Save

Multi-level marketing system with industrial food processing in Africa: smallholder tomato farmers in Tanzania

In: International Food and Agribusiness Management Review
Authors:
Nasibu Yusuph Mwinyiheri Graduate Student, College of Economics and Management, Fujian Agriculture and Forestry University Fuzhou, Fujian 350002 China

Search for other papers by Nasibu Yusuph Mwinyiheri in
Current site
Google Scholar
PubMed
Close
,
Victoria Salin Professor, Department of Agricultural Economics, Texas A&M University College Station, TX 77843 USA

Search for other papers by Victoria Salin in
Current site
Google Scholar
PubMed
Close
,
Yu Yvette Zhang Professor, Department of Agricultural Economics, Texas A&M University College Station, TX 77843 USA

Search for other papers by Yu Yvette Zhang in
Current site
Google Scholar
PubMed
Close
,
Benxi Lin Professor, College of Economics and Management, Fujian Agriculture and Forestry University Fuzhou, Fujian 350002 China

Search for other papers by Benxi Lin in
Current site
Google Scholar
PubMed
Close
, and
Chileshe Chewe Graduate Student, College of Economics and Management, Fujian Agriculture and Forestry University Fuzhou, Fujian 350002 China

Search for other papers by Chileshe Chewe in
Current site
Google Scholar
PubMed
Close
Open Access

Abstract

There are few challenges greater than the puzzle of how to move perishable goods from smallholders’ fields to final consumers, particularly where transportation barriers abound. Industrial processing can relax the perishability constraint and improve smallholders’ options. In Iringa, Tanzania, multiple tomato processing establishments and traditional marketing agents are available for farmers to use. Information about the channels is expected to be vital for producers to choose optimally. In this study, we collected field data from 286 smallholder farmers and analyzed their market channel choice using the random utility model implemented with multinomial logit regression. Revenue performance was further evaluated. Our results confirm that access to market information and extension services were associated with higher farm revenue. Women farmers had more concerns than men about lacking information. Market participation differed by size of the business. Farmers with higher harvested quantity were more likely to choose industrial processing. Producers with small farms preferred full marketing service and mid-size farms chose wholesale markets. Agribusiness management scholars may follow similar research design with future attention paid to elicitation of producers’ information about the marketing channels.

1. Introduction

Transformation of the smallholder agriculture sector from subsistence farming to profitable business has been the key objective of development specialists in government and the private sectors (Dome and Prusty, 2016). Smallholder farmers’ market participation is deemed important to economic growth and poverty reduction policy (Barrett, 2008; Collier and Dercon, 2014; Jayne et al., 2010). As expressed by the president of the African Development Bank, Dr. Akinwumi Adesina (CNBC Africa, 2020),

“Agriculture is the most important profession and business in the world. The size of food and agriculture in Africa will rise to $1 trillion by 2030. The population of Africa, now at 1.2 billion, will double to 2.5 billion by 2050. They all must eat. and only through food and agribusiness can this be achieved.”

Tomato is the most popular and widely consumed among vegetables globally (Anang et al., 2013; Ugonna et al., 2015). It is an important source of minerals, vitamins and healthy acids and a crop that requires a very short duration (3–4 months) to maturity with high yields (Dome and Prusty, 2016). Therefore, tomato production is an important economic activity for smallholder farmers and provides an opportunity for poverty reduction (Ugonna et al., 2015; Nyamba et al., 2016; Maspaitella et al., 2017; Mutayoba, 2018). However, tomato production is challenged by small-scale farmers’ lack of access to markets in sub-Saharan Africa with significant loss of food crops post-harvest due to poor storage systems, poor transportation and lack of processing enterprises (Ugonna et al., 2015). In this paper, we study the marketing system of tomato in a district of Tanzania, where tomato is one of the most widely cultivated horticultural crops, the production is dominated by small-scale producers (Dome and Prusty, 2016; Tanzania Horticultural Association (TAHA), 2011), and there are multiple processing establishments.

In the Iringa region of Tanzania, multiple marketing channels for an estimated 60 000 tomato farmers coexisted: specifically, the industrial channel, direct-to-consumer, using wholesalers, or using assemblers (marketing agents). The Iringa region is unique in that a tomato processing company, Darsh Industries Ltd, entered the market in 2015 as part of a regional economic development effort. Two other large-scale processing firms, Dabaga Vegetable and Fruit Canning Company Ltd. and Iringa Vegetable Oils and Related Industries (IVORI) Ltd., were already sourcing and processing tomatoes in the region. In addition, small-scale food producers and retailers buy the fresh produce. There are farmers associations, market centers, government agricultural officers, village leaders and buyers from processing companies interacting in the marketing system.

The active participation of smaller holder farmers in the marketplace is vital for stimulating economic growth and alleviating poverty in Tanzania. For smallholder tomato producers, who face challenges due to the perishable nature of their product and have access to various market outlets with different revenue implications, it is imperative to identify the option that guarantees them maximum rewards. Therefore, having timely and accurate market information is essential for farmers to make informed decisions and improve their profitability. The objectives of this study are to improve marketing systems by identifying the factors that affect smallholders’ choice among formal and informal market outlets, and to investigate whether choice of the market channel influences revenue outcomes for the farmers. Our study investigates differences among farmers who use the coexisting marketing alternatives. We describe the drivers of choice and the variation in engagement with the market alternatives based on an extensive survey of smallholder farmers. One key element of our research is to understand the importance of the institutions that provide better market information to farmers. There are three hypotheses: (1) Information, knowledge, and trust are key factors in the market channel choice; (2) social and demographic composition of the household are significant factors constraining the more formal channels; and (3) revenue outcomes are higher for producers who market to the industrial channel. Because we had access to data on the farmers’ characteristics, market information and advisory services, access to transportation, and farm revenue, results about the outcomes for farmers can be obtained.

2. Related literature

Marketing systems in rural and peri-urban Africa are often characterized by micro-entrepreneurs and smallholder farmers (Etienne et al., 2019; Goldsmith, 1985; Ingenbleek, 2020). Both formal and informal exchanges are meaningful to the smallholders. The flexibility of the marketing system has been an emergent theme from the studies that investigated barriers to marketing and responses to those barriers, and whether impediments to marketing impact differently by age, education, or gender. There is vast evidence that the formal economy in sub-Saharan Africa has rigidities due to transaction costs, bureaucracy, corruption, or lack of knowledge among the traders (Kahiya and Kadirov, 2020).

Farmers will find commercial opportunities if there are institutions and service providers to surmount the difficulties of bringing goods to the market, including foods that are highly perishable. The marketing challenges are so great that marketing specialists are necessary. Those agents, known locally as “market women” or “middlemen,” are often blamed for inefficiency and opportunism in dealing with smallholder farmers (Mwagike and Mdoe, 2015; Tembachako et al., 2015). In some instances, some farmers in developing countries can access higher-end markets and receive higher returns by contracting with a large agribusiness firm (Barrett et al., 2012; Wang et al., 2014).

Marketing systems for fresh agricultural produce are particularly vulnerable to transportation barriers, thus raising the importance of institutions that provide better market information to sellers (Khasa and Msuya, 2016; Mwagike, 2015; Mwagike and Mdoe, 2015; Nyamba et al., 2016). Good intelligence about the prices being paid or quantities needed will save on food loss and conserve the sellers’ valuable time. Research has also documented the importance of collectivist culture and networks in sub-Saharan Africa for cross-border trade and in the success of women business owners (Ntseane, 2004). Some of the socially derived patterns are artifacts of patriarchy, while family connections and female mentors were found to support women in business (Ntseane, 2004).

A number of researchers studied multi-level food marketing systems in East Africa using approaches similar to the survey used in this study. For example, in Ethiopia, survey participants included smallholder farmers, wholesalers from central markets, retailers, processors, and exporters (Temesgen et al., 2017), thus providing information from a number of agents in the business-to-business exchanges in the marketing system. Dome and Prusty (2016) estimated food loss at five urban vegetable markets in Tanzania, contrasting the outcomes for wholesalers at large central markets with those for retailers selling from temporary stalls at minimarkets.

In those quantitative choice models, there was little connection discovered between the selection of marketing channels and farm revenue, a gap that is addressed in this study. Many studies have investigated how smallholder farmers’ market participation was impacted by characteristics of the household or by policies, such as tax or subsidy, education and training, or other rural development policies that were intended to strengthen a value chain (e.g., Bellemare and Barrett, 2006; Jari and Fraser, 2009; Schipmann and Qaim, 2011). For example, different gender categories played different roles in the tomato value chain (Khasa and Msuya, 2016). Men and women smallholder farmers often have different access to production resources and information (Agunga et al., 2018; Davis et al., 2012; Isaya et al., 2018; Quisumbing and Pandolfelli, 2010).

Other factors found in the previous literature to be significantly related to farmers’ participation in the formal channels included: education (Magogo et al., 2015); cultivated area (Tura and Hamo, 2018); produce price (Zivenge and Karavina, 2012); distance from farm gate to the market (Magogo et al., 2015; Ola and Menapace, 2020; transportation (Negi et al., 2018); access to information (Aker and Fafchamps, 2015; Birthal et al., 2015; Jensen, 2007; Goyal, 2010; Minten, 1999; Nakasone, 2013; Negi et al., 2018); extension services (Abdul-Rahaman and Abdulai, 2018; Due et al., 1997; Isengildina et al., 2006; Mmbando et al., 2015; Rahman, 2003) and membership in farmers groups (Abdul-Rahaman and Abdulai, 2018; Barrett, 2018; Lemeilleur and Codron, 2011; Fischer and Qaim, 2012; Xaba and Masuku, 2012; Hao et al., 2018). These studies use different marketing channels and various explanatory variables making it hard to generalize the results, but they provide implications about important factors that affect farmer choices of marketing channel.

Farmers of any scale must be concerned with the requirements of the consumer. When consumers are distant from the grower, formal standards for quality, such as grades, are part of the producers’ requirements and sometimes serve as barriers to successful participation in a marketing system. Research on agricultural value chains originating in Africa and serving consumers in Europe emphasized the high quality standards and the collective action required to meet the standards for exporting (Okello, 2015). The marketing system that Okello studied is resonant of Ingenbleek’s observations of North-South “biogeography” on the African continent (Ingenbleek, 2020). Unlike the extended cross-border marketing systems that Ingenbleek and Okello described, our study is of a marketing system intended to become more formalized from the entry of processing companies.

3. Market area and institutions

The study area is located in the Ilula Township of Kilolo district, which is one of the four districts of the Iringa Region of Tanzania. Agriculture plays a huge role in the economy of the Kilolo District, contributing to 81% of the Gross Domestic Product and providing employment to more than 80% of the population (URT, 2019a,b). Presently, the Kilolo district population is 263 559 whereby 48.6% are males and 51.4% are females (URT, 2022a,b). Tomato is an important cash-earning commodity for the entire Iringa region whereby In Mainland Tanzania, the Iringa region had the largest area planted with tomato (9390 ha; 19.1%), followed by Dodoma (6054 ha; 12.3%) and Tanga (3533 ha; 7.2%) (URT, 2021). Furthermore, the region had the highest production of tomato (50 358 tons; 15.8%), followed by Dodoma (44 600 tons; 14.0%) and Dar es Salaam (28 631 tons; 9.0%) (URT, 2021). The market extends to the major metropolitan port city, Dar es Salaam, which is more than 400 kilometers from the farms and is the final destination for around 50% of the tomatoes grown in the Iringa region (URT, 2013).

The research team responsible for field work in our study was part of an active rural development initiative focused on the tomato value chain. Similar industrial processing zones are common in emerging market economies (Ahmed et al., 2016; Maspaitella et al., 2017; Mmbando et al., 2016). There were three tomato processing firms in the region — Dabaga Vegetable and Fruit Canning Company Ltd. established in 1979, Iringa Vegetable Oils and Related Industries (IVORI) Ltd. founded in 2006, and Darsh Industries Ltd. that started operation in 2015, three years prior to the data collection. The industrial processing zones were the outgrowth of policies to promote locally owned manufacturing with associated job creation. With three processing firms buying tomatoes, the farmers had an alternative to selling the crop that was not available in all the rural areas in sub-Saharan Africa.

If the grower did not deal with the industrial buyers, the produce that was sold for fresh consumption could be marketed through assemblers, wholesalers, or be sold directly to consumers in roadside stalls (Figure 1). There were distinct service levels along the farm-to-market continuum for those alternative channels. The assemblers were market agents who were very familiar with the market area and they negotiated prices, organized transport, and completed all the transactions beyond the farm gates. Other functions that assemblers may have performed include aggregating supply from multiple farmers, temporarily storing tomatoes at markets, or distributing products to multiple wholesalers at some distance from the grower location.

Alternatively, farmers could take on more of the responsibility for marketing functions and sell the produce to wholesalers. Growers dealing with wholesalers were responsible for the cost of transportation to the wholesale market. Wholesalers in the study area purchased tomatoes at formal market centers: Tanzania Social Action Fund (TASAF) and Mlamke markets. These two market centers were located close to the main road that connects Ilula with Dar es Salaam (Figure 2). While it was not common in this region, farmers could sell directly to households in roadside stands.

Figure 1.
Figure 1.

Marketing system for tomatoes, Ilula, Tanzania. Prepared with BioRender.

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Figure 2.
Figure 2.

Roadway distance from Ilula to urban market. Source: Map, Google Earth Pro. Population and population statistics are from the 2012 Census General Report: 2012 Population and Housing Census published by the Development Partners Group in Tanzania for the Unite Republic of Tanzania (http://www.tzdpg.or.tz/fileadmin/documents/dpg_internal/dpg_working_groups_clusters/cluster_2/water/WSDP/Background_information/2012_Census_General_Report.pdf)

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

4. Survey methodology

This study was conducted in 2018 with respondents drawn randomly from lists of tomato smallholder farmers obtained from the Ward Agricultural Officers as well as the Village Executive Officers. Two wards, Ilula and Nyalumbu, were purposely selected due to the large number of tomato growers in the households. The stratified random sampling technique was applied to obtain the proportionate balance of tomato smallholder respondents in the study area. Random selection of farmers within these wards gave equal probability of respondents’ selection and maintained heterogeneity of the study population across village locations. From each village list, 5% of farmers were drawn randomly to participate in the survey. A total of 290 farmers in Masukanzi, Madizini, Lugalo and Ikokoto villages were randomly selected and 286 of them responded to the survey with 70, 65, 80 and 71 respondents, respectively, in each of those villages.

Prior to launching the farmer surveys, the field research team conducted 10 key informant interviews of government and industry representatives to help develop the survey questionnaire and choose the study area. The private sector key informants were two marketing officers from processing companies, Dabaga Ltd. and Darsh Industries, producer of the Red Gold brand. From the Kilolo District government, interviews were conducted with one District Executive Director and one District Agriculture, Irrigation, and Cooperative Officer (DAICO). Other informants were two Ward Executive Officers and a Village Executive Officer from each of the four villages. The farmer surveys and the key informant interviews were conducted by two authors of this study who have prior experience with social research in this region.

5. Analytical models

We assume that a farmer will choose the marketing channel that will maximize his/her utility based on the theoretical framework of the random utility model of consumer choice developed by McFadden (1974), which was widely applied in the studies of marketing channels for smallholder farmers (e.g. Blandon et al., 2009; Schipmann and Qaim, 2011). In our study, each farmer i chooses one main marketing channel from the four marketing alternatives (j=1, 2, 3, and 4), sales to assemblers, wholesalers, processors, and consumers, which provided a certain level of utility Uij. A farmer chooses channel j if and only if Uij > Uik for any kj. Farmer i’s utility from choosing the i-th alternative can be decomposed as

Uij = Vij + εij,

Where Vij is the systematic component of the utility and εij captures the factors that affect utility but cannot be observed thus treated as random.

To implement the random utility model, a multinomial logit regression using assemblers as the base case category was estimated. Multinomial logit regression model is suitable for analyzing unordered responses with more than two choices (Wooldridge 2010), which is commonly used in research on marketing channel choices (Bezabih et al., 2015; DeLong et al., 2019; Magogo et al., 2015; Mmbando et al., 2016; Park et al., 2014; Xaba and Masuku, 2012; Temesgen et al., 2017). In the multinomial logit model, the probability that a farmer i chooses jth market channel can be written as

FIG000003

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

where xi is a vector of characteristics of individual i, J is the total number of market channels, βk measures the contribution of personal characteristics to the probability of selecting the k-th market channel, where k takes the value from 1 to J, and in particular, βj is for the j-th market channel. We can compute marginal effects to determine the percentage change in the probability of farmer i with characteristics xi choosing a market channel j for a one-unit change in explanatory variables as follows,

FIG000004

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

where is the probability-weighted average of the β coefficients.

Furthermore, we construct a revenue model based on a standard model of supply choices under uncertainty (Hanemann and Tsur, 1982) to identify the important factors that affect the performance of farmers. Ideally, a farmer’s performance should be measured by the profit. We choose to use the revenue instead due to the lack of cost data. A farmer’s profit, π, is given by

π = pqc(w, q),

where p is the product price, q is the amount of product supplied, and c(.) is a cost function generated by some production function with input price vector w. Assume a farmer faces uncertainty with respect to the product price with density function f(p) with mean p0 and chooses an output level, q, to maximize the expected utility,

max u(q) = max ∫ u(π(q, p); x)f(p)dp,

where x is the vector of the farmer’s individual characteristics. The solution to the farmer’s maximization problem will be denoted by q(p0, w; x). A farmer’s revenue R can then be expressed as,

R = pq(p0, w; x).

In our data, we have farmers’ revenue according to their sales from the market channels, but unfortunately do not have the cost. Therefore, we choose to use a simple reduced-form model to explore the effect of a farmer’s individual characteristics on revenue. We do not use price p as an explanatory variable in the regression analysis, as a farmer’s revenue is directly linearly associated with the price, with the coefficient q(p0 , w; x), noting that p0 is exogenously given. What we are most interested in are the important factors in x that affect a farmer’s supply choice, q, which in term yields the revenue. Therefore, our revenue model is a reduced-form linear model as follows,

,

FIG000005

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Where Ri is farmer i’s revenue, xjk is the farmer i’s kth characteristics, and εi is the error term.

6. Descriptive findings about the marketing system

We first summarize the descriptive findings of our data using summary statistics (Table 1). Assemblers were the dominant market channel used to sell tomatoes. Wholesalers were the next most common marketing agent for the farmers, followed by the industrial processing channel. Very few farmers engaged in marketing directly to consumers. Only 38 out of 286 (13.3%) respondents were female, which is consistent with the finding of a previous tomato marketing study in the same area that men dominated marketing activities for this cash crop in Tanzania although women contributed significantly to production of staple foods (Khasa and Msuya 2016). Among female respondents, 63% marketed through assemblers. A smaller proportion, but still a majority, of the male respondents (60.9%) used assemblers.

Farmers relied on local marketers but attitudinal questions revealed dissatisfaction with the system. A majority of the farmers (57%) indicated that they would prefer to change the way in which they market their tomatoes. Other responses directly confirmed mistrust in the transactions. By a large majority, the tomato farmers did not trust the market actors to measure the quantity of tomatoes in the transactions (85.7%). Most farmers (86.0%) also reported that the buyer had the decision-making power over the measurements used in the market, namely whether the quantity was based on weight or by volume and who controlled the scales.

Table 1.
Table 1.
Table 1.

Survey responses by gender of the farmer.

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

There was a significant gender gap in access to market information. Nearly 35% of men reported to have access to market information, whereas only 18.44% of women reported so. Men and women also had somewhat different perceptions of trust and lack of control. While large majorities of both genders reported lack of knowledge of the laws and regulations governing the marketing of produce, female respondents indicated greater confidence than males (18.4% of women compared with 16.9% of men reporting being knowledgeable). The female farmers expressed knowledge of the institutions in spite of the lower level of formal education among women. Women expressed trust that the marketing agents used proper weights and measures more frequently than men (18.4% for women relative to 13.7% for men). Although both genders were largely mistrustful of the measurements, and other survey questions that related to trust indicated gaps in trust in marketing agents, it is to be expected that the farmers preferred to have control over how the product quantity was measured. While most farmers were unable to control the measurement, relatively more women than men reported that they controlled the measuring approach used in marketing (15.8% of women in control compared with 13.7% of men).

Revenue and prices received in the alternative market channels varied as expected by the different levels of service provided to the growers (Table 2). The average revenue from tomato production was lowest for the farmers marketing through the assembler channel, which was the channel used by the majority of farmers. Farmers’ average revenue when using assemblers was 11.5% lower than average revenue of farmers marketing to wholesalers. The wholesalers paid higher prices than assemblers by 8.9% on average. The processing industry paid the lowest price, 1.1% lower than assemblers and 10% lower than wholesalers, yet growers using industrial channels reported the highest revenue. Clearly the volume purchased by the industrial buyers contributed to larger total sales revenue for farmers. This observation is suggestive of the incentives for a transition toward farm consolidation spurred by industrial food production. Although at the time of this study, none of the farmers operated large farms. The farmers who marketed directly to consumers reported the highest average price and revenue, as expected given that direct marketers must be compensated for performing all the marketing functions and bearing all the risk that the perishable crop might be lost. However, the price difference was not statistically distinguishable from prices paid in other marketing channels (Table 2), likely because few farmers sold direct-to-consumers and the sample size impeded statistical significance.

7. Hypothesis and explanatory variables

Utility theory implies that a number of factors affect utility from a market channel. Furthermore, there are constraints on the farmers’ business, possibly due to the social disadvantages of the smallholder farmer, that need to be in the model as control. In order to understand the important factors influencing market channel choices, we estimated two quantitative models. The first model is a multinomial logit model that has the objective of explaining drivers of a choice among market channels. The second quantitative model is a regression to specify revenue as a function of the market channel, socioeconomic characteristics of the farmer, and information sources available to the farmer. It is a direct quantitative measure of the factors associated with marketing success for the growers.

Table 2.
Table 2.

Price and revenue by market channel.

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Explanatory variables were chosen based on the findings in the descriptive statistics as discussed in the section above and the section on Related Literature, including basic demographic information and social conditions of farmers, such as age, gender, marital status, household size, quantity harvested, access to information and extension service, farmer association membership, and trust. The reason to include these variables is because previous studies have shown that these factors can affect farmers’ choice of marketing outlet. For example, Magogo et al. (2015) studied market outlet choices of vegetable farmers in Kenya. They found that, using no market participation as the base, farmers with higher education level preferred to sell to brokers; female farmers and farmers with larger household size preferred to sell the products at local open-air markets; whereas farmers with higher quantity sold were less likely to sell at the farm gate. Furthermore, farmers’ access to information and extension service influenced their choice of profitable market channels and improved profitability (e.g., Aker and Fafchamps, 2015; Goyal, 2010; Jensen, 2008; Minten, 1999; Nakasone, 2013). Many studies have shown that gender categories played different roles in the tomato value chain (e.g. Khasa and Msuya, 2016). Studies have also found that farmers’ decisions may depend on their marital status (e.g. Yusuf and Li, 2016; Van Aelst and Holvoet, 2016). As transportation barriers affect smallholder farmers (Magogo et al., 2015 Negi et al., 2018), we included a transportation variable, specifically, the type of motor vehicle owned by farmers in our analysis. The hypothesis behind the choice model are farmers would choose the channels with the highest price conditional on sociodemographic conditions. Household size and education are hypothesized to contribute to formal channels, i.e., industrial and wholesale channels. Marketing information including extension services are also expected to positively affect the choice of formal marketing systems. Correlations between all explanatory variables are weak (lower than 0.2) except marital status vs. age (0.3745) and household size vs. age (0.2658) that are still not large enough to cause the multicollinearity problem in the regression analysis. Marketing systems may involve trust-building to engender loyalty from the final consumer (Layton, 2009). Does trust play a role at the primary producer level as well? Farmers need to find ways to build trust in business relationships (such as improving cooperation and communication) in order to enhance their bargaining power and reduce transaction costs. (Mmbando et al., 2016). In particular, weights and measures is a problem for smallholder farmers in Africa particularly in Tanzania. Most of the farmers in Tanzania were using volume instead of weights as recommended by the regulators in the market. To our knowledge, there is a gap in the literature to document issues surrounding weights and measures. In our analysis, we include three questions related to farmers’ knowledge and trust in the marketing network: (1) Knowledge of weights and measures; (2) Who is the decision maker about the measuring approach used in marketing? (3) Do you trust the weights and measures?

8. Regression results

The results of the multinomial logit model and the marginal effects are presented in Table 3. Since only 2.1% farmers chose directly selling to consumers, we dropped this choice in the regression and chose assemblers as the base. A majority of the growers sold at the farm gate and delegated all marketing functions to the assemblers. What explains the use of wholesalers or agro-processors instead? The findings indicate that quantity and farm size were key factors for those using processors or wholesalers. Farmers with higher quantity harvested were more likely to choose the processor channel over assemblers, with one extra unit of output adding 0.4% chance.

Farmers who were members of larger households were more likely to choose wholesalers compared to assemblers with one extra household member adding 4.39% chance, while one extra household member lowered the chance of choosing assemblers by 3.31%. Farmers with pickups or trucks were 10.26% less likely to choose processor channels compared to the growers using bicycles to market, which suggests that proximity to the processing facility was important to the selection of the industrial buyers. Information networks and trusts in the exchange as reported by the farmers turned out to be connected to market channel choices. When farmers believed that sellers — farmers themselves — are the ones who decide about the measure to be used, they were 7.05% more likely to choose the processor channel. This result implies that control over volume was important to the grower. Prices did not emerge as a statistically strong effect on the farmers’ choice between assemblers and processors, yet price did correlate for those marketing to wholesalers. Every 1000 shillings in price paid (around 5% of the average price) increased the probability of choosing the wholesaler market option by 0.848%. A number of other factors were also controlled in the multinomial logit model of the marketing channels, including transportation and logistics factors, knowledge level of the farmer, and family characteristics, but none of them were statistically significant.

We also ran a separate regression using size of farm instead of quantity harvested as the regressor (Table 4). The farm size in our data is a categorical variable (small-size farms <1 acre, mid-size farms 1–3 acres, large-size farms >3 acres). Using mid-size farms as the base, we found that compared to mid-size farms, large-size farms are more likely to choose the industrial processing channel, which is in line with the regression results in Table 3 that farms with higher harvest quantity preferred the industrial processing channel compared to assemblers. Additionally, compared to mid-size farms, small-size farms were less likely to choose the wholesalers. Large-size farms were also less likely to choose wholesalers although p value is not significant. So, another way to interpret this result is that, operators of mid-size farms were most likely to choose the wholesalers; small-size farms preferred assemblers; and large farms preferred the industrial processing channel.

The revenue drivers were estimated with a log linear model adjusted for the non-constant variances of the residuals (heteroscedasticity), a common problem with cross-sectional data. Breusch–Pagan test (p =0.0000) rejected the null hypothesis of constant variance and justified the log linear transformed model (p=0.9235). We have also estimated a heteroscedastic linear regression with two-step generalized least squares estimation, which yielded similar results to the log linear model. Therefore, we chose to report results of the log linear regression (Table 5), where the exponentials of the coefficients represent the percentage change in revenue for a one-unit change in explanatory variables.

The regression model results confirm the suggestions from the descriptive statistics regarding revenue differences across market channels (Table 5, Model 1). To test the value of a marketing service company to take on the risk and expense of transportation, we also constructed a model including the interaction of the transportation mode with the market channel choice (Table 5, Model 2). The Likelihood-Ratio Test showed that the bigger model makes the better use of the information in the data compared to the smaller model significantly (p=0.0510). The results showed that motorcycle users in the wholesale channel had higher revenue than other groups. Factors such as marital status, age, gender, off-farm income, grading, education, and membership in farmer associations did not have statistically significant effects on the revenue. Wald test (p=0.7665) showed that these variables did not create statistically significant improvement in the fit of the model. We hence performed a regression dropping these variables (Table 5, Model 3) and found similar results to Model 2.

Table 3.
Table 3.
Table 3.

Multinomial logistic regression of market channel choice model (quantity harvested) (n=249).

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Revenue from the industrial channel was significantly higher than revenue for growers marketing with assemblers. The revenue analysis provided insights about the value of information networks, where the market information was sourced, and whether the information was widely available to all farmers. Those farmers who indicated having adequate market information comprised 32.5% of the survey respondents and women farmers were less likely than men to report having market information. Farmers who reported to have access to market information had significantly higher revenue at the mean than those who felt uninformed. There was a clear divide about which sources of information were beneficial. Publicly-provided agricultural extension services were a relevant source of information and made a positive contribution to revenue. Growers who relied on middlemen (or market women) for market information had lower revenue than farmers who used other sources, such as radio, extension agents, or farmers’ associations.

These results confirm that information was a driver of revenue benefits in the tomato marketing system. Those farmers who relied on traditional marketing agents (middlemen/market women) had lower revenue. It is very important to be sure that market information is available from unbiased sources, such as the public sector extension services. Efforts to have extension advice widely available and focused on concerns for smallholders and for women within the smallholder producer sector must continue. The results suggest that the gender gap about market information and the consequent revenue impact should be carefully distinguished from the survey result that women farmers understood laws and regulations. Statutory or institutional knowledge did not correlate with revenue, whereas actionable marketing information about prices, quantity bids, or location of buyers mattered.

It is noteworthy that information flowing from the market channel itself may be confounded with the explanatory variable “market information.” In our survey, we asked respondents “Do you have market information?” From our team’s experience as field researchers collecting the survey data, the meaning of the market information variable is whether the respondents had access to the standard independent sources of information about market demand and prices, for example, radio, extension service reports, and ministry of agriculture publications. Information sources such as these were available at all relevant times of the year for growers to choose when and how much to plant, and what prices might be at certain markets. These were the information sources that assist in price discovery, in negotiations, and in making a selling decision. As such, these are valid factors to use to explain which channel was used. The use of the market channel is the physical delivery of the produce and is not in itself a means of price discovery. While it is true that agents in the various channels may quote a spot price, the channel choice is a selection of transactions, handling and distribution and did not serve the purpose of unbiased information flow. To confirm, we performed a reverse causality test using information as dependent variable and marketing channels as regressors and found that market channels did not affect information differently (data not shown).

Table 4.
Table 4.

Multinomial logistic regression of market channel choice model (farm size) (n=249).

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Table 5.
Table 5.

Log linear model of revenue drivers (n=253).

Citation: International Food and Agribusiness Management Review 2024; 10.22434/ifamr2022.0152

Household size was an important explanatory factor in revenue with the number of household members positively associated with revenue. However, because of the way this variable was collected, it is impossible to conclude whether the larger households contained more working-age people who could specialize in marketing tasks, or whether the larger households had more youth or elders in need of family care. The modest size of the effect suggests that additional household members were engaged in farm activities that marginally increased production volume, but they did not permit the household to engage fully in marketing activities that captured value. The other household members might have been doing off-farm work or might have been younger than working age.

Formal education did not emerge as a contributing factor in revenue differences among farmers. Education was not statistically significant in the regression model, probably because the study participants generally had low education levels with only 64 out 286 respondents having education beyond primary school. Women were disproportionately represented in the poorly educated groups, but there was not a clear disadvantage in revenue associated with lower education.

9. Conclusion and implications

The marketing system for fresh tomatoes in the Iringa region of Tanzania was sufficiently developed to provide alternative market channels for smallholder farmers. A large majority of the farmers used the full-service marketing agent, a local assembler who provided all the marketing functions beyond the farm gate. Consistently with previous literature (e.g. Aker et al., 2015; Birthal et al., 2015), we found that information sources were influential in the marketing system and in revenue outcomes. Access to market information and extension services improved smallholder farmers’ revenue. The variety of buyers and agent types allowed farmers to specialize in production and avoid the need to engage in complex marketing tasks. There were observable differences in producers’ revenue associated with the marketing services provided by the marketing agents, suggesting market efficiency in an economic sense and a potential opportunity for a smallholder farmer who wished to engage in more marketing activities.

In our study, access to market information was uneven, in line with previous literature (Agunga et al., 2018; Davis et al., 2012; Hadebe and Msuya, 201; Isaya et al., 20186; Quisumbing and Pandolfelli, 2010). The percentage of women farmers who reported that they lacked market information was higher than men. The women farmers felt that they understood the formal rules and laws yet the key transactional information that was statistically correlated with higher revenue was not available to them. To compensate, like most growers in the region, the women farmers hired marketing services from the assembler. If the women farmers only had middlemen/market women to rely on for marketing information, then there was an economic penalty judging by revenue comparisons. Given the possible constraints on women from family / household work, the revenue deferred by using full-service marketing agents in the assembler channel was an acceptable tradeoff.

Extension services are potential avenues to equitably offer information services that might lead to enhancing the position of women farmers (Isaya et al., 2018). It turned out that the proportion of women who had access to extension services exceeded the proportion of men with access, suggesting that public information sources contributed to valuable information flows but were not entirely satisfactory. To ensure that women farmers have commercial opportunities, all farmers should have equal access to market information. Information did matter to revenue and women reported that they had a disadvantage in accessing that information.

Regarding trust in the marketing system, it is common for small business operators to feel mistrustful of the marketing agents who act as gatekeepers to the consumer. Farmers’ level of perceived trust in buyers could significantly affect their choices of marketing channels (Mmbando et al., 2016). Mistrust was so widespread in this study that trust, generally stated, was not a statistically significant differentiating factor in the farmers’ revenue streams. Weights and measures was a problem for smallholder farmers in Africa particularly in Tanzania where most of the farmers used volume instead of weights as recommended by the regulators in the market. To our knowledge, this problem has not been documented in any previous studies and our study has added to the body of literature. Our results show that farmers who controlled the means of measuring the quantity of their goods sold were more likely to sell to the industrial processors and better off with higher revenue. This finding suggests an opportunity to ameliorate the mistrust issues simply by buyers accepting the sellers’ determination of weights and measures. In Iringa, the sharing of control over measures was accomplished without an extensive public sector role in the marketing system. The sole publicly-funded agent in the market was the extension advisory service, which was positively associated with farmers’ revenue.

Farmers with pickups or trucks were less likely to choose processor channels compared to growers using other transportation vehicles to market. Interestingly, motorcycle transportation in the wholesale channel was associated with higher revenue than all the other groups. In contrasting the logistics of motorcycles compared with trucks and bicycles, as a vehicle adapted to the weak transportation infrastructure, motorcycle delivery may have had a timing advantage which allowed more of the produce to be salable upon arrival at the wholesale market. Motorcycles may have been more flexible in navigating poor quality roads or bypassing traffic congestion. In addition, the packing method for the motorcycle may have been advantageous relative to shipping in the bed of a truck. This suggestion about packing relies on Kitinoja and Kader (2015), who described the damage to fresh produce shipped in the traditional baskets like the Tanzanian tenga, and from using plastic crates without a cloth wrapping. These results point out that transportation vehicles carrying larger quantity shipments might not improve marketing outcomes if there is damage during transit.

Among smallholder farmers, the volume incentives that an industrial buyer offers overcame the disincentive of lower prices paid by the company. Many studies have shown that quantity sold and land area affected farmers’ marketing decisions (Davis et al., 2012; Fischer and Qaim, 2012; Xaba and Masuku, 2012). We found that farmers with higher harvest quantity were more likely to choose the industrial processing channel. Moreover, when categorized by farm size, our results show that small-size farms preferred assemblers, mid-size farms preferred the wholesalers, while large farms preferred the industry channel. From a risk management perspective, this makes sense, because a guarantee of high volume would be preferred to the risk and the extra effort it would take to deliver the perishable goods to a distant wholesale market. Noting that there were three processing firms in the marketing system, rather than a sole source, the competition among the buyers may have contributed to the prices being acceptable. The fact that an industrial buyer was part of this marketing system indicates that the transition to commercial farming was proceeding. There might be social change in the future linked to the volumes sold to the processing channel as the business relationship develops and some farms consolidate based on the opportunity to specialize in production for processing. Our results can provide important policy implications for scaling agribusinesses in the developing world.

There are some limitations in this study. For example, the sample size is relatively small that may not be representative of the 60 000 tomato farmers in the area. Although the performance of farmers is normally measured by profit, revenue is used as an outcome variable in our analysis due to the lack of cost data. There are potential endogeneity problems in the analysis that are not examined due to limits on data and information. For example, do farmers first decide how much to produce and then choose the marketing channel? Future research should take into account these factors in the study of the smallholder farmers’ marketing choices.

References

  • Abdul-Rahaman, A. and A. Abdulai. 2018. Do farmer groups impact on farm yield and efficiency of smallholder farmers? Evidence from rice farmers in northern Ghana. Food Policy 81: 95105.

    • Search Google Scholar
    • Export Citation
  • Agunga, R., C.A. Sanga and E. Isaya. 2018. Empowering women farmers in Tanzania through communication for development. World 5 (2): 13758.

    • Search Google Scholar
    • Export Citation
  • Ahmed, Y.E., A.B. Girma and M.K. Aredo. 2016. Determinants of Smallholder Farmers Participation Decision in Potato Market in Kofele District. Oromia, International Journal of Agricultural Economics 1: 4044.

    • Search Google Scholar
    • Export Citation
  • Aker, J. C. and M. Fafchamps. 2015. Mobile phone coverage and producer markets: Evidence from West Africa. The World Bank Economic Review 29 (2): 262292.

    • Search Google Scholar
    • Export Citation
  • Anang, B.T., Z.A. Zulkarnain and S. Yusif. 2013. Production constraints and measures to enhance the competitiveness of the tomato industry in Wenchi Municipal District of Ghana. American Journal of Experimental Agriculture 3 (4): 824838.

    • Search Google Scholar
    • Export Citation
  • Barrett, C.B. 2008. Smallholder market participation: Concepts and evidence from eastern and southern Africa. Food Policy 33: 299317.

    • Search Google Scholar
    • Export Citation
  • Barrett, C.B., M.E. Bachke, M.F. Bellemare, H.C. Michelson, S. Narayanan and T.F. Walker. 2012. Smallholder participation in contract farming: comparative evidence from five countries. World Development 40 (4): 715730.

    • Search Google Scholar
    • Export Citation
  • Bellemare, M.F. and C.B. Barrett. 2006. An ordered Tobit model of market participation: Evidence from Kenya and Ethiopia. American Journal Of Agricultural Economics 88 (2): 324337.

    • Search Google Scholar
    • Export Citation
  • Bezabih, E., K. Mengistu, M. Jeffreyson and Y. Jemal. 2015. Factors Affecting Market Outlet Choice of Potato Producers in Eastern Hararghe Zone, Ethiopia. Journal of Economics and Sustainable Development 6: 159173.

    • Search Google Scholar
    • Export Citation
  • Blandon, J., S. Henson and T. Islam. 2009. Marketing preferences of small‐scale farmers in the context of new agrifood systems: A stated choice model. Agribusiness 25 (2): 251267.

    • Search Google Scholar
    • Export Citation
  • Birthal, P.S., S. Kumar, D.S. Negi and D. Roy. 2015. The impacts of information on returns from farming: evidence from a nationally representative farm survey in India. Agricultural Economics 46 (4): 549561.

    • Search Google Scholar
    • Export Citation
  • CNBCAfrica. 2020. ‘Agriculture is the most important business in the world,’ African Development Bank President Akinwumi Adesina tells students. Available online at https://www.cnbcafrica.com/africa-press-office/2020/01/29/agriculture-is-the-most-important-business-in-the-world-african-development-bank-president-akinwumi-adesina-tells-students (accessed 29 January 2020)

    • Search Google Scholar