Article Text
Abstract
Background The tobacco industry has used the alleged negative impacts on economic livelihoods for tobacco farmers as a narrative to oppose tobacco control measures in low/middle-income countries. However, rigorous empirical evidence to support or refute this claim remains scarce. Accordingly, we assess how much money households earn from selling tobacco, and the costs they incur to produce the crop, including labour inputs. We also evaluate farmers’ decision to operate under contract directly with tobacco manufacturers and tobacco leaf-buying companies or to operate as independent farmers.
Methods A stratified random sampling method was used to implement a nationally representative household-level economic survey of 585 farmers across the three main tobacco growing regions in Kenya. The survey was augmented with focus group discussions in all three regions to refine and enrich the context of the findings.
Results Both contract and independent farmers experience small profit margins per acre, with contract farmers operating at a loss. Even when family labour is excluded from the calculation, income levels remain low, particularly considering the typically large households. Generally, tobacco farmers enter into contracts with tobacco companies because they have a ‘guaranteed’ buyer for their tobacco leaf and receive the necessary agricultural inputs (fertiliser, seeds, herbicides and so on) without paying cash up-front.
Conclusions Tobacco farming households enter into contract with tobacco companies to realise perceived economic benefits. The narrative that tobacco farming is a lucrative economic undertaking for smallholder farmers, however, is inaccurate in the context of Kenya.
- eonomics
- low/middle income country
- tobacco industry
Statistics from Altmetric.com
Introduction
The emergence, promotion and expansion of tobacco leaf cultivation in many low and middle income countries (LMICs) have been supported by the tobacco industry narrative that tobacco production is lucrative for the economy, including benefits to government and tobacco farmers.1–9 This narrative is deployed against tobacco control measures, with arguments that such measures result in the loss of export earnings,5 6 8 jobs for cigarette manufacturing workers,7 tax earnings to governments as a result of reduction of tobacco consumption and,5 7 more pertinent to this research, that control measures can negatively affect the economic livelihoods of farmers dependent on tobacco as a cash crop.3 6 These arguments resonate with some governments, like Malawi’s and Zambia’s, which have even challenged novel tobacco control efforts in international economic fora,10 11 creating a major barrier to tobacco control in many countries where tobacco is grown. Although there is a small emerging literature suggesting that smallholder tobacco farmers do not make adequate returns from tobacco farming and that the contribution of tobacco earnings to gross domestic product in most LMICs is small,8 9 12–14 there is still a paucity of empirical evidence across countries and time to systematically counter the tobacco industry’s prosperous livelihood narrative. Country-specific empirical studies are important because of a need to build a wide and deep body of evidence, and policy makers’ increasing demand for country-specific evidence to justify enhancing tobacco control to their constituencies.
This research is a rigorous household-level economic assessment of tobacco farming in Kenya using a nationally representative sample to survey farmers and follow-up focus groups drawn from the sample. It builds on earlier work in Kenya but unlike previous research that covered only one region and was not based on an extensive household survey,12 13 this research uses original data from households in the three regions and over four counties where tobacco is most widely grown, making results nationally representative. The study also differs because it further elaborates the value chain that makes tobacco an attractive commodity in regions where it is grown.
This research makes important contributions to the emerging discussion about tobacco farming in three ways. First, it uses analysis of nationally representative data on the economics of tobacco farming to inform policy makers in developing a national policy on tobacco farming in Kenya. Second, and unusual in such studies, it accounts for production costs more comprehensively by systematically incorporating a monetised value of family labour. Finally, it builds upon other studies in low/middle-income countries that examine both contract and independent farmers (see below) to help determine if there are differences in economic livelihoods between these groups.8
Tobacco farming in Kenya
The number of tobacco farmers in Kenya has been increasing recently from 35 000 in 2007 to 55 000 in 2015.13 15 16 Much of the farming takes place in the Western region, where over 24 000 are involved in tobacco farming.13 In recent years, cigarette manufacturers like British American Tobacco (BAT) and Mastermind Kenya (MTK) and leaf-buying companies like Alliance One have been the major firms participating in the tobacco leaf market. However, Alliance One exited the market in 2016 because of divestment from flue-cured Virginia tobacco, the dominant tobacco leaf grown in Kenya. BAT-Kenya, which commenced cigarette manufacturing in Kenya in 1957 is the dominant firm, with Mastermind, which started operations in the 1980s being the second largest.17 The two firms are engaged in manufacturing, and with value addition to the leaf, their profits have soared over the years. However, smallholder farmers have not experienced a commensurate growth in income and hence benefit of this value addition. Smallholder tobacco farmers generally prefer to engage in tobacco production under contract, wherein they receive inputs and agricultural extension services from the tobacco companies, with input costs deducted from the earnings upon sale of the leaf.18 19 Independent farmers source and pay for inputs and do not receive extension services from the tobacco companies. After harvest, independent farmers have freedom to sell the tobacco leaf unlike contract farmers who can only sell to the firm that contracts them. Each season the farmers choose to enter into contracts with the firms. However, if the farmers are in debt at the end of a season because of poor harvest and/or low prices, and lack income to repay inputs, they are obligated to another season with the contractor. The tobacco industry’s consistent narrative in Kenya as in other countries is that contract farming is a better economic choice for farmers,8 but there has been little empirical research comparing contract and independent farmers’ economic livelihoods, and no rigorous examination of why some farmers choose to contract while others remain independent.
Evidence from Kenya suggests that tobacco production is associated with serious social and environmental impacts. These include, among others, deforestation, poverty, negative ecological balance and food insecurity, with many farmers using most of their land for tobacco farming instead of food crops.18–21 Research suggests that tobacco has been less profitable to farming households than other crops such as passion fruit, soybeans, water melon, pepper and pineapple, when cost and return analyses of different crops grown in the tobacco growing region of Kuria are compared.22 Studies have also demonstrated that the socioeconomic status of tobacco growing households is lower than non-tobacco farming households in the same areas, with a higher prevalence of child labour, polygamy and large family sizes.20 Despite having a well-developed value chain compared with alternative crops in the same region, tobacco farming households demonstrated less access to financial resources, had lower incomes and depended more on remittances from other family members employed in formal sectors locally and abroad than did non-tobacco farming households.12 13 20 Accordingly, this research seeks to add to this literature by determining the impacts of tobacco farming on farmers’ economic welfare, by examining the economic costs and returns of tobacco farming, under both contract farming and independent farming in Kenya’s three main tobacco growing regions.
Methods
To examine the economic conditions of tobacco growing in Kenya, we implemented a quantitative household-level economic survey supplemented by qualitative focus groups. First, we collected primary survey data. The survey was developed by a multidisciplinary, international research team and implemented in January 2015 in three regions spread over the four Kenyan counties where tobacco is most widely grown. The survey questionnaire was divided into nine sections: household characteristics; livelihood, income and assets; land ownership and crop production; tobacco production generally; tobacco production under contract farming; tobacco marketing; farmer debt and credit; household food security and the future of tobacco production. The survey used a multistage and stratified random sampling procedure. To determine the sample size, we first defined the population size N of tobacco farmers in Kenya to be 55 000.13 We used a simple random sampling process adopting the conservative SD p̂ to be 0.5, confidence level as 95% (Z=1.96) and allowed the margin of error e to be 4.5%.
(1)
Based on equation (1), the study established that the unadjusted sample size needed to be 494. To adjust for population size, equation (2) was then considered.
(2)
As the population size is large, the adjusted sample size remained at 494. Based on previous agricultural surveys in the country, the expected response rate was between 85% and 90%. Therefore, we sought to reach out to 600 tobacco farmers to reach our target sample size. In the end, we obtained a sample size of 585 (a response rate of ~97.5%).
We implemented the study evenly surveying 200 households in each of the three main tobacco growing regions. These regions were purposefully selected based on production data from the Kenya Ministry of Agriculture. Data were collected in Migori County in Nyanza region, in Bungoma and Busia Counties in Western region and Meru County in Eastern region. The first step was to identify the main tobacco growing areas in each region based on production records and government agricultural staff. Four administrative units (counties) were selected from the regions and between 4 and 32 villages from each county. Enumerators randomly selected a tobacco farming household in each selected village, and then moving along a predetermined transect route that eventually converged to the village centre, selecting every other tobacco farming household along the route. Because of lack of availability of data on the actual tobacco farming households in a village, the selection also depended on the concentration of tobacco farming households witnessed by enumerators and the number of tobacco growing villages identified by the government agricultural staff.
The survey interviews were implemented by one team across all counties over a period of 1 month. The team of 10 enumerators was trained in data collection, interviewing approach and ethics in data collection to standardise the data collection. The first author supervised the team during data collection to ensure correct implementation of protocols. The data from the completed questionnaires were inputted into Stata.
Following the survey, we implemented focus group discussions (FGDs) with survey participants in each of the tobacco growing regions to contextualise and enhance our interpretation of the findings. Overall, we conducted one FGD in each region (Migori, Busia and Meru counties), for a total of three. The FGDs took place in a village centre or school in a randomly selected village with a high concentration of tobacco farming households. A subsample of surveyed farmers were randomly selected from the area (n=10–15 farmers per FGD). An FGD tool was developed by the research team based considerably on previous work in Malawi. The FGDs were led by the first author and a research assistant and used either audio recording or written notes.
Data analysis
Building on previous related research and using pertinent variables of all 585 observations,8 we estimate two related but distinct types of profit. First, for the perceived profit, we estimate average annual gross margins from tobacco growing enterprises that incorporate the revenues from tobacco sales and all costs associated with growing tobacco, including all physical inputs (cost after deducting resell of equipment or recycled parts to other crops), fees, transportation, levies and hired labour. Second, we estimate a cost–profit analysis that incorporates in the costs a monetised value of family labour based on 2013/2014 monthly minimum wage measures of agricultural day labourers from the Kenya Ministry of Labour and using Kenyan and Economic Intelligence Unit (EIU) exchange rate data. We define this enhanced measure as the actual profits that the household earns from engaging in tobacco production, by accounting for the foregone labour earnings of household members in producing the tobacco crop.
Smallholder tobacco farmers have another choice on whether to grow tobacco as a contract or as an independent farmer. This choice determines the level of interaction they will have with tobacco firms, including access to inputs and market, thereby possibly affecting the level of profit earned. Accordingly, we first compare perceived and real average annual tobacco-specific profits between contract and independent farmers.
We then further examine the social–economic factors associated with farming under a contract. Both complete case analysis and imputation for missing data using the hot deck nearest-neighbour method for all covariates were adopted, resulting in the analytic sample sizes of 462 and 585, respectively. The dependent variable is a dichotomous variable called contract farming. Specifically, participants who indicated that they had a written contract or some kind of marketing agreement were defined as contract farmers. Logistic regression was used to estimate the association between contract farming and the social–economic characteristics of farmers. Additionally, random effect logistic regression models were used to control for possible regional difference. To select covariates for analysis, this research drew from previous work,8 much of which used the machine learning method, random forest (RF). The FGD data were also considered to inform variables that may have been overlooked. RF is helpful as a complementary analysis tool for the logistic regression model, because it handles a considerable number of input variables without variable deletion, ranks the importance of explanatory variables and its recursive partitioning process brings in new perspectives in terms of exploring the feature space.23–25 Focus group data were analysed systematically for salient themes pertaining to the daily lives of tobacco farmers, including their reflections on the social and economic aspects of tobacco growing.
Results
Descriptive statistics
The sample of 585 tobacco growing households comprises 107 independent and 478 contract farmers. Tobacco farmers are predominantly men, are married, have a primary education and have farming as their primary source of income. The annual income earned from tobacco farming accounts for approximately 65% of the total household income on average. The results suggest that independent and contract farmers are similar in many socioeconomic aspects. However, the results also show that contract farmers are more experienced in tobacco farming with the average years of growing tobacco being 10 years compared with 7 years for independent farmers. Further, contract farmers have larger land sizes: 3.67 acres, of which an average of 2.8 acres is cultivated land and 1.87 acres is devoted to tobacco. This compared with independent farmers who have an average of 3.03 acres, of which 2.4 acres is cultivated and 1.56 acres devoted to tobacco.
Profit analysis
Table 1 presents the costs, prices and production of smallholder tobacco farmers. The average prices offered to the two groups varied, with contract farmers on average offered a higher price for tobacco leaf per kilogram, although the difference is not statistically significant. Farmers’ non-labour costs are also presented in table 1. Note that for the input costs we include the principal variable costs such as tools, fertiliser, herbicide, pesticide and seeds, but not the fixed costs such as land rental (where applicable, though importantly, land rental is not a large part of most farmers’ production). The results indicate that contract farmers have 25.11 per cent higher per-acre input costs than independent farmers. This difference was significant at p<0.05 for per-acre input costs although not statistically significant for per-kg costs. It is important to note that many of the farmers involved in the FGDs recognised the exploitative dependency the contractual relationship created for them: ’Tobacco companies are in business. They make money from selling inputs to contract farmers. If you want to do business with them, even when you have the ability to purchase inputs, you have to be a contract farmer. Otherwise they won’t do business with you.’ Contract farmers incur higher levies (local tax) per acre (p<0.05 for both per acre and per kg measures). Finally, transport costs are more for contract farmers in per-acre measures but less per kg, although neither difference is statistically significant.
Table 1 also presents the average labour hours—combined total of all household members—needed to produce an acre and a kilogram of tobacco leaf. Note that the kilogram measure used in this table is the amount actually sold in the 2013/2014 season (not necessarily the amount produced, which is typically more because some tobacco is not sold for a variety of reasons, which can include poor quality or a lack of demand). Per-acre labour hours from household members are lower for contract farmers (253) than individual farmers (339), although this difference is not statistically significant.
We calculated a profits-per-acre measure that includes personal and family labour so that we could compare the actual profit with the perceived profit of the tobacco farmers who, as our focus group data affirmed, were not typically incorporating this significant set of costs in their profit estimations. A monetisation estimate for family labour was computed by first summing all of the labour hours dedicated to tobacco by all household members. We then multiplied these hours by the daily minimum wage in US$ (using Kenyan government and EIU exchange rates) based on 2013/2014 monthly agricultural minimum wage measures from the Ministry of Labour Office. The choice of using the agricultural minimum wage is that any able-bodied farmer could choose to work as a labourer on another local farm. We argue that our results are a conservative estimate of the value of their labour because most farmers would have a higher skill level than the average labourer because they manage their farm and could likely find higher-paid work at least some of the time. While the two columns on the far right of figure 1 suggest a perceived profit for contract farmers of US$254/acre and US$394/acre for independent farmers, the actual profits drop precipitously once labour is included: a US$13/acre net loss for contract farmers and a decrease in profit for independent farmers to US$43/acre.
Independent farmers versus contract farmers: decision to become a contract farmer
Table 2 presents the results of four logistic regression models of the decision to become a contract farmer. Models 1 and 2 are logistic regression models in complete case analysis with dependent variable of the dichotomous variable contractor (1=contract farmer; 0=independent farmer). Models 3 and 4 are logistic regression models with missing values computed using hot deck nearest neighbour method. Further, models 2 and 4 are random effect models, while models 1 and 3 did not fix regional effects.
One of the most pronounced relationships between variables is that of marital status. Those who are married monogamously are at least three times more likely in the complete case sample to grow tobacco as contract farmers than single individuals, and at least 2.7 times more likely in the imputed case sample. In both cases, the coefficient is positive and statistically significant. Larger households were also more likely to engage in tobacco farming as contract farmers compared with relatively smaller households, with all four models suggesting that this is statistically significant. This finding likely indicates the importance of family labour in tobacco farming. Experienced farmers are more likely to be contract farmers with the coefficients being positive and statistically significant in all four models, suggesting that as a tobacco farmer increases growing experience by 1 year, the likelihood of him or her deciding to be a contract farmer increases. Statements from FGD participants speculated about this relationship:
We have fixed expenses like school fees for the children. Having certainty in income, however low is better than no income at all.
Tobacco is the only crop in the area where farmers are assured of some income. Other crops in the area have no money or cannot sustain a family consistently. To draw income from tobacco you need to be a contract farmer.
Legal entitlement of the farmer to the farm is also an important variable and its coefficient is positive and statistically significant; land ownership increases the likelihood of a farmer being a contract farmer by 28.8%. The coefficient for the need of credit is also positive and significant, suggesting that those who do not have access to existing capital to finance agricultural activities are between three and four times more likely to become contract farmers.
Discussion
The results provide important insights into the economic livelihoods of tobacco farmers in Kenya, including differences in contract versus independent tobacco farming. The results suggest that farmers in contractual relationships with tobacco companies generate limited actual profits, particularly when compared with their perceived profits, which suggests little to no monetary reward for choosing to enter into contract. The contract farmers also indicate significant dissatisfaction with the price that they receive for the leaf that they sell, with less than a third reporting that they believe they are receiving a fair price (p<0.05). This could be because the assignment of leaf grade and price is at the discretion of an official from the tobacco companies at the leaf-buying centres, with the farmers or farmer representatives to the tobacco companies having no part in the decision. Evidence from Malawi has shown that prices are persistently and systematically lowered, with very little recourse for tobacco farmers.26 Where a particular farmer voices disagreement with the grade and price allocated, the tobacco officials simply reject their produce, creating a situation where the farmer could either fail to sell his crop altogether or have his earnings delayed. The FGD participants indicated that contract farmers are given inputs at higher prices than they would ordinarily buy from shops, consistent with the survey findings: said one FGD participant: ‘The contract price for fertilizer was USD$40 compared to USD$20 in the retail shops’. This finding is also consistent with previous research that has demonstrated instances where the price for contracting farmers was higher than retail shops.26–28
Generally, both contract and independent farmers earn low gross margins from tobacco farming. Once accounting for family labour, independent farmers have slightly higher earnings than contract farmers. Three factors help explain the difference in the adjusted profit margins between the two categories of farmers: (a) non-labour inputs, (b) family labour and (c) tobacco leaf prices at the collection areas. The average tobacco price per acre for contract farmers is 12.8% higher than that of independent farmers. This suggests that tobacco companies might be encouraging all tobacco farmers to become contract farmers by purchasing their crop at higher prices, consistent with a finding from a study of tobacco farming in Malawi.26 The logic of contract farming is also tied to efficiency and quality gains, where companies introduce structural supports—for example, by helping farmers with effective chemical applications (eg, fertiliser, pesticide, herbicide and so on)—to ensure that farmers are growing a higher quality product.29 At the same time, tobacco companies appear to be exploiting the farmers through downgrading the quality of tobacco leaf while also by increasing the profit margins on their sale of inputs to contracted farmers.
The results suggest that farmers who have more experience in growing tobacco and who have the legal and permanent title to their farmers are more likely to be contract farmers, possibly explained by a greater awareness of their likelihood of selling their crop compared with that of independent farmers.
Opportunity costs also appear to play a part in the contracting decision of farmers. Notably, older farmers are less likely to be contract farmers while those with higher education levels are likely to be contract farmers. Older farmers generally/typically have larger land sizes and more experience in growing other crops which help in income diversification, affording them financial security outside tobacco growing. This can be seen particularly in Meru County, the most fertile tobacco growing area, where farmers participate in other economic activities that generate sizeable amounts of income when compared with tobacco, and where they reported relatively fewer complaints about the tobacco companies during the FGDs. With more education, farmers are generally more likely to be rational in making economic decision on farming tobacco as opposed to other crops. This is because tobacco growing areas are characterised by unstable markets for other alternative crops, while farming tobacco ensures a guaranteed buyer.
Family labour forms a critical part of tobacco growing, as it does with other crop farming. However, tobacco is a particularly labour-intensive crop that would generally be expensive to engage in if one depended purely on hired labour. Many household members actively contribute considerable time towards tobacco growing activities, an aspect generally not considered when tobacco farmers or researchers compute the costs of tobacco production. By monetising family labour as we have done in this study, we imply that tobacco only becomes minimally profitable through the use of free family versus paid labour. Our findings in this regard are particularly strong because our monetising of this household labour is conservative and likely underestimates its value.
These findings illustrate an important labour dynamic. Tobacco companies are exploiting what amounts to ‘free’ (to the companies) or at least unaccounted for—by the farmers—labour in smallholder tobacco growing. Because farmers do not incorporate household labour into their cost calculations in any way, their perceived profits are much higher than if they were to incorporate even a fraction of such costs into their cost calculations. At the same time, the exclusion of household labour costs in rural low and medium income countries is not unusual.30 This is particularly true when other sources of employment and other income opportunities are scarce. This is an important point for tobacco control proponents who are targeting the control of tobacco supply, whereby the local economy, which is often tied to the global economy, must be considered a key factor in policy interventions aimed at creating opportunities for other sources of income. Research has found that local opportunities are a key determinant of farmers’ decisions to pursue non-farm employment.31
Conclusion
This study demonstrates that tobacco farming is not a particularly lucrative enterprise for most smallholder tobacco farmers in Kenya, either independent or contract. Most farmers are making only a tiny profit at best. Moreover, once even a conservative estimate of the value of their own and family labour is incorporated, their actual profits diminish, suggesting that tobacco farming is even less lucrative than farmers generally conceptualise. Earnings from tobacco faming are typically low and unlikely to help most farmers move out of poverty. This ‘free’ family labour also indirectly contributes to high earnings for the tobacco firms. It therefore makes considerable economic sense for the government to aggressively seek viable alternative livelihoods in line with Article 17 of the WHO Framework Convention on Tobacco Control. This includes improving supply and value chains for other agricultural products that the farmers grow locally, increasing farmers’ access to credit and improving agrarian and farm management education for these households. Finally, although it is not possible to immediately provide a similar production model as one for the tobacco industry, farmers could organise themselves into formal groups and tap into existing agricultural development programmes and demand services that would facilitate income generation and diversification of their production systems.
What this paper adds
The tobacco industry often uses economic arguments to counter tobacco control efforts although there is a paucity of empirical evidence to prove otherwise. This paper provides quantitative analysis of economic livelihoods of tobacco farmers. It also provides one of the first cross-sectional studies of economic livelihoods of a representative sample of tobacco farmers in Kenya. The results suggest that the narrative that tobacco farming is a lucrative economic undertaking for smallholder farmers, however, is inaccurate in the context of Kenya.
References
Footnotes
Contributors PM, DM, JD, FG, RZ, RLe and RLa contributed to the study design. DM drafted the survey tool. PM, JD, FG, RZ, DM, RLe and RLa contributed to subsequent refinement of the survey tool. PM collected the survey data. PM, RLe and DM contributed to the focus group design. PM collected focus group data. QL, PM and JD completed the statistical analysis. PM, RLe, RLa and JD wrote the first draft of the manuscript. All team members contributed to the writing of the manuscript. JD led the revision. PM, JD and RLa led the resubmission. PM submitted the manuscript on behalf of the team.
Funding This research was supported by the Office of the Director, National Institutes of Health (OD) and the National Cancer Institute (NCI) under Award Number R01TW010898; and the National Institute on Drug Abuse, the Fogarty International Center and NCI under Award Number R01DA035158.
Competing interests None declarerd.
Patient consent Not required.
Ethics approval Institutional review board of Morehouse School of Medicine.
Provenance and peer review Not commissioned; externally peer reviewed.