milk, because 

its perishability and other technical characteristics generate many transaction costs in its 

production, processing, and marketing (Jaffee, 1995). Most of these studies are from either 

Kenya (Burke et al., 2015; Olwande et al., 2015) or Ethiopia (Holloway et al., 2005; 2004; 

2000). Their salient finding, consistent with findings from other African markets, is that 

household-specific variables (such as the level of education of a household’s head) and 

geography-specific variables (especially access to milk buyers such as traders, cooperatives, and 

processors) strongly influence the probability and value of sales among milk producers, 

reflecting the pervasive impacts of transaction costs in milk markets (Barrett, 2008). These 

studies also confirm the importance of productivity-enhancing technologies and conditions, 

namely household ownership of crossbred dairy cows, access to extension, and favorable 

agricultural-ecological (or agro-ecological) conditions.  

This study focuses on Mali because the milk market participation literature has largely 

overlooked West Africa and especially the Sahel region, which differs distinctively from the East 

African context in terms of policy history, market structure, and agricultural ecosystems. For 

example, the Kenyan dairy sector benefits from more favorable climactic conditions and, since 

2002, has also enjoyed the protection of imports tariffs of up to 60% (Orasmaa et al., 2016). In 

contrast, the West African Economic and Monetary Union (UEMOA) has established a Common 

External Tariff (CET) schedule that taxes imported powdered milk at only 5%, and other 

products at up to 35% (UEMOA, n.d.). 

To my knowledge, no other research has conducted econometric analysis of milk producer 

market participation in the Sahel region, and only one working paper has examined this issue in 

 

102 

the broader West Africa region. Balagtas et al. (2007) use a Heckman selection model to study a 

limited sample of households in Ivory Coast. However, this paper defines milk market entry as 

household ownership of any cattle (versus household ownership of dairy cattle specifically or the 

production of milk) and uses a small nonrandom sample. This study uses a more precise 

definition of milk market participation, exploiting a randomly-sampled and nationally-

representative household dataset, and by employing a rich set of control variables. Additionally, 

Mali has the third-largest cattle population in West Africa (after Nigeria and Niger) and a similar 

ranking in terms of annual milk production (FAO, 2019); thus, it provides an excellent case 

study for understanding commercial behaviors in the region’s high-potential milk supply basins.  

In West African countries, livestock plays a critical role in household incomes and the 

national economy. In Mali specifically, three-quarters of households own livestock of some kind, 

and in 2011 the livestock sector contributed to about eight percent of the national gross domestic 

product (FAO, 2017). Focusing on the dairy sub-sector, consumer demand in Mali and the region 

has been climbing steadily with income growth and urbanization (Zhou and Staatz, 2016). While 

this poses a huge opportunity to producers, regional supply has not been keeping apace. Zhou 

and Staatz (2016) estimate, even under conservative assumptions of future income growth, that 

by 2040 regional dairy supply will fall short of demand by a magnitude of five. Unless 

production growth increases, this deficit will have to be made up by a commensurate increase in 

imports, in order to avoid real price increases. Figure 6 provides a picture of this pattern for the 

case of Mali. It presents linear trends of the per capita supplies of domestic milk and imported 

dairy, based on the past decade of milk output and population growth.27 Over time, domestic 

 

27 The large surge in domestic supply, followed by a drop, during the 2011-13 period might be explained by very 
good rain and pasture conditions in 2012 and, in contrast, poor agricultural and security conditions in 2013 
(Government of Mali, 2013; 2012). The opposite inflection of imported milk supply during this same period 
suggests the substitutionary relationship that domestic and imported dairy supply have with one another.   

 

103 

supply has been decreasing while imports have been rising to meet the deficit. At the household 

level, low rates of market participation help to explain this pattern: although 20% of Malian 

households produce milk, only 3% market any volume milk during the year (authors’ 

calculations from World Bank, 2015). 

 
Figure 6: Per capita supply of domestic milk and dairy imports in Mali (L/per person) 
30
25
20
15
10
5
0

2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023

Availability p.c. of imported cow dairy
Linear (Availability p.c. of imported cow dairy)

Availability p.c. of domestic cow dairy
Linear (Availability p.c. of domestic cow dairy)

Sources: Authors’ calculations from Government of Mali (2007-2011; 2012; 2013; 2014-2016; 2017) and FAO 
(2019) 
  

 

 

I investigate this household-level pattern in Mali. Section two presents the conceptual 

framework for understanding an agricultural household’s participation in milk markets when 

transaction costs are present. Section three describes the data and empirical approach for 

applying this framework to the Malian context. In section four I present the results, then 

conclude by highlighting key findings and their policy implications for increasing milk supply in 

the Malian market.  

 

4.2 Conceptual framework 

Because the focus is understanding the supply behavior of milk producer households, the 

theoretical framework is the agricultural household model (Singh et al., 1986). If I could 

 

104 

reasonably assume that Malian producers had perfect access to markets for milk and all 

necessary inputs, then household milk supply boils down to a profit-maximization problem in 

which decision-making is guided only by exogenously-determined prices and conditioned on the 

given production technology. However, because Malian dairy producers face significant 

transaction costs in these markets, I must extend the model to account for market imperfections. 

Specifically, below I draw mainly from Barrett (2008)’s articulation of the non-separable 

agricultural household model.28  

Assume that a household maximizes its utility over a bundle of commodities, subject to a 

budget constraint involving farm production, sales, and non-farm income; a production 

technology constraint; and a vector of unobservable “decision prices” (Key et al., 2000; p. 248). 

The decision prices for selling (or purchasing) a given commodity equals its observable local 

market price (P) minus (plus) the transaction costs that a household faces to participate in that 

market. The transaction costs themselves depend on household-specific characteristics (Z) and 

physical and institutional infrastructure (G), which together affect the search, information, 

transportation, and negotiation costs associated with carrying out a transaction. The transaction 

costs also depend on household-level productive assets (A), liquidity from non-farm income (W), 

and net sales (NS). The latter variable affects transaction costs when there is a fixed costs-

component wherein the per-unit amount of total transaction costs drops as volumes increase and, 

consequently, there exists a threshold quantity below which market participation is infeasible 

(Barrett, 2008; Holloway et al., 2004). Net sales can also capture purchase arrangements in 

which buyers pay differentiated prices based on volumes in a given sales lot.  

28 Olwande et al. (2015) similarly drew from Barrett (2008)’s model in their analysis of farmer participation in milk 
(and other) commodity markets in Kenya. 

 

 

105 

The defining feature of this model is that, because household-specific variables determine 

transaction costs, the decision prices that producers face are likewise household-specific. 

Consequently, in a given milk market I expect to observe differentiated participation in markets 

across households.29 For milk-producing household i, the market participation decision has two 

parts. The first is the decision to participate (or not) as a seller, denoted by M which equals one 

for market entry and zero otherwise. Second is the decision of sales volumes, denoted by the 

continuous variable Q, which is positive if and only if !=1. I can express the reduced-form 
 $%=$%(!,(,),*,+,,,-.).                                                                                                  (1) 

equation as:  

 

4.3 Materials and methods  

4.3.1 Data sources 

I apply the household market participation model to the Malian context by using data from 

the 2014 Mali Living Standards Measurement Study-Integrated Survey on Agriculture (LSMS), 

a 4,009 household cross-sectional survey that was implemented by the Planning and Statistics 

Unit of the Malian Ministry of Rural Development (CPS/SDR) and the World Bank (World 

Bank, 2015). The population-based survey has national coverage, with the exception of the 

northern region of Kidal which surveyors could not access due to insecurity at the time of data 

 

29 In addition to household-specific transaction costs, Barrett (2008)’s market participation model specifies a second 
layer of transaction costs that are commodity and geography-specific. Because I focus on participation in milk 
markets, it is not necessary to control for commodity type. The geography-specific nature of transaction costs arises 
because local markets “are differentially integrated into the global economy because of spatial differences in costs of 
commerce, in the degree of competition among market intermediaries, or both” (Barrett, 2008; p. 301). The 
incorporation of observed local market prices—which are determined by these factors—accounts for the 
geographically-differentiated nature of transaction costs.  

 

106 

collection. Government of Mali (2016) provides detailed information on the stratified random 

sampling approach of the LSMS survey. 

The analysis is based on 717 households that reported owning a female cow. Of these milk 

producers, 126 households participated in milk markets as sellers. Data from the livestock 

modules of the LSMS were collected in a single round from December 2014 to February 2015 

(Government of Mali 2016). Other modules cover household and community (i.e., enumeration 

area)-level characteristics, and were collected between July 2014 and February 2015. 

 

4.3.2 Econometric model and estimation  

The econometric model must account for the two-staged nature of market participation, as 

depicted in Eq. (1), as well as for the large share of nonparticipants in the dataset (i.e., 

households that produce, but do not sell, milk). Nonparticipation in markets results in a corner 

solution problem, in which the outcome variable (in this case, milk sales) is zero for a nontrivial 

number of observations but is continuous otherwise. Applying an ordinary least squares 

estimator on such a truncated dataset would result in biased and inconsistent estimates 

(Wooldridge, 2015).  

The Tobit model (Tobin, 1958) represents one solution for addressing the corner solution 

problem (see Holloway et al. (2004; 2000) for different applications of the Tobit approach in the 

Ethiopian milk market). However, applying this model to the subsample of milk producers 

involves estimating simultaneously the determinants of the probability and magnitude of the 

market participation outcome. This imposes the restrictive constraint that the processes driving 

these two stages be the same, i.e., that the set of significant explanatory variables, and the 

 

107 

directions of their effects, be the same (Burke, 2009). Other studies have showed that this is an 

unreasonable assumption in the context of milk marketing (e.g., see Burke et al., 2015).  

There are two commonly-used alternative models that more flexibly address the corner 

solution problem through a two-step procedure, but which also nest the Tobit model as a special 

case (Lin and Schmidt, 1984). One approach, the Heckman (1979) sample selection model, treats 

nonparticipants as unobserved data resulting from nonrandom sample selection (see Balagtas et 

al., (2007) for an application in Ivoirian milk markets). However, because the data is a random 

sample, and because the theoretical model views the outcomes of both stages (including 

nonparticipation outcomes) as the result of a household decision-making process, the Heckman 

model is not appropriate for the analysis.  

The other approach is the Cragg (1971) two-tiered (or double-hurdle) model, which treats 

zeroes as observed outcomes and allows for two different processes to estimate different 

parameters for the probability and value of sales separately (see Olwande et al. (2015) and Burke 

et al. (2015) for application in the Kenyan milk market, and Holloway et al. (2005) for an 

application in Ethiopia). Because these features are better suited to the theoretical model and 

dataset, I adopt the Cragg model for the analysis.  

 

 

Specifically, I estimate regressions of the following form:  

Stage 1:    ((!%=1)=(($%>0)=	3%4+6%,   and                                                                         (2) 
Stage 2:   $%=	789+	:%.		                                                                                                             (3) 

 

 

108 

household makes any milk sales and zero otherwise. Eq. (3) defines the household’s decision 

Eq. (2) defines the milk market entry decision for household i, where !% takes on unity if the 
regarding its level of market participation, in terms of the quantity of milk sales. 3% and 78 are 
and 4 and 9 are the marginal effects of these vectors of explanatory variables, for the first and 

the two vectors of explanatory variables, according to the theoretical model depicted in Eq. (1), 

second stage, respectively. I estimate Eq. (2) using maximum likelihood estimation and a probit 

model. I can estimate Eq. (3) by fitting the data to either a truncated normal distribution or a 

lognormal distribution (Cragg, 1971). I assume that the errors in both equations are normally and 

independently distributed. 

 

4.3.3 Variable definitions  

Table 13 defines the dependent and explanatory variables that I select for the model. In 

addition to the theoretical model, the choice of variables is guided by a review of the other 

empirical studies investigating household participation in milk markets, and data availability 

from the LSMS survey. The first stage dependent variable is binary, taking on unity when a 

household reports any milk sales made in the previous year, and zero otherwise. The second 

stage dependent variable represents the liters of milk that each household sold in the previous 

year, which I calculate based on the number of months in the year that households reported milk 

offtake and the average milk quantities that they reported selling in each of these months.  

 
 

 

 

109 

Table 13: Variable definition and summary statistics  

 

Definition 

HH made any volume of milk sales (Level of participation > 0) during 12-month survey period 
Volume of milk (L/year) sold by HH during 12-month survey period 
 

No. female cows (of local breed) raised by HH 
No. female cows (of mixed/exotic breed) raised by HH 
Share of HH cattle herd vaccinated in past 12 months (%) 
Share of HH cattle herd treated for internal parasites (%) 
Share of HH cattle herd treated for external parasites (%) 
A pond or stream was a primary or secondary herd water source during dry season (dummy) 
Oilseed cake was a primary or secondary source of herd nutrition in past 12 months (dummy) 
Household owns a feed/drinking trough (dummy) 
Number of functioning cell phones owned by household 
Number of functioning radios owned by household 
Household owns at least one means of transportation: bicycle, motorcycle, or car (dummy) 
Total hectares of land that is cultivable by HH  
Total annual non-farm income (1,000 FCFA/year) of HH during 12 mos. prior to survey period 
Total annual non-farm income (1,000 FCFA/year) of HH during 12 mos. during survey period 
Number of household members who are adult males  
Number of household members who are male and over the age of 18  
Number of household members who are under the age of 18  
HH head is male (dummy) 
HH head reported Fulani ethnicity (dummy) 
Number of years of formal schooling completed by HH head 
 

 

Variable 
 
Dependent Variables 
Participation 
Level of participation 
 
 
Household (HH)-specific explanatory variables 
No. local dairy cows 
No. foreign dairy cows 
% vaccinated 
% treated for parasites 
% treated for ticks 
Water source 
Oilseed cake 
Trough 
No. cell phones 
No. radios 
Transport 
Ha. Land 
Log (nonfarm income, lagged) 
Log (nonfarm income) 
No. adult males 
No. adult female 
No. children 
HH head gender 
HH head Fulani 
HH head yrs of edu. 
 
 
Location-specific explanatory variables 
Log (milk price) 
Urban 
No. collection centers 
 % electricity access 
Dist. weekly market 
Dist. daily market 
Dist. training center 
Dist. financial institution 
Dist. motorable road 
Average temperature 
Annual rainfall 
Semi-arid 
Arid 
Notes: N(sample) =718; N(market participants) = 127; HH refers to household 
 
 

Ave. (of two survey rounds) of local median price (FCFA/L) of packaged fluid milk 
Community is located in an urban area (dummy) 
Number of collection centers inventoried at the cercle level 
Share of households sampled in community that have access to electrical grid (%) 
Distance (km) to nearest periodic market 
Distance (km) to nearest permanent market 
Distance (km) to nearest agricultural training center 
Distance (minutes) to nearest micro finance institute 
Distance (km) to nearest clay or paved road 
Average annual temperature (multiplied by 10 °C) during 1960-1990 period 
Total annual precipitation (mm) during 1960-1990 period 
Community is in semi-arid (vs arid or sub-humid) agro-ecological zone (dummy) 
Community is in arid (vs semi-arid or sub-humid) agro-ecological zone (dummy) 

4.3.3.1 Household-specific explanatory variables 

Household-level productive assets and production technology (represented by A in Eq. (1)) 

raise farm output and productivity, thereby increasing marketable surplus (NS) and reducing per-

unit production and transaction costs. Thus, I expect that household access to such resources will 

positively influence market participation. The number of female cows raised by the household is 

clearly a critical asset for milk production, and I distinguish between local-breed cows and mixed 

(and foreign) breed cows to also capture the yield-enhancing benefits of the latter type.  

 

110 

To capture household management of dairy herd health and nutrition, I include variables 

measuring (separately) the shares of the total cattle herd in the past year that was vaccinated, 

treated for internal parasites, and treated for ticks and other external parasites. As a measure of 

herd access to water throughout the year, I include a dummy variable indicating whether a 

natural water source (e.g. pond or stream) was one of the primary water sources during the dry 

season. Two other dummies indicate access to resources that are associated with intensive milk 

production: use of an oilseed cake as a primary type of feed, and household ownership of feeding 

(or drinking) troughs. I also include household ownership of other assets that improve access to 

information and markets. Cell phones and radios could be a means for accessing information on 

markets, prices, and production practices. Ownership of a means of transportation should greatly 

reduce the time to market.  Finally, I include the number of household adults and the number of 

farm hectares cultivated by the household, which may serve as important sources of labor and 

animal feed, respectively, in the context of imperfect factor markets. Similarly, in a context of 

imperfect credit markets, I also include estimates of the total nonfarm income earned during two 

periods: the past twelve months (i.e. concurrent to the milk production period examined), and the 

twelve to twenty-four months (i.e., preceding the milk production period examined).  

Other household-level characteristics (Z) can influence market participation by influencing 

productivity, and/or by generating or attenuating transaction costs. Household characteristics are 

also determinants of milk consumption which, in the non-separable agricultural model, enters the 

market participation decision by constraining net surplus (NS). I attempt to capture gender 

effects by including a dummy variable for male or female household headship, and by 

 

111 

disaggregating the total household adults variable by male and female.30 However, there is some 

ambiguity around the expected net effects of gender and household size. First, although women 

typically have more limited access to inputs and greater time and mobility constraints, compared 

to men, in traditional West African agricultural households they tend to be more involved with 

dairy herd maintenance, milking, intrahousehold milk distribution, and milk marketing (Salla, 

2016). Second, although an increase in the number of household adults increases access to labor 

(as mentioned above), it may also increase household consumption of milk, thereby reducing net 

surplus, all other factors held constant. Increases in the number of household children, which 

increases milk demand without improving labor, should have a less ambiguous negative effect on 

market participation. Another household-level variable indicates whether the household head is 

Fulani, which is the largest pastoral ethnic group in Mali. I also include a continuous variable 

indicating the years of formal schooling completed by the household head, as a measure of his or 

her human capital, with the expectation that greater human capital has a positive effect on market 

participation.  

 

4.3.3.2 Location-specific explanatory variables 

Local market prices (P) are important determinants in a household’s vector of decision 

prices. Price differences across markets, furthermore, reflect a second layer of location-specific 

transaction costs that are determined by market integration and concentration. I obtained prices 

from the LSMS community-level dataset, in which survey enumerators recorded three price 

observations (for each of a select number of consumer commodities) from the local market of 

 

30 In the sample, 61% of households reported that female cows were primarily managed by the household head. 
Another 27% reported collective management by multiple household members, which presumably includes partial 
management by the household head. Thus, focusing on the gender of this individual is appropriate. 

 

112 

each enumeration area and in both survey rounds.31 Consumer prices should be a sufficient 

indicator of households’ market incentives because milk supply chains in Mali are relatively 

short, and many households do sell their milk directly to consumers. To approximate the local 

output milk price faced by producer households, I took the median consumer prices of packaged 

fluid milk for each enumeration area, then averaged these medians from both survey rounds. 

Because households produce and market milk on a daily basis throughout the year, taking the 

average from two periods is appropriate. Also, for these reasons, estimating expected prices 

(instead of the use of realized prices) is less a concern for milk, compared to agricultural 

products associated with delayed production cycles. However, in an effort to capture longer-term 

expectations regarding milk demand and prices I include a dummy for whether the household is 

located in an urban area, versus a rural area.32 Because urban areas contain large and growing 

consumer markets (Zhou and Staatz, 2016), this variable should encourage market participation. 

However, higher population densities place pressure on land that livestock might otherwise 

access for grazing or foraging; thus, I cannot predict the net effect of this variable.  

Access to various physical and institutional infrastructure (G), can facilitate the adoption of 

productivity-enhancing technology and directly reduce the transaction costs to market 

participation. For example, milk collection centers provide a market outlet for local producers, 

and often facilitate access to other services and inputs such as veterinary care, vaccinations, and 

feed. I include a variable indicating the total number of centers at the cercle-level, which I expect 

 

31 Due to insecurity and other reasons, LSMS survey teams were unable to access 197 of 1,073 of the enumeration 
areas, which corresponds to about 15% of the sample of milk producers. I addressed missing community data (i.e., 
prices and infrastructure) by imputing median values from the next-largest geographic units.  
32 Following government census definitions (Government of Mali, 2012), the LSMS defines an enumeration area as 
urban if it both has at least 5,000 inhabitants and is administratively classified as an urban commune. In Mali, there 
are thirty-six urban communes and 667 rural communes. 

 

113 

to positively affect market participation.33 Access to electricity enables such centers to run lights, 

cooling tanks, and refrigerators, while relying less on more costly gas generators. Electricity also 

enables retailers to store milk in refrigerators, which should increase milk demand from these 

intermediary buyers. Thus, I estimate the share of households in each community that reported 

having access to electricity and include this variable in the regressions. 

To capture access to other various public goods, I include variables representing community 

distance to the associated infrastructure.  I hypothesize that access to a weekly market, and 

especially to a permanent (i.e., daily) market, will positively affect market participation, as each 

should reduce the transaction costs that households incur to transport milk, search for buyers, and 

negotiate prices. Access to agricultural training centers should also positively affect market 

participation insofar as these improve access to extension agents and, thereby, encourage the 

adoption of productivity-enhancing technologies. Access to a financial institution might also 

positively affect market participation, by allowing producers to invest in lumpy assets (e.g., 

additional dairy cows). Household access to credit might smooth income during stressed periods, 

thereby stabilizing demand for milk while helping producers to avoid destocking as a negative 

coping mechanism. Access to a motorable road (defined here as a clay or paved road) should 

reduce transportation costs and overall access to markets and services. Finally, I control for 

location-specific climactic and agro-ecological conditions by including several variables that 

geo-reference rainfall and temperature data from secondary databases.   

 

 

33 This data comes from a commune-level government inventory that distinguishes between (but does not define) 
collection points, collection centers, and dairies. For simplicity, I refer to all of these, collectively, as “collection 
centers.” In Mali, cercle and commune are the second and third administrative units in Mali, respectively. Among 
Mali’s eight regions, there are forty-nine cercles and 703 communes.  

 

114 

4.4 Results and discussion 

4.4.1 Descriptive statistics  

In Table 14, I summarize the milk sales for each quintile of the weighted sample. The top 

twenty percent of household milk sellers account for eighty-five percent of all sales, while the 

bottom twenty percent account for only one percent. This distribution of sales is quite 

concentrated. For example, Olwande et al. (2015) calculated that the top quintile of milk sellers 

in Kenya accounted for 59% of sales in 2010.  

 
Table 14: Distribution of milk sales across quintiles, weighted 
  

Quintiles, based on annual household milk sales  
5th 
1st  
144 
12,926 
85% 
1% 

4th 
1,147 
9% 

 

 
Mean household sales (L/year) 
Share of total sales  

 

 

2nd 
256 
1% 

3rd  
437 
3% 

Table 15 reports summary statistics of the explanatory variables for the LSMS sample, and 

also disaggregates these statistics between milk market participants and non-participants. The 

average size of the local breed herd is less than eight cows for the entire sample, but it is twenty-

two cows among the sub-sample of market participants. For both the sample and market 

participants, ownership of mixed-bred dairy cows is very low. There are reports that farmer 

adoption of cross-bred cattle is increasing in peri-urban Bamako, through artificial insemination 

programs, direct breeding of local herds with cross-bred bulls, and the direct importation of 

breed stocks (Government of Mali, 2017). However, this adoption does not appear to be 

widespread in Mali as a whole.   

 
 
 

 

 

115 

Table 15: Summary statistics  

Sample  
(N=717) 

  

  
 
 

Variable 

 

Mean 

 

Std. Dev. 

 

Min 

 

Max 

 

 
 

 
 

Non-

participants  
(N=591) 
Mean 

5.93 
0.24 
0.51 
0.37 
0.23 
0.38 
0.12 
0.12 
2.14 
1.24 
0.68 
12.30 
7,333.85 
25,483.99 
2.95 
3.32 
7.81 
0.98 
0.15 
0.73 

435.89 
0.03 
1.72 
0.03 
11.87 
36.86 
29.73 
48.34 
20.89 
276.19 
708.81 
0.79 
0.15 

  
 
 
 

 
 

  

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 

  
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 
 
 
 
 
 
 
 
 
 
 
 
  

Participants 
(N=126) 
Mean 

 
 

 
 

21.89 
0.40 
0.68 
0.41 
0.30 
0.56 
0.22 
0.26 
2.26 
1.08 
0.61 
15.60 
5,651.84 
18,951.23 
2.75 
2.90 
6.83 
0.96 
0.44 
0.67 

463.93 
0.02 
1.96 
0.05 
11.69 
44.06 
33.96 
60.21 
22.39 
274.93 
749.45 
0.85 
0.10 

0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
 
133.29 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
0.00 
261.00 
78.00 
0.00 
0.00 

208.00 
22.00 
1.00 
1.00 
1.00 
1.00 
1.00 
1.00 
15.00 
8.00 
1.00 
239.45 
84,888.00 
3,116,750.00 
15.00 
22.00 
47.00 
1.00 
1.00 
16.00 
 
1,225.00 
1.00 
12.00 
1.00 
130.00 
200.00 
240.00 
600.00 
185.00 
300.00 
1,299.00 
1.00 
1.00 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

8.73 
0.27 
0.54 
0.38 
0.24 
0.41 
0.14 
0.15 
2.16 
1.21 
0.67 
12.88 
7,038.26 
24,335.98 
2.91 
3.25 
7.64 
0.98 
0.20 
0.72 

 
Household (HH)-specific explanatory variables 
No. local dairy cows 
No. foreign dairy cows 
% vaccinated 
% treated for parasites 
% treated for ticks 
Water source 
Oilseed cake 
Trough 
No. cell phones 
No. radios 
Transport 
Ha. land 
Nonfarm income, lagged  
Nonfarm income 
No. adult males 
No. adult female 
No. children 
HH head gender 
HH head Fulani 
HH head yrs of edu. 
Location-specific explanatory variables 
Milk price 
 
Urban 
 
No. collection centers 
 
 % electricity access 
 
Dist. weekly market 
 
Dist. daily market 
 
Dist. training center 
 
Dist. financial institution 
 
Dist. motorable road 
 
Average temperature 
 
Annual rainfall 
 
Semi-arid 
 
Arid 
  
Notes: HH refers to household 
 

440.81 
0.03 
1.76 
0.03 
11.83 
38.13 
30.48 
50.43 
21.15 
275.97 
715.95 
0.80 
0.15 

15.57 
1.46 
0.43 
0.44 
0.39 
0.49 
0.35 
0.35 
2.34 
1.27 
0.47 
26.77 
8,479.96 
167,358.40 
2.00 
2.35 
5.77 
0.14 
0.40 
2.42 
 
126.15 
0.16 
2.30 
0.14 
11.83 
33.00 
31.92 
53.65 
22.77 
7.34 
293.23 
0.40 
0.35 

 

 

 

Adoption of other productivity-enhancing technologies is also quite low. However, the 

average household land holding is almost 13 hectares. Average nonfarm household income was 

about 24 million FCFA during the year covered by the survey, and 7 million FCFA in the 

preceding year. The average household size is almost 14 people, with about half of that number 

made up of children under the age of sixteen. Household heads are almost always male, and on 

average have less than one year of formal education. Twenty percent of producer households 

have a Fulani household head, but this share jumps to forty-four percent among market 

participants.  

 

116 

Turning to location-specific characteristics, only three percent of producer households live in 

areas classified as urban, and about the same share has access to electricity. On the other hand, 

80% of producers live in the semi-arid zone. On average, producer households live about twelve 

kilometers (km) from a weekly market, thirty-eight km from a daily market, thirty km from a 

training center, fifty km from a financial institution, and twenty-one km from a motorable road. 

Sixty-seven percent of households report owning some mechanical or motor-driven means of 

transportation for accessing these infrastructures and institutions, while the remainder would 

presumably walk or else use public, borrowed, or animal-powered transport. On average, 

households live in a cercle with 1.76 milk collection centers; however, due to the nature of this 

data it was not possible to estimate distances to a collection center.   

 

4.4.2 Econometric results   

Table 16 presents results from the two stages of the Cragg model. I estimated each using 

robust standard errors. In order to facilitate interpretation of the MLE results of the probit 

regression, which is nonlinear, I compute the average partial effects (APE) of each explanatory 

variable on the probability of market entry.34 I first fitted the 2nd stage with a truncated normal 

distribution; however, it was not sufficiently smooth to obtain MLE convergence. In the final 

model, in order to smooth out the distribution of the 2nd stage dependent variable, I fitted the data 

with a lognormal distribution. The results of the 2nd stage regression are already interpretable as 

conditional average partial effects (CAPE), representing the APE of each explanatory variables 

on the quantity of milk sold, conditioned on market entry. Further, because the dependent 

variable in the second stage is in logarithmic form, the estimated coefficients represent 

34 I estimate standard errors and derive significance for the APE statistics (from the probit model) and CAPE 
statistics (from the lognormal model) via the delta method.  

 

 

117 

elasticities for explanatory variables that are also in logarithmic form (i.e., income and price) and 

semi-elasticities for all others.  

Lastly, I estimated the unconditional APE (UAPE) in order to understand the net effect of 

each explanatory variable. The UAPE is dependent on both stages of the estimation and, thus, 

represents an overall effect across the entire population of milk producers. For these reasons, it is 

a helpful summary statistic and is especially useful for policy analysis.35  

To test for robustness, I also fitted the data with two alternative models. The first was a Tobit 

model with the same explanatory variables as the original model. The second was a Cragg model 

that included regional dummies as explanatory variables. The results, displayed in Appendix A, 

show that the sign and significance of the parameters estimated in the original model are largely 

robust to these alternative specifications, with only a few minor differences. 

 
 

 

35 To obtain UAPE standard errors, I followed Burke (2009)’s bootstrapping method using 100 replications. 
However, to use this method I had to fit the data with a truncated lognormal distribution in the 2nd stage. A 
comparison of the 2nd stage coefficient estimates using lognormal and truncated lognormal shows that they are the 
same in significance and in value up to at least two decimal places.  
 

 

118 

Table 16: Cragg model results and average partial effects 
  

Probit (1st Stage) 

  

 

Regression Results 

Robust 

APE 

Lognormal (2nd Stage) 
Regression Results /  
Conditional APE 

Robust 

 
 
 

 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Coef. 
 

Coef. 
 

** 
* 
** 

* 

** 
* 
** 

* 

Coef. 

 

SE 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
*** 
 

** 

 
 
*** 
 

** 

0.001 
0.006 
0.029 
0.032 
0.032 
0.024 
0.031 
0.028 
0.006 
0.014 
0.030 
0.000 
0.007 
0.006 
0.009 
0.009 
0.004 
0.077 
0.028 
0.006 
 

0.006 
0.007 
0.068 
-0.026 
0.042 
0.065 
0.060 
0.067 
0.008 
-0.017 
-0.041 
0.001 
-0.008 
-0.004 
-0.002 
0.000 
-0.004 
-0.141 
0.128 
-0.002 

0.020 
-0.010 
-1.173 
0.724 
1.005 
0.073 
-0.763 
0.099 
-0.019 
-0.206 
0.395 
-0.002 
-0.037 
-0.180 
0.162 
-0.079 
0.051 
-1.215 
0.097 
0.031 

0.032 
0.039 
0.370 
-0.140 
0.230 
0.352 
0.325 
0.366 
0.042 
-0.091 
-0.222 
0.004 
-0.045 
-0.021 
-0.013 
-0.002 
-0.019 
-0.765 
0.697 
-0.013 
 

SE 
 
0.009 
0.032 
0.161 
0.173 
0.171 
0.132 
0.167 
0.152 
0.033 
0.074 
0.161 
0.002 
0.039 
0.035 
0.050 
0.048 
0.021 
0.418 
0.146 
0.031 

 
 
Household (HH)-specific explanatory variables 
No. local dairy cows 
No. cross-bred dairy cows 
% vaccinated 
% treated for parasites 
% treated for ticks 
Water source 
Oilseed cake 
Trough 
No. cell phones 
No. radios 
Transport 
Ha. land 
Log (nonfarm income, lagged) 
Log (nonfarm income) 
No. adult males 
No. adult female 
No. children 
HH head gender 
HH head fulani 
HH head yrs of edu. 
 
 
Location-specific explanatory variables 
Log (milk price) 
Urban 
No. collection centers 
 % electricity access 
Dist. weekly market 
Dist. daily market 
Dist. training center 
Dist. financial institution 
Dist. motorable road 
Average temperature 
Annual rainfall 
Semi-arid 
Arid 
 
Constant 
 
 
Pseudo R-squared 
Observations 
Notes: Dependent variable of the probit model is 1 if household sold milk and 0 otherwise.  
Dependent variable of truncated normal model is liters of milk sold. ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively

0.207 
2.201 
0.012 
1.633 
0.003 
0.002 
-0.001 
0.001 
0.009 
-0.031 
-0.002 
-0.213 
-1.073 
17.472 
 
0.4694 
126 

0.603 
-0.179 
0.008 
0.462 
-0.004 
0.005 
0.001 
0.001 
0.000 
-0.018 
0.001 
1.086 
0.986 
-0.783 
 
0.286 
717 

0.042 
0.085 
0.006 
0.098 
0.001 
0.000 
0.000 
0.000 
0.001 
0.002 
0.000 
0.098 
0.117 
 
 
 
  

0.111 
-0.033 
0.002 
0.085 
-0.001 
0.001 
0.000 
0.000 
0.000 
-0.003 
0.000 
0.200 
0.182 

0.231 
0.465 
0.032 
0.542 
0.007 
0.002 
0.002 
0.001 
0.003 
0.013 
0.000 
0.525 
0.634 
4.192 

*** 
 
 
 
 

* 

*** 
 
 
 
 

* 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

* 
*** 
 
 

* 
*** 
 
 

 
 

 
 
 

 
 
 
 
 

 
 
 
 
 

 

 
 

 
 
 

 
 
 
 
 

 
 
 
 
 

 
 

** 

 
 
 
  

** 

 
 
  

 

 

 

  

 

 
 
 
  

 

 

  

SE 
 
0.005 
0.060 
0.390 
0.353 
0.497 
0.253 
0.327 
0.328 
0.078 
0.112 
0.290 
0.003 
0.072 
0.090 
0.138 
0.133 
0.045 
0.556 
0.290 
0.057 

0.522 
1.630 
0.052 
1.299 
0.011 
0.004 
0.004 
0.001 
0.007 
0.029 
0.001 
0.639 
1.018 
8.724 

 

 

 
 
*** 
 
*** 
** 
** 

 

 
 

 
 
 

 
 
 

 
 

 
 
 
 
 
 
 
 
 
 

 
 
 

** 

* 

** 

** 

 

** 

 
 
  

  

 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

Net effects 

Unconditional 

 APE 

Coef. 
 

0.040 
0.043 
0.223 
-0.035 
0.440 
0.421 
0.243 
0.439 
0.046 
-0.142 
-0.190 
0.004 
-0.058 
-0.056 
0.014 
-0.017 
-0.013 
-1.099 
0.821 
-0.010 
 

0.737 
0.171 
0.016 
0.819 
-0.005 
0.006 
0.000 
0.001 
0.002 
-0.026 
0.000 
1.210 
0.940 

 
 
 
  

SE 

 
0.013 
0.081 
0.204 
0.237 
0.258 
0.168 
0.194 
0.215 
0.049 
0.095 
0.217 
0.003 
0.059 
0.051 
0.079 
0.074 
0.030 
0.546 
0.197 
0.046 
 

0.287 
0.838 
0.032 
0.594 
0.010 
0.003 
0.003 
0.002 
0.004 
0.018 
0.001 
0.984 
1.072 
 
 
 
  

 

 
*** 
 
 
 

* 
*** 
 

** 

 
 
 
 
 
 
 
 
 

** 
*** 
 
 

*** 
 
 
 
 

** 

 
 
 
 
 
 
 

 
 
 
  

 

119 

First, the overall pattern of results confirms the hypothesis that each stage of market 

participation is driven by a different process: the signs and significance of almost all explanatory 

variables vary across the two equations. The two variables that represent exceptions—the 

number of local breed female cows and the gender of the household head—are discussed further 

below. Thus, the data justifies the use of a two-stage model as opposed to a one-stage Tobit. I 

also conducted a formal specification test of the Tobit model against the Cragg model, using a 

post-estimation likelihood ration test (Lin and Schmidt, 1984), which confirmed that the Cragg 

model is the better fit.    

The number of female cows of local breed is positive and statistically significant in both 

stages. Across both stages and for the entire population of milk producers, the UAPE estimate 

indicates that the acquisition of one additional local breed cow increases milk sales by an 

average of 4%. However, the number of foreign or cross-bred cows is not a significant variable 

in either stage. This is surprising, given that every other market participation study that includes 

a similar variable finds the estimated coefficient to be significant and larger than the effect of 

local breed cows (Olwande et al., 2015; Balagtas et al., 2007; Holloway et al. 2005; 2000). The 

result may be due to the overall low level of adoption of cross-bred cows in Mali: in the sample, 

only 8% of households own such a cow, and only half of these own more than one. 

Herd vaccination rates, use of oilseed cakes in feed rations, and access to a trough, to land 

and to a year-round natural water source each have a positive and significant effect on the 

probability of market participation. However, none of these variables positively influence milk 

sales once households enter the market; further, oilseed cakes and vaccination have a negative 

effect on volumes sold. The herd share that is treated for ticks and the share that is treated for 

internal parasites each have a positive and significant effect on volumes sold. Of these 

 

120 

productivity-enhancing resources and technologies, three have a net-positive effect (i.e. 

significant and positive UAPE) across the entire sample of milk producers. A one percent 

increase in the herd share that is treated for ticks, dry season access to a natural water source, and 

ownership of a trough are each associated with increases in milk sales of 4%, 42%, and 44%, 

respectively.  

Surprisingly, the number of radios and nonfarm income have a negative effect on volumes 

sold for households that have entered the milk market. It could be that milk sales are a less-

preferred means of income-generation compared to other livelihoods (including nonfarm 

activities) that are more accessible to wealthier households, and that the number of radios 

partially captures this wealth effect. The usefulness of a radio or cell phone in facilitating access 

to market or production information depends on the availability of such information, which may 

in fact be limited in Mali. For example, the Malian government does not currently monitor, 

publish, or report the market prices of milk, as it does for other agricultural commodities.  

Other household characteristics that influence market participation are the gender and 

ethnicity of the household head. In additional to the number of local breed cows, the gender of 

the household head is the only other variable to have a significant positive effect on both stages 

of market participation. Female-headed households are 14% more likely to participate in milk 

markets and are associated with a 122% increase in milk sales, compared to male-headed 

households.36 Overall, the UAPE estimate indicates that such households are associated with a 

110% increase in milk sales. Household heads that are of Fulani ethnicity are also 13% more 

likely to participate in milk markets, compared to others, and the UAPE estimate indicates that 

Fulani households are associated with an 82% increase in milk sales.  

36 In the sample of 717 producer households, fifteen households had a female head. 

 

 

121 

Increases in the number of male, female, or children household members do not significantly 

influence market participation. I expected some ambiguity with respect to the net effect of the 

number of household adults, since more adults potentially means more household demand for 

milk (thereby reducing net surplus) as well as more labor (thereby potentially increasing milk 

output). However, the insignificance of the coefficient for children is surprising, assuming that 

these individuals only factor into the consumption aspect of household decision-making.  

A 1% increase in the price of packaged milk is associated with an 11% increase in the 

probability of market participation. Although price does not have a significant effect in the 

second stage, its net effect on all producers (the UPAE) is a .7% increase in sales for every 1% 

increase in price, ceteris paribus. Although being located in an urban zone and community 

electrification shares are only statistically significant in the second stage at the 18% and 21% 

levels of confidence, respectively, their estimated effects are quite significant. 

Surprisingly, none of the variables capturing access to market institutions or infrastructure 

are significant, with the exception of distance to a daily market. However, its estimated 

coefficient on the probability of market participation is positive. The UAPE estimate is also 

positive and significant, suggesting that, as a household’s distance to a daily market increases by 

each additional kilometer, its milk sales also increase by an average of .6%, other factors held 

constant. It could be that a primary motivation for producing milk is to meet household milk 

demand, in which case it is the unconsumed surplus that is marketed. Access to daily markets 

could allow such households to outsource their milk supply, obviating the need to produce it 

themselves. If milk marketing is a less preferred means of earning income, compared to other 

activities that suddenly become more feasible with the presence of daily markets, it would 

strengthen this line of reasoning. The muted effects of access to a training center or financial 

 

122 

institution might be explained by the low quality of services offered by these institutions (or their 

limited relevance to milk producers), even if they are nearby. I might understand the lack of 

significance of access to a weekly market in light of the fact that milk producers require a more 

regular market outlet to sell daily output. If a large share of milk is sold at farmgate or at the 

homes of neighbors, this would further mute the effects of better access to markets, milk 

collection centers, or motorable roads.  

Finally, households located in the semiarid agro-ecological zone—as opposed to arid or sub-

humid—are 20% more likely to participate in milk markets. This indicator primarily 

characterizes the water availability conditions—and, by extension, vegetative conditions—that 

best supports rainfed dairy cattle production, i.e. an annual length of growing period of 70-180 

days (Sebastian, 2016). This zone covers most of the southern half of Mali, including all regions 

except Gao, Kidal, and Timbuktu.  

 

4.5 Conclusion 

Substantial growth in the market supply of Malian milk will be necessary to meet the rising 

demand for dairy products, while also improving the livelihoods of milk producers and 

strengthening the competitiveness of the Malian dairy sector against imports. In this study, I have 

utilized a nationally-representative household dataset to investigate the factors that can 

encourage such growth. Following other recent papers that have focused on East Africa, I use a 

double-hurdle econometric model, which allows me to examine separately the probability and 

the value of milk market participation. Because this is the first study of its kind to analyze milk 

marketing in a major milk producing countries of West Africa, I expect the results to provide 

fresh policy insights for this region. In particular, four key findings emerge.  

 

123 

First, despite the great yield-enhancing potential of mixed-bred dairy cattle, the adoption of 

this technology has been extremely limited in Mali. So much so, that the variability of mixed-

bred cattle ownership in the dataset appears insufficient to allow me to estimate the marginal 

effects. However, taking the statistically significant UAPE for the number of local breed cows as 

a rough lower-bound estimate of the marginal effect of each additional mixed bred animal, I can 

conclude that the impacts on market participation should indeed be substantial. The Malian 

government should continue to increase producer access to mixed-bred cattle.  

Second, improvements to the health and nutrition of dairy herds also have great potential to 

improve market participation through increased productivity. The findings indicate that pest and 

disease control, access to zero-grazing technologies (such as feeding troughs), and to year-round 

water sources are especially key. Although the particular measure for improved feed did not have 

a significant net effect across both stages of market participation, the significant and positive 

effect of being located in semi-arid zone, which partly reflects grazing conditions, points to the 

importance of herd nutrition. Improving the availability of high-quality feed will be especially 

critical with the dissemination of mixed-bred cattle, which have more complex nutritional needs 

compared to local breeds.  

Third, gender has great influence on a household’s participation in milk markets. Assuming 

that the household head plays a primary role in the management of milk production and use, 

female decision-makers market more than twice the volumes of males, other factors held 

constant. This result, combined with the reality that women currently face unequitable access to 

productive resources, suggests that milk commercialization policies could make substantial gains 

by focusing on female producers. However, other research conducted in Mali has cautioned that 

 

124 

women may get displaced milk value chains as they modernize (Schneider et al., 2007). Overall, 

this finding highlights the importance of mainstreaming gender into any milk-related policies.  

Fourth, the results provide evidence that Malian milk producers are responsive to price 

incentives, despite the considerable asset specificity and transaction costs that are present in milk 

marketing. This suggests that macroeconomic policies, such as stronger import duties that 

increase the domestic price of fresh milk relative to that of imported substitutes, should have a 

positive pull on milk supply. This result also underlines the importance of market price 

information. The Malian government should prioritize the inclusion of milk prices in its regular 

market monitoring and information products.  

 

 

 

125 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

APPENDIX 

 

126 

 

 

APPENDIX 

 
 
Table 17: Robustness checks: alternative Tobit and Cragg model specifications 
  
Alternative model #2: Cragg with regional dummies 

 
 

Alternative model #1: Tobit  
Regression Results 
Robust SE 
Coef. 
 
 

 
 

  
 
 
*** 

** 

 

 
 

  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

Coef. 

Probit (1st Stage) Regression Results 
 
 

Robust SE 

 
 

Lognormal (2nd Stage) Regression Results 
 
 

Robust SE 

Coef. 

 
 

  
 
 
*** 

 

 

 
 

*** 

** 
* 

* 
** 
** 
** 

*** 
** 
** 

*** 
* 
*** 

 
 
*** 
 
 
 

0.041 
0.177 
1.127 
1.171 
1.154 
0.927 
1.113 
1.007 
0.232 
0.490 
1.110 
0.013 
0.270 
0.247 
0.364 
0.342 
0.131 
2.837 
0.949 
0.221 

0.009 
0.033 
0.166 
0.175 
0.176 
0.137 
0.172 
0.156 
0.034 
0.075 
0.170 
0.002 
0.038 
0.035 
0.049 
0.048 
0.021 
0.427 
0.148 
0.032 

0.004 
0.053 
0.335 
0.303 
0.417 
0.232 
0.269 
0.281 
0.068 
0.102 
0.268 
0.004 
0.062 
0.075 
0.115 
0.106 
0.034 
0.490 
0.289 
0.048 

0.019 
-0.027 
-1.118 
0.681 
0.979 
-0.051 
-0.698 
0.108 
-0.006 
-0.242 
0.465 
-0.002 
-0.028 
-0.163 
0.115 
-0.056 
0.050 
-1.383 
-0.052 
0.048 

0.032 
0.032 
0.379 
-0.101 
0.235 
0.388 
0.326 
0.418 
0.044 
-0.090 
-0.189 
0.003 
-0.040 
-0.027 
-0.008 
-0.017 
-0.019 
-0.833 
0.747 
-0.013 

0.156 
0.211 
1.753 
-0.255 
1.964 
2.155 
2.389 
1.993 
0.163 
-0.656 
-1.151 
0.021 
-0.322 
-0.379 
0.164 
-0.031 
-0.091 
-5.418 
4.933 
-0.117 
 

 
Household (HH)-specific explanatory variables 
No. local dairy cows 
No. foreign dairy cows 
% vaccinated 
% treated for parasites 
% treated for ticks 
Water source 
Oilseed cake 
Trough 
No. cell phones 
No. radios 
Transport 
Ha. land 
Nonfarm income, lagged 
Nonfarm income 
No. adult males 
No. adult females 
No. children 
HH head sex 
HH head fulani 
HH head yrs of edu. 
 
 
Location-specific explanatory variables 
Milk price 
Urban 
No. collection centers 
 % electricity access 
Dist. weekly market 
Dist. daily market 
Dist. training center 
Dist. financial institution 
Dist. motorable road 
Average temperature 
Annual rainfall 
Semi-arid 
Arid 
Constant 
Kayes 
Koulikoro 
Sikasso 
Segou 
Mopti 
 
Pseudo R-squared 
Observations 
Notes: Dependent variable of the probit model is 1 if household sold milk and 0 otherwise. Dependent variable of truncated normal model is liters 
of milk sold. ***, **, and * indicates p<0.01, p<0.05, and p<0.1, respectively. In Alternative model #2, three regional dummies (Tombouctou, 
Bamako, and Gao) were omitted due to multicollinearity.

0.602 
-0.155 
0.017 
0.340 
-0.005 
0.005 
0.001 
0.001 
0.000 
-0.030 
0.000 
1.157 
1.219 
2.600 
0.343 
-0.135 
0.051 
0.139 
-0.093 
0.292 
717 

-0.141 
2.085 
-0.012 
1.559 
0.004 
0.002 
0.000 
0.000 
0.007 
-0.003 
-0.002 
0.115 
-0.742 
11.349 
0.150 
0.017 
0.813 
0.086 
0.753 

4.403 
-0.972 
0.064 
5.370 
-0.019 
0.030 
-0.001 
0.007 
0.014 
-0.119 
0.003 
5.732 
4.738 
-3.753 
 
 
 
 
 
 

0.239 
0.471 
0.033 
0.565 
0.007 
0.002 
0.002 
0.001 
0.003 
0.016 
0.001 
0.524 
0.695 
4.842 
0.450 
0.464 
0.497 
0.480 
0.428 
 

0.486 
1.339 
0.038 
1.109 
0.010 
0.003 
0.003 
0.001 
0.006 
0.029 
0.001 
0.675 
0.936 
8.549 
0.470 
0.576 
0.843 
0.659 
0.544 

1.623 
3.279 
0.215 
3.228 
0.044 
0.013 
0.015 
0.007 
0.021 
0.091 
0.003 
3.055 
3.965 
28.639 

* 
*** 
 
 
 
*** 
 
 

* 
*** 
 
 
** 

 
 

 
 
 

 
 

 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 
 
 

0.127 
717 

 
 
 
 
 
 
 
  

 
 
 
 
 
 
 
 
 

 

 
 
 
 

 
 
 

 

 
 
 
 
 
 

** 
* 

 
 
 
 
 
 

  

* 

** 

 
 
 
 
 
 
 
 
 

 

 
 
 
 
 

 
 

126 

** 

* 

 
 

 

 
 

 
 

** 

*** 

 
 

*** 

 
 

 
 
  

 
 

 

  

* 

 

 
 
  

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  

 
 
  

 

127 

 

 

 

 

 

 

 

 

 

 

 

 

 

REFERENCES 

 

128 

 

REFERENCES 

Balagtas, J.V., Coulibaly, J.Y., Jabbar, M., and Negassa, A. (2007). Dairy Market Participation 

with Endogenous Livestock Ownership: Evidence from Cote d’Ivoire. Paper presented at 
2007 American Agricultural Economics Association Meeting, Portland, Oregon, July 29-
August 1, 2007.   

 
Barrett, C. B. (2008). Smallholder market participation: Concepts and evidence from eastern and 

southern Africa. Food Policy, 33(4), 299–317.  

 
Bellemare, M. F., and Barrett, C. B. (2006). An Ordered Tobit Model of Market Participation: 
Evidence from Kenya and Ethiopia. American Journal of Agricultural Economics, 88(2), 
324–337. 

 
Burke, W. J. (2009). Fitting and interpreting Cragg's tobit alternative using Stata. The Stata 

Journal, 9(4), 584-592. 

 
Burke, W. J., Myers, R. J., and Jayne, T. S. (2015). A Triple-Hurdle Model of Production and 

Market Participation in Kenya’s Dairy Market. American Journal of Agricultural Economics, 
97(4), 1227–1246.  

 
Cragg., J.G. (1971). Some Statistical Models for Limited Dependent Variables with Applications 

to the Demand for Durable Goods. Econometrica 39: 829–44.  

 
UEMOA. (n.d.) Tarif Exterieur Commun de l’Union Economique et Monetaire Ouest Africaine.  
 
Food and Agriculture Organization (2019). FAOSTAT. Retrieved from 

http://www.fao.org/faostat/en/#data 

 
Government of Mali, MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 

NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. (2007-2011). 
RAPPORTS ANNUELS 2007-2011. 

 
Government of Mali. (2012). MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 

NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. 2012. 
RAPPORT ANNUEL, 2012. 

 
Government of Mali. (2013). MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 

NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. 2013. 
RAPPORT ANNUEL, 2013. 

 
Government of Mali, MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 

NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. (2014-2016). 
RAPPORTS ANNUELS 2014-2016. 

 

 

129 

Government of Mali, MINISTERE DE L’ELEVAGE ET DE LA PECHE, DIRECTION 
NATIONALE DES PRODUCTIONS ET DES INDUSTRIES ANIMALES. (2017). 
RAPPORT ANNUEL, 2017. 

 
Government of Mali, Ministry of Rural Development, Planification and Statistics Unit 

(September, 2016). Basic Information Document, ENQUÊTE AGRICOLE DE 
CONJONCTURE INTÉGRÉE AUX CONDITIONS DE VIE DES MÉNAGES 2014.  

 
Government of Mali, Ministry of Rural Development, Planification and Statistics Unit (2016). 
Enquête Agricole de Conjonture Intégrée aux Conditions de Vies des Ménages 2014: Basic 
Information Document. Bamako.  

 
Government of Mali, Minstry of Economy, Finance, and Budget, National Statistics Institute, 

Central Census Bureau. (December, 2012). QUATRIEM RECENSEMENT GENERAL DE LA 
POPULATION ET DE L’HABITAT DU MALI (RGPH-2009), ANALYSE DES RESULTATS 
DEFINITIFS, THEME URBANISATION.  

 
Heckman, J. J. (1979). Sample Selection Bias as a Specification Error. Econometrica, 47(1), 

153–161.  

 
Holloway, G. (2004). A revised Tobit procedure for mitigating bias in the presence of non-zero 

censoring with an application to milk-market participation in the Ethiopian highlands. 
Agricultural Economics, 31(1), 97–106.  

 
Holloway, G. J., Barrett, C. B., and Ehui, S. K. (2005). Bayesian Estimation of the Double 
Hurdle Model in the Presence of Fixed Costs (SSRN Scholarly Paper No. ID 2633551). 
Rochester, NY: Social Science Research Network.  

 
Holloway, G., Nicholson, C., Delgado, C., Staal, S., and Ehui, S. (2000). Agroindustrialization 

through institutional innovation Transaction costs, cooperatives and milk-market 
development in the east-African highlands. Agricultural Economics, 23(3), 279–288. 

 
Jaffee, S., (1995). “Perishable Profits: Private Sector Dairy Processing and Marketing in Kenya.” 

In Jaffee, S., and Morton, J. F., ed. Marketing Africa’s high-value foods: comparative 
experiences of an emergent private sector. Dubuque, IA: Kendall/Hunt Pub. milk – Sci-
Afrique. (n.d.). 

 
Key, N., Sadoulet, E., and de Janvry, A. (2000). Transactions Costs and Agricultural Household 

Supply Response. American Journal of Agricultural Economics, 82(2), 245–259. 

 
Lin, T.-F., and Schmidt, P. (1984). A Test of the Tobit Specification Against an Alternative 

Suggested by Cragg. The Review of Economics and Statistics, 66(1), 174–177. 

 
North, D.C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge: 

Cambridge University Press. 

 

 

130 

Olwande, J., Smale, M., Mathenge, M. K., Place, F., and Mithöfer, D. (2015). Agricultural 

marketing by smallholders in Kenya: A comparison of maize, kale and dairy. Food 
Policy, 52, 22-32. 

 
Orasmaa, T., Duteurtre, G., and Corniaux, C. (2016). The end of EU milk quotas-Implications in 

West Africa. Literature review and future perspectives. Report for CIRAD. 

 
Reardon, T., and Timmer, C. P. (2012). The Economics of the Food System Revolution. Annual 

Review of Resource Economics, 4(1), 225–264. 

 
Salla, Abdou. (2016). Review of the livestock/meat and milk value chains and policy influencing 

them in West Africa. Published by the Food and Agriculture Organization of the United 
Nations and the Economic Community of West African States.  

 
Schneider, M., Kouyaté, H., Fokou, G., Zinsstag, J., Traoré, A., Amadou, M., and Bonfoh, B. 

(2007). Dynamiques d’adaptation des femmes aux transformations des systèmes laitiers 
périurbains en Afrique de l’Ouest. Revue d’Elevage et de Médecine Vétérinaire des Pays 
Tropicaux, 60(1–4), 121–131. 

 
Singh, I., Squire, L., and Strauss, J. (1986). Agricultural household models: Extensions, 

applications, and policy. 

 
Theriault, V., Vroegindewey, R., Assima, A., and Keita, N. (2018). Retailing of Processed Dairy 

and Grain Products in Mali: Evidence from a City Retail Outlet Inventory. Urban Science, 
2(1), 24.  

 
Tobin, J. (1958). Estimation of Relationships for Limited Dependent Variables. Econometrica, 

 
Wooldridge, J. M. (2015). Introductory econometrics: A modern approach. Nelson Education. 
 
World Bank. (2015). Living Standards Measurement Survey – Integrated Survey on Agriculture 

 
Zhou, Y., and Staatz, J. (2016). Projected demand and supply for various foods in West Africa: 

Implications for investments and food policy. Food Policy, 61, 198–212. 

131 

26(1), 24–36.  

2014.  

 

 

 

 

CHAPTER FIVE: CONCLUSIONS 

This research has identified and analyzed key challenges that have constrained supply and 

demand along the Malian dairy value chain, with a focus on the urban markets of Bamako. Five 

key findings emerge, along with several implications for improving the competitiveness of the 

dairy value chain through new investments and enhanced policies.  

First, this research has provided evidence that end-buyers prefer fresh milk to imported 

powdered milk. Bamako consumers are willing to pay a price premium of about fifty-five cents 

per liter for pasteurized milk that is made purely from fresh milk (a premium of almost twenty 

percent on the current price of the most popular dairy product), as opposed to reconstituted 

powdered milk. Even after disaggregating consumers into three preference-based classes, each 

class still has a higher WTP for 100% fresh milk than for 100% powdered milk. Very similar 

results hold for retailers. Given the current limited availability of fresh milk-based dairy products 

in Bamako, these findings point to an important market opportunity for fresh milk producers and 

processors. 

However, in order to realize this potential demand, the fresh milk value chain must better 

differentiate its products from those made from powdered milk. The second set of findings shed 

light on promising sources of differentiation. One general opportunity is to provide more 

information to consumers. The clear and consistent labeling of milk ingredients is one upgrade 

that manufacturers should make, and a business practice that government should more 

effectively regulate. There is also a significant WTP among consumers and retailers for quality 

certification. Government-backed certification appears to be the preference, conditioned on 

better government enforcement and consumer recognition of this system. However, about two-

thirds of consumers and retailers also have a significant WTP for private certification. Future 

 

132 

research should seek to understand the dairy quality issues that consumers and retailers care 

about most, in order to identify the attributes that should be verified. Although enhanced 

packaging remains an important source of differentiation, consumers tend to associate 

“traditional” (and low-quality) packaging with “traditional” fresh milk-based products. Thus, 

there is very limited demand for certain forms of enhanced packaging, such as plastic pouches 

and cardboard cartons. Because enhanced packaging remains an important source of 

differentiation—not to mention potential implications for protecting contents and reducing 

environmental waste—future research should seek to identify packaging upgrading options that 

are still effective in signaling the fresh milk contents of a product.  

The third finding highlights another means for value chain actors to improve information 

flows between themselves and consumers, by working more closely with retailers. Not only are 

retailers another conduit for providing product information to consumers, this research has 

showed that they also have accurate knowledge of consumer preferences and can thus provide 

such information to upstream value chain partners. In terms of geographic placement and total 

commercial volumes, boutiques and alimentations currently have the widest distribution 

throughout Bamako. However, it is mostly large powdered milk processors that partner with 

these retail shops, perhaps partly because these retailers source dairy products through 

wholesalers, which implies a requisite scale that most fresh milk processors cannot meet. 

Alternatively, many fresh milk processors retail directly, from their own homes or through 

owned kiosks. Despite any advantages of being able to directly interact with consumers, the 

informality and resource-intensiveness of these retail formats constrain the scalability of fresh 

milk distribution in Bamako. The peri-urban fresh milk value chain should explore investments 

 

133 

(e.g., in logistical and organizational capacity) that enable it to compete in more common retail 

formats, even while continuing to distribute through specialized formats.  

Fourth, if the fresh milk value chain is to reach greater scale in distribution then it must also 

improve the supply of fresh milk from producers. Policy should improve access to technologies, 

practices, and resources that improve productivity and animal health, including access to mixed-

bred dairy cattle, veterinary care, nutritious feed, zero-grazing technologies, and year-round 

water sources. Not only should policy seek to increase the total supply, but it should also seek to 

smooth supply throughout the year and improve quality, as these outcomes would substantially 

reduce the transaction costs in procuring fresh milk. Given that female decision-makers market 

more than twice the volumes of males, but face unequitable access to productive resources, 

mainstreaming gender consideration into dairy policy should be a government priority. 

Finally, despite the presence of market imperfections, all three studies have underlined the 

important role of price throughout the Malian dairy value chain. Currently, processors are 

discouraged from procuring fresh milk instead of powdered milk, due to its higher purchase price 

and transaction costs. This research has highlighted three levers that could influence the decision 

prices faced by processors, in favor of fresh milk. First, through the adoption of quality signaling 

mechanisms (e.g., clear ingredients labeling), processors could capture price premiums that 

Bamako consumers appear willing to pay for fresh milk-based products. Second, government 

policy (i.e., raising tariffs in dairy imports) could increase the relative price of powdered milk 

with respect to fresh milk. Third, policy and value chain investments that improve cattle 

productivity and/or reduce the transaction costs of marketing milk should increase the market 

supply of fresh milk; holding other factors constant (including demand), this would reduce fresh 

milk prices. Additional research should estimate the costs of implementing these measures and 

 

134 

investments, and their expected welfare impacts on various actors of the dairy value chain, 

including consumers and smallholders.  

 

135