Estimating public and private costs and benefits of implementing a milk quality assurance system in Kenya: A case study

ABSTRACT

Assuring food safety is increasingly important in emerging and developing economies. Despite being a public good, food safety in these economies is often addressed through limited private sector innovations while the public health costs and impact of food-borne diseases are largely unknown. This study estimates annual private and public costs and benefits of a quality-based milk payment system (QBMPS) in Kenya. Private costs and benefits were estimated using a partial budget analysis. From a private perspective, results show that participating farmers benefit most from the QBMPS, with a net profit of about USD 0.02 USD/kg) from an additional investment cost USD 0.0125/kg to produce Grade A milk. The cooperatives and processor have a net loss of 0.025 USD/kg milk, mainly driven by testing and initial investment costs. Disease burden was calculated using the Disability Adjusted Life Years (DALYs), while direct and indirect health costs were calculated using an incidence-based analysis. We estimate an annual disease burden of 53,000 DALYs from milk-related infectious diseases and a generated health benefit of 13 KES/kg due to reduction in disease incidence if 15% of Kenya’s milk goes through the QBMPS. These health benefits justify public and private sector support for a QBMPS.

KEYWORD:

• Quality-based milk payment
• partial budget analysis
• DALY
• disease burden
• food safety

1. Introduction

1.1. Public and private sector approaches to improve global food safety and quality

Managing and assuring the quality and safety of food products in emerging and developing markets are increasingly receiving significant attention, especially in sub-Saharan Africa. Food safety is considered an important aspect of food and nutrition insecurity in developing countries (L. Unnevehr, 2015; Ortega & Tschirley, 2017) especially in relation to how it contributes to the global food-borne disease burden (WHO, 2015). The dynamics of transforming agri-food systems in Africa are amplifying attention to food safety issues. Like in other developing markets, agrifood systems transformation is associated with rapid urbanization and expanding modern food supply chains, linked to changing dietary patterns and rising incomes, resulting in increased consumption of products that are of higher value and more food safety risk-prone, such as milk, meat and vegetables (Reardon, 2015). Additionally, increasing consumer awareness and demand for food safety triggers changes in how food markets seek to assure their target consumers of the safety of products (Grace, 2015; Ortega & Tschirley, 2017; Roesel et al., 2014; Staatz & Hollinger, 2016).

While food safety is considered a public good, there are increased efforts to address it through private sector innovations (L. J. Unnevehr, 2007). The public good nature of food safety is linked to the impact of food-borne diseases on public health in terms of morbidity and mortality. The socio-economic costs of such diseases are still largely unknown (Havelaar et al., 2015). It is argued that public interventions – including policies, standards and regulations – are important to improve food safety and reduce resultant socio-economic losses, including loss of productivity and life (L. Unnevehr, 2015). While public interventions have existed as a response to what are characterised as market failures – including limited effective incentives and information asymmetry (L. J. Unnevehr, 2007; L. Unnevehr, 2015) – studies show the expanding role of the private sector in innovating food safety assurance systems (Fagotto, 2014; Henson & Hooker, 2001; Henson & Reardon, 2005; Mensah & Julien, 2011). However, these mainly apply to developed countries, and experiences from developing countries are predominantly linked to export markets (Idowu Kareem, 2019; L. Unnevehr, 2015; Wang, 2018). For this reason, the state of food safety in domestic markets in developing countries remains a concern warranting investigation (Mwambi et al., 2020).

Economic insights on effective market incentives and appropriate interventions for the provision of safe food remain limited, particularly in resource constrained and fragmented domestic African markets (Ortega & Tschirley, 2017). Recognising that assuring food safety involves significant costs, such contexts need effective interventions to address food safety and quality concerns that are economically rational, based on evidence of risk, cost effective and sustainable in the long term, from both a public and private goods perspective (Ortega & Tschirley, 2017; Smit, 2016; Valeeva et al., 2004). The private sector is known to play a key role in assuring food quality by its willingness to implement process changes and to kindle the demand for safe food (Ortega & Tschirley, 2017). The changes are usually not by a single private sector actor, but require concurrent and coordinated efforts and investment of multiple actors in order to develop integrated approaches to managing food safety across a value chain (Valeeva et al., 2004). For example, a recent study in Kenya found that smallholder farmers who are members of producer organisations were more likely to adopt food safety practices than non-members, and also highlighted the need for both social and economic incentives to improve milk safety Mwambi et al. (2020).

QBMPS has been successful in monitoring and improving milk quality globally, though especially in developing countries, implementation challenges might restrict the success of the system (Botaro et al., 2013; Ndambi et al., 2019; PAŠIĆ et al., 2016). A number of countries have succeeded to improve milk quality by considering various parameters. In Ecuador, the milk fat, protein, total solids, total bacterial count and somatic cell counts are used (Contero et al., 2021). In New Zealand, it is mainly based on fat and protein content, but it was suggested that farmers could gain more if the lactose content was also considered (Edwards et al., 2022). In Brazil, Busanello et al. (2020) found that there was seasonality in the values of all four quality parameters (fat, protein, total bacterial count and somatic cell counts) that were used in their payment scheme, with the farmers getting better bonuses in the winter months compared to the summer months.

1.2. Milk safety and quality considerations in the Kenyan context

In the 2021 Dairy Industry Act, the government of Kenya has emphasized on improving milk quality and safety mainly by increasing the proportion of milk being pasteurized and sold through a formal channel. Adherence to the government’s licencing demands has been challenging, especially to smaller informal businesses processing and/or trading milk. According to Blackmore et al. (2022), a milk vendor is required to pay a fee and a cess on milk volumes traded, and also needs a public health licence and a medical permit, and finally an annual business permit from their county of operation. The authors further stated that less than a half of such vendors possessed the required licences, and a majority finding the requirements too demanding to meet and too costly to afford, while keeping the business profitable. Milk vendors interviewed in a study by Alonso et al. (2018) mentioned that the regulatory authorities would have achieved more safety by combining these fees with more trainings, but also by enforcing the set quality and safety regulations. The economics of private costs and benefits of such coordinated efforts remains a knowledge gap, especially in the context of developing countries’ food chains. Moreover, there is paucity of knowledge on how to link the private costs and benefits of investing – or not – in improving food safety, with the public good dimensions related to the resultant health costs and benefits. Positive effects of private investments on public health could warrant public support to such investments. Linking these private and public good perspectives may offer new insights into the rationale for practices of food chain actors and their willingness to co-invest in socio-economically viable approaches to assuring food safety.

This paper seeks to contribute to this understanding by studying a pilot intervention that sought to introduce a quality-based milk payment system in the smallholder-dominated dairy sector in Kenya. The research question guiding the paper was: what insights in food safety investments emerge from considering both the private and public costs and benefits associated with a quality assurance system along the dairy chain in the developing economy context of Kenya?

1.3. The case of piloting a quality-based milk payment system in Kenya

The growing Kenyan dairy sector, one of the largest in sub-Saharan Africa (Makoni et al., 2014; Hemme et al., 2021) provides a good case for understanding the link between agri-food sector transformation and increased attention to enhance food safety in the domestic markets in developing countries. Limited attention to milk quality results in safety concerns related to milk-borne diseases resulting from microbial contamination, high aflatoxin levels, antimicrobial residues, peroxides and other adulterants as noted in numerous studies (Kang’ethe et al., 2017; Ondieki et al., 2017; Ongarora & Karwimbo, 2019; Wafula et al., 2016). Similar to other developing and emerging countries’ food systems contexts, quality assurance in Kenya’s dairy sector is complicated by the sector being dominated by smallholders, who sell their milk to traders that operate mainly in informal markets and by a growing number of processors who compete for milk volumes (Ortega & Tschirley, 2017; Rademaker et al., 2016). In 2020, Kenya had a total of about two million dairy farms with an average herd size of 2.6 cows per farm and average yield of about 800 kg milk (SCM; Solid Corrected Milk) per cow per year and a total national milk production of about 4 million tons SCM from cows only (Hemme, 2021). Weak enforcement of quality and safety regulations worsen the milk quality and safety issues in Kenya (Ndambi et al., 2019; Özkan Gülzari et al., 2020). Threats faced by industry include growing food safety concerns of local consumers, increased imports and challenges in accessing export markets due to sub-standard quality (Bebe et al., 2020; FAOSTAT, 2022; Foreman & De Leeuw, 2013).

A quality-based milk payment system (QBMPS) aims to improve milk quality along the chain by basing payment for milk not only on volume but also on specific quality parameters; mainly physical, chemical and microbial quality (Ndambi et al., 2020). Smallholders play a key role in the Kenyan dairy sector as they comprise 80% of all dairy producers and contribute 56% to total milk production in Kenya (Odero-Waitituh, 2017). Such smallholder-dominated chains (Odero-Waitituh, 2017; VanLeeuwen et al., 2012) present unique challenges: milk from smallholders is more likely to be contaminated than from larger producers, the high number of intermediaries along the chain exposes the milk to a higher risk of contamination (Paraffin et al., 2018; Shitandi & Sterenesjö, 2004), high costs for testing individual milk samples and quality assurance of milk collected from smallholder farmers is particularly expensive due to the high transaction costs associated with small volumes collected from many sources, such as high costs for testing individual milk samples and training of high numbers of farmers and transporters (Kiwanuka & Machethe, 2016; Ortega & Tschirley, 2017). Therefore, the design of milk quality assurance systems for smallholders needs special consideration for keeping transaction costs low enough to ensure a final product of good quality that is affordable to consumers and offers suppliers a competitive price. Kenyan consumers tend to buy cheaper raw and pasteurized milk than more expensive value added dairy products (Bebe et al., 2020).

The QBPMS study involved one processor, Happy Cow Ltd. (HC) in Nakuru, which smallholder suppliers each supply an average of 8–10 kg of milk daily. To improve the quality of the milk supplied, farmers in the pilot received three different prices per kg of milk, based on its quality. They had to improve hygienic practices and feed quality to get a better quality grade hence a better price for their milk. Grade C milk, the lowest acceptable quality of milk, was bought at a base price of about 35 KES (0.34 USD) per kg, varying with the market. This base price was comparable to the prices offered by other main processors such as Brookside and New KCC, but lower than offered by informal traders. Grade B milk, of better quality than Grade C milk, attracted a bonus (additional) payment of 1 KES (0.01 USD) per kg of milk, while the best quality, Grade A milk, attracted a bonus payment of 2 KES (0.02 USD) per kg. It should be noted that the rejected milk of farmers, not meeting the standard for grade C milk, was often still marketable in markets with less stringent quality requirements or without proper testing facilities. Milk quality parameters used were total bacterial count, presence of antibiotic residues, adulteration with water (measured by freezing point), and total solids (including fat and protein). The payment module is based on a summation of the quality scores for these four parameters, as shown in Annex 1. Aflatoxin was also included as a parameter for milk acceptance, though it was not a parameter for milk pricing. Milk with aflatoxins exceeding the maximum acceptable level for East Africa (500 ng/kg) was rejected. To reduce testing costs, samples were collected from a 50 kg milk can that combines milk from 5 to 10 farmers, meaning that if one farmer had low-quality milk, all the others likely suffered. Random sampling ensured that each milk can was tested twice a month for the above-mentioned parameters. This pilot led to progress in the adoption of some good practices by all supply chain actors leading to improvement in the safety and quality of milk. However, there were some challenges with infrastructure and behavioural change of some actors which have been elaborated in the full report by Ndambi et al. (2019)

2. Materials and methods

2.1. Data collection

Data was collected through interviews, focus group discussions (FGDs) and secondary sources. Interviews were conducted with a 90 individual farmers (being members of six dairy cooperatives), six cooperative employees (one from each cooperative), twelve milk transporters, six representatives of two milk processing companies, four medical doctors, fourteen medical/public health researchers, and eight dairy industry stakeholders employed in the public sector. In addition, two FGDs were conducted with farmer groups.

Ninety farmers (25 each from cooperatives piloting the QBMPS and 10 each from the other cooperatives) were randomly sampled from these cooperatives and interviewed using a semi-structured questionnaire to generate both quantitative and qualitative information mainly on their milk handling practices and cost implications. The major variable collected details on their participation in the QBMPS, the changes they made while participating, the costs incurred, the benefits they realized and the challenges they faced. Other chain actors interviewed were selected based on recommendations from the cooperative management. They provided information on milk quality management along the dairy chain including the investments and costs incurred which were used in creating Figure 1. Medical doctors, health researchers and public sector stakeholders were selected based on their familiarity with public health issues and the QBMPS. Medical experts shared their experiences and provided information on milk-related diseases and hazards, as well as their severity and treatment. The outcome of the FGDs was transcribed and used in understanding group dynamics issues which could not be captured in the individual farmer interviews, and for understanding group perceptions on costs and benefits of the QBMPS.

Figure 1. Overview of private and public costs and benefits in the calculations (authors’ elaboration).

2.2. Approach for estimating costs and benefits

Both public and private costs and benefits were considered in calculations, as illustrated in Figure 1.

The framework used in estimating various dimensions of public and private costs and benefits for key actors along the dairy supply chain is shown in Figure 1. Private sector actors considered in this study include milk producers, cooperatives (also referred to as CBEs – Collection and Bulking Enterprises) and processors. Farmers who participate in a QBMPS received several trainings, including trainings on production topics such as animal feeding and husbandry, whose effect would give them yield benefits in addition to improving milk quality. These were, however, disregarded as they were too difficult to quantify or their effect in the case study appeared to be minimal. Some of the benefits not considered in this study include, for example, farmers’ time saved from faster milk collection due to shorter waiting time, due to a more efficient collection system; safer milk for farm household consumption; increased milk yield due to better animal feeding; and lower animal health costs due to better animal feeding and farm management. Valeeva et al. (2004) confirmed that though most of these benefits might seem obvious, it is usually challenging to assign a monetary value to them, and there is also missing literature on their clarification. For consumers we concentrated on health costs and benefits.

2.2.1. Calculating private costs and benefits of major chain actors

To calculate the private costs, a real-resource compliance cost method (Valeeva et al., 2004) was applied considering the partial cost approach as recommended by Walker et al. (2019) for situations where several actors share costs and benefits. This analysis focused on the additional and reduced costs and additional and reduced benefits that were incurred by each chain actor from their participation in the QBMPS (Figure 1), compared to a situation where milk payment is based on milk volume only (Annex 2). All costs, both running and investments that were incurred as a result of the introduction of the QBMPS, were considered as additional costs. For farmers who adopted the QBMPS, information on additional costs and benefits was obtained through interviews during farm visits. For the cooperatives and the processor, the actual additional costs were calculated by comparing costs before and during implementation of the QBMPS, expressed per kg of milk. Whenever the difference between additional benefits and additional costs for a chain actor was positive, it was considered that this actor profits from the QBMPS, while (s)he was considered to make a loss when this difference was negative.

2.2.2. Estimating public health costs and benefits

The following indicators were used to estimate the public health costs and benefits of the QBMPS. It is assumed that, once the QBMPS is in place, the future public benefits equal the current public costs. Public health costs for milk-related diseases was taken as estimate for current public costs:

(i) Disease incidence – Data on disease incidence attributed to consumption of poor quality milk were obtained from the Kenya Dairy Board (KDB, 2017) and were expressed as disease incidence cases per 100,000 and cases per 10,000 of the Kenyan population size; considered as 48.46 million inhabitants at the time of this study. Due to lack of reliable information on disease incidences in Kenya at national level and the complications related to attributing these diseases to poor milk handling, we adopted data as estimated by KDB (2017).

(ii) DALYs – Disability Adjusted Life Years indicate the burden of disease across the population and indicate the gap between current health status and the ideal situation where people reach their life expectancy free from diseases or disabilities (Devleesschauwer et al., 2014; Larson, 2013; World Health Organisation, 2020). DALYs were calculated to estimate benefits of reduced health burden due to using a QBMPS system (relative to not using such a system).

DALYs were calculated using the following equations (McDonald et al., 2018):

(a) $DALY=\hspace{0.17em}YLL+YLD$(a)

(b) $YLL\hspace{0.17em}=\hspace{0.17em}N\hspace{0.17em}x\hspace{0.17em}L$(b)

(c) $YLD\hspace{0.17em}=\hspace{0.17em}I\hspace{0.17em}x\hspace{0.17em}DW\hspace{0.17em}x\hspace{0.17em}L$(c)

Where:

YLL  = number of years lost because of (earlier) death

N = number of deaths due to milk-related illnesses per year

L= difference between the life expectancy and the average age at which death occurs due to a particular milk-related illness. Data on N and L was adopted from the Kenya Dairy Board (KDB, 2017), the average life expectancy was based on the World Bank (World Bank, 2017), and the average age at which death occurs was obtained from secondary data and expert interviews.

YLD (Equation (3)(c) $YLD\hspace{0.17em}=\hspace{0.17em}I\hspace{0.17em}x\hspace{0.17em}DW\hspace{0.17em}x\hspace{0.17em}L$(c) ) = is the number of milk-related incidences in a year multiplied by the duration of the illness and a weight factor.

I = the number of annual incidences, which were adopted from (KDB, 2017)

L = the average duration of the case until remission or death in years, and

DW = the disability weight, which reflects the severity of the disease on a scale from 0 (perfect health) to 1 (worst possible health state). The disability weight indicates the proportional reduction in good health due to an adverse health state (Devleesschauwer et al., 2014). Data on the DW was based on previous research (Salomon et al., 2012, 2015). For tuberculosis only, a differentiation was done between HIV-infected patients and non-HIV-infected patients, as the severity for these two types of patients differs and thus affects the disability weights.

(iii) Direct public health costs - Direct costs of being ill “represent the value of goods, services and other resources consumed in providing care due to an illness” (McLinden et al., 2014, p. 2). These costs include basic medical care expenditures, such as for diagnosis, treatment, continuing care, rehabilitation, terminal care and transportation among others (Hodgson & Meiners, 1982; McLinden et al., 2014). To estimate the direct costs of milk-related illnesses, an incidence-based costs approach (Sundström, 2018) was used, which is often used for analyses that aim at calculating the benefits of preventing the occurrence of a new case of the disease (Hodgson & Meiners, 1982).

The direct costs were calculated using the equation below:

Total direct public health costs = incidences of specific illnesses x direct costsper incidence(d)

(iv) Indirect public health costs - Indirect costs account for losses in productivity due to an illness or death (McLinden et al., 2014). These indirect costs were measured using the human capital approach, which is commonly used to measure indirect costs of illness. These losses were calculated by multiplying the total life years lost (DALY) by the average productivity (output per worker) per year as shown in formula (e). The average productivity for Kenya reported as 1455.36 USD (KES 148,447) (World Bank, 2020) was applied.

$Total\hspace{0.17em}indirect\hspace{0.17em}public\hspace{0.17em}health\hspace{0.17em}costs\hspace{0.17em}=\hspace{0.17em}DALYs\hspace{0.17em}milk\hspace{0.17em}related\hspace{0.17em}illnesses\hspace{0.17em}x\hspace{0.17em}average\text{break}productivityper\hspace{0.17em}capita\hspace{0.17em}per\hspace{0.17em}year(e)$

(v) Total public health costs - These were calculated by summing direct and indirect costs for all milk-related illnesses for the entire population of the country.

2.2.3. Health cost reduction scenarios

Three scenarios were defined to compare the impact of increasing coverage of QBMPS and increasing formalisation of the dairy sector on the public health costs for milk-related diseases.

Scenario 1: In the first scenario, it is assumed that half of the current formal marketed milk (representing 7.5% of total production and 15% of total marketed milk)) goes through the QBMPS.

1. Scenario 2: In the second scenario, it is assumed that all of the current formal marketed milk (15% of production, 30% of traded milk) goes through the QBMPS.

2. Scenario 3: In the third scenario, it is assumed that the formally marketed milk volume doubles to 30% of milk produced, 60% of milk traded) and all of it goes through the QBMPS.

In these scenarios, it is assumed that for each percent of milk produced that goes through a QBMPS, incidence rates of milk-related diseases will reduce by 1%, as will total public health costs. The reduction in total public health costs is then expressed as “Health benefit of QBMPS per kg of milk produced” through dividing cost reduction by the amount of milk produced (in kg).

3. Results

3.1. Private costs and benefits

Partial private costs and benefits applying to farmers, cooperatives, and processors as business entities are detailed as follows.

3.1.1. Costs and benefits for farmers

In analysing the additional costs and benefits of the QBMPS, it was assumed that farmers would differ in their level of investment into the piloted QBMPS, and therefore receive dissimilar benefits from it, by attaining a certain quality grade of milk (Table 1).

Table 1. Variation in costs and benefits for farmers involved in the QBMPS#.

	Unit	Additional costs and revenue for different milk grades, per kg of milk
		Grade A	Grade B	Grade C
Milk quantity considered	kg	1.00	1.00	1.00
QBMPS payment	KES	2.00	1.00	0.00
Revenue from forgone milk rejection	KES	1.86	1.09	0.00
Total additional revenue	KES	3.86	2.09	0.00
Additional feed costs (e.g. for mycotoxin binder)	KES	0.15	0.15	0.00
Milking equipment costs	KES	0.08	0.08	0.08
Water costs	KES	0.17	0.08	0.08
Improved housing costs	KES	0.53	0.53	0.00
Additional time for cleaning equipment and attending trainings	KES	0.62	0.33	0.04
Total additional cost	KES	1.55	1.17	0.20
PROFIT/LOSS (net additional cost)	KES	2.31	0.92	−0.20
PROFIT/LOSS if milk rejection is not considered (net additional cost)*	KES	0.45	−0.17	−0.20

^#Calculated using data from the survey of 90 farmers.

*Considering that farmers might sell rejected milk to less stringent buyers.

Using the market price at the time of the study (35 KES/kg; USD 0.343/kg), a farmer would incur additional costs of 1.55 KES/kg (USD 0.0125/kg) of milk in investments in order to continuously meet the standards for a premium payment of + 2 KES (Grade A milk). The bonus of + 2 KES would also bring the milk price level to be comparable with the price offered by milk vendors from the informal sector. The same farmer would also get an additional 1.86 KES/kg as revenue from forgone milk rejection, resulting in a total additional profit of 2.31 KES/kg of milk. If the forgone milk rejection is not considered, as in the case where farmers can sell their rejected milk to buyers with less stringent quality, the profit is reduced to 0.45 KES/kg. Forgone milk rejection is considered as milk which could have been rejected due to poor quality, but is now of acceptable quality due to improved hygienic practices at farm level. If milk of poor quality is detected by traders, transporters and cooperatives, it is returned to the farmer. However, once poor quality milk is detected by the processor, it is discarded and not returned to the seller.

A farmer (from a group of 5–10 farmers, together filling a 50-litre milk can) producing Grade B milk incurs an additional cost of 1.17 KES/kg in investments and gets a benefit of 2.09 KES/kg. Because there is no extra payment for farmers with Grade C milk, they make a net loss of 0.20 KES/kg of milk due to costs they incur in order to be paid under the QBMPS scheme.

3.1.2. Costs and benefits for processor and cooperatives

3.1.2.1. Costs

A breakdown of cost per kg of milk for various investments in the QBMPS made by the processor and the cooperatives is presented in Figure 2. considering a situation where Grade B milk is delivered. Meanwhile, the costs for grade A and C milk are shown in Table 2. The processor spends an average additional 3.05 KES and the cooperatives 0.56 KES per kg of milk that goes through the QBMPS.

Figure 2. Estimates on additional costs for processor and cooperative.

Table 2. Summary of private costs and benefits (in KES per kg of milk).

Grade	A	B	C
Farmers
Costs	1.55	1.17	0.20
Benefits	3.86	2.09	0
Profit/loss	2.31	0.92	−0.20
Cooperatives
Costs	0.56	0.56	0.56
Benefits	0.32	0.19	0.00
Profit/loss	−0.24	−0.37	−0.56
Processor
Costs	3.09	3.05	3.02
Benefits	0.93	0.54	0.00
Profit/loss	−2.16	−2.51	−3.02

About 40% of the processor’s total costs were used for consumables in the laboratory, 20% were used for depreciation of hardware (milk analyser, refrigerator, alcohol guns, etc.) and about 15% each for training of farmers and benefits for project staff. Only 8% of the total cost was used to make bonus payments to farmers. For the cooperatives, 90% of costs were for additional staffing, while the other 10% were almost equally distributed between laboratory consumables and software development.

3.1.2.2. Benefits

Milk rejected by the processor is not paid for and farmers with better quality milk have fewer cases of milk rejection. For every additional litre of milk collected by the cooperative that would otherwise have been rejected due to poor quality, the cooperative receives about 6 KES. The costs and benefits of farmer, cooperative and processor due to participation in the QBMPS are summarized in Table 2. If all milk collected were Grade B milk, the cooperative would make an extra benefit of 0.19 KES/kg of milk after incurring a cost of 0.56 KES/kg milk, giving it a net loss of 0.37 KES/kg milk.

Reduced product losses due to better fermentation of yoghurt and cheese as well as longer product shelf life result in an average benefit of 0.93 KES/kg milk received (Table 2). The table shows that cooperatives and processors jointly have a net additional cost of about 3 KES/kg milk, mainly due to the significant initial costs of laboratory consumables, additional staffing and training of farmers.

3.2. Public health costs and benefits

3.2.1. DALYs (Disability Adjusted Life Years)

The incidences of milk-related infectious diseases per year and their DALYs are presented in Table 3. The table shows that the impact of listeriosis, brucellosis and E. coli are especially substantial. Although salmonellosis occurs almost as frequently as brucellosis, the former has a low mortality rate and only a short duration, leading to fewer healthy life years lost from salmonellosis annually compared to Brucellosis. In total, an estimated 53,093 healthy life years are lost annually in Kenya due to milk-related infectious diseases, considering the average productivity of 1,455 USD (World Bank, 2020).

Table 3. Estimated annual costs of milk-related health hazards in Kenya.

Illness	Incidence (cases/year)*)	DALY (years)	Cost per case (million KES) *)	Total direct costs (million KES)	Total indirect costs (million KES)	Total costs (million KES)
Tuberculosis	3,392	16,045	0.0224	76.12	2,381.85	2,457.97
Brucellosis	28,107	19,259	1.2078	33,944.26	2,858.92	36,803.18
Listeriosis	8,238	3,521	43.0183	354,384.72	522.67	354,907.39
Salmonellosis	26,653	563	0.4355	11,608.45	83.59	11,692.03
E. coli infections	23,745	2,089	1.0009	23,766.99	310.17	24,077.16
Campylobacteriosis	2,423	10,694	0.1641	397.66	1,587.54	1,985.20
Q fever	1,890	922	0.0010	1.93	136.86	138.79
Antibiotics			-	4,346.51	-	4,346.51
Aflatoxins			-	8.73	2.49	11.22
Total		53,093		428,535.37	7,884.09	436,419.45

Exchange rate: 1 USD = 102 KES;

*)Source: KDB (2017) : The rest of the data is based on authors‘ calculations.

3.2.2. Direct and indirect health costs

Total direct and indirect health costs are estimated at 436 billion KES/year (Table 3). This is an underestimation of actual costs, as costs due to hydrogen peroxide adulteration, a potential cause of cancer and gastrointestinal disorders, are not included. The highest total costs for milk-related health hazards were due to listeriosis, followed by brucellosis and E. coli.

3.2.3. Health cost-reduction scenarios

The three scenarios for the extent to which public health costs could be reduced due to reduced incidence rates of milk-related illnesses are shown in Table 4. With reductions in incidence cases, public health costs also decrease. In the 15% reduction scenario, if all formally traded milk in Kenya would go through a QBMPS, a health benefit of 13 KES per kg of milk produced would be achieved from the reduction of incidences of milk-related illnesses. In case half of the current formal milk would go through the QBMPS (7.5% scenario), avoided costs would be 32.7 billion KES per year or circa 6.5 KES per kg of milk produced. In the 30% scenario (doubling of formal milk trade, all under QBMPS), avoided costs would be 130.9 billion KES or 26.2 KES per kg of milk produced.

Table 4. Health cost-reduction under three scenarios (million KES).

	Current	7.5% reduction	15% reduction	30% reduction
Total direct costs	428,535	396,395.22	364,255.06	299,974.76
Total indirect costs	7,884	7,292.78	6,701.48	5,518.86
Total costs	436,419	403,687.99	370,957	305,494
Avoided costs	-	32,731	65,463	130,926
Avoided costs (KES per kg of milk produced)		6.55	13.09	26.19

4. Discussions

Assuring food safety in the dairy sector entails significant costs, incurred by different actors along the supply chain. These high costs have also been found in other studies where innovations are being introduced to assure food safety such as Viator et al. (2017) in the US and food crop improvement (Wesseler et al., 2017) in Kenya. Individual (private sector) actors will only incur such costs if these are offset by significant benefits to their business. In developing countries, private actors along food value chains including smallholders need to be incentivised to deliver safe food products through premium prices relative to other alternative markets they would engage in, for example informal channels. This is coupled with cost of compliance that is lower than the benefits accrued. Furthermore, allocating resources to support capacity building to the different value chain actors rather than a punitive approach to meet food safety can also be an incentive (Hoffmann et al., 2019). Understanding these costs and benefits is necessary to incentivize and guide these actors to make the necessary investments. The findings show differentiated costs and benefits of quality assurance for the major dairy chain actors, as associated with preventing the incidence of milk-related diseases.

4.1. Private costs and benefits

Heterogeneity in the adoption of techniques and practices by farmers led to a range of returns to quality assurance from the farmer perspective, illustrated by the three quality levels, Grades A, B and C. The results show that smallholder producers who consistently invest in milk quality (producing Grade A milk) obtain the highest benefits from a QBMPS. Meanwhile, farmers who are hesitant and inconsistent in their investments in milk quality (producing Grade B and C milk) might not benefit from participating in the QBMPS. The benefit for farmers producing Grade A milk increases if forgone milk rejection due to poor quality is considered. However, this calculation has been done separately because under the Kenyan conditions farmers who sense quality issues with their milk could as well sell it in alternative markets with less stringent quality requirements. Also, milk rejected at the cooperative level and returned to the farmer could still be sold to less stringent markets. Nevertheless, whenever milk of unacceptable quality reached the processor, it was discarded and the cooperative and farmers were not paid for it. With the relatively low incentive level and availability of alternative, less stringent markets, it can be concluded that the QBMPS may be more effective if the same milk rejection regulations were applied by other buyers, or if the price incentives were higher.

Farmers also benefited from several trainings on milk hygiene, animal husbandry and feeding, which may have contributed to higher production next to improved milk quality. These benefits were not investigated in this study. A study by Nyokabi et al. (2021) identified several poor milk handling practices at farm level including, milking cows in open environments, using calves to stimulate milk let-down, using same cleaning water and towels for many cows, not dipping teats, use of plastic containers for milk storage, poor cleaning of animal housing and many others, which if addressed in such farmer trainings would contribute greatly to quality improvement. The piloted QBMPS also encouraged grouping of farmers, which, on the one hand, gave an opportunity to some farmers with poor quality milk to free ride on others in the group with better quality milk, but, on the other hand, made striving farmers to lose their chance of getting a bonus if their group had farmers with poorer quality milk. Also, grouping was seen to strengthen both horizontal and vertical integration along the chain and make the chain more robust. The farmer to farmer and the cooperative to cooperative (horizontal) bonds were strengthened and at the same time the bond across farmers and the cooperatives (vertical). Mwambi et al. (2020) found that Kenyan smallholder farmers in groups or cooperatives were more likely to adopt food safety measures than their counterparts that did not belong to a group or cooperative. In this study it was noticed that sometimes when group milk was of unacceptable quality, the farmers could trace the responsible group member and discourage him/her from poor practices. It would, however, be fairer if the QBMPS tests individual milk samples and pays each farmer based on this. In smallholder systems, this can only be achieved if cheaper tests are used and/or fewer parameters are tested. This could be more attainable if all milk buyers were obliged to test for milk quality, which would create equal opportunities to engulf the current cost disadvantage incurred by buyers who test milk before buying.

Some changes beyond the farmer level could certainly contribute to milk quality improvement. For example, milk transporters, especially those transporting milk from collection points to cooperatives, could contribute to milk quality. This is important because transporters are paid per litre of milk they deliver and would strive to increase their milk volumes in order to increase their income. In this pilot study, milk transporters were trained to observe quality issues, and costs for their trainings were included in the processor’s costs. The additional costs of the transporters were considered to be negligible, though it was recommended for them to also receive financial incentives to encourage them to become more punctual in milk collection and delivery.

Compared to farmers, processors and cooperatives make significantly higher investments, especially at the onset of the QBMPS. This makes them incur a net loss from participating in the QBMPS. Some of these losses could be avoided, by externally funding some one-off investments and by technological innovations that lead to cheaper milk testing options and efficiency gains in quality assurance (Ndambi et al., 2018; Poonia et al., 2017).

Consumers could pay higher product prices to cover the additional costs incurred by cooperatives and processors if they would understand the health risks associated with consumption of poor quality milk and are willing to pay for better quality milk.

4.2. The public good dimension of supporting a QBMPS

Looking at the public health side, in this study, the estimated total of 53,093 healthy life years lost annually in Kenya due to milk-related infectious diseases is a call for public concern, as was also concluded by Nyokabi et al. (2021) in a study on milk quality along the dairy chain in Kenya. Considering an average lifespan of 62.13 years (World Bank, 2017), this gives an average annual loss of 855 full lives due to milk-related infectious diseases, which can be avoided by improving milk quality, in addition to the total health cost of over 436 billion KES incurred annually due to milk-related health hazards. These results confirm the suggestions of Özkan Gülzari et al. (2020), where the importance of QBMPSs were seen vital in assuring milk quality in East Africa. These figures provide a strong justification for sustained public investments to support the introduction of QBMPS but also to ensure a functional enabling environment through a regulatory framework that has sufficient enforcement capacity and that instils confidence in the private sector to innovate and invest in quality assurance systems. Together, these play a key role in creating a level playing field for the various actors in the dairy chain (Ndambi et al., 2019).

4.3. Government action to support and regulate milk quality and safety

According to Ndambi et al. (2019), the Kenyan dairy sector is controlled by a number of regulatory frameworks: (i) Milk safety and animal feed safety is principally governed by the Dairy Industry Act (Cap. 336) and the Public Health Act (Cap. 242), (ii) Animal Diseases Act (Cap. 364), (iii) Fertilizer and Animal Feedstuffs Act (Cap. 345) and (iv) Standards Act (Cap. 496). They also highlighted challenges with compliance to these regulatory frameworks. To address these regulatory gaps, the Kenya Dairy Board (KDB) developed the Kenya Dairy Industry Regulations, which were to replace the Dairy Industry Act (Cap. 336), from February 2021. The act covers four main sections: preliminary, registration of primary producers, licencing of dairy business operators, regulatory permit and consumer safety levy. Muunda et al. (2021) stipulate that these regulations are likely to lead to increased milk price, hence reduced consumption, especially for low-income consumers.

4.4. Limitations in the study scope and methodology

This study attempted to calculate the costs and benefits of a QBMPS based on a pilot study by the processor Happy Cow Ltd. in Nakuru, Kenya. Contextual and methodological limitations encountered include:

• Calculation of processor benefits only consider yoghurt and cheese as final products, since these were the main products processed at time of study. Considering other dairy products like pasteurized milk and UHT may alter the current results.

• Calculated benefits do not consider (i) trade benefits that might arise due to access to new markets as a result of selling better quality dairy products; (ii) productivity benefits that farmers participating in the QBMPS get from trainings; (iii) environmental benefits arising from reduced loss (waste) of milk, more productive cows, healthier cows and a more efficient dairy industry (faster collection and transportation).

• Costs required for regulation, for example, enforcement of minimum standards of aflatoxins in feed, have not been considered.

• In calculating health costs, for the number of years lost due to an earlier death, “the average age at which death occurs” is compared with “the average life expectancy”. However, if mortality is selective, the average life expectancy may not be representative for the people that die, and this might affect our estimates.

• The public health costs might have been underestimated considering that some diseases might not be identified or not reported as some sick people might not seek medical assistance.

• On the other hand, the contribution of post-purchase behaviour of consumers, for example milk boiling, has not been considered. Increased awareness on better post-purchase practices might also reduce health hazards related to harmful microbes in milk.

• Similarly, the impact of antimicrobial residues and aflatoxins in milk on human health is controversial in literature and hence our results should be interpreted with care.

5. Conclusions and recommendations

This study indicates that implementing a QBMPS provides a potential to address milk safety and quality issues by incentivizing private sector to make necessary investment and contribute to public health by reducing milk-related diseases. The QBMPS leads to improved revenues for committed farmers who achieve the required of quality grade of milk within the formal market. On the other hand, processors and cooperatives, the initial higher investments necessary for QBMPS, result in a net cash loss from the system at the onset, though subsequently the benefits may accrue. These net loss points to the need for more support to these actors as incentives for implementation of the QBMPS. Thus undistributed benefits of such systems require measures that ensure that all actors benefit. The large public health cost savings suggest that public investment in QBMPS at the onset would be justified and would be valuable in nurturing this market innovation.

Considering that food safety is a public concern, a more inclusive and nationally effective approach to milk quality assurance needs to be considered to create a more level playing field. Effective implementation of QBMPS requires a stronger commitment to milk quality assurance among involved stakeholders, including processors, cooperatives, farmers, and transporters. This requires that regulatory authorities create an enabling environment for milk quality improvement by regulation and enforcement of required quality standards. Such enforcement should ensure that initiatives to improve milk quality along the dairy chain are not hampered by unfair competition from other actors who do not comply with required quality procedures. Streamlining the dairy sector towards a formal sector is essential, since quality assurance can only be implemented in conditions where traceability is possible and sale of sub-standard milk is penalized. This study shows that introducing food quality systems require upfront investment. Perhaps suitable models of public–private partnerships could be created for such investments.

Based on the findings from the pilot, we recommend that the QBMPS starts off with fewer test parameters to lower the initial cost. Moreover, investing in development of cheaper tests could reduce running costs and could be applied to individual suppliers, which would reduce free-riding behaviour. Processors will need to provide a more attractive price incentive to farmers and other chain actors, to create a quality brand product and to sell products at a higher price that compensates for the additional quality assurance costs. This will, however, only attract consumers who are able and willing to pay for more expensive and better quality milk. There is need to also better understand the effect of the informal market that offers competitive pricing to the formal channel that would be the target of the QBMPS. The public good dimension of food safety illuminates the need for targeted public investment that are forward-looking to stem off the risks associated with unsafe food. We recommend fine-tuning QBMPS in the Kenya dairy sctor, and other similar sectors in developing countries to be a more targeted preventive investment that can incentivize and leverege more actors in the formal and informal market to consider making investments for a safe and quality sector development.

Acknowledgments

The authors are grateful to the staff of Happy Cow Ltd Kenya and its suppliers, ILRI Kenya, KEMRI, SNV Kenya, University of Nairobi Kenya, Wageningen University Netherlands, Kenya Dairy Board, and to the medical doctors and medical research officers interviewed during this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was conducted under the 3 R Kenya project, funded by the Embassy of the Kingdom of the Netherlands in Nairobi, Kenya, as part of the Agriculture and Food & Nutrition Security program.

Annex 1.

Parameters used in the pilot QBMPS and KEBS¢ standards

Table

Test	Grade	QBMPS standard‡	KEBS^¢ standards	Premium/penalty Score^a
Total plate count(units in cfu/ml)	A	0–2,000,000	<200,000	50
	B	2,000,001–10,000,000	200,001–1,000,000	0
	C	>10,000,000	≤ 2,000,000	−50
Antibiotic residues	All	Negative	Negative	15*
Freezing point	All	−0.500	−0.525 to−0.565	20#
Total solids	All	>11%	>11.75%	15#

Note: ¢KEBS = Kenya Bureau of Standards.

^aPremium or penalty score given to milk of the corresponding QBMPS standard (column 3); *milk positive for antibiotic residues is discarded; ^#otherwise a 0 score; ‡Source: Happy Cow Ltd..

Based on HC criteria, Grade A (premium) milk would have a score between 70 and 100 and would receive a bonus of 2 KES/kg of milk above the base price, Grade B (standard) milk would have a score between 40 and 69 and would receive a bonus of 1 KES/kg of milk above the base price, while milk with a score of below 40 would be rejected or get the base price. Rolling out the QBMPS required the various actors to make some investments which in turn would result in positive returns.

Annex 2.

Partial cost analysis to estimate profit and loss for individual actors

Profit and loss was calculated as:

(1) $\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{i}\mathrm{t}/\mathrm{l}\mathrm{o}\mathrm{s}\mathrm{s}=\mathrm{A}\mathrm{d}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{b}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{i}\mathrm{t}\mathrm{s}-\mathrm{a}\mathrm{d}\mathrm{d}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{s}$(1)

Empirically this model was specified as outlined below:

(2) $\pi =\sum _{i=1}^{i=n}{P}_{y}Y-\sum _{i=1}^{i=n}{P}_{x}X$(2)

Where: $\pi$ = Profit or loss

$P_y$ = Price of kg milk equivalent sold by the actor

$Y$ = Quantity of milk sold

$\sum _{i=1}^{i=n}{P}_{y}Y$ = Total additional benefit from sales of milk and milk products

$P_x$ = Price of additional inputs (1, 2, 3 … .n)

$X$ = Quantity of additional inputs (1, 2, 3 … … n)

$\sum _{i=1}^{i=n}{P}_{x}X$ = Total costs incurred by the actor

Equation (22) illustrates the situation the actors faces in practice anyhow. In order to identify the additional costs and benefits of the QBMPS pilot, the study considered those costs and benefits that yield additional profit or loss to the actor:

$\mathrm{\Delta }\pi =\mathrm{\Delta }\sum _{i=1}^{i=n}{P}_{y}Y-\mathrm{\Delta }\sum _{i=1}^{i=n}{P}_{x}X$ (3)

Where:

$\mathrm{\Delta }\pi$ = Additional profit/loss made from participation in the QBMPS

$\mathrm{\Delta }\sum _{i=1}^{i=n}{P}_{y}Y$= Additional revenue made by the actor in the QBMPS

$\mathrm{\Delta }\sum _{i=1}^{i=n}{P}_{x}X$= Additional costs incurred by the actor in the QBMPS

The partial cost and benefit analysis was broken down as:

$\frac{\mathrm{\Delta }\pi }{Y}=\frac{\sum _{i=1}^{i=n}{P}_{y}Y}{Y}-\frac{\sum _{i=1}^{I=n}{P}_{x}X}{Y}$

Where: $\frac{\mathrm{\Delta }\pi }{Y}=$ Additional profit/loss per kg of milk made due to participation in the QBMPS

$\frac{\sum _{i=1}^{i=n}{P}_{y}Y}{Y}$ = Additional revenue per kg of milk made by the actor in the QBMPS

$\frac{\sum _{i=1}^{I=n}{P}_{x}X}{Y}$ = Additional costs per kg of milk incurred by the actor in the QBMPS

The model was specified as follows:

For the farmer:

(5) $\begin{array}{rl}& \pi =(BONUSPAYMENT+LESSMILKREJECTION)\\ & -(\mathrm{\Delta }FEEDCOST+\mathrm{\Delta }MIKEQUIPMENTCOST+\mathrm{\Delta }WATERCOST+\mathrm{\Delta }HOUSINGCOST\\ & +\mathrm{\Delta }CLEANING+\mathrm{\Delta }TRAIN)\end{array}$(5)

For the cooperative:

(6) $\pi =(LESSMILKREJECTION)-(STAFFCOST+SOFTWARECOSS+CONSUMEABLES)$(6)

For the processor:

(7) $\begin{array}{rl}& \pi =(YIELDGAINSINCOME+LESSRETURNS+LESSMISPRODUCTION)\\ & -(STAFF+TRAINING+SOFTWARE+CONSUMEABLES+BONUSPAY\\ & +HARDWARE)\hspace{0.17em}\end{array}$(7)

Fixed costs

A straight-line annual depreciation was calculated for fixed investments. An allocation factor was used to assign each investment to the QBMPS based on the proportion of milk that was assigned to the QBMPS.

Specific calculations for farmers

Three levels of milk quality were considered: Grades A, B and C, Grade A being the premium quality of milk that attracts a 2 KES bonus payment, Grade B being moderate quality that attracts a 1 KES bonus and Grade C being milk of acceptable quality that attracts no bonus.

The revenue from forgone milk rejection considers farmers’ benefits due to reduced rejection of milk by the processor. It should be noted that all milk rejected by the processor is discarded. Based on data from the processor, it was estimated that farmers supplying Grade A milk can reduce milk rejection rates to 0.5% compared to 5.8% for those supplying Grade C milk. This reduction rate was applied to the average daily sales per farm to calculate the economic loss of farmers due to rejection of poor quality milk.