Impact of Integrated Pest Management Technology on Food Security among Mango Farmers in Machakos County, Kenya


Mango (Mangifera indica) is one of the leading tropical fruits grown in Kenya and is ranked third after banana and pineapples in terms of acreage and total production volume. However, production has fallen below consumption due in part to fruit fly (Bactrocera invadens) infestation. About 40 percent of annual mango production in Kenya estimated at US$ 32 million, is lost due to direct damage of fruit flies. In an effort to improve production, the International center for Insect Physiology and Ecology (ICIPE) has developed a set of Integrated Pest Management (IPM) technologies aimed at controlling fruit fly infestation in mangoes. However, the impacts of these technologies on the food security are not well understood.  This study evaluated the impact of IPM technology for mango fruit fly control on food availability and accessibilty among 600 mango farmers in Mwala and Kangundo sub-Counties selected using a stratified sampling procedure. A seven-day recall was used to elicit Per Capita Calorie Intake while a 30-day recall was used to measure household dietary diversity. To evaluate the impact of IPM on food security the difference-in-difference method (DD) was used. The results indicate that 67 percent of IPM participants in Mwala and 75 percent of non-participants in Kangundo were food secure as they had attained the 2,250 Kcal threshold recommended by the Kenya National Bureau of Statistics (KNBS). The OLS regression results show that the IPM technology had a positive impact on per capita calorie intake but not on the quality of food intake (HDDI) estimated by the poison regression. This suggests that farmers using IPM technology benefit from income gains, and higher incomes improve the economic availability to food but not food access. The study recommends that the government should promote IPM technology for the control of mango fruit fly as it is likely to improve the food security of smallholder farmers.


2SLS Two-Stage Least Square
AIEI African Impact Evaluation Initiative
AR Autoregressive
ASDS Agricultural Sector Development Strategy
CIDP County Integrated Development Plan
DD Difference-in-difference
FAO Food and Agriculture Organization of the United Nations
FFS Farmer Field Schools
FSD Financial Sector Deepening
GOK Government of Kenya
HCDA Horticultural Crop Development Authority
HDDI Household Dietary Diversity Index
ICIPE International Centre for Insect Physiology and Ecology
IPM Integrated Pest Management
IV Instrumental Variable
KES Kenya Shilling (Currency)
KNBS Kenya National Bureau of Statistics
MOA Ministry of Agriculture
OLS Ordinary Least Square
PCCI Per Capita Calorie Intake
PSM Propensity Score Matching
SSA Sub Saharan Africa
STDF Standards and Trade Development Facility
USD United States Dollar (Currency)
VIF Variance Inflation Factor



1.1 Background

Mango production is a major income-generating activity in the country where it is produced by both large- and small-scale farmers for both export and domestic consumption. With regard to production, the fruit is ranked third after banana and pineapple in terms of acreage and quantity produced in Kenya (Korir et al., 2015). It is estimated that more than 200,000 small-scale farmers in Kenya derive their livelihood from mango production (ICIPE, 2009). In addition to income-generation opportunities, mango is important in fighting nutritional disorders as it contains almost all the known vitamins and essential minerals (Griesbach, 2003).

Kenya grows 32 mango varieties; however, only seven are grown on commercial scale (Ministry of Agriculture [MoA], 2010). These include Apple, Boribo, Kent, Tommy Atkins, Ngowe, Dodo and Van Dyke (Muthini, 2015). Over the years, mango production has been increasing owing in part to the growing global demand for mangoes which increased by 22 percent from 2007 to 2011 (Financial Sector Deepening [FSD] Kenya, 2015).

Approximately 98 percent of mangoes produced in Kenya are consumed in the domestic market while the rest is exported (Government of Kenya [GOK], 2010). Mangoes account for 26 percent of export earnings from fruits, which is second only to avocados at 62 percent in Kenya. In 2010, mangoes earned Kenya US$70 million and US$10.1 million in the domestic and export markets respectively (GOK, 2012). The major export destinations for Kenyan mangoes in 2010 were United Arab Emirates, Tanzania, and Saudi Arabia with each accounting for 53, 20 and 22 percent respectively (HCDA, 2010).

Voor de Tropen et al. (2006) identify several factors constraining mango production and marketing in Kenya. These include high perishability, poor quality planting material, pest and disease infestation, high cost of inputs, limited adoption of improved technologies, seasonal gluts, poor post-harvest handling, and poor market infrastructure. According to Korir et al. (2015), pest and diseases constitute the most debilitating constraints in mango production especially among the resource-poor smallholder farmers in Kenya. Ekesi et al. (2010) observe that about 40 percent of annual mango production in Kenya, estimated at US$ 32 million, is lost due to direct damage by native fruit fly species such as Ceratitis cosyra, Ceratitis rosa and Ceratitis capitate.

Infestation of mangoes by insect pests limits Kenya’s access to profitable export markets where such insects are considered quarantine pests (Korir et al., 2015). Thus, mango exporters in Kenya incur huge losses due to rejection and subsequent destruction of the fruit fly-infested mangoes. According to Horticultural News (2010), Kenya’s fruit industry losses up to KES 478 million annually from ban of fruits exports to South Africa due to fruit fly infestation. Kenya’s mango exports to the United States, Europe, Japan and Middle East must meet stringent phytosanitary standards to access those markets (Mitcham and Yahia, 2010). Locally, Muchiri (2012) reported 56 percent mango yield loss due to fruit fly damage in Embu County.

In order to stem the huge losses in mango production, farmers in Kenyan predominantly use chemical broad-spectrum pesticides to ameliorate the problem (Amata et al., 2009). Although chemical pesticides have been employed as the primary pest control strategy by mango farmers, there is increasing evidence of pest resistance to available pesticides (Korir et al., 2015). Additionally, the larvae of some insect pests, which is the most destructive stage, develop inside the fruit tissue and are not reached by pesticides applied on the surface of the fruit (ibid.).

To address this limitation, farmers tend to increase the frequency of spraying thereby increasing both the production cost and the likelihood of developing pest resistance to available pesticides (Macharia et al., 2009). In an effort to enable mango farmers reduce production losses and minimize the incidence of pest resistance, the International Center for Insect Physiology and Ecology (ICIPE) has developed a set of Integrated Pest Management (IPM) technologies for mango fruit fly control in several sub-Saharan countries (ICIPE, 2009).

In Kenya, the strategy has been implemented in the major mango growing areas of eastern and coast counties. Trials on the IPM technology package were conducted at Mwala sub-County in 2015 through a project in which farmers were enrolled and trained on use of the mango fruit fly IPM package components at designated lead mango orchards. After each training session, participants were issued with starter kits of the IPM technology for trial at their orchards. These technologies were based on baiting and male annihilation techniques (MAT), fungal application, orchard sanitation, use of weaver ant (Oecophylla longinoda) and biological control using parasitoids (Korir et al., 2015; Kibira et al., 2015; Muriithi et al., 2016).

The mango fruit fly IPM technology uses a combination of interventions that complement each other rather than work as a stand-alone management strategy (Ekesi and Billah, 2007). The spray food bait is a food protein bait (DuduLure®) developed by ICIPE and is combined with an insecticide named spinosad (Muriithi et al., 2016). The food bait is applied as localized spots at a rate of 50 ml solution on 1 m2 of mango canopy. Both adult male and female fruit flies are attracted to the confined area on the canopy of the mango tree where the food bait is sprayed (Ekesi et al., 2015). The fruit flies ingest the bait along with the toxicant, which kills them before they infest fruits (Ekesi et al., 2014).

The male annihilation technique (MAT) involves deployment of high-density trapping stations consisting of a male attractant (in this case methyl eugenol), combined with a toxicant (malathion) to trap and kill male flies thus reducing their populations to very low levels such that mating does not occur or is greatly reduced (Ekesi and Billah, 2007). The strategy employees 7 Lynfield trap stations per ha recharged after 6 weeks of exposure (Muriithi et al., 2016).

The bio-pesticides used in the IPM package are fungus-based formulations that targeted pupariating larval stages of the fruit flies and emerging adults but do not have any effect on beneficial parasitoids (Ekesi et al., 2015). Instead they complement the beneficial parasitoids in significantly reducing the fruit fly populations. Orchard sanitation is achieved using an Augmentorium (Klungness et al., 2005). This is a tent-like structure that sequesters fruit flies that emerge from fallen rotten fruits collected from the field and deposited in the structure, while at the same time conserving their natural enemies by allowing parasitoids to escape from the structure through a fine mesh at the top of the tent (ibid.).

The IPM mango fruit fly control package is aimed at reducing economic losses at the farm level, reducing insecticide use and enhancing the supply of high quality mangoes to the market, raising profit levels for the producers thus improving their livelihood. The current mango fruit fly IPM technology dissemination and promotional activities have shown success with many farmers rapidly taking up the strategy (Korir et al., 2015). Kibira et al. (2015) and Muriithi et al. (2016) have shown that the use of IPM technology can lead to a reduction in magnitude of mango losses due to fruit fly infestation with associated reduction in expenditure on insecticides and increased net farm income.

The expected increase in net income will increase farmers’ food purchasing power, which in turn, is hypothesized to increased food security. On the other hand, it is possible that the innovations may be unsuccessful or do not produce immediate result, hence, has negative effect on household income and food security. For example, an increase in income can lead to households’ expenditure on food devoted to cereal staples alone such as millet, maize and sorghum. Since the introduction of the IPM package in Machakos County no work has been done to evaluate the intervention in terms of its effects on smallholder household food security. The current study assesses the impact of Mango IPM technologies for controlling fruit fly on household food security.

1.2 Statement of the research problem

The adoption and extensive use of improved agricultural technologies is vital for poverty reduction and improved food security in developing countries (Barrett and Lentz, 2010). Agricultural technologies can boost crop productivity, allowing higher production and lower food prices, directly contributing to alleviate food insecurity. ICIPE has developed and implemented a set of Integrated Pest Management (IPM) technologies aimed at reducing mango losses and the cost of production. This in turn will lead to an increase in marketable produce or save labour for non-farm activities and subsequently increase household income and food security.

Previous studies on the effect of mango fruit-fly IPM technologies have concentrated on pesticide expenditure and income (first order effects). Although these studies have shed some light on IPM adoption, reduced pesticide expenditure and increased farm income, they have not examined the impact of IPM on food security (second order effects) in Machakos County. Consequently, the existing literature is unable to inform policy makers on the impact of IPM on food security. In addition, there is limited knowledge on the factors influencing food security in Machakos County. Since the introduction of the mango fruit fly IPM technology in Kenya, no research has been done to evaluate the intervention in terms of its impact on smallholder household food security.

1.3 Purpose and Objectives

The purpose of this study is to evaluate the impact of IPM technology on food security among smallholder mango farmers in Machakos County.

The specific objectives of this study are;

  1. To assess characteristics of smallholder mango growers in Mwala and Kangundo sub-counties.
  2. To assess the impact of IPM technology on food availability and accessibility among mango producers in Machakos County.

1.4 Hypotheses tested

  1. The IPM technology has no impact on food availability and accessibility of smallholder mango producers in Machakos County.

1.5 Justification

In Kenya, Mango is ranked third among tropical fruits in terms of acreage and total production and accounts for 26 percent of foreign earnings from fruits. However, it is confronted with a major threat of fruit fly infestation that causes reduction of quality and quantity of marketable fruit and hence considerable produce losses. Use of IPM technologies has been shown to reduce magnitude of mango losses due to fruit fly infestation, reduction in insecticide expenditure and increased net farm income.

Understanding the impact of IPM technology is important for generating information to policy makers (National and County governments), mango IPM project funders (ICIPE) and farmers on technology effectiveness for future adjustment and up scaling to other mango producing areas. Knowledge about factors influencing food security points out areas of policy intervention that need to be emphasized in order reduce food insecurity in the country. The information generated by this study will contribute to the growing body of knowledge on impact assessment particularly focusing on other mango producing areas.

1.6 Organization of this thesis

This thesis is organized into five chapters. Chapter one introduces the background of the study, the statement of the research problem, purpose and objectives, hypotheses and justification of the study. In Chapter two, relevant studies are reviewed. These revolve around on impact assessment and the general approaches/methods used to operationalize them. Chapter three presents the methods and data used in this study. This chapter presents the analytical and empirical frameworks as well as the type and sources data used, and sampling procedures. Chapter four presents the results and discussion. Finally, Chapter five presents the summary of major findings, conclusions and policy recommendations.



2.1 Food security concept

Food security is a broad concept that is generally defined as physical and economic access to adequate, safe and nutritious food by all people at all times for an active and healthy life (FAO, 1996). The broad definition implies that food security is more than food production and accessibility. Generally, this definition has four dimensions that constitute the four pillars of food security: food access, availability, utilization and stability of food supply (Gross et al., 1999).

Food access is ensured when all members in a household have enough resources to acquire food to meet their nutritional and dietary requirements. Access reflects the demand side of food security, as manifest in uneven inter- and intrahousehold food distribution and in the sociocultural limits on what foods are consistent with prevailing tastes and values within a community (Barrett, 2010). Availability is achieved when sufficient quantities of food are available to all individuals (Latham, 1997). Food utilization requires a diet that provides sufficient energy and essential nutrients, along with access to potable water and adequate sanitation. Stability, on the other hand, concerns the balance between vulnerability, risk, and insurance to food access and availability, which are often termed as security (Jones et al., 2013).

In an effort to reduce the proportion of people suffering from hunger by half, world leaders committed themselves to the Millennium Development Goals (MDGs) aimed at eradicating poverty and hunger. Despite the tremendous progress towards the goal of halving the number of hungry people in the world, food security remains a major risk for 815 million worldwide according to the FAO and WFP report. The food security situation has worsened sharply in parts of sub-Saharan Africa, South-Eastern Asia and Western Asia due to conflict, climate change, drought and increase in population (FAO, 2017).

Studies by Babatunde (2007), Oriola (2009) and Fayeye and Ola (2007), have documented that despite the growing food production globally, malnutrition, hunger and famine are prevalent in many parts of Africa. This is partly due to domestic policies in many countries in sub-Saharan Africa having contributed very marginally to food security. These authors argue that improvement in food production in sub-Saharan Africa will boost per capita GDP, raise purchasing power and access to food. These studies conclude that research is needed on improved technologies that are output-driven, ecologically friendly, acceptable and affordable to the resource-poor farmers. To increase food security especially in developing countries, good governance and stable political governance system are emphasized by these studies.

2.2 Measurement of food security

While food security encompasses the four dimensions, the time and cost involved in collecting data on all the dimensions may be prohibitively high. This is evident from previous studies, where different researchers adopt different measures of food security. In estimating the impact of technology adoption on food security, many authors have often used indirect monetary (income and expenditure) and/or production measures (farm production and yields) of food security (e.g., Mason and Smale, 2013; Shiferaw et al., 2008). Other authors have used poverty intensity indexes to measure food security (e.g., Kassie et al., 2012; Kabunga et al., 2014). The use of monetary and production indicators partially captures the impact of the technology on food availability and food access and assumes a causal relationship with food utilization and stability (Magrini and Vigani, 2016).

Other studies that directly estimate the effects of agricultural technologies on household food security in sub-Saharan Africa (SSA) use subjective indicators based on household surveys with self-assessment questions on own-food security status combined with monetary proxies (e.g., Kabunga et al., 2014; Kassie et al., 2012; Shiferaw et al., 2014). The main problem of the subjective approach is not standard (Magrini and Vigani, 2016). Moreover, the presence or absence of particular strategies is often not a standard indicative of food security status. Subjective indicators are also likely to be influenced by measurement errors due to biased self-perception of the respondents of their food security status (Kabunga et al., 2014).

Orewa and Iyangbe (2009) and Bashir et al. (2010) used household calorie consumption method to measure food security. Orewa and Iyangbe (2009) used a 48-hour recall method while Bashir et al. (2010) used a 7-day recall period in obtaining information on the type and quantity of food each household member consumed over the relevant period. The calorie content in each food item consumed was determined and used in estimating the total food intake of the household members. A minimum level of per capita calorie below which a household was considered food insecure was set.

Other measures or indicators of food security include the Household Dietary Diversity Index (HDDI) and the household food insecurity access indicator (HFIAI). HDDI is calculated by summing data on the consumption of 12 food groups (i.e., cereals, roots and tubers, fish, meat, fruits, eggs, vegetables, dairy products, pulses and nuts, oils and fats, sugar, and condiments). The HFIAI score is a continuous measure of the degree of food insecurity (access) in the household in the past four weeks. HFIAI is based on the idea that the experience of food insecurity (access) causes predictable reactions and responses that can be captured and quantified through a survey and summarized in a scale (Coates et al., 2007). These methods are preferred to calorie intake due to the simplicity of survey administration and the fact that they can be used in combination with other measures (Chege et al., 2015b; Coates et al., 2007).

The current study adopts the calorie intake method with a 7-day recall together with HDDI. Per capita calorie intake is the most widely used method of assessing food availability. However, literature points to the intrinsic limitation of this method in assessing calorie intake indicating that it does underestimate calorie intake in that it does not take into consideration the different age and activity levels of the household members and is thus at fault (Claro et al., 2010). However, it is easy and less expensive to calculate thus used in this study.

Food security definition includes food consumption in enough quantity to meet for energy and nutrient requirement which is the main focus of calorie intake. Its error structure is also far well understood than for any other method employed for assessing food security (Chege et al., 2015b). It has thus been used in validating other food security measures. However, it is not without shortcomings, which include possibility of underreporting, logistic complexity and prohibitive cost of survey (ibid.). HDDI is an attractive proxy indicator of food accessibility because obtaining these data is relatively straightforward.

2.3 Determinants of food security

Literature on factors affecting household food security in various developing countries especially in Africa have been documented. These determinants or factors are most often not location-specific (i.e. different determinants were found to influence food security differently in the study areas with some determinants recurring). The study conducted in Nigeria by Oluwatayo (2008) using probit model found out that age, educational level, sex of household head, and income have positive influence on food security whereas household size has negative influence on household food security.

Orewa and Iyangbe (2009) attempted to identify the factors that have major influence on the level of food calorie intakes of rural and low-income urban households in Nigeria using OLS regression analysis. The result revealed a significant positive relationship between daily per capita calorie intake and age, household size, sex, education level and salaried income earners. On the other hand, dependency ratio and non-engagement in farming had a negative influence on daily per capita calorie intake.

Sikwela (2008) documents that fertilizer application, access to irrigation, per aggregate production and cattle ownership have positive effect on household food security in South Africa. The study on the other hand, showed that household size and farm size have negative effect on household food security. Oni et al. (2010) assessed the socio- economic factors affecting smallholder farming and food security in Thulamala, South Africa. The study found out that total income, education level, household own food production, number of people living in a household and spending patterns significantly affected food security.

Babatunde et al. (2007) utilized a three-stage random sampling technique to obtain a sample of 94 farm households in Nigeria. Using the recommended calorie required approach; the study revealed that 36 per cent and 64 per cent of the households were food secure and food insecure respectively. A logit regression model estimated showed that household income, household size, educational status of household head and quantity of food obtained from own production were found to determine the food security status of farming households in the study area.

Determinants identified in the above studies are not identical. Different factors were found to influence food security in different areas. The current study adds to this existing literature, by assessing the factors influencing food security in the Mwala and Kangundo sub-Counties, Kenya.

2.4 Approaches to assess the impact of food security interventions

Impact evaluation aims to establish whether or not an intervention produces its intended effects (AIEI, 2010). One of the most enduring challenges in undertaking impact evaluation is the failure by the evaluator to systematically and objectively gauge what would have happened to the beneficiaries of a program, project or policy in the absence of the intervention, or the so-called the counterfactual problem (Khandker et al., 2010). The problem of evaluation is that while the program’s impact (independent of other factors) can truly be assessed only by comparing actual and counterfactual outcomes, the counterfactual is not observable (ibid.). Therefore, the main challenge in impact assessment is that of finding an appropriate counterfactual.

Two approaches exist to overcome the counterfactual problem in impact assessment. These are the ‘before and after’ and the ‘with and without’ approaches (Gittinger, 1984). The ‘before and after’ approach compares key indicators before and after the intervention (Wainaina et al., 2012). A baseline survey of participants and non-participants is done before the intervention and a follow up survey done after. Statistical methods are then used to assess whether there is a significant difference in the essential variables overtime (Gittinger, 1984). According to Gittinger (1984), the ‘before and after’ comparison fails to account for all the changes that would occur without the intervention and thus leads to an erroneous statement of the benefit attributable to the intervention.

The ‘with and without’ approach, on the other hand, is more comprehensive in its capture of the changes attributable to the intervention (Gittinger, 1984). It compares the behavior of key variables in a sample of beneficiaries of the intervention (or treatment) with their counterparts in non-intervention (or control) group (Wainaina et al., 2012). This approach uses the comparison group as a proxy to gauge what could have happened in the absence of the intervention. It is particularly useful when the baseline is missing (ibid.).

The challenge of using this approach is the tendency of beneficiaries to allocate themselves to one intervention group or the other, which leads to self-selection bias (Khandker et al., 2010). This problem could also arise due to ethical reasons where the program implementer subjectively allocates potential beneficiaries to one intervention group or the other. That is, programs are designed based on the needs of the communities and individuals, who in turn select themselves according to program design and placement (ibid.). Self-selection could be based on observed characteristics, unobserved factors, or both.

In order to overcome the self-selection bias problem, three impact evaluation designs have been proposed in the literature namely experimental, quasi-experimental and non-experimental (Baker, 2000). In experimental or randomized designs, a well-defined sample of beneficiaries is randomly selected into treatment and control groups (ibid.). In this case, the only difference is that the treatment group has access to the program (“treatment” or intervention). When it is impossible to construct treatment and control groups through experimental designs, the quasi-experimental designs are used (ibid.). In this case, comparison groups are generated that resemble the treatment group based on observed characteristics. Non-experimental designs are used when it is impossible to randomly select a control group (ibid.). In such situations, program participants and non-participants are compared using statistical methods.

Depending on the nature of the counterfactual and self-selection bias problems, various econometric techniques are used to undertake impact evaluation. These include reflexive comparison, instrumental variable methods, matching methods and difference-in-difference (DD) methods (Baker, 2000). Reflexive comparison is a quasi-experimental design in which a baseline survey is conducted before and a follow up survey after the intervention.

The counterfactual is constructed on the basis of intervention participants before the intervention. This design is useful in evaluating the full coverage of an intervention where the entire population participates and therefore there is no control group. The major drawback with reflexive comparison method is that the situation of the participants may change due to reasons independent of the intervention (ibid.). In such cases, the method may not differentiate between intervention and external effects leading to unreliable results (Morton, 2009).

The instrumental variable (IV) approach involves the use of at least one variable in the treatment equation as instrument of participation. This also serves as its major limitation since finding such instruments remains a difficult task in empirical analyses (Chege et al., 2015b). The other two limitations of the IV approach include the fact use of instrumental variables that explain little variation in the endogenous explanatory variables can lead to large inconsistencies in coefficient estimates even if only a weak relationship exists between the instrument and the error term in the structural equation (Bound et al., 1995). Secondly, coefficient estimates are biased in the same direction as those produced through the ordinary least squares (OLS) estimator in finite samples (ibid.).

Matching methods include one-to-one matching, radius matching, weighting and sub-classification (Khandker et al., 2010). These methods involve the pairing beneficiaries and non-beneficiaries of an intervention with similar observable characteristics believed to affect program participation (ibid.). During matching, a statistical comparison group is constructed based on a model of the probability of participating in the treatment using observed characteristics (ibid.). The matching only controls for the differences on observed characteristics and there may be some biases resulting from unobserved variables that could affect program participation (ibid.).

The DD method is used on panel or longitudinal data. It entails comparing a treatment with a control group before and after an intervention (Baker, 2000). In this case, the “first difference” constitutes of the difference between the treatment and control groups before the intervention while that after the intervention is the “second difference” (ibid.). Thus, the total difference is the difference between the first and second differences (ibid.). The DD estimator compares program participants and non-participants before and after the intervention (Khandker et al., 2010). The difference of observed mean outcomes for the treatment and control groups is then calculated before and after program intervention (ibid.).

The main advantage of DD is that it removes biases coming from permanent differences between those groups (Kibira et al, 2015). In addition, biases from comparisons over time in the treatment group coming from trends are removed. Thus, the DD method solves the problems arising from non-random selection as well as the non-random placement of program participants (Ravallion, 2005). Time-invariant selection bias has been deemed as the main limitation of DD (Kibira et al, 2015). Despite its shortcomings, DD estimator is intuitively appealing, simple and can be used with panel data (Khandker et al., 2010) as is the case with the current study.

2.5 Studies on the impact of agricultural innovations on food security

Agricultural technologies have a special role in developing countries, boosting production in the agriculture sector, hence driving the overall growth and lowering food prices. While analyzing the potential impact of improved wheat varieties on household food consumption in South eastern Ethiopia, Mulugeta and Hundie (2012) employed a propensity score matching (PSM) method. The authors used a purposive sampling technique on 200 selected farm households. The results showed that the adoption of improved wheat varieties had a positive impact on households’ food availability.

Magrini and Vigani (2016) assessed the impact of new technologies on food security among maize producers in Tanzania. The study selected 543 households were selected using multi-stage, stratified, random sampling. Using matching techniques to estimate impact, the authors found a positive and significant impact on use of improved seeds and inorganic fertilizer on all dimensions of food security. The study reported mixed findings on determinants of food security, for example, household size had positive effect on food security when a household used improved seeds but negative in terms of inorganic fertilizer.

Assessing the impact of improved dairy cow breeds on nutrition in Uganda, Kabunga et al. (2014) employed matching techniques on a random sample of 906 households. The study found out that the adoption of improved dairy cows considerably increased milk yield, household’s milk market orientation, and expenditure on food. In addition, the adoption of improved cow breeds considerably reduced stunted growth amongst children below five years of age. The study used subjective indicators to assess households’ perception of their food security. Despite being cost-effective, subjective indicators are particularly prone to errors especially when long term stability is analyzed.

2.6 Studies on IPM technology

Several studies have been done on adoption and use of integrated pest management strategies. For example, Fernandez-Cornejo et al. (1994), Dasgupta et al. (2004) and Garming et al. (2007) suggest that IPM is a knowledge intensive technology and dissemination of accurate information, to create awareness among farmers, about IPM enhances adoption. Korir et al. (2015) found that education, the number of mature mango trees planted, whether or not a farmer kept records of the mango enterprise, use of protective clothing during spraying, and participation in IPM technology training had a positive influence on the intensity of adoption.

Muchiri (2012) used stratified sampling to select 257 IPM participants and non-participants from the intervention and control areas in Embu County. The study revealed substantial losses in mangoes amounting to KES. 3.2 million per acre due to fruit fly infestation. In addition, 66 percent of respondents were willing to pay KES 1,100 per acre for the IPM mango fruit fly control package. Studies by Isoto et al. (2008), Kibira et al. (2015), Muriithi et al. 2015 and Njankoua et al. (2007) have found that IPM use leads to increase in income.

Kibira et al. (2015) also reports that, on average, recipients of the IPM technology recorded a 55 percent reduction in mango rejection relative to non-recipients. In addition, recipients of IPM spent 46 percent less on insecticides per acre compared to their counterparts. Further, the participants received 22 percent more income than non-participants. These findings are consistent with those of Njankoua et al. (2007), who reported that IPM training had a reduction in the frequency of spraying fungicides and the number of sprayers applied per treatment by 47 and 17 percent respectively in Cameroon.

2.7 Studies using difference-in-difference method

Feder et al. (2004) evaluated the impact of Farmer Field School (FFS) on yields and pesticide use in Indonesia using the DD approach. The data were obtained from a sample of 268 households of which 112 had participated in the training while 156 households had not attended the training. The evaluation considered direct impact on participating farmers and secondary benefits through farmer to farmer diffusion from previous FFS beneficiaries to other farmers. The study found no significant differences in performance between FFS graduates and exposed farmers in terms of pesticide use and yields.

Omilola (2009) estimated the impact of improved agricultural technology on poverty reduction in Nigeria using double difference approach. A multistage random sampling approach was used to select a total of 200 adopters and 200 non adopters for the study. The analysis showed that participants received statistically significant and higher increases in agricultural income than non-participants. Non-adopters had larger changes in other income sources than adopters. The overall findings revealed that the differences between the adopters and non-adopters’ poverty status of the new technology were fairly small, demonstrating that the adoption of agricultural technology did not considerably translate to poverty reduction for its adopters.

Yamano and Jayne (2004) used the DD approach to assess the impact of mortality of the working age group on crop production of small-scale farmers in Kenya. The study used a two-year panel of 1,422 randomly selected Kenyan households surveyed in 1997 and 2000. The findings indicated that: the effects of death of an adult on crop production was sensitive to age, gender and position of the deceased; death of a working male household head greatly affected household off-farm income negatively; households coped with the death of a working adult by selling particular types of assets.

2.8 Summary

Based on the reviewed literature, many authors have used different methods to measure food security. The determinants of food security are not location specific which emerges as a gap. On the other hand, many methods have been used to assess impact of IPM technologies including PSM, IV and DD. The DD estimator is intuitively appealing, simple and can be used with panel data. Although a few studies have been undertaken on impact of IPM technologies on food security, none has been done in Machakos County. To fill this gap, the current study assessed the impact of mango fruit fly IPM technology on food security in the county using the DD method.



3.1 Analytical framework

The effect of a technology on household food security is transmitted through three main linkages; (i) reallocation of farm resources between enterprises as a result of technology adoption, (ii) changes in household income, and (iii) changes in food consumption patterns as a result of changes in the income derived from the proceeds of technology adoption (von Braun, 1988). The technology impacts the profits derived from increasing farmers’ knowledge on a technology.

The second link is through possible changes in household income. Kibira et al. (2015) and Muriithi et al. (2016) have shown that an agricultural technology can cause significant income gains. Higher incomes improve the economic access to food, which may result in higher calorie consumption, especially in previously undernourished households (third link). Moreover, rising incomes may contribute to better dietary quality and higher demand for more nutritious foods, including vegetables, fruits, and animal products. When technological change raises income and income raises food consumption, the positive effects of this change can be identifiable. The relationships are, however, not straight forward (von Braun, 1988).

3.2 Empirical framework

3.2.1 Objective 1: Characterizing mango farmers in Mwala and Kangundo sub-Counties

To achieve objective one descriptive statistics is used. Socio-economic characteristic differences between adopters and non adopters are tested using t test for differences of the means.

3.2.2 Objective 2: Assessment of impact of IPM technology on food security

This is achieved in two stages. The first stage measures the food security status of households while in the second stage assesses the impact of IPM on food security. Food security is measured using two methods namely, per capita calorie intake and household dietary diversity index (HDDI). Based on the average dietary energy requirement in Kenya of 2,250 kcal per adult equivalent, households are categorized as either food secure or insecure as used by the Kenya National Bureau of Statistics.

To assess impact of IPM on food security the difference in difference (DD) method is used. The DD estimator for per capita calorie intake (Yi), a continuous covariate, is estimated with ordinary least squares (OLS) (Omilola, 2009). On the hand, a truncated poison regression is estimated to assess the impact of fruit fly IPM on Household Dietary Diversity Index (HDDI) a measure dietary quality. The higher the diversity index so is the quality of diet and vice versa. Measures of food security adopted in this study

Household food security is measured using (i) per capita calorie intake and (ii) Household Dietary Diversity Index (HDDI) following Hoddinott and Yohannes (2002).

a) Per capita calorie intake

The calorie intake is estimated from data collected through a 7-day recall of consumption of all significant sources of calories consumed in the household. The household member that prepared the food or another adult who was present and ate the food in the household during the 7 days is

2011). An increase in the average number of different food groups consumed provides a quantifiable measure of improved household food access (Swindale and Bilinsky, 2006). Justification for inclusion of independent variables

The independent variables chosen for the empirical model are based on previous empirical review on technology adoption and food security interlinkage studies mentioned in Chapter two. Table 3.2 presents the descriptions and expected signs of the variable used in the model.

Household Head Age: The age of household head is expected to impact on his or her labour supply for food production (Babatunde et al., 2007). Young and energetic household heads are expected to cultivate larger farms compared to the older and weaker household head. Age is measured by the years of the household head. The square of age is included in the model as result of the nonlinear relationship between age and food security. Age is hypothesized to be positively associated with the quantity and quality of food consumed by households in Machakos County, Kenya.

Household Head Education: The education level determines the number of opportunities available to enhance livelihood strategies, improve food security and reduced poverty levels (Amaza et al., 2009). It is hypothesized that the more the years of education of the household head the better the food security situation of the household. This is because education is positively attributed to uptake of improved technology, improved managerial capacity even at the farm level and more probability of off farm employment opportunities either self-employment or otherwise (Pankomera et al., 2009). Education is measured by the number of years of formal schooling completed by the household head. The current study hypothesizes education to be positively related with food security.

Experience. Refers to the number of years the household head has been engaged in mango farming activities. It’s expected that an experienced household head to have more insight and ability to diversify his or her production to minimize risk of food shortage. Research findings by Feleke et al. (2003) and Oluyole et al. (2009) have shown a positive relationship between food security status and farming experience. This variable is measured as number of years that the household head has been practicing mango production. The expected sign for experience on food security is positive.

Gender: Gender of household head looks at the role played by the individuals in providing households’ needs including acquisition of food. Kassie et al. (2012) have documented an increased food security of male headed households compared to female headed household stating that female headed households are mostly single parented and have limited access to productive resources. Gender of the household head is a dummy variable taking 1 if the household is a man headed and 0 if a woman. In this study, Gender is hypothesized to be positively related to the food security of households

Household size: Household size determines the amount of labor available for farm production, farm produce kept for own consumption, and agricultural marketable surplus of farm harvest (Amaza et al., 2009). Households with large family members are mostly associated with a high dependency ratio and more food requirements, depicting a negative effect on food security. However, an increase in a household size could translate to an increase in the number of income earning adults depicting a positive effect on food security (Iyangbe and Orewa, 2009). Therefore, the expected sign for household size can be either positive or negative. This variable is measured as number of people living in the household. Household size is expected to have either a positive or negative effect on household’s food security

Group membership: Agricultural groups provide social network platforms within which participants share new information and experiences such as IPM strategies and proper pesticides use. Group membership also increases farmers bargaining power in terms of credit and market access. Belonging to a group also acts as a form of social capital which Martin et al. (2004) found to be significantly positively associated with food security. Sseguya (2009) found that households that had membership in one or more groups were more food secure. The dummy variable takes the value of 1 if the respondent is a member of a group and 0 if not a member. Group membership is expected to be positively associated with food security.

Farm income: It improves access to food for those who earn the income. The higher the income, the higher the expected per capita calorie intake and the more diverse a household diet is expected to be. Anderson (2002) found a positive impact between farm income and food security. In this study, annual farm income is hypothesized to be positively associated with household’s economic access of food in the Machakos County, Kenya. Farm income is a continuous variable measured in KES.

Access to extension service: Field extension officers are important in dissemination of improved technology. It is important that the contact between extension officers informing on the innovation and farmers occurs before the adoption in order to avoid any reverse causality problem. Kassie et al. (2012) and Lewin (2011) found that government investment in agricultural extension has a significant impact in food security status. Lewin (2011) found that at least one visit to each household from an agricultural extension agent during each cropping season would reduce food insecurity by 5.2 percent. The dummy variable takes the value of one if the farmer had accessed formal agricultural extension services and zero if they did not

Tropical Livestock units owned: Livestock play a number of roles which include; income generation, provision of inputs and providing a buffer against environmental and economic shocks (FAO, 2009). Livestock act as a source of food for instance, milk, eggs and meat and can also be considered as assets thus a form of wealth indicator. Animals provide manure and are used as a form of traction hence increasing output. Households with more livestock units are expected to have more per capita calorie intake and diverse diets. The tropical livestock unit is commonly taken to be an animal of 250kg live weight (Jahnke, 1982).

Distance to the Agricultural market: Long distances to the market centre and input shops translate to high transport and fare paid by farmers, most importantly when sourcing important inputs for farming. Longer distances discourage farmers from visiting markets frequently hence less likely in getting market information (Staal et al., 2002; Fekele et al., 2003; Matchaya and Chilonda, 2012). Hence farmers may sell their produce at times when prices are low and buy when prices are high. It is hypothesized that distance to the market is negatively related to food security. The variable is measured in kilometers (KM) between the respondent’s farm and the mango inputs market.

Farm size: This is the logarithm of the household land cultivated under mango. It is hypothesized that as the size of the farm increases, the level of food production increases as well. Mwanaumo et al. (2005) and Deininger (2003) establishes a positive relationship between farm size and food security. Larger land sizes are associated with more mango produced. Increase in mango production is hypothesized to increase income available for household to purchase of food. Therefore, the expected effect of farm size on food security is positive. The area under mango production is measured in acres.

(b) Heteroscedasticity

Heteroscedasticity occurs when the variance of the error term is non-constant in which case the OLS estimator, although still unbiased, is inefficient and the hypothesis tests are not valid (Wooldridge, 2002). If present in the data the estimates will not be the Best Linear Unbiased Estimates (Gujarati, 2009). In this study, the Breusch-Pagan/Cook-Weisberg test was used to test for heteroscedasticity under the null hypothesis of a constant variance (homoscedasticity). According to Coenders and Saez (2000), a significant parameter estimate of the Breusch-Pagan/Cook-Weisberg test leads to the rejection of the null hypothesis of homoscedasticity.

(c) Autocorrelation

Autocorrelation occurs when members of series of observations ordered in time are correlated (Gujarati, 2012). It is a violation of the assumption that the size and direction of one error term has no bearing on the size and direction of another. This results to inefficient estimation (Gujarati, 2012). This study used panel data which can be prone to autocorrelation.

3.4 Study Area

The study was conducted in Machakos County, which is ranked fourth in terms of mango production in Kenya. Mwala and Kangundo sub-counties (Figure 3.1) have been specifically selected by the African Fruit Fly Programme in Kenya. In Mwala sub-County, the study was conducted in three wards (Mwala, Mbiuni and Miu) while in Kangundo sub-County four wards were selected (Kangundo North, Kangundo Central, Kangundo South and Kangundo East). Mwala sub-County has a population of 89,211 persons and covers an area of 1017.9 km2 (Machakos County Intergrated Development Plan [CIDP], 2015). The local climate is semi-arid (average annual rainfall of 500mm with an average altitude of 1400 meters above sea level.

Kangundo sub-County has a total area of 177.2 km2 and lies at an average altitude of 1555 meters above sea level (Machakos CIDP, 2015). According to 2009 national population and housing census, Kangundo sub-County had 94,367 persons. Temperature in Kangundo ranges between 12oC and 28oC annually while the average annual rainfall is 958 mm (ibid.). The main economic activities/industries include dairy farming, beekeeping, trade, limited coffee, eco-tourism, businesses and manufacturing. The primary agricultural products in Mwala and Kangundo sub-counties include mangoes, maize, pawpaws, watermelons, beans, cow peas, pigeon peas, lentils and livestock.

3.5 Data sources and Sampling procedure

The study used primary data collected from mango farmers using semi-structured questionnaire (Appendix 1). Information on farmer demographics, socio-economic characteristics, mango production and marketing and food security indicators was collected. Secondary data were obtained from government data sources such as MoA, HCDA, journals and sessional papers, previous studies and internet sources. Data on the acreage of mango production and the volume of marketing for previous years and volume conversion rates used in the areas were obtained from the sub-county agricultural offices in Mwala and Kangundo, Machakos.

The study used a stratified sampling procedure to select the farmers to be interviewed. All the mango farmers in Machakos County constituted the study population. Mwala and Kangundo sub-counties were purposively selected on the basis of being the leading mango producing areas in Machakos County. Because ICIPE had implemented the mango fruit fly IPM project in Mwala sub-county, it was designated as the treatment site while Kangundo constituted the control area. The sample size each of the two study sites was calculated using the Cochran sample size formula for continuous data (Bartlett et al., 2001):

A structured questionnaire (Appendix 1) was administered to 600 sampled mango producers in their farms; 300 IPM control package participants (intervention group) and 300 non-participants (control group), from the selected sub-counties. Prior to questionnaire administration, the enumerators were trained and the tool pre-tested in Embu County. Data were collected in two scenarios; ‘before’ and ‘after’ the IPM control package intervention. A baseline survey was undertaken in both study sites in February and March 2015 to collect baseline information on the 600 households on mango production during the 2014 growing season. After the baseline survey, farmers in Mwala sub-County were trained on how to apply the IPM technology on their mango orchards.

They were then given the various components of the IPM. A follow-up survey targeting the same households was undertaken in December 2015 to capture information on IPM technology used during the 2015 mango season. During this follow-up survey, 4 percent of the 600 households were not readily available for interviewing. Hence, the sample size dropped to 566 households of which 289 were in the treatment site (Mwala sub-County) and 277 were in the control site (Kangundo sub-County). A final sample for the analysis was 1147 households including 588 IPM farmers and 559 control farmers.

As is the case in many household surveys, the current study encountered a few problems during the data collection process. In a few cases, the respondents were unwilling to respond to certain questions such as income and asset value. Most households in the small farm sector do not keep written records of their transactions. Hence, most of the answers given were based on recalls. But overall, the survey went on smoothly and without any major problems.



4.1 Socio-economic characteristics of mango farmers in Mwala and Kangundo sub-Counties

The socio-economic characteristics of mango farmers in Mwala and Kangundo sub-Counties are presented in Table 4.1. Half of the mango farmers in Mwala were in the 41-60 year age bracket while 45 percent of sampled households in Kangundo were in this age bracket. The IPM participants (Mwala) had significantly a lower average age of 58 years while non-participants (Kangundo) had mean age of 61 years (p<0.05).

In Kangundo sub-County, the household heads’ average number of years of formal education was 10 years, which was significantly higher than Mwala’s 9 years (p<0.05). This literacy level would imply that mango farmers are likely to synthesize information and appreciate the new technology. Education enables farmers to interpret and respond to new information faster than those without education (Kibira et al, 2015).

Eight seven percent of the households in IPM adopters and non adopters are male headed (Table 4.1). However, there was no significant difference in gender between Mwala and Kangundo sub-Counties (p>0.05). The average household size was 5 people among the sampled groups. The average number of years of experience in mango farming in Mwala was 14 years while that in Kangundo was 12 years with significant difference between the two (p>0.05). Experienced household heads have more insight and ability to diversify their production to minimize risk (Feleke et al., 2003).

Three quarters of the mango farmers did not personally seek advice or assistance on mango production from extension service providers (Table 4.1). However, they consulted during organized training fora such as field days, demonstrations, seminars and workshops. The number of times participants and non-participants attended such events was 97 percent and 21 percent respectively, with significant difference between the two groups (Table 4.1). Extension officers are important in dissemination of improved agricultural technologies and also provide marketing information (Lewin, 2011).

The average number of Tropical Livestock Units (TLUs) was 2.5 and 2.7 in Mwala and Kangundo sub-counties respectively (Table 4.1). The main livestock species reared in the two counties were cattle, goats, sheep, poultry, donkey, rabbit and pigs. These livestock were used as food and non-food sources such as manure, animal traction and transportation. On average, mango farmers in Kangundo sub-County traveled significantly longer distances (10 km) to the market compared to those in Mwala sub-County (5 km). Access to input and output markets is known to increase the uptake of new agricultural technologies in rural areas of Africa (Asfaw et al., 2012).

IPM participants and non-participants had statistically similar acreages of land of 4.21 and 4.29 respectively. However, on average, IPM participants allocated significantly more land to mango production than non-participants at 1.1 and 0.75 acres respectively (p<0.05). At KES 46,533 in Mwala and KES 28,640 in Kangundo, the average annual farm income between IPM participants and non-participants was not significantly different (p>0.05). Farm income enables farmers to procure farm inputs necessary for mango production.

Overall, most (97 percent) of the respondents had no access to credit specifically targeted to mango production. However, significantly more IPM participants (5 percent) than non-participants (1 percent) had access to credit. Majority of those who did not access credit expressed fear of default due to unreliable and unstreamlined mango marketing system as the reason for their unwillingness to go for credit. Access to credit has been shown to increase farmers’ purchasing power thus enabling them to procure farm inputs and cover operating costs (Guirkinger and Boucher, 2005; Eswaran and Kotwal, 1990; Komicha and Öhlmer, 2007).

4.2 Food security situation in the study sites

Table 4.2 presents the average per capita food intake and the household dietary diversity indices for the two study sites during the baseline and follow-up survey. On average, the per capita food intake was higher in Kangundo at 3,007 Kcal/day than in Mwala at Kcal 2,840/day during the baseline survey. This shows that they were above the standard average dietary energy requirement for Kenya cutoff of 2250 Kcal (as used by the (Kenya National Bureau of Statistics)). This suggests that households in the two study areas were food secure.

Based on the means presented in Table 4.2, 72 and 67 percent of survey respondents in Mwala sub-County were food secure during the baseline and follow up surveys respectively as they exceeded recommended per capita calorie intake of 2250 kcal. In Kangundo sub-County, 81 percent and 75 percent of the respondents were food secure during the baseline and follow up respectively. With regard to HDDI, households in Mwala sub-County had almost similar averages between the baseline and follow-up surveys (Table 4.2). A similar pattern is repeated in Kangundo sub-County. Overall, there was no difference in the dietary diversity scores between IPM participants and non-participants.

4.3 Impact of mango fruity fly IPM technology on food security

4.3.1 Model Diagnostic Tests

Before estimating the factors influencing food security situation by use of regression analysis, preliminary tests were carried out on the data. The tests included; multicollinearity, heteroscedasticity and autocorrelation. To check for the presence of multicollinearity problem among the independent variables the Variance Inflation Factor (VIF) was computed. The results of the VIF for the variables included in all the models were less than 10 (Appendix 3) and the pairwise correlations were less than 0.7 (Appendix 4), hence no independent variables were dropped from the estimated model.

To test for heteroscedasticty, the Breusch-Pagan was used. As shown by the results in Appendix 8, the Breusch-Pagan/Cook-Weisberg test was not statistically significant (p=0.512), implying that heteroscedasticity was not a problem in the dataset. Autocorrelation test (actest) in Stata presented in Appendix 5 detected the presence of autocorrelation in the data (p<0.00). Iterative Prais-winsten method was used to correct for autocorrelation. The Prais-Winsten estimation procedure takes into account serial correlation of type Autoregressive (1) in a linear model (Prais and Winsten, 1954). The procedure is an iterative method that recursively estimates the beta coefficients and the error autocorrelation of the specified model until convergence of rho, i.e. the AR(1) coefficient, is attained (Wooldridge, 2013).

4.3.2 Impact of IPM on per capita calorie intake

Table 4.3 presents the estimate of the difference-in-difference per capita calorie intake in the two study areas derived from equation (3.6). As shown, the average difference in per capita calorie intake was negative in each group of respondents for baseline and subsequent survey; i.e., -109 and -164 Kcal/person/day in Mwala and Kangundo sub-counties respectively (Table 4.3). The difference in per capita calorie intake was negative among the two groups of respondents during baseline and follow-up surveys; i.e., -112 and -167 Kcal/person/day in Mwala and Kangundo sub-counties respectively (Table 4.3). The reduction in per capita calorie intake in follow up period can be attributed to dry spell in the area during the study period leading to decreased food availability.

Table 4.3: Average IPM technology effect on per capita calorie intake among mango farmers in Mwala and Kangundo sub-Counties, Kenya

The difference in per capita calorie intake between IPM participants and non-participants during baseline and follow up survey was positive 55 Kcal/person/day. This total difference (or DD) indicates that, on average, IPM participants received only 1.93 percent more per capita calorie intake than their counterparts. This suggests that the mango fruit fly IPM technology contributed a small but positive increase in per capita calorie intake among IPM participants in Mwala sub-County. Unconditional treatment effect of IPM technology on per capita calorie intake

Table 4.4 presents the results of unconditional treatment effect of adopting the mango fruit fly IPM technology by fitting equation (3.13) into the data using OLS. The unconditional treatment effect was evaluated for the sole purpose of assessing the impact of IPM on a strong assumption that the IPM users and nonusers have no other differences apart from the fact that the former adopted the new technology. The coefficient of the unconditional treatment effect of IPM (Time*IPM) was positive but not statistically significant (p>0.05) (Table 4.4).

Table 4.4: OLS parameter estimates of unconditional effect of IPM technology on per capita calorie intake among mango farmers in Mwala and Kangundo sub-Counties, Kenya Conditional treatment effect of IPM technology on per capita calorie intake

Table 4.5 present the OLS parameter estimates of the conditional effect of IPM technology on per capita calorie intake using equation (3.14). The coefficient of the conditional treatment effect of IPM (IPM*Time) is positive and statistically significant implying that adoption of IPM technology led to an increase in per capita calorie intake among survey households. Hence, the second hypothesis that IPM had no impact on per capita calorie intake in Mwala and Kangundo sub-counties was rejected. Suggesting that the technologies lead to an increase in food availability.

Farm income, access to extension services, wealth category and distance to agricultural input market had a positive and significant impact on per capita calorie intake. On the other hand, household size had a negative effect. An additional household member was associated with a 171 Kcal decline in per capita intake. This could be due to the fact that households with many members are mostly associated with a high dependency ratio and more food requirements, depicting a negative effect on food security, ceteris paribus. This finding is consistent with Goshu et al. (2013)’s who observed that family size was negatively related to food security in rural Ethiopia.

Contrary to the a priori expectation, an additional kilometer in distance to agricultural input markets increased the per capita calorie intake by 14 Kcal, ceteris paribus. This can be a case that households that travelled long distances to agricultural input market consumed more calories from own production and gifts as compared to purchases (Tembo and Simtowe, 2009). This finding is however inconsistent with available literature (see e.g., Staal et al., 2002; Fekele et al., 2005; Matchaya and Chilonda, 2012), that suggests long distances to input markets reduces the amount of food consumed.

Access to agricultural extension had a positive impact on per capita calorie intake. Thus, holding other factors constant, a shift from no access to agricultural extension increased the per capita food intake by 164 Kcal. This finding corroborates those of Kassie et al. (2012) and Lewin (2011) who reported that government investment in agricultural extension has a significant impact in food production and subsequently food security. Agricultural extension services provide farmers with important information, such as patterns in food prices, new technologies, crop management, and marketing. Such information is intended to increase households’ ability to increase food production or increased income which in turn increase consumption levels (per capita calorie intake).

Ceteris paribus, a shift from not wealthy to a moderate wealth category had a positive and significant effect on per capita calorie intake. Thus, a shift from not wealthy to a moderate wealth category led to a 166 Kcal increase in the household per capita calorie intake. Additionally, a movement from not wealthy to a wealthy category increased the per capita calorie intake by 188 Kcal, all else being equal. Wealthy households do not face entry barriers in access to markets and subsequently food access due to high levels of physical and financial assets (Holloway et al., 2001).

4.3.3 Impact of IPM technology uptake on household dietary diversity index

The HDDI was lower during the follow-up survey than during the baseline (Table 4.6). The country experienced a dry spell during the reference period which affected the different varieties of food accessible in the market. Accordingly, the difference in HDDI during the two periods was negative for both IPM participating and non-participating households. Across time, the difference was small but positive. Hence, the total difference in HDDI between IPM participants and non-participants was only 0.001 (or 0.01 percent)3 across time (Table 4.6).

Table 4.6: Difference in Difference (DD) estimate of average IPM technology effect on HDDI among mango farmers in Mwala and Kangundo sub-Counties, Kenya Unconditional treatment effect of IPM technology on household dietary diversity index

The coefficient of the unconditional treatment effect of IPM technology (IPM*Time) on HDDI not statistically significant (p>0.05) (Table 4.8). This could be explained by the fact that the household dietary diversity behavior adjusted only slightly because income was subjected to temporal variability (Chege et al., 2015a). This slight income increments leads to households diversifying food within groups and not between groups.

Table 4.7: Marginal effects of unconditional effect of IPM technology uptake on HDDI among mango farmers in Mwala and Kangundo sub-Counties, Kenya Conditional treatment effect of mango fruit fly IPM technology on household dietary diversity index

Controlling for possible influences in the conditional effects model did not improve the results (Table 4.9). Hence, the coefficient of the conditional treatment effect of IPM technology (IPM*Time) was not statistically significant (p>0.05). This suggests that the income benefits of IPM technologies do not necessarily translate into nutritious diets. Thus, the increased food consumption reported earlier is related to availability, and not diversity of food. In fact, the focused group discussions indicated that a large share of the expenditure on food was devoted to cereal staples such as maize, wheat and rice.

Table 4.8: Marginal effects of conditional effect of IPM technology on HDDI among mango farmers in Mwala and Kangundo sub-Counties, Kenya

Because of lack of effect of IPM technology uptake on HDDI (the outcome of interest), all the other regressors, some of which were statistically significant, were not relevant to this study and warrant further discussion.



5.1 Summary

Mango production is a major source of income for both medium and small-scale farmers in Kenya. However, it is confronted with a major threat of fruit fly infestation that causes reduction of quality and quantity of marketable fruit and hence considerable produce losses. As a result, the country’s horticultural industry loses out on huge revenues that could be derived from higher trade volumes in local urban and export markets. In addition, the increased use of pesticides in the effort to reduce fruit losses has led to a rise in production costs. Use of Insecticides has been shown to be ineffective in controlling the fruit flies.

This study evaluated the impact of integrated pest management (IPM) technology for mango fruit fly control on food security among smallholder mango producers in Machakos County using a difference-in-difference (DD) method. The study found both IPM participants and non-participants to be food secure with per capita calorie intake above the 2250 Kcal threshold for Kenya. In Mwala sub-County, 72 and 67 percent of survey respondents in were food secure during the baseline and follow up surveys respectively while, 81 percent and 75 percent of the respondents were food secure in Kangundo sub-County during the baseline and follow up respectively

Although there were disproportionately more food insecure households among the participants than the non-participants both before and after technology adoption, the participants fared slightly better than the non-participants in terms of food insecurity reduction. The DD method shows that IPM had a positive impact on per capita calorie intake. Farm income, access to extension services, wealth category and distance to agricultural input markets positively influenced the per capita calorie intake. On the other hand, household size had a negative effect. The Poisson model found that IPM had no impact on HDDI, implying IPM does not lead to increased food diversification.

5.2 Conclusion

This study found that uptake of mango fruit fly IPM technology control has a positive influence on household food security and therefore, it concludes that scaling up the mango fruit fly IPM technology could be an option to improve the welfare of rural communities constrained by mango fruit fly infestation. However, the uptake of mango fruit fly IPM technology does not improve household dietary diversity. This could be as a result that an increase in income from mango marketing wasn’t enough for households to diversify their food. The per capita calorie intake a measure of food availability could be improved by increasing farm income and wealth category and also access to extension services.


  1. The study found out that high farm income and wealth status improves households’ food consumption. Hence, policies promoting income and wealth generation such as value addition and group marketing among mango producers should be emphasized.
  2. Improving access to extension services may enhance adoption of IPM. The current extension services are faced with many challenges which include: inadequacy and instability of funding, poor logistic support for field staff, use of poorly trained personnel at local level, ineffective agricultural research extension linkages, insufficient and inappropriate agricultural technologies for farmers, disproportionate Extension Agent: Farm Family ratio. Hence, policies addressing the above-mentioned challenges should be encouraged.


Ahmed, A. U., Rabbani, M., Sulaiman, M., & Das, N. C. (2009). The impact of asset transfer on livelihoods of the ultra poor in Bangladesh. BRAC, monograph Series, 39.

AIEI [African impact evaluation initiative] (2010). Impact evaluation methods. Available at (Accessed August 1, 2016).

Amata, R., Otipa, M., Waiganjo, M., Wabule, M., Thuranira, E., Erbaugh, M., and Miller, S. (2009). Incidence, prevalence and severity of passion fruit fungal diseases in majorproduction regions of Kenya. Journal of Applied Biosciences, 20, 1146–1152.

Amaza, P., Abdoulaye, T., Kwaghe, P., Tegbaru, A. (2009). Changes in household food security and poverty status in PROSAB area of Southern Borno State, Nigeria. International Institute for Tropical Agriculture IITA Promotion of Sustainable Agriculture Borno State PROSAB IITA Niger.

Anderson, A. (2002). The effect of cash cropping, credit and household composition on household food security in southern Malawi. African Studies 6, 175–202.

Asfaw. S, Kassie, M., Simtowe, F. & Leslie Lipper, L. (2012). Poverty Reduction Effects of Agricultural Technology Adoption: A Micro-evidence from Rural Tanzania. Journal of Development Studies, 48(9): 1288-1305.

Arimond, M., Ruel, M.T. (2004). Dietary diversity is associated with child nutritional status: evidence from 11 demographic and health surveys. Journal of Nutrition 134, 2579–2585.

Babatunde, R. O., Omotosho, O. A. and Sholotan, O.S. (2007). Factors Influencing Food Security Status of Rural Farming Households in North Central Nigeria. Agricultural Journal, 2(3): 351-357.

Baker, J.L. (2000). Evaluating the impact of development projects on poverty: A handbook for practitioners. World Bank Publications.

Bartlett, J.E., Kotrlik, J.W., and Higgins, C.C. (2001). “Organizational Research: Determining Appropriate sample Size in Survey Research”. Information Technology, Learning, and Performance Journal, 19(1), 43-50 (Spring).

Barrett, C. B. (2010). Measuring food insecurity. Science, 327(5967), 825-828.

Bashir, M. K., Naeem, M. K., & Niazi, S. A. K. (2010). Rural and peri-urban food security: a case of district Faisalabad of Pakistan. World Applied Sciences Journal, 9(4), 403-411.

Bound, J., Jaeger, D. A., & Baker, R. M. (1995). Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variable is weak. Journal of the American statistical association, 90(430), 443-450.

Caliendo, M. and Kopeinig, S. (2008). Some Practical Guidance for the Implementation of Propensity Score Matching. Journal of Economic Surveys, Vol. 22, (2008), 31–72.

Chege, C.G., Andersson, C.I., Qaim, M. (2015a). Impacts of supermarkets on farm household nutrition in Kenya. World Development. 72, 394–407.

Chege, J.W., Nyikal, R.A., Mburu, J., Muriithi, B.W. (2015b). Impact of Export Horticulture Farming On Per Capita Calorie Intake Of Smallholder Farmers In Eastern And Central Provinces In Kenya. International Journal of Food Agricultural Economics. 3, 65.

Claro, R. M., Levy, R. B., Bandoni, D. H., & Mondini, L. (2010). Per capita versus adult-equivalent estimates of calorie availability in household budget surveys. Cadernos de Saúde Pública, 26(11), 2188-2195.

Coates, J., Swindale, A., & Bilinsky, P. (2007). Household Food Insecurity Access Scale (HFIAS) for measurement of food access: indicator guide. Washington, DC: Food and Nutrition Technical Assistance Project, Academy for Educational Development, 34.

Coenders, G. and Saez, M. (2000). Collinearity, heteroscedasticity and outlier diagnostics in regression. Do they always offer what they claim? New Approaches in Applied Statistics, 16:79-94.

Connell, J. P., & Kubisch, A. C. (1998). Applying a theory of change approach to the evaluation of comprehensive community initiatives: progress, prospects, and problems. New approaches to evaluating community initiatives, 2(15-44), 1-16.

County Integrated Development Plan (CIDP). (2015). Machakos County Integrated Development Plan, 2015. Available at (Accessed June 18, 2017).

Dasgupta, S., Meisner, C., and Wheeler, D. (2004). Is environmentally friendly agriculture less profitable for farmers? Evidence on integrated pest management in Bangladesh.Applied Economic Perspectives and Policy, 29(1): pp 103-118.

Deininger, K. W. (2003). Land policies for growth and poverty reduction. World Bank Publications.

Ekesi, S., Billah, M.K. (2007). A Field Guide to the Management of Economically Important Tephritid Fruit Flies in Africa, ICIPE Science Press, Nairobi. ISBN 92 9064 209.

Ekesi, S., Mohamed, S., Hanna, R. (2010). Rid fruits and vegetables in Africa of notorious fruit flies. CGIAR SP-IPM Technical Innovation Brief 4.

Ekesi, S., Mohamed, S., Tanga, C.M., 2014. Comparison of food-based attractants for Bactrocera invadens (Diptera: Tephritidae) and evaluation of mazofermespinosad bait spray for field suppression in mango. Journal of Economic Entomology 107, 299-309.

Eswaran, M. and Kotwal, A. (1990). Implications of credit constraints for risk behavior. Oxford Economic Papers, 42(2):473-482.

FAO. (1996). World Food Summit Plan of Action. Available at (Accessed June 06 2017).

FAO. (2009). Value chain analysis of the tropical fruit subsector: The case of mango production, processing and trade in Kenya. Rome: FAO.

FAO, IFAD, UNICEF, WFP and WHO. (2017). The State of Food Security and Nutrition in the World 2017. Building resilience for peace and food security. Rome, FAO.

Fayeye, T.R. & Ola, D.J. (2007). Strategies for Food Security and Health Improvement in the Sub-Saharan Africa. World Journal of Agricultural Sciences, 3(6):808-814.

Feder, G., Murgai, R., & Quizon, J. B. (2004). Sending farmers back to school: The impact of farmer field schools in Indonesia. Applied Economic Perspectives and Policy, 26(1), 45-62.

Feleke, S., Kilmer, R. L., & Gladwin, C. (2003). Determinants of food security in southern Ethiopia. A selected paper presented at the American Agricultural Economics Association Meetings in Montreal, Canada.

Fernandez-Cornejo, J., Beach, E. D., and Huang, W. Y. (1994). The adoption of IPMtechniques by vegetable growers in Florida, Michigan and Texas. Journal ofAgricultural and Applied Economics, 26(01), 158-172.

Financial Sector Deepening (FSD). (2015). 2014 annual report. Nairobi, Kenya: FSD Kenya. Available at (Accessed January 23, 2016).

Garming, H., and Waibel, H. (2007). Do farmers adopt IPM for health reasons? The case of Nicaraguan vegetable farmers. In Proceedings of the Tropentag Conference Utilisation of Diversity in Land Use Systems: Sustainable and Organic Approaches to Meet Human Needs: pp. 9-11.

Government of Kenya [GoK]. (2010) Agricultural Sector Development Strategy 2010–2020. Government Printers, Nairobi. Available at May 21, 2016)

Government of Kenya (GoK). (2012). Kenya’s LAPSSET Corridor Mango Production Investment Opportunity. Government Printers, Nairobi. (Accessed February 6, 2016)

Goshu, D., Kassa, B., Ketema, M. (2013). Is food security enhanced by agricultural technologies in rural Ethiopia? African Journal of Agricultural Resource Economics,8, 58–68.

Greene W. H. (2007). Econometric Analysis, 6th edition. Prentice Hall, NJ. 1216 pp.

Griesbach, J. (2003). Mango growing in Kenya. World Agroforestry Centre (ICRAF), Nairobi.

Gross R.W Schultink and A.A Kielmann (1999). Community Nutrition: Definition and Approaches in M. Sadler, J.J. Strain and B. Caballero (Eds.) (1999), Encyclopedia of Human Nutrition, Academic Press Ltd., London, U.K, pp. 433-441.

Guirkinger, C. and Boucher, S. (2005). Credit constraints and productivity in Peruvian agriculture. Department of Agricultural and Resource Economics, University of California-Davis. Available at (Accessed May 30, 2017).

Gujarati, D. N. (2009). Basic econometrics. Tata McGraw-Hill Education.

Gujarati, D.N. (2012). Basic econometrics. Tata McGraw-Hill Education.

Heckman, J. J. (1979). Statistical models for discrete panel data. Chicago, IL: Department of Economics and Graduate School of Business, University of Chicago.

Hoddinott, J., & Yohannes, Y. (2002). Dietary diversity as a food security indicator. Food consumption and nutrition division discussion paper, 136(136), 2002.

Horticultural Crops Development Authority (HCDA) (2010): National Horticulture Validated Report 2009. Available at (Accessed February 18, 2014).

Horticultural News (2014). Resistance management through IPM.May-June, 2014. No.33

Holloway, G.J., Barrett, C.B. and Ehui, S.K. (2001). The Double Hurdle Model in the Presence of Fixed Costs Applied Economics and Management Working Paper, Cornell University.

ICIPE. (2009). Biennial report highlights, 2008-2009. ICIPE Science Press.

Irungu, P., Omiti, J. M., & Mugunieri, L. G. (2006). Determinants of farmers’ preference for alternative animal health service providers in Kenya: a proportional hazard application. Agricultural economics, 35(1), 11-17.

Isoto, R.E., Kraybill, D.S., Erbaugh, M.J. (2008). Impact of integrated pest management technologies on farm revenues of rural households: The case of smallholder Arabica coffee farmers. African Journal of Agricultural Resource Economics. 9, 119–131.

Iyangbe, C., Orewa, S. (2009). Determinants of daily protein intake among rural and low-income urban households in Nigeria. American-Eurasian Journal of Science Resource. 4, 290–301.

Jahnke, H. E. (1982). Livestock production systems and livestock development in tropical Africa (Vol. 35). Kiel: Kieler Wissenschaftsverlag Vauk.

Jones, A. D., Ngure, F.M., Pelto, G., & Young, S. L. (2013). What are we assessing when we measure food security? A compendium and review of current metrics. Advances in Nutrition, 4, 481–505.

Kabunga, N. S. (2014). Improved dairy cows in Uganda: Pathways to poverty alleviation and improved child nutrition (Vol. 1328). International Food Policy Research Institute.

Kassam, L. (2014). Aquaculture and food security, poverty alleviation and nutrition in Ghana: Case study prepared for the Aquaculture for Food Security, Poverty Alleviation and Nutrition project. WorldFish.

Kassie, M., Jaleta, M., Shiferaw, B.A., Mmbando, F., De Groote, H. (2012). Improved Maize Technologies and Welfare Outcomes In Smallholder Systems: Evidence From Application of Parametric and Non-Parametric Approaches. Presented at the 2012 Conference, August 18-24, 2012, Foz do Iguacu, Brazil, International Association of Agricultural Economists.

Kennedy, G., Ballard, T., & Dop, M. C. (2011). Guidelines for measuring household and individual dietary diversity. Food and Agriculture Organization of the United Nations.

Khandker, S.R., Koolwal, G.B., Samad, H.A. (2010). Handbook on impact evaluation: quantitative methods and practices. World Bank Publications.

Kibira, M., Affognon, H., Njehia, B., Muriithi, B., Ekesi, S., 2015. Economic evaluation of integrated management of fruit fly in mango production in Embu County, Kenya. African Journal of Agricultural Resource Economics 10, 343-353.

Klungness, L., Jang, E.B., Mau, R.F., Vargas, R.I., Sugano, J.S., Fujitani, E. (2005). New sanitation techniques for controlling tephritid fruit flies (Diptera: Tephritidae) in Hawaii. Journal of Applied Science in Enviromental Management. 9, 5-14.

KM, K. J., Suleyman, D. M., & PK, A. D. (2013). Analysis of food security status of farming households in the forest belt of the Central Region of Ghana. Russian Journal of Agricultural and Socio-Economic Sciences, 13(1).

Komicha, H. H. and Öhlmer, B. (2007). Influence of Credit Constraint on Technical Efficiency of Farm Households in Southeastern Ethiopia. International Conference on African Development Archives. Paper 125.

Korir, J. K., Affognon, H. D., Ritho, C. N., Kingori, W. S., Irungu, P., Mohamed, S. A., & Ekesi, S. (2015). Grower adoption of an integrated pest management package for management of mango-infesting fruit flies (Diptera: Tephritidae) in Embu, Kenya. International Journal of Tropical Insect Science, 35(2), 80-89.

Kotrlik, J., Higgins, C. (2001). Organizational research: Determining appropriate sample size in survey research appropriate sample size in survey research. Information Technology Learning Performance Journal. 19, 43.

Latham, M. C. (1997). Human nutrition in the developing world (No. 29). Food & Agriculture Organization.

Lewin, P.A. (2011). Three essays on food security, food assistance, and migration. Graduate Theses and Dissertations. 20692. Available at http;// (Accessed September 20, 2015).

Macharia, I. N., Mithöfer, M., & Waibel, H. (2009). Potential environmental impacts of pesticides use in the vegetable sub-sector in Kenya. African Journal of Horticultural Science, 2 (138-151).

Magrini, E., & Vigani, M. (2016). Technology adoption and the multiple dimensions of food security: the case of maize in Tanzania. Food Security, 8(4), 707-726.

Martin, K.S., Rogers, B.L., Cook, J.T., Joseph, H.M. (2004). Social capital is associated with decreased risk of hunger. Social Science and Medicine, 58, 2645–2654.

Mason, N.M. and Smale, M. (2013). Impacts of subsidized hybrid seed on indicators of economic well-being among smallholder maize growers in Zambia. Agricultural Economics, 44, 659–670.

Matchaya, G., & Chilonda, P. (2012). Estimating effects of constraints on food security in Malawi: policy lessons from regressions quantiles. Applied econometrics and international development, 12(2), 165-191.

Mbo’o-Tchouawou, M. Karugi, J. Mulei, M. Nyota, H. (2016). Assessing the participation of men and women in cross-border trade in agriculture: Evidence from selected East African countries. ReSAKSS Working Paper No. 38. International Food Policy Research Institute (IFPRI) and the International Livestock Research Institute (ILRI).

Ministry of Agriculture. (2010). Embu East District Annual Report. Unpublished.

Mitcham, E., Yahia, E. (2010). Alternative treatments to hot water immersion for mango fruit report to the national mango board.

Morton, M. (2009). Applicability of impact evaluation to cohesion policy. Presented at the An Agenda for a Reformed Cohesion Policy. A place-based approach to meeting European Union challenges and expectations. Independent Report.

Mulugeta, T., Hundie, B. (2012). Impacts of Adoption of Improved Wheat Technologies on Households’ Food Consumption in Southeastern Ethiopia. Presented at the 2012 Conference, August 18-24, 2012, Foz do Iguacu, Brazil, International Association of Agricultural Economists.

Muriithi, B. W., Affognon, H. D., Diiro, G. M., Kingori, S. W., Tanga, C. M., Nderitu, P. W., & Ekesi, S. (2016). Impact assessment of Integrated Pest Management (IPM) strategy for suppression of mango-infesting fruit flies in Kenya. Crop Protection, 81, 20-29.

Muthini, D.N. (2015). An Assessment of Mango Farmers’ choice of Marketing Channels In Makueni, Kenya. Unpublished MSc Thesis, University of Nairobi, Nairobi.

Mwanaumo, A., Jayne, T. S., Zulu, B., Shawa, J. J., Haggblade, S., & Nyembe, M. (2005). Zambia’s 2005 Maize Import and Marketing Experiences: Lessons and Implications(No.54615). Michigan State University, Department of Agricultural, Food, and Resource Economics.

Oluwatayo, I. B. (2008). Explaining inequality and welfare status of households in rural Nigeria: evidence from Ekiti State. Humanity & Social Science Journal, 3(1), 70-80.

Oluyole, K. A., Oni, O. A., Omonona, B. T., & Adenegan, K. O. (2009). Food security among cocoa farming households of Ondo State, Nigeria. Journal of Agricultural and Biological Science, 4(5), 7-13.

Omilola, B. (2009). Estimating the impact of irrigation on poverty reduction in rural Nigeria. IFPRI Discussion Paper 00902. Washington, DC: International Food Policy Research Institute.

Oni, S.A., Maliwichi L. and Obadire, O. S. (2010). Socio-Economic Factors Affecting Smallholder Farming and Household Food Security: A Case of Thulamela Municipality in Vhembe District, South Africa. African Jounal of Agricultural Research, AJAR Vol. 5(17), pp. 2289-2296

Oriola, E. O. (2009). A framework for food security and poverty reduction in Nigeria. European Journal of Social Sciences, 8(1), 132-139.

Pankomera, P., Houssou, N., Zeller, M. (2009). Household Food Security in Malawi: Measurement, Determinant, and Policy Review. Presented at the Conference on International Research on Food Security, Natural Resources Management and Rural Development.

Prais, S. J. and Winsten, C. B. (1954): Trend Estimators and Serial Correlation. Cowles Commission Discussion Paper No. 383 (Chicago)

Ravallion, M. (2005). Evaluating anti-poverty programs. Paper Prepared for the Handbook of agricultural economics (Vol 4), ed. R. Evenson and TP Schultz.

Sikwela,M. M (2008): Determinants of Household Food security in the semi-arid areas of Zimbabwe: A case study of irrigation and non-irrigation farmers in Lupane and Hwange Districts. Thesis for the degree of Master of Science in Agriculture. Department of Agricultural Economics and Extension. University of Fort Hare, Republic of South Africa.

Shiferaw, B., Kebede, T.A., You, L. (2008). Technology adoption under seed access constraints and the economic impacts of improved pigeonpea varieties in Tanzania. Agricultural Economics, 39(3): 309–323.

Sseguya, H. (2009). Impact of social capital on food security in southeast Uganda. Graduate Theses and Dissertations. 10747. Available at http;//

Staal, S. J., Baltenweck, I., Waithaka, M. M., DeWolff, T., & Njoroge, L. (2002). Location and uptake: integrated household and GIS analysis of technology adoption and land use, with application to smallholder dairy farms in Kenya. Agricultural Economics, 27(3), 295-315.

Swindale, A., Bilinsky, P. (2006). Household dietary diversity score (HDDS) for measurement of household food access: indicator guide. Washington DC Food Nutrition Technical Assistance Project and Academic Education Development.

Tembo, D., & Simtowe, F. (2009). The effects of market accessibility on household food security: Evidence from Malawi. In Conference on International Research on Food Security, Natural Resource Management and Rural Development (pp. 6-8).

Von Braun, J. (1988). Effects of technological change in agriculture on food consumption and nutrition: rice in a West African setting. World Development. 16, 1083–1098.

Voor de Tropen, F. M. IIRR (2006) Chain Empowerment: Supporting African farmers to develop markets. Royal Tropical Institute, Amsterdam.

Wainaina, P.W., Okello, J.J., Nzuma, J. (2012). Impact of contract farming on smallholder poultry farmers’ income in Kenya. Selected Paper prepared for presentation at the International Association of Agricultural Economists (IAAE) Triennial Conference, Foz do Iguaçu, Brazil, 18-24 August, 2012.

Wooldridge, J. M. (2013): Introductory Econometrics. A Modern Approach. 5th ed. Mason, OH: South-Western Cengage Learning Cengage.

Yamano, T., & Jayne, T. S. (2004). Measuring the impacts of working-age adult mortality on small-scale farm households in Kenya. World Development, 32(1), 91-119.

Leave a Comment Here