Impact_Eval_RP

Received: 29 July 2017 Revised: 27 January 2019 Accepted: 6 March 2019
DOI: 10.1111/1477-8947.12191
ARTICLE
Sustainable development in East Africa: impact

evaluation of regional agricultural development
projects in Burundi, Kenya, Rwanda, Tanzania,
and Uganda
Enock Warinda1 | Dickson M Nyariki2 | Stephen Wambua3 |

Reuben M Muasya4 | Munir A Hanjra5
1
Association for Strengthening Agricultural Research in Eastern and Central Africa (ASARECA), Entebbe, Uganda
2
Department of Business and Economics, Murang'a University of Technology, Agricultural Economics, Muranga, Kenya
3
Department of Agricultural Economics and Agribusiness, South Eastern Kenya University, Kitui, Kenya
4
Dryland Agriculture, South Eastern Kenya University, Kitui, Kenya
5
International Centre of Water for Food Security, Charles Sturt University, Wagga, New South Wales, Australia
Correspondence
Enock Warinda, Association for Strengthening
Abstract
Agricultural Research in Eastern and Central Investments in agricultural technologies, capacity building
Africa (ASARECA), Entebbe, Uganda.
and policy harmonization are needed to support sustainable
Email: [email protected]
development across Africa. Regional development projects
can facilitate the adoption of agricultural technologies and
innovations across nation-state borders and generate bene-
fits for shared prosperity. This paper uses panel data from
1,160 smallholder households including beneficiaries and
non-beneficiaries from five countries in East Africa - Burundi,
Kenya, Rwanda, Tanzania, and Uganda. From a pool of over
90 projects implemented over a span of 15 years, 23 regional
projects are included in this study. The major economic bene-
fits to smallholder participants are higher crop productivity
and income, access to adaptable technologies, access to mar-
kets, higher livestock and milk production, gender equality,
enhanced food security, resilience and capacity building for
uptake and scaling up of future innovation platforms. For
example, the adoption of low-cost tissue culture banana
by the beneficiaries increased their incomes by 15% in
Rwanda while the adoption of appropriate land and water
Nat Resour Forum. 2020;44:3–39. wileyonlinelibrary.com/journal/narf © 2020 United Nations 3

4 WARINDA ET AL.
management technologies increased the potato yields from

2.8 tons (USD 2,840/ha) to over 7.5 tons (USD 7,410/ha) in
Kenya. The beneficiaries ensured value added to commodi-
ties like bananas (for export) and orange-fleshed sweet pota-
toes (a nutrition-sensitive intervention for enhancing
domestic intake of vitamin A) in Burundi, Kenya, Rwanda and
Uganda. Additionally, milk production increased from 6 to
11 L per cow per day in Uganda and from 6 to 8 L per cow
per day in Tanzania, with increases in sales generating USD
115 per cow per month above the non-beneficiaries. These
results are supported by previous studies on technology
adoption, investments in agriculture and well-being out-
comes. Our findings with the higher farm income Difference-
in-Differences (DiD) estimator for the female beneficiaries
compared to male beneficiaries might have important impli-
cations for investing in regional development projects that
will close the gender gap in agricultural productivity in Africa.
Regional projects can also support post-conflict development
efforts for food security and peace in fragile contexts such as
in Burundi. Our findings might serve as an input to the Afri-
can Union's Comprehensive Africa Agriculture Development
Programme, localization of the United Nations Sustainable
Development Goals related to food security and agriculture
in Africa and an input to monitoring, evaluation and learning.
KEYWORDS
beneficiaries, capacity building, economic benefits, food security,

fragility, gender, innovation platforms, propensity score matching,
smallholders
1 | I N T RO DU CT I O N
Africa has abundant resources including land, water and human sources to support sustainable development in
agriculture and enhance food security (Dixon et al., 2017; Jayne, Mason, Burke, & Ariga, 2018; Sadoff et al., 2015).
Though research shows that reducing poverty requires investments in agricultural technologies, capacity building
and harmonization of policies that support economic growth and regional transformation (Abro, Alemu, & Hanjra,
2014; Ferede, Ayenew, & Hanjra, 2013; Hanjra & Gichuki, 2008; Wichelns, 2015), investments to support agricul-
tural transformation remains a challenge (Hanjra, Ferede, & Gutta, 2009a; Williams, 2015). Africa also has the
fastest growing population in the world, projected to increase from 1.1 billion in 2015 to 2.5 billion people by
2050 (Nieves et al., 2017) with huge implications for natural resources, agriculture, food security, investments and
public policy.
WARINDA ET AL. 5
Governments and development partners across Africa have launched many initiatives to transform agriculture
and reduce poverty. The African Union launched the Comprehensive Africa Agriculture Development Programme
(CAADP) in 2003 and approved the CAADP Results Framework in 2014, with an investment target of 10% of
national expenditure on agriculture. The Malabo Declaration, specifying seven commitments aims at achieving agri-
cultural transformation for sustainable growth and shared prosperity, while the Science, Technology and Innovations
Strategy for Africa 2024 aims at allocating 1% of gross domestic product (GDP) towards research and development.
The Africa Union Agenda 2063 and African Development Bank's Feed Africa Strategy 2025 (AfDB, 2016) are being
implemented to address food and nutrition security issues (Williams, 2015). Effective implementation of these strate-
gies requires a regional approach and multi-stakeholder collaboration (Hanjra, Noble, Langan, & Lautze, 2017; HLPE,
2015; UN, 2016), including adoption of agricultural technologies by farmers, and equally involving women and youth
who grow 80% of staple foods (AfDB, 2016). Yet within the East African Community (EAC), actual investments in
National and Regional Agricultural and Food Security Investment Plans continue to fall short of commitments.
Burundi is the only country in the EAC to achieve the CAADP target of 10% spending in the agriculture sector
(AfDB, 2016).
Regional agricultural projects can facilitate adoption of agricultural technologies to help African countries
achieve their targets. These projects offer opportunities in policy harmonization and program coordination with mul-
tiple host governments and non-state actors across regional countries to achieve better outcomes. Although Agenda
2063 builds on country development plans, there are few efforts to identify agriculture development projects to tar-
get farmers across nation-state boundaries. Despite this recent emphasis, the impacts of regional projects on small-
holder productivity and income have rarely been examined in empirical studies (Lynum, Beintema, Roseboom, &
Badiane, 2016; World Bank, 2008).
This paper supports the African Union's CAADP and Agenda 2063 through impact evaluation of regional agricul-
tural projects in East Africa. The focus countries are Burundi, Kenya, Rwanda, Tanzania and Uganda. The analysis
focuses on impact evaluation of selected regional projects in contributing to the generation of benefits for small-
holder farmers. Panel data is taken from 1,160 smallholder households including beneficiaries (61%) and non-
beneficiaries (39%) with gender disaggregated data including women (40%) from two spells in 2010 and 2015.
Instead of a singular commodity, the projects examined priority food commodities important to livelihoods and food
security of millions across multiple countries. The commodities include quality protein maize (QPM), beans, mosaic
resistant cassava and sorghum, orange-fleshed sweet potato (OFSP), bananas, and milk.
The paper finds some interesting empirical results regarding the regional project outcomes. Major economic ben-
efits to smallholder participants are higher crop productivity and income, access to adaptable technologies, access to
markets, higher livestock and milk production, gender equality, enhanced food security, and capacity development
for uptake and upscaling of future development interventions. Our findings are supported by previous literature on
technology adoption (Alwang et al., 2017; Feder, Just, & Zilberman, 1985; Foster & Rosenzweig, 2010), investments
in agriculture and wellbeing outcomes (Dawson, Martin, & Sikor, 2016; Hanjra, Ferede, & Gutta, 2009b; Jayne,
Chamberlin, & Headey, 2014; Pingali, 2012; Stevenson, Villoria, Byerlee, Kelley, & Maredia, 2013). Findings of this
analysis may serve as an input to monitoring, evaluation, and learning and support future innovation platforms across
sub-Saharan Africa.
2 | LITERATURE REVIEW
Key challenges affecting high impact issues in agriculture require a different mind-set and science-based approach.
National agriculture research systems have local mandates and are best suited to implement specific projects
addressing low-impact local issues. Common issues affecting multiple countries such as limitations in agriculture
technology, human capacity, cross border biosafety, market access, agriculture policy and response to climate change
can be best tackled by deploying a regional approach for enhanced shared prosperity (Ferede et al., 2013). Greater
6 WARINDA ET AL.
benefits and synergies are possible through a regional focus on agricultural research for development (World Bank,
2008), partnerships that involve host governments and non-state actors (Orenstein & Shach-Pinsley, 2017; Shin,
Kim, & Sohn, 2017), and sustainable intensification in agriculture for shared prosperity (Ahlerup, Baskaran, & Bigsten,
2017; Rockström et al., 2017).
Focus is now shifting from country-specific projects to scalable projects that can be implemented in more than one
neighboring country, thereby generating wider benefits and tackling regional challenges. Regional projects have provided
platforms for packaging appropriate technologies that enhance incomes and food security for smallholder farmers. Data
generated from farming systems in Borana, Ethiopia; Nyando, Kenya; Lushoto, Tanzania; and Hoima, Uganda (Asfaw,
Shiferaw, Simtowe, & Lipper, 2012; Hisali, Birungi, & Buyinza, 2011; Shikuku et al., 2017) show farmers strongly appreci-
ate timely availability of information on new crops, changes in crop varieties, and planting regimes. However, non-farm
income needs to be in place to enhance production intensification (Hammond et al., 2017; Kristjanson et al., 2012).
Prioritizing context-specific agriculture technologies remains the greatest challenge. Works in Tanzania and
Uganda reveal heterogeneity not only across the sites, but also among different gender groups and agro-ecological
zones to target interventions (Mwongera et al., 2017), such that consensus-driven frameworks are needed to pro-
mote a regional approach to agriculture for development (Brandt, Kvakic, Butterbach-Bahl, & Rufino, 2017). Domains
to target technological innovations and development efforts to address the problem of poor productivity of small-
holder agricultural livestock systems must be designed to target socially diverse and spatially heterogeneous farms
(Tittonell et al., 2010). For example, ex-ante impact evaluation for two different regions in Kenya shows that the
introduction of an improved maize variety or a low-yielding, dual-purpose sweet potato variety in Machakos would
be sufficient to offset the negative impacts of climate change, while in the Vihiga area, dual-purpose sweet potato
varieties combined with improved livestock feed and breeds would be needed (Claessens et al., 2012).
Evaluation of the Mama SASHA project (Sweet potato Action for Security and Health in Africa) that is promoting
consumption of a bio-fortified, orange-fleshed sweet potato in western Kenya shows that feedback meetings are
critical to build understanding across actors and to problem-solve in real time to respond to the array of stakeholders,
program funders and disciplines involved – in agriculture, nutrition and healthcare (Cole et al., 2016). In as much as
agronomic biofortification can increase yields and micronutrient contents of crops, it requires integrated soil fertility
management. In East Africa, lack of knowledge and access to inputs hamper its implementation (de Valença, Bake,
Brouwer, & Giller, 2017; Murekezi, Oparinde, & Birol, 2017).
Data from 8,600 households in 10 countries in Africa shows the benefits of sequential cropping systems for
improved yields (Bussmann, Elagib, Fayyad, & Ribbe, 2016; Fermont, van Asten, Tittonell, van Wijk, & Giller, 2009;
Tachie-Obeng, Akponikpè, & Adiku, 2013; van Asten, Wairegi, Mukasa, & Uringi, 2011; Waha et al., 2013); Jarvis,
Ramirez-Villegas, Herrera Campo, & Navarro-Racines, 2012). Improved cassava harvesting and processing technol-
ogy (such as chippers) adds value and reduces post-harvest losses by up to 20%, while detoxified cassava leaves
serves as a safe nutrient source rich in protein and vitamins (Latif & Müller, 2015). Introducing high-yielding legumes
into existing pasture in smallholder systems improves cow diets, milk production and offers higher returns (Mottet
et al., 2017; Notenbaert, Pfeifer, Silvestri, & Herrero, 2017).
Data from 884 households across Rwanda shows that improved livestock feeding benefitted nearly 40% of the
households and is the most promising option to achieve triple win outcomes – food security, adaptation and emis-
sions (Argent, Augsburg, & Rasul, 2014; Paul et al., 2017; Rawlins, Pimkina, Barrett, Pedersen, & Wydick, 2014;
REMA, 2015). Assorted benefits of capacity building and farmer training programs from regional projects, leading to
improved crop productivity, empowerment and household agricultural incomes have been documented in East Africa
(Davis et al., 2012; Friis-Hansen & Duveskog, 2012; Williams, 2017).
African countries look beyond state boundaries and pursue regional harmonization of agricultural policies to sup-
port regional market development and regional public goods that can be readily developed in one country and
deployed in others. These policies include: multiplication and distribution of quality seeds in Southern and Eastern
Africa (Langyintuo et al., 2010; Rohrbach, Minde, & Howard, 2003), intra-regional trade (Geda & Seid, 2015), and
food standards (Heggum, 2016; Kaitibie, Omore, Rich, & Kristjanson, 2010; Walker, Ryan, & Kelley, 2010).
WARINDA ET AL. 7
Harmonization of public policies in Rwanda have incentivized smallholder farmers to adopt and uptake modern
seeds, inputs and credit facilities (Dawson et al., 2016), notwithstanding the challenges of sole cropping (Isaacs,
Snapp, Chung, & Waldman, 2016), and landlessness, and inequality and disruption to local trade (Ali, Deininger, &
Goldstein, 2014; Habyarimana & Nkunzimana, 2017; Rushemuka, Bizoza, Mowo, & Bock, 2014).
In as much as impact evaluations of local agricultural projects are widely reported in Africa, their availability to
the public is limited. Some of the available evaluations include: “Health of people and environment in the Lake Victo-
ria Basin Project in Uganda and Kenya” (Ghiron et al., 2014), “Gender, assets and agricultural development: lessons
from eight projects” in Uganda, Kenya, Tanzania, Mozambique, Burkina Faso, India and Bangladesh (Johnson,
Kovarik, Meinzen-Dick, Njuki, & Quisumbing, 2016), “Do different implementing partnerships lead to different pro-
ject outcomes? Evidence from the World Bank project-level evaluation data” (Shin et al., 2017), and “Policy misfits,
climate change and cross-scale vulnerability: how development projects undermine resilience in Tanzania and
Mozambique” (Bunce, Brown, & Rosendo, 2010). Yet, impact evaluations are rare for regional agricultural projects,
particularly those implemented across multiple neighboring countries.
3 | DATA AND METHODS
3.1 | Study region
This study covers the five countries of EAC, with a population of 157 million, land surface area of 172 million ha,
10 million ha under water bodies and a combined GDP of USD 146.9 billion (USD 1,014 per capita) (Table 1) (EAC,
2015; REMA, 2015; USAID, 2013). Agriculture is among the leading economic sectors in the region and provides
livelihoods to the populace (80%) and accounts for a significant share of GDP – Burundi (35%), Kenya (28%), Rwanda
(33%), Tanzania (28%) and Uganda (23%).
3.2 | Data collection and analysis
This study uses a mixed methods approach and a stratified random sampling to collect qualitative and quantitative
data. At least three project districts within the project catchment areas where seven selected food commodities are
TABLE 1 Economic profile and contribution of agriculture to East African Community economy, 2015
Economic indicator Burundi Kenya Rwanda Tanzania Uganda Overall

Population (million) 10.8 44.9 11.3 51.8 37.8 156.6
Population growth (% annual) 3.3 2.6 2.4 3.2 3.3 3.0
GDP (US$ billion) 3.1 60.9 7.9 48.1 27 146.9
GDP growth (% annual) 4.8 4.7 4.6 7.0 4.7 5.2
GDP per capita (US$2015) 286 1,355 696 955 715 1,014
Land surface area (million ha) 2.78 58.27 2.63 48.1 16.9 172
Agricultural land (million ha) 1.13 56.9 2.29 48.1 16.9 125.3
Irrigated land (%) 0.72 0.18 3.0 2.0 3.8 1.94
Size of holding/household (ha) 0.6 2.5 0.72 2.0 2.5 1.6
Share of smallholder in production (%) 86 75 80 75 75 78
Agriculture value added to GDP (%) 35 28 33 28 23 31
Source: EAC (2015) and others.

8 WARINDA ET AL.
grown are first selected, leading to the sampling of at least 21 administrative units. Out of these units, two villages
are randomly selected, leading to 42 villages. At least 30 farmers are randomly selected from each village (estimated
total populations is about 18,000). A total of 1,160 households comprising of beneficiaries (61%) and non-
beneficiaries (39%), with 40% being women are selected. Using a pre-tested questionnaire, two survey rounds are
conducted to collect two sets of panel data between June and September of 2010 (Panel 1) and June and September
of 2015 (Panel 2).
From a pool of over 90 projects implemented over a span of 15 years within four domains (Wood & Chamberlin,
2003), 23 regional projects focusing on seven priority commodities (maize, beans, cassava, sorghum, potatoes,
bananas and milk) are included into this study (Table A1). Primary data is collected through on-farm household sur-
veys, face-to-face interviews, observations, focus group discussions and workshops with policy makers. Key datasets
collected include source, use of, and challenges with the technologies shared with the farmers, annual change in pro-
ductivity, access to markets and market information, capacity development and sustainability aspects. Selection of
these variables is based on their influence among the respondents participating in the regional projects. Secondary
data on agricultural policies and regulations are collected from relevant government institutions, country research
institutes and agencies such as USAID, FAO, IFAD, World Bank and International Food Policy Research Institute
(IFPRI).
Data is analyzed using independent t-test. Covariates include sex, age, type of farmer, level of education, house-
hold size, land ownership, membership in farmer groups, earnings from farming, value productivity per hectare, and
access to agricultural technologies. Prior to running the t-test, propensity score, e(x) is estimated using logistic regres-
sion. The propensity score method provides the conditional probability of assigning respondents to a particular treat-
ment based on the selected vector of observed covariates. Each respondent in the treatment group is matched with
another respondent not in the treatment group, but whose observed covariates indicate a high likelihood of being in
the treatment group.
The propensity of assignment of the respondent to the regional project (treatment), given the covariate is den-
oted by:
eðxÞ = pðz = 1jxÞ ð1Þ
We assume:
Y
N
pðz1 , …, zn jx1 ,…, xn Þ = eðxi Þzi f1 −eðxi Þg1− zi ð2Þ
i=1
We estimate the propensity score from the logistic regression as:
expðβ0 + β1 x1 + β2 x2 + … + βn xn
eðxÞ = pðz = 1jxÞ = ð3Þ
1 + expðβ0 + β1 x1 + β2 x2 + … + βn xn
where x1, x2, … xn are individual covariates; β0, β1 … βn are the corresponding regression coefficients, with coefficient
β0 representing the influence of absolute component (value of probability when all covariates are zero). The parame-
ters β0, β1 to βn are estimated from the data using the maximum likelihood method. The generated propensity score
is used to create a pseudo-randomized dataset, thus allowing an unbiased estimator of the treatment effect
(Littnerova et al., 2013). The estimated propensity score is matched using simple 1:1 nearest neighbor matching. The
caliper of 0.15 of the standard deviation of the logit of the propensity score is used to include unmatched respon-
dents as recommended (Thoemmes, 2012). Associated methods and steps are as indicated in the following Sections
3.3 and 3.4.
WARINDA ET AL. 9
3.3 | Impact evaluation approach
This study uses an integrated approach that includes (a) impact approach, (b) capacity approach, and (c) policy
approach for impact evaluation while using the “with/without” approach. Impact approach refers to the economic
benefits pertaining to the beneficiaries in terms of gains in yield, productivity, income, value added, and sustainable
land and water management practices to enhance food security and reduce poverty. Capacity approach considers
capacity building impacts for farmers participating in the projects. Policy approach considers supportive policies,
investments and policy harmonization to enhance agricultural productivity and regional integration. We use
difference-in-differences (DiD) and propensity score matching methods (Gitonga, De Groote, Kassie, & Tefera, 2013;
Mendola, 2007) to minimize selection bias with individual level panel data (Abadie, 2005; Donald & Lang, 2007).
The DiD estimator is defined as the difference in average outcome in the treatment group before and after
treatment minus the difference in average outcome in the comparison group before and after treatment. We assume
that the difference between “before” and “after” in the comparison group is an optimal counterfactual for the treat-
ment group, as it accounts for potential sources of selection bias and compares treatment and comparison groups in
terms of temporal trends in outcome relative to the outcome observed for a pre-intervention baseline. The impact of
the regional projects is estimated as:
Y it = α + δT i1 t + πT i1 + γt + ϵit ð4Þ
Where Yit = benefits (income, yield or productivity) accrued from engagement in the projects by farmer i at time
t; Ti1 = engagement in intervention; T = 1 if farmer engages in intervention, for example, adoption of new technolo-
gies, and T = 0 if otherwise; t = survey round (t = 0 for panel 1 in 2010; t = 1 for panel 2 in 2015); δ = impact of inter-
vention (DiD), representing interactions between post-project engagements (Ti1) and time (t = 1,…, n years); and
ε = the error term.
Letting Y Bt and Y NB
t represent the respective impacts to the regional project beneficiaries and non-beneficiaries
in time t, we use DiD method to estimate the benefits/impact of the project to the beneficiaries as:
n o n o
DiD = E YB1 −YB0 jT1 = 1 −Ef YNB NB
1 −Y0 jT1 = 0 ð5Þ
The DiD estimator mimics unobserved heterogeneity that mostly leads to selection bias and is based on the
assumption of time-invariant linear selection effects such that differences between beneficiaries and non-
beneficiaries eliminate the bias (Abadie, 2005; Heckman, Ichimura, Smith, & Todd, 1998; Mayoux & Chambers,
2005; Stuart, 2010; Stuart et al., 2014). In this paper, the DiD estimator considers that: (a) the fitted model is cor-
rectly specified, with all the additive structures correctly imposed; and (b) the error term is uncorrelated with other
variables in the fitted model, such that the error term is characterized as:
8
>
> Covðϵit , T i1 Þ = 0
<
Covðϵit , tÞ = 0 ð6Þ
>
>
:
Covðϵit , T i1 tÞ = 0
The paper considers the case of two scenarios, noted as 1 and 0. It stipulates that the ith individual among
N smallholder farmers under study has both a response r1i that would have resulted if (s)he had received treatment
1 (beneficiary), and a response roi that would have resulted if (s)he had received treatment 0 (non-beneficiary). It fur-
ther posits that the causal effects of participation in regional projects are shown as comparisons between r1i and r0i
(for example r1i – r0i or r1i/r0i). Since each respondent receives only one treatment, either r1i or r0i is observed, but
10 WARINDA ET AL.
not both. Comparisons of r1i and r0i represent some degree of speculation since determination of the causal
effects of treatments is a data problem of missing true baselines, given that either r1i or r0i is missing for each
individual.
The beneficiaries and comparable non-beneficiaries are matched using propensity score. Only the beneficiaries
and non-beneficiaries with comparable propensity scores are used to determine the effect of these regional projects.
The parameter of interest is the average treatment effect of these regional projects (ATTj), hereby calculated as the
mean difference in outcome across these two groups (Equation 7):

ATTj = E½I1j RPj = 1 −E½I0j RPj = 0 ð7Þ
Where,
ATTj = Impact of regional projects (average treatment effect of the treated for each project);
I1j = Value of the outcome of farm household after participation in the project;
I0j = Value of the outcome of the same farm household j if he/she had not participated in the project
(or participated in a similar national project);
RPj = Regional projects, where 1 indicates participation, and 0 otherwise.
Predicted values from the standard logit model are used to estimate the propensity score for each observation
in the participant and the comparison-group samples. Using the estimated propensity scores, Ṕ(X), matched-pairs are
constructed on how close the scores are across two samples. The nearest neighbor to the ith participant is hereby
defined as the non-participant that minimizes [p(X) – p(Xj)]2 over all j in the set of non-participants, where p(Xk) is the
predicted odds ratio for observation k, that is,
ṔðXk Þ
pðX k Þ = ð8Þ
1− ṔðXk Þ
Letting ΔYj denote the benefits for the jth unit attributable to the regional projects, the PSM estimator of mean
impact is estimated as:
X
T X
C
=
ΔY ωj Y j1 −Xj β^0 − W ij Y ij0 −Xi β^0 ð9Þ
j=1 i=1
Where,
Yj1 = post-intervention indicator of success; Yij0 = outcome indicator of the ith non-treated matched to the jth
treated; T = total number of treatments; C = total number of non-treated households; ωj = sampling weights used to
construct the mean impact estimator; Wij = weights applied in calculating the average benefits of the matched non-
participants; and β^0 = the OLS estimate for the comparison group sample.
3.4 | Impact evaluation model specification
In addition to DiD and propensity score matching model, we evaluate the impact of regional projects (Δ) by
estimating value added in agricultural activities by beneficiaries (VAα) and non-beneficiaries (VAβ) using their house-
hold survey data. We estimate value added in agricultural activities by letting i represent a household that depends
on self-employed agricultural income (ySE L
α,i ), agricultural labor income ( yα,i ), and agricultural capital income from
renting of land, storehouse or farm equipment and interest payments (yKα,i ) for beneficiaries; and ySE L K
β,i , yβ,i and yβ,i for
non-beneficiaries, respectively.
WARINDA ET AL. 11
We determine value added in agriculture for beneficiaries as:
X X X
VAα = ySE
α,i + yLα,i + yKα,i ð10Þ
i i i
Based on Equation (10), the self-employed agricultural income for household i, less variable and fixed costs is
determined as:
!
X
J X
J X
J
ySE
α,i = Pj xHome
i,j + xMarket
i,j + xInvestment
i,j −Costsα,i ð11Þ
j j j
where, j represents the targeted agricultural commodities supported by regional projects; Pj is the farm-gate price of
each of the targeted commodities, j; and xi, j are quantities of the same commodities, j used at home, sold at the mar-
ket, and invested for revenue generation, and includes gifts and other in-kind payments.
P
The authors define agricultural labour income for beneficiary household i ( yLα,i ) as all the income in currency or
i
in-kind for labour services rendered. Wage income is measured at the individual level and then aggregated and
discounted to the household level by gender. Similarly, the authors define agricultural capital income for the benefi-
P
ciaries ( yKα,i ) as all income earned in cash or in-kind for rental of storehouses, land or equipment, plus interest pay-
i
ments. We estimate it directly from reported values and indirectly from proxy values, and aggregate by gender at
household level.
We present costs incurred by beneficiaries, Costsα, i as the summation of all fixed costs (FCα, i) and variable
expenditures (VCα, i) on intermediate goods purchased, including hired labour, land and products and services used
for production of commodity j:
Costsα,i = FCα,i + VCα,i ð12Þ
We determine value added in agriculture for non-beneficiaries as:
X X X
VAβ = ySE
β,i + yLβ,i + yKβ,i ð13Þ
i i i
Based on Equation (13), the self-employed agricultural income for non-beneficiary household i, less variable and
fixed costs is determined using Equation (13) and adjusting for non-beneficiaries as:
!
X
J X
J X
J
ySE
β,i = PNBj xHome
i,j + xMarket
i,j + xInvestment
NBi,j −CostsNBβ,i ð14Þ
j j j
Thus, we estimate the impact of regional projects (Δ) as the net benefits accrued to beneficiaries and non-bene-
ficiaries, viz.:
ðΔÞ = VAα −VAβ ð15Þ
We collected qualitative data through focus group interviews with randomly selected participants in each of the
target villages in selected countries.
12 WARINDA ET AL.
4 | RESULTS AND DISCUSSION
4.1 | Economic benefits of regional projects
Economic theory suggests that farmers are rational individuals and tend to respond to price and market signals.
However, most agricultural markets operate under imperfect competition as farmers lack relevant capacity and
access to information and technologies to make optimal choices (Mogues, Fan, & Benin, 2015). Not all farmers par-
ticipate optimally in the agricultural development projects to generate best impacts. Those who adopt new agricul-
tural technologies have higher productivity and returns to land, water, family labor and capital resource (Feder et al.,
1985). The differential adoption rates among these farmers seem to disappear once higher productivity per hectare
and higher returns per worker is realized (Timmer, 2017). Improved farmer engagement and learning have the poten-
tial to improve impacts through coordinated assessment of technology adoption, monitoring, evaluation and learning
(Alwang et al., 2017; Dawson et al., 2016; Feder et al., 1985; Foster & Rosenzweig, 2010; Hanjra et al., 2009b; Jayne
et al., 2014; Pingali, 2012; Stevenson et al., 2013).
4.1.1 | Dynamics of income from regional projects
We calculate incomes accruing to beneficiary farmers using the framework shown in Table 2 and Equation 16:
Y it = α + δT i1 t + πT i1 + γt + ϵit ð16Þ
Where Yit = farmer's benefits (increased on-farm income); α = the intercept; Ti1 = engagement in intervention
(T = 1 if farmer engages in intervention, and T = 0 if otherwise); t = survey round (t = 0 for panel 1 in 2010; t = 1 for
panel 2 in 2015); δ = impact of intervention (DiD), representing interactions between post-project engagements (Ti1)
and time (t = 1,…, n years); πTi1 = time varying covariates; and ϵit = the error term.
By performing direct logistic regression on selected covariates assumed to influence change in farm incomes
among the respondents using Equation (16), the paper shows that all the covariates are statistically significant (χ 2
(7, N = 941) = 888.31, p < .001. The equation explains between 53.5% (Cox and Snell R-Square) and 85.5%
(Nagelkerke R-squared) of the variance in increased farm incomes and also correctly classifies 96.9% of the cases.
Following regression iterations, 12 covariates generate a unique statistically significant contribution to the equation
(Table 3). The odds ratios for selected covariates show that the beneficiaries who hire farm laborers, own land and
keep livestock have higher probability of increased farm incomes compared to non-beneficiaries (controlling for all
other factors in the model).
We calculate the propensity score by matching, 424 treatment cases with 424 control cases. From the control
group, 16 respondents are unmatched, compared to 260 for treated farmers. None of the samples are outside the
common support. The output shows that the overall Chi-square balance test is not significant (χ2 (11) = 17.37,
TABLE 2 Framework for calculating Difference-in-Differences (DiD)
Pre (2010) Post (2015) Post-pre difference (2015–2010) = (ϕ^1 )

Treatment (beneficiary) Y RP
0 Y RP
1 Y RP RP
1 − Y0
Control (non-beneficiary) Y NP
0 Y NP
1 Y NP NP
1 − Y0
Treatment – Control (T-C) Difference (ϕ^2 Þ Y RP NP

0 − Y0 Y RP NP
1 − Y1
(Y RP NP RP NP ^
1 − Y 1 ) – (Y 0 − Y 0 ) (ϕDID )
Source: Author's conceptualization.

WARINDA ET AL. 13
TABLE 3 Logistic regression predicting increase in farm incomes
95% C.I. for Exp)B)
B S.E. Wald df Sig. Exp(B) Lower Upper

Variables 6.2 1.432 18.8 1 .000 494 29.89 8,179.13
Sex −.58 .371 2.4 1 .121 .562 .272 1.16
Education −.11 .042 7.0 1 .008 .894 .823 .97
Hires laborers 7.86 .651 145.4 1 .000 2,578 718.96 9,242.13
Land size 2.01 .297 45.7 1 .000 7.463 4.17 13.37
Land under technology −2.05 .364 31.9 1 .000 .128 .063 .26
Owns livestock .42 .072 34.6 1 .000 1.527 1.33 1.76
Constant −11.10 1.224 82.3 1 .000 .000
Source: Authors, based on survey data (2015).
p = .097), thus suggesting that matching has helped reduce the bias associated with observable characteristics. Simi-
larly, the larger multivariate imbalance measure (L1 = .980) before matching compared to L1 = .971 after matching
indicates that matching improves the overall balance. The univariate balance test also shows that the standardized
mean differences for all covariates are balanced at |d| ≤ 0.25. The distribution of propensity scores shown in
Figure A1 indicates some overlaps in the treatment and control groups, while the plots of individual units shown in
Figure A2 show that the treatment and control groups are comparable. The standardized differences before and after
matching are slightly skewed from zero, indicating higher propensity scores among the groups (Figure A3), while the
magnitude of standardized differences before and after matching for each covariate is shown in Figure A. Score
improvements after matching is observable compared to before matching (Figure A5) with bold lines illustrating
increased standardized differences after matching. Figure 1 shows the propensity scores based on farm incomes for
matched respondents.
Household income from farm and non-farm activities vary among the countries, respondents and over time. By
using the framework in Table 2, the estimated DiD for various respondent categories shows an overall positive
impact realized by beneficiaries of regional projects (ϕ^DiD = USD 223) above their non-beneficiary counterparts
(Figure 2). Beneficiaries in Uganda record the highest average income gains of USD 372 per household, while Kenya
and Rwanda have the least average income gains of USD 174 each.
Increases in farm income generated by beneficiaries in Burundi are notable (Figure 3, Table 4). Although, much
of the country has been economically unstable and fragile, participation in regional projects made the beneficiaries
better off than their non-beneficiary counterparts. Through the adoption of technologies availed by the regional pro-
jects, the beneficiaries generate an average income gain of USD 194 per household, while the non-beneficiaries
record an income decrease of USD 12 per household. This decrease is because these non-beneficiaries either did
not access new technologies or abandoned/downscaled their farming in pursuit of other non-farm activities.
This significant increase in farm incomes by beneficiaries in Uganda (USD 372), Burundi (USD 206) and Tanzania
(USD 205) is attributed to heightened advocacy, membership in farmer organizations, enhanced access to market
information, access to credit facilities (from the banks, micro-credit institutions, government credit schemes as well
as informal savings and credit groups), participation in assorted trainings (on good agronomic practices, crop and live-
stock management, pest and disease control, and added value), and engagement in community-based awareness
campaigns and extension programs for scaling up new technologies and innovations, especially early-maturing and
climate-resilient varieties.
The paper shows an overall increase in non-farm income by beneficiaries (USD 159) above non-beneficiaries,
with Rwanda and Burundi recording the highest and lowest increases (USD 231 and USD -20) respectively
(Figure 4). These increases are attributed to beneficiaries' access to new markets for their commodities, credit
14 WARINDA ET AL.
FIGURE 1 Propensity scores based on farm incomes for matched respondents
facilities, trainings and access to market information. The main type of information include: commodity prices in dif-
ferent markets (88%), list and details of commodities in demand (82%), status of prevailing supply in different mar-
kets (73%) and alerts on mobile phones whenever the commodities are in demand (37%). The beneficiaries accessed
the market information from other farmers (88%), radio (92%), farmer organizations and groups (79%), SMS text mes-
sages (56%), extension officers (68%), and market-place posters and posted bulletins (51%).
Overall, the estimated DiD shows that the beneficiaries of the regional projects have higher gains in total house-
hold income generated from on- and non-farm activities (Table 5). These revenues are generated from the sales of
surplus commodities, implying that enhancing the implementation of regional projects through regional cooperation
is likely to lead to improved productivity, incomes, economic and climate resilience as well as competitiveness of
WARINDA ET AL. 15
FIGURE 2 Average country level DiD (US$/household/year) in farm incomes. Source: Own calculation
F I G U R E 3 Average post-pre differences in farm incomes (US$/household/year) between respondents. Source:

Own calculation
T A B L E 4 Average post-pre differences in farm incomes at household level (US$/household/year) between

respondents
Burundi Kenya Rwanda Tanzania Uganda
Year Ben NBen Ben NBen Ben NBen Ben NBen Ben NBen
2015 587 374 954 783 1,096 906 895 689 988 615
2010 393 386 770 773 782 766 528 527 531 530
Change 194 −12 184 10 314 140 367 162 457 86
Abbreviations: Ben, Beneficiaries; NBen, Non-beneficiaries.

Source: Own calculation.
selected agricultural commodities. Compared to non-beneficiaries, the beneficiaries in Uganda and Rwanda have the
highest average total income (USD 510 and USD 406, respectively), with Burundi and Kenya recording USD 186 and
USD 360 respectively above the non-beneficiaries (Figure 5).
4.1.2 | Gender and income dynamics
Significant difference in incomes generated by male and female beneficiaries engaged in on- and off-farm activities is
observed (Table 6). Overall, the female beneficiaries record an average of USD 226 (compared to male, USD 218)
above the non-beneficiaries (Figure 6). In as much as the female beneficiaries in Burundi and Tanzania generate USD
108 and USD 95, respectively from farming above the male beneficiaries, they are still better off than the non-
beneficiaries by the same amount. These benefits have prompted some of the women to sign contracts with private
sector companies to ensure stable on-farm incomes. Even though male dominance in income generation is observ-
able in Kenya and Rwanda (with average income gap of USD 104 and USD 167 respectively, between male and
female), the female beneficiaries are still better off than their non-beneficiary counterparts by USD 106 and USD
16 WARINDA ET AL.
F I G U R E 4 Average post-pre differences in non-farm incomes (US$/household/year) between respondents.

Source: Own calculation
TABLE 5 Total income DiD between beneficiaries and non-beneficiaries (US$/household/year)
Post-pre difference (2015–2010) = (ϕ^1 )

Beneficiaries 368 548 540 554 669
Non-beneficiaries 182 188 134 238 159
DiD = T-C Difference (2015–2010) = (ϕ^2 ) 186 360 406 316 510
82, respectively. The generally high average incomes among female beneficiaries is attributed to their increased
access to credit facilities (33.3%), training opportunities (27.6%), adoption of available technologies such as QPM
(31.7%) and OFSP (29.8%), active engagement in farmer organizations (35.5%) and hiring of more human laborers
(33.8%) to meet their increasing farm demands (Figure 7). The findings confirm the fact that farming forms the pre-
ferred occupation by women within the rural areas where women supply up to 50% of farm labor in the five coun-
tries (Christiaensen, 2017) and grow the bulk of staple foods.
Comparison between male and female farmers shows significant variations (Figure 6). Uganda records the
highest DiD income (USD 371) generated by female beneficiaries above the non-beneficiaries, followed by Burundi
(USD 270), Tanzania (USD 257), Kenya (USD 106), and Rwanda (USD 82). The main source of farm incomes includes
cultivation of maize, beans, sweet potatoes, cassava and bananas, and active engagement in women groups, attend-
ing field demonstrations and national trade fairs, and participating in assorted trainings on good agricultural practices,
soil and water management and marketing. Over 20–25% gender gap in agricultural productivity is observable in
Africa (Christiaensen, 2017; Doss, Meinzen-Dick, Quisumbing, & Theis, 2017). The higher female DiD provide impor-
tant reasons for further investments in raising female productivity in agriculture.
4.1.3 | Comparison between projects and agriculture innovations
Higher farm incomes influence beneficiaries' access and adoption decisions on various technologies and innovations.
The OFSP root and vines producers in western Kenya signed a contract with Farm Concern International, linking
them with better markets in Eldoret and Kitale, and thus increasing their average income by USD 340 per farmer per
season, compared to USD 160 for non-beneficiaries. These linkages with markets have enabled the beneficiaries to
understand market needs and to produce market-oriented commodities.
The beneficiaries (compared to non-beneficiaries) have better linkages with the private sector, especially agri-
businesses involved in input and output markets for priority commodities. The banana wilt control intervention in
Uganda and Rwanda have contributed to increased production and provided platforms for implementing new income
WARINDA ET AL. 17
F I G U R E 5 Average total income (DiD, US$/household/year). ON, On-farm incomes; OF, Off-farm incomes.
Source: Own calculation
generating activities. From an initial production of 9,600 crates in 2009, the 2015 production reached 57,600
thereby generating USD 10,000 weekly compared to USD 1,700 weekly in 2009. This linkage with the private sector
has enabled the beneficiaries to receive an average of USD 0.3 per kg of banana delivered to their cooperative,
thereby linking them to a steadier market than the non-beneficiaries.
Farmers adopting improved and resistant banana varieties in Uganda recorded an increase in production from
5% of average monthly production (in 2010) to over 60% (in 2015). Given that the average production of bananas
under optimal conditions and appropriate agronomic practices is 4,000 kg per ha, each of these farmers increases
their production to nearly 2,500 kg per month (up from 200 kg). Based on prevailing farm-gate price of USD 0.3 per
kg, each farmer generates an average monthly income of USD 300 (up from USD 24). Similarly, some of the orga-
nized farmer groups in western Uganda increase their income from USD 30 per month per household from their
infected farms in 2009 to USD 600 within 4 years. The adoption of the low-cost tissue culture banana in Rwanda
led to an increase in the total gross margin by 14.6% above the non-beneficiaries (additional income of USD 30.77
per ha in 2010 to USD 35.25 per ha in 2015). In these countries, beneficiaries received government/local legal
expert support through input subsidies, by dealing with emerging contractual issues and in interpretation of relevant
by-laws and contracts.
T A B L E 6 Average post-pre differences in farm incomes between male and female respondents
(US$/household/year)
Male Female Male Female Male Female Male Female Male Female
B2015 565 623 929 1,006 1,119 1,075 846 960 1,007 973
B2010 406 372 716 885 782 796 492 577 541 523
Δ 159 251 213 121 337 279 354 383 466 450
NB2015 487 292 767 795 854 963 746 622 629 600
NB2010 490 311 764 780 766 766 554 496 537 521
Δ −3 −19 3 15 88 196 192 126 91 79
DiD 162 270 210 106 249 83 162 257 375 371
Notes: B2010, B2015 = Beneficiaries' average farm incomes in 2010 and 2015.
NB2010, 2015 = Non-beneficiaries' average farm incomes in 2010 and 2015.
Δ = Post-pre differences in farm incomes between male and female respondents.
DiD = Observed double difference between respondents.
18 WARINDA ET AL.
F I G U R E 6 Average post-pre differences in farm income (US$/household/year) between male and female
respondents. Source: Own calculation
FIGURE 7 Factors contributing to increased incomes (%) of female respondents. Source: Own calculation
4.1.4 | Dynamics of value productivity for selected crops
In this paper, value productivity is used to measure the efficiency of production, and is determined using farmer esti-
mates/records on yields, harvest prices and land area, such that:
0PN
1
½Ai * Yi * Pi
Bi=1 C
Cj = @ P N A (17)
Ai
i=1
Where, Cj = value productivity per ha for each crop; Ai = area under the ith crop (ha); Yi = yield per ha of the ith
crop (metric tons); and Pi = farm harvest price of the ith crop.
Value productivity per hectare varies significantly within countries and commodities (Table 7; Figures 9 and 10),
due to high dependence on rainfed agriculture, access levels to water for livestock and irrigation, and emergence of
pests and diseases. Beneficiaries experience greater increases in value productivity compared to non-beneficiaries
by an average of USD 1,248 annually per country (Figure 8). Given variations in the levels of adoption of available
technologies and access to credit facilities, training opportunities and market information, Kenya records the highest
annual value productivity per hectare (USD 1,578), while Burundi has the lowest (USD 815). Notwithstanding the
farmers' fear of re-emergence of cassava mosaic disease in Uganda, Tanzania, Kenya and Rwanda, the commodity's
value productivity is ranked highest (USD 259 per ha), while the annual value productivity per hectare for beans,
banana, sorghum and millet record USD 183, USD 156, USD 154, and USD 147, respectively.
To enhance value productivity, beneficiaries adopt strategic management practices. Maize beneficiaries select
disease and pest resistant varieties, avoid sowing the same varieties in the same fields repeatedly, diversify crop pro-
duction, practice intercropping, and use natural enemies such as the ladybug against aphid. Banana beneficiaries are
WARINDA ET AL. 19
trained in careful uprooting and burying affected plants, using healthy germplasm and corm paring at planting, and
managing black sigatoka resistant varieties.
4.1.5 | Sustainable land and water management
The percentage yield impact from adoption of a technology is measured as follows:

YieldB − YieldNB
%ΔY = max *100 ð18Þ
YieldNB
Where,
%ΔY = percentage change in yield impact from adoption of available technologies,
YieldB = estimated yield from regional projects using available technologies,
YieldNB = estimated business-as-usual yield from non-beneficiaries (baseline value).
The percentage change in yield of commodities (3.3%) is way below the required 6% annual production to meet
the CAADP target. Focus group discussions with farmers and other stakeholders attribute the main causes of this
slower growth productivity to soil nutrient and water management constraints, persistent uncertainties of rainfall
affecting sowing dates, crop choices, cropping decisions, climate variability, insufficient water harvesting technolo-
gies, long distances to water points, and poor or no adaptable irrigation technologies.
Adoption of integrated water and soil fertility management technologies varies among countries (Table 8)
despite latent potential to enhance productivity. Over 76% of the beneficiary farmers adopt improved soil fertility
management compared to 8% among the non-beneficiaries (leading to a DiD of 68%). Similarly, 70% of the beneficia-
ries (compared to 10% non-beneficiaries) adopt improved water management technologies (showing a DiD of 60%
in favor of beneficiaries). These findings show evident spillover effects within countries. In Kenya, adoption of appro-
priate land and water management technologies increase OFSP yield from 2.8 tons (USD 2,840 per ha) to over 7.5
tons (USD 7,410 per ha) in two seasons. In Uganda and Tanzania, utilization of integrated soil fertility and water
management technologies for production of livestock feed lead to increase in milk production from 6 to 11 and from
6 to 8 L per cow per day, respectively. Beneficiaries generate an average monthly income of USD 150 per cow in
Uganda and USD 60 per cow in Tanzania above the non-beneficiaries due to adoption of available technologies.
Conflicts over land, water and livestock feeds escalate, thus reducing the ability of governments, development
agencies and local communities to manage impacts of droughts and other shocks (Cervigni & Morris, 2015). Over
84% of the respondents are affected by droughts and experience an average of 11 and 8 weeks of food
unavailability in 2010 and 2015 respectively. Improved crop production technologies have delivered more reliable
incomes and enhanced resilience by boosting rainfed agriculture productivity. Simulation of the impact of best-bet
crop intensification technologies like drought-resistant and heat-tolerant varieties, improved fertility management,
and rainwater harvesting on productivity of key staples within drylands (maize, sorghum, and millet) suggest that the
number of drought-affected poor households could be reduced by 10–80% compared to a “business as usual” sce-
nario where regional projects are less undertaken (Bussmann et al., 2016; Cervigni & Morris, 2015).
4.1.6 | Dynamics of agricultural value addition
Significant increases in value addition on commodities are observable. On average, beneficiaries receive an annual
added value of USD 2,027 from priority commodities, compared to USD 1,118 by non-beneficiaries, showing an
increase (DiD) of USD 909 (Table 9, Figure 11). Tanzania records the highest DiD in value add (USD 1,250), followed
by Kenya (USD 980), Uganda (USD 791), Rwanda (USD 768), and Burundi (USD 754). Burundi's experience shows
20
TABLE 7 Value productivity per hectare for food commodities (US$/ha/year)
Maize Sorghum Millet Beans Sweet potato Cassava Banana Irish potato TOTAL
2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 Post-pre DiD
Ken Ben 216 552 275 529 145 417 219 534 245 568 369 754 215 564 125 335 1,809 4,253 2,444 1,578
NBen 210 381 277 313 140 365 220 248 248 366 372 395 212 387 120 210 1,799 2,665 866
Rwa Ben 140 380 190 354 159 339 178 337 125 279 222 751 166 489 80 177 1,260 3,106 1,846 1,264
NBen 146 210 192 228 163 224 176 262 124 201 230 332 161 302 84 99 1,276 1,858 582
Tan Ben 343 623 306 530 92 414 87 602 120 325 207 579 137 410 75 163 1,367 3,646 2,279 1,498
NBen 350 542 210 255 95 120 86 382 122 176 210 283 140 198 72 110 1,285 2,864 781
Uga Ben 146 398 120 299 230 461 57 251 125 332 222 461 110 350 140 312 1,150 2,864 1,714 1,085
NBen 151 208 122 154 233 364 55 100 120 239 219 313 106 203 143 197 1,149 1,778 629
Bur Ben 95 198 141 295 37 274 45 312 441 642 80 188 69 153 59 127 967 2,189 1,222 815
NBen 93 151 140 194 35 101 43 123 438 490 78 122 71 88 62 98 960 1,367 407
Total Ben 940 2,151 1,032 2,007 663 1,905 586 2,036 1,056 2,146 1,100 2,733 697 1,966 479 1,114 6,553 16,058 9,505 1,248
Total NBen 950 1,492 941 1,144 666 1,174 580 1,115 1,052 1,472 1,109 1,445 690 1,178 481 714 6,469 9,734 3,265
Ben – Nben −10 659 91 863 −3 731 6 921 4 674 −9 1,288 7 788 −2 400 84 6,324 6,240
DiD 134 154 147 183 134 259 156 80 1,248
Abbreviations: Ken = Kenya; Rwa = Rwanda; Tan = Tanzania; Uga = Uganda; Bur = Burundi; Ben = Beneficiaries; NBen = Non-beneficiaries.
Source: Survey data (2015).
WARINDA ET AL.
WARINDA ET AL. 21
FIGURE 8 Average value productivity per hectare by each country. Source: Own calculation
that regional agricultural projects can support post-conflict development efforts for food security and peace in fragile
contexts (USAID, 2013). Income from farming is directly affected by inadequate post-harvest handling processes
capable of extending shelf life of perishable commodities like OFSP and dairy products. Non-beneficiaries experience
36% higher post-harvest losses, whereas beneficiaries reduce these losses by adopting post-harvest handling tech-
nologies and achieve higher added value. Beneficiaries ensure added value to commodities like cassava and bananas
(for export) and OFSP (for enhancing domestic and local intake of vitamin A) in Burundi, Kenya, Rwanda and Uganda,
leading to increased returns and nutrition outcomes.
Value addition for bananas (in Uganda and Rwanda), cassava (in Kenya and Uganda), and OFSP (in Kenya and
Tanzania) is attributed to adoption of quality- and safety-enhancing technologies during processing, packaging, trans-
portation, storage and marketing. Cassava value addition includes adoption of chippers and cleaners that reduce
over-peeling. Beneficiaries applying this technology fetch up to 38% increase in sales. Banana value addition involves
packaging pre-peeled bananas into airtight bags that reduce transport and storage costs.
Beneficiaries in Kenya, Rwanda, Tanzania and Uganda signed contracts with agro-based companies whereby the
contractors specify the required standards/grade of the commodity, length of the contract, quantity to produce and
the expected prices. The contracts allow beneficiaries to sell excess produce. In nearly all contracts, the contractor
bears the risk of marketing and provides credit inputs and technical advice to beneficiaries, while farmers take the
risk of production. Afribanana Products value chain incubators for bananas in Uganda, and the Pineapple Cooperative
Society value chain (COAFGA) in Rwanda contribute to the development of the banana and pineapple sub-sectors,
respectively. Through MoU with the private sector, the beneficiaries not only sell the banana, but also make biode-
gradable bags, banana fabrics, banana juice, banana wine, charcoal briquettes and tissue culture planting materials.
The unit prices of specific commodities produced by non-beneficiaries are generally lower than their beneficiary
counterparts, mainly because they are not able to act collectively to negotiate better unless they use middlemen.
FIGURE 9 Average value productivity per hectare for each commodity. Source: Own calculation
22 WARINDA ET AL.
This price differential is clear where the unit price of a cow in Tanzania for beneficiaries is USD 158 and USD 195 in
2010 and 2015 respectively, compared to USD 127 and USD 152 in 2010 and 2015 respectively for non-beneficia-
ries. The average livestock ownership (calculated as tropical livestock units – TLU) by beneficiaries increased from
4.3 to 6.4 compared to 1.6 to 2.3 for non-beneficiaries in 2010 and 2015 respectively.
Results on the disaggregated average value added per household reported in Table 9 show that agricultural labor
incomes are higher by USD 1,170 for beneficiaries in all countries compared to non-beneficiaries (USD 585). This
shows a DiD of USD 585 in favor of beneficiaries. Agricultural capital income for beneficiaries in all the countries is
higher by an average of USD 186 for beneficiaries compared to USD 28 for non-beneficiaries (DiD of USD 158). The
quantity of commodities used at home for beneficiaries and non-beneficiaries is not significantly different between
the respondents, such that beneficiaries sell an average of 105 kg compared to 97 kg by non-beneficiaries
(DiD = 8 kg). On average, beneficiaries invest USD 164 in agriculture compared to USD 83 by non-beneficiaries
(DiD = USD 82). These results show that regional agricultural development projects contribute to greater income,
investments and average savings of USD 182 for the beneficiaries above the non-beneficiaries.
4.2 | Capacity building benefits of regional projects
Adoption of quality protein maize and bean technologies by beneficiaries in Rwanda and Tanzania identify existing
problems and provide beneficiaries with a myriad of options to engage with other non-state actors to afford and
access high nutrition food for their households. Such collaborative needs assessment also lead to identification of
priority staples like cassava, bananas, maize, sorghum and potatoes. These provide households with good cushion
during droughts and are more reliable sources of economic returns. Some of the farmers formalize their groups into
farmers' organizations, thereby giving them greater voice and bargaining power as evidenced in Kenya's potato
farmers and Uganda's cassava growers. Members of these organizations access facilities like loans and negotiate bet-
ter with traders and private sectors to enhance sales of outputs and purchase of inputs. Beneficiaries of OFSP pro-
ject in Kenya bargained with Farm Concern International and One Acre Project for the sale of their produce with
each farmer receiving USD 340 during that season compared to USD 140 in the previous season.
4.2.1 | Dynamics of partnerships for enhancing productivity
Tanzania has 13 new partnerships between Maruku Agricultural Research and Development Institute and local gov-
ernments to manage banana bacterial wilt. Local governments moderate these partnerships and convene regular
meetings with beneficiaries. Uganda strengthens its partnerships with District Farmers Associations of Kumi,
Katakwi, Soroti and Abim and with Uganda National Farmers' Federation, East African Farmers' Federation, district
agricultural offices, sub-county local governments, and Grow More Seeds Company. This leads to increased produc-
tion and sale of seeds of potatoes, beans, maize, and sorghum by about 23%. Majority of beneficiaries that began
partnerships with multiple organizations, cooperative societies, women groups, community-based organizations, and
other self-help groups in 2008 have continued as members (Figure 12). In 2015, there were fewer new memberships
because most of the regional projects ended, and thus shifted focus to strengthen existing partnerships for project
sustainability, develop exit strategies, mobilize resources and improve project management strategies.
4.2.2 | Satisfaction with membership in organizations and technologies
Nearly 76% of beneficiaries express satisfaction with benefits gained by their membership status (Table 10). Their
higher degrees of satisfaction point to positive effects of diffusion of skills, technologies and specialized knowledge
WARINDA ET AL. 23
FIGURE 10 Value productivity per ha for each commodity. Source: Survey data, 2015
and information products. Delays in addressing emerging challenges faced by beneficiaries, especially timely access
to seeds, recommended varieties, and access to market and climate information contributes to some dissatisfaction.
Beneficiaries are 7.5 times more likely to be members of associations than non-beneficiaries, while members of these
associations are 1,440 times more likely to adopt availed technologies, a clear indication of the benefits of regional
projects.
The impacts and sustainability of these projects are also determined by assessing the level of farmer aware-
ness regarding various technologies and innovations, whether they have used them, the year when they had
their first contact with them and whether they are still using them. Using the Likert scale, with 1 showing very
24
TABLE 8 Adopters of soil and water management technologies across the study settings
Kenya Rwanda Tanzania Uganda Burundi Post-pre difference
Technologies Ben NBen Ben NBen Ben NBen Ben NBen Ben NBen Ben NBen DiD
Improved water management (%) 84.7 9.2 78.6 9.8 47.4 15.8 67.9 10.7 68.9 2.2 69.5 9.5 60.0
Improved soil fertility management (%) 88.2 9.8 78.0 11.9 63.6 9.1 72.7 9.1 76.5 0.0 75.8 8.0 67.8

WARINDA ET AL.
WARINDA ET AL.
TABLE 9 Estimates of average annual value added (US$/household/year)
Agricultural labour Agricultural Quantity used Quantity sold at Quantity for Cost of Agricultural Value
income capital income at home markets investment goods Added
J
P J
P
P L P K J
P
y a,i y a,i xHome
i,j xMarket
i,j
Post-pre differences
xInvestment
i,j Costsa, i VAa
i i j j between Ben
j
Country Status 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 2010 2015 & NBen DiD
Kenya Ben 4,186 5,738 264 359 524 747 1,307 1,793 411 358 197 244 6,889 9,239 2,350 980
NBen 2,779 3,540 119 171 426 676 887 1,123 62 106 92 119 4,365 5,735 1,370
Rwanda Ben 4,109 4,637 611 865 68 80 1,349 1,507 78 237 218 299 6,433 7,625 1,192 768
NBen 2,356 2,597 18 33 128 138 779 855 66 123 64 89 3,411 3,835 424
Tanzania Ben 5,080 6,853 223 259 1,188 1,324 1,254 1,865 501 1,073 226 294 8,472 11,668 3,196 1,250
NBen 3,682 4,606 79 102 1,058 1,293 1,535 2,101 129 304 58 81 6,541 8,487 1,946
Uganda Ben 5,969 7,408 305 672 189 285 949 1,281 16 78 112 179 7,540 9,903 2,363 791
NBen 3,188 4,111 441 459 516 496 1,843 2,316 550 645 69 152 6,607 8,179 1,572
Burundi Ben 4,856 5,412 148 324 645 703 1,569 1,696 132 216 235 266 7,585 8,617 1,032 754
NBen 401 476 6 38 55 67 132 146 90 133 132 234 816 1,094 278
Average Ben 4,840 6,010 310 496 523 628 1,286 1,628 228 392 198 256 7,384 9,410 2,027 909
NBen 2,481 3,066 133 161 437 534 1,035 1,308 179 262 83 135 4,348 5,466 1,118
Ben - NBen 2,359 2,944 178 335 86 94 250 320 48 130 115 121 3,036 3,944 909
DiD 585 158 8 70 82 7 909

25
26 WARINDA ET AL.
FIGURE 11 Average value added (US$/household/year). Source: Own calculation
high level of satisfaction and 5 the least, an overall score of 1.9 for beneficiaries and 3.0 for non-beneficiaries is com-
puted and the difference is highly significant. Higher satisfaction is mainly attributed to the fact that all beneficiaries
have more interest in accessing and utilizing the best technologies and innovations than non-beneficiaries (Table 11).
Significant variations in the level of respondents' satisfaction with assorted technologies are evident. Beneficiaries
are more satisfied with improved water management technologies (1.7), soil erosion management technologies (1.8)
and quality protein maize varieties (1.8), than their non-beneficiary counterparts who indicate indifference on the
same technologies (3.5, 3.2 and 3.1 respectively). Higher value productivity per hectare for food commodities
(Table 7) and economic gains are associated with the recorded higher satisfaction.
4.2.3 | Participation in potential future projects
Besides awareness and satisfaction with technologies, beneficiaries seem more confident to undertake similar pro-
jects alone and to participate in future projects, and not heavily rely on external support. Using the same Likert scale,
beneficiaries' level of confidence is 2.1, compared to non-beneficiaries (3.0) who are indifferent. Such capacity build-
ing outcomes might help in supporting priority projects in the region – for instance, scaling up agricultural innovation
platforms for shared benefits and prosperity for smallholder women farmers, extending adoption of technologies to
urban and peri-urban agriculture, and engaging youth in agribusiness.
4.3 | Benefits in agriculture policy and planning
Regional projects contribute to the creation of new opportunities for sharing information and analysis of policies,
laws, regulations and procedures. Policy analyses focus on pathways for enhancing free trade of essential agricultural
FIGURE 12 Number of new member joining the group in each year. Source: Survey data, 2015
WARINDA ET AL. 27
T A B L E 1 0 Level of satisfaction with

Number of household
membership in groups and organizations
Beneficiary Non-beneficiary Overall
Very satisfied 186 (22.7%) 14 (1.7%) 200 (24.4%)
Satisfied 439 (53.5%) 109 (13.3%) 548 (66.8%)
Indifferent 13 (1.6%) 29 (3.5%) 42 (5.1%)
Unsatisfied 8 (1.0%) 11 (1.3%) 19 (2.3%)
Very Unsatisfied - (−) 11 (1.3%) 11 (1.3%)
Overall 646 (78.8%) 174 (21.2%) 820 (100.0%)
commodities, assessment of trends in food prices and outlook, and provide impetus for joint sector reviews and
science-policy linkages. Despite being linked to other trading blocs like COMESA and SADC, the EAC block faces
non-tariff barriers that affect exchange of essential technologies, seeds and germplasm. However, these challenges
are being reduced by projects like Eastern Africa Agricultural Productivity Program (EAAPP), which aims to foster
agricultural development in Kenya, Uganda, Tanzania (and Ethiopia). Plans are also underway to introduce e-tags on
trucks for rapid clearance of emergency shipments at border crossings and faster food trade, leading to lower food
loss and rapid food aid to affected member states.
4.3.1 | Agriculture policy harmonization
Regional projects help build a unified and shared approach for enhancing policy coordination, such as harmonization
of quality standards for products, handling of pest and disease risks, streamlining border procedures and paperwork
and advocating with policy-makers to improve overall enabling environment for regional trade and market access. By
building on the momentum of these projects, a significant breakthrough has been achieved, thus helping increase
harmonization of policies and regulations to de-bottleneck trade in the targeted priority crops and addressing policy
bias in commodity production such as maize policies. This provides a level playing field and market-oriented
approach for crop diversification and market integration across EAC. Some of these standards are currently under
implementation and enforcement at national level, especially the 7 harmonized standards for cassava and 4 for pota-
toes. These are approved by East African Standards Committee (EASC) and are already in force.
Implementation of EAAPP has led to distribution of improved heifers to organized farmer groups and individual
farmers in Kenya, Uganda, Tanzania and Rwanda as well as procurement and distribution of sexed semen and syn-
chronization hormone with 5,000 doses each to Uganda and Congo. It has enhanced exchange of bulls and distribu-
tion of improved forage seeds between Kenya, Uganda and Tanzania, thereby enhancing milk production among
beneficiaries by 2 to 5 L per cow per day which is above the non-beneficiaries whose average milk production is 6 to
8 L per cow per day. The conducive policy environment has led to exchange of 421 heifers with Uganda and Tanza-
nia and enhanced artificial insemination services to 45,277 cows with 1,751 cows fertilized at the bull service sta-
tions. Conception is recorded for 17,201 cows with 15,309 calves born between 2011 and 2015.
On the other hand, reduced intra- and inter-country and regional trade barriers have contributed to farmers' will-
ingness to expand their cropped areas for adoption of productivity-enhancing technologies such as maize, cassava,
banana and sorghum varieties. Kenya, Uganda, Tanzania and Rwanda have customized and piloted food price-
forecasting model to act as early warning and response system for farmers and to assist policy-makers on price
trends and pattern of movement of essential commodities within and across the borders. The East African Food
Prices data portal, a web-based information system for food and input prices for these countries contains updated
food prices of several commodities including maize, beans and milk. The portal can be accessed at www.eafdp.org.
28
TABLE 11 Level of satisfaction with selected technologies and innovations among sample households based on Likert scale
Overall level
of
Very satisfied 1 Satisfied 2 Indifferent 3 Dissatisfied 4 Very unsatisfied 5 satisfaction
Availed Technologies &
Innovations Ben NBen Ben NBen Ben NBen Ben NBen Ben NBen Ben NBen
Quality Protein Maize 208 (31%) 15 (2%) 193 (29%) 42 (6%) 54 (8%) 56 (8%) 22 (3%) 64 (9%) 8 (1%) 13 (2%) 1.82 3.09
Quality Seed Potato 100 (19%) 25 (5%) 115 (22%) 77 (15%) 40 (8%) 87 (17%) 16 (3%) 54 (11%%) 1.84 2.52
Orange-Fleshed Sweet Potato 189 (26%) 27 (4%) 214 (29%) 81 (11%) 66 (9%) 97 (13%) 35 (5%) 23 (3%) 1.74 2.79
Crop Protection 22 (18%) 2 (2%) 23 (19%) 11 (9%) 13 (11%) 18 (15%) 6 (5%) 20 (15%) 5 (4%) 2.05 3.27
Bean innovation 73 (16%) 8 (2%) 241 (52%) 9 (2%) 49 (11%) 41 (9%) 6 (1%) 28 (6%) 6 (6%) 1.97 3.16
Crop-livestock integration 84 (23%) 15 (4%) 124 (34%) 44 (12%) 28 (8%) 46 (13%) 8 (2%) 16 (4%) 1.84 2.52
Striga-resistant sorghum 88 (24%) 15 (4%) 127 (24%) 44 (12%) 28 (8%) 46 (12%) 8 (2%) 16 (4%) 1.82 2.52
Striga-resistant maize 192 (21%) 13 (1%) 269 (30%) 29 (3%) 151 (17%) 72 (8%) 90 (10%) 60 (7%) 23 (3%) 2.20 3.26
Banana Varieties 192 (21%) 13 (1%) 269 (30%) 29 (3%) 151 (17%) 74 (8%) 90 (10%) 67 (7%) 25 (3%) 2.20 3.30
Cassava Variety 189 (26%) 27 (4%) 214 (29%) 81 (11%) 66 (9%) 97 (13%) 35 (5%) 23 (3%) 1.74 2.79
Soil erosion management 36 (22%) 13 (8%) 32 (21%) 13 (8%) 2 (1%) 20 (12%) 6 (4%) 21 (13%) 1 (1%) 17 (11%) 1.75 3.19
Post-harvest handling 27 (18%) 12 (8%) 17 (12%) 14 (10%) 10 (7%) 21 (14%) 6 (4%) 21 (14%) 1 (1%) 17 (12%) 2.07 3.20
Integrated Water Management 245 (30% 9 (1%) 266 (33%) 24 (3%) 32 (4%) 90 (11%) 10 (1%) 97 (12%) 3 (0%) 30 (4%) 1.67 3.46
Integrated Soil Fertility Mgt 278 (33%) 52 (6%) 261 (31%) 70 (8%) 33 (4%) 68 (8%) 6 (1%) 44 (5%) 30 (4%) 1.60 2.73

WARINDA ET AL.
WARINDA ET AL. 29
4.3.2 | Dynamics of spillover effects of technologies and innovations
Regional projects significantly contribute to wider spillover effects to other farmers and across sectors. Key com-
modities with faster spillover and spill-ins across countries include cassava, millet, striga-resistant sorghum, climbing
and bush beans, and low-cost tissue culture banana varieties. In some cases, spill-ins contributes to over 100%
increase in productivity compared to previous varieties planted before adoption of these technologies. Key factors
impacting spillovers and spill-ins are: village characteristics (more heterogeneous villages with mixed feelings regard-
ing the proposed agricultural practices lag behind in adoption of availed technologies and innovations); household size
(larger households adopting new food production-enhancing technologies faster); education (more educated farmers
opting for better technologies faster); farmers' skills (improved skills resulting in better outcomes) and community
meetings (more instructive meetings, higher adoption of new technologies). Field work shows that the most effective
approaches for farmer-to-farmer exchange of ideas and materials and freely sharing among community members is
during the local community meetings. These meetings provide platforms for farmers to showcase their successes
and share their challenges.
5 | CONCLUSIONS AND POLICY RECOMMENDATIONS
The paper shows that regional agricultural projects generate economic and capacity building benefits to farmers and
other stakeholders and public benefits in agriculture policy environment. The economic benefits include gains in crop
yield, livestock productivity, farm- and non-farm income and food security outcomes.
Benefits in capacity building include enhanced farmer participation in innovation platforms, leading to access to
improved technologies, early maturity and harvest, better soil fertility and effective water management, greater value
addition and participation in markets and increased awareness, confidence and better outlook towards adoption of
improved agricultural technologies. Fortified maize, OFSP, striga-resistant sorghum, and low-cost tissue culture
banana varieties significantly benefit farmers, with productivity exceeding 100% above those varieties planted
before adoption of new technologies. Through adoption of effective land and water management technologies,
OFSP yield increased from 2.8 tons (USD 2,840 per ha) to over 7.5 tons (USD 7,410 per ha) in two seasons in Kenya.
Utilization of integrated soil fertility, water management and livestock feed-enhancing technologies in Uganda and
Tanzania improve milk production from 6 to 11 L, and from 6 to 8 L per cow per day, respectively with greater milk
sales and consumption at household level. Beneficiaries generated monthly incomes of USD 115 per cow above
non-beneficiaries and record greater increases in the number of improved cattle breeds compared to non-beneficia-
ries. Given that beneficiaries in Tanzania and Uganda sell at least 75% of the milk produced, these findings show that
beneficiary households retain up to 2.75 L per cow per day (up from 1.5 L) in Uganda, and 2.0 L per cow per day
(up from 1.5 L) in Tanzania for own consumption. This brings beneficiary household members in Uganda and Tanza-
nia to a daily milk supply of 0.46 L and 0.33 L, respectively. This compares closely with the European minimum daily
threshold of 0.5 L. The findings further show that milk production from regional projects cross national milk produc-
tion threshold in Uganda and Tanzania (currently standing at less than 0.038 L per day in Tanzania compared to
0.066 L per day in Uganda).
Our results corroborate with previous findings regarding economic benefits of interventions in QPM (Gunaratna,
Groote, Nestel, Pixley, & McCabe, 2010), OFSP (Cole et al., 2016), biofortified iron beans (Murekezi et al., 2017),
bananas (Shimwela et al., 2016), crop-livestock integration (Mottet et al., 2017), improved livestock breeds (Argent
et al., 2014) and farmer capacity building (Davis et al., 2012) to sync sowing time with rainfall (Mottet et al., 2017),
investments in sustainable soil and water management (Shikuku et al., 2017) and greater participation of non-state
actors (Shin et al., 2017).
Further gains in crop-livestock productivity are possible but require investments to tackle bigger challenges fac-
ing sustainable agriculture and natural resource management. Value productivity per hectare varies significantly
30 WARINDA ET AL.
among crops and is mainly attributed to a farmers' dependence on rainfed agriculture, limited access to water for
supplemental irrigation, weather-related changes and fear of pest and disease outbreaks. This shows there is need
for greater investments that target innovation platforms like water management solutions, sustainable intensification
of crop-livestock system, banana-coffee system, maize-bean system, and sorghum-small stock development.
Several capacity building impacts are notable beyond smallholder farmers' level, especially improved national
human resources for regional development through short- and long-term trainings on emerging regional and national
agricultural challenges such as maize lethal necrosis disease, strengthened national institutions for research through
provision of appropriate infrastructure and equipment like essential laboratory chemicals for conducting research,
and strengthened regional, national and local institutions to establish and host innovation platforms.
Regional agricultural development interventions also deliver notable benefits in agriculture policy environment
through coordinating inter-country trade policies and harmonization of standards, regulations and streamlined cross-
border procedure for rapid transit of food commodities and food aid. Instead of operating in knowledge silos, the
regional approach to agricultural development provides a shared platform to discuss emerging challenges including
drought mitigation plans, cross-border disease management, biosecurity and food security challenges.
The main challenges faced in implementing regional projects in EAC are low crop and livestock productivity, high
(over 30%) post-harvest handling losses by non-beneficiaries, limited access to credit and markets, limited knowledge
and information, inadequate suitable water and soil fertility management technologies and nascent policy
framework.
Strategic recommendations include investing in: (a) regional agricultural systems for adoption and scaling up of
improved technologies; (b) local and region-wide partnerships and capacity building on sustainable land and water
management through private sector engagement; (c) MEL system during the design of regional projects; and
(d) national and regional policy analysis, bi-annual joint sector reviews and regional water policy dialogues.
ACKNOWLEDGEMEN TS
We thank several reviewers for comments and colleagues in many universities and institutes for their support.
RE FE R ENC E S
Abadie, A. (2005). Semiparametric difference-in-differences estimators. Review of Economic Studies, 72, 1–19.
Abro, Z. A., Alemu, B. A., & Hanjra, M. A. (2014). Policies for agricultural productivity growth and poverty reduction in rural
Ethiopia. World Development, 59, 461–474.
AfDB. (2016). Feed Africa - Strategy for agricultural transformation in Africa: 2016 – 2025. African Development Bank.
Ahlerup, P., Baskaran, T., & Bigsten, A. (2017). Regional development and national identity in sub-Saharan Africa. Journal of
Comparative Economics, 45, 622–643.
Ali, D. A., Deininger, K., & Goldstein, M. (2014). Environmental and gender impacts of land tenure regularization in Africa:
Pilot evidence from Rwanda. Journal of Development Economics, 110, 262–275.
Alwang, J., Gotor, E., Thiele, G., Hareau, G., Jaleta, M., & Chamberlin, J. (2017). Pathways from research on improved staple
crop germplasm to poverty reduction for smallholder farmers. Agricultural Systems, 172, 16–27. https://doi.org/10.
1016/j.agsy.2017.1010.1005
Argent, J., Augsburg, B., & Rasul, I. (2014). Livestock asset transfers with and without training: Evidence from Rwanda. Jour-
nal of Economic Behavior & Organization, 108, 19–39.
Asfaw, S., Shiferaw, B., Simtowe, F., & Lipper, L. (2012). Impact of modern agricultural technologies on smallholder welfare:
Evidence from Tanzania and Ethiopia. Food Policy, 37, 283–295.
Brandt, P., Kvakic, M., Butterbach-Bahl, K., & Rufino, M. C. (2017). How to target climate-smart agriculture? Concept and
application of the consensus-driven decision support framework “targetCSA”. Agricultural Systems, 151, 234–245.
Bunce, M., Brown, K., & Rosendo, S. (2010). Policy misfits, climate change and cross-scale vulnerability in coastal Africa:
how development projects undermine resilience. Environmental Science & Policy, 13, 485–497.
Bussmann, A., Elagib, N. A., Fayyad, M., & Ribbe, L. (2016). Sowing date determinants for Sahelian rainfed agriculture in the
context of agricultural policies and water management. Land Use Policy, 52, 316–328.
Cervigni, R., & Morris, M. (2015). Confronting drought in Africa's drylands: Opportunities for enhancing resilience. Washington,
DC: IBRD/World Bank.
Christiaensen, L. (2017). Agriculture in Africa – Telling myths from facts: A synthesis. Food Policy, 67, 1–11.
WARINDA ET AL. 31
Claessens, L., Antle, J. M., Stoorvogel, J. J., Valdivia, R. O., Thornton, P. K., & Herrero, M. (2012). A method for evaluating cli-
mate change adaptation strategies for small-scale farmers using survey, experimental and modeled data. Agricultural Sys-
tems, 111, 85–95.
Cole, D. C., Levin, C., Loechl, C., Thiele, G., Grant, F., Girard, A. W., … Low, J. (2016). Planning an integrated agriculture and
health program and designing its evaluation: Experience from Western Kenya. Evaluation and Program Planning, 56,
11–22.
Davis, K., Nkonya, E., Kato, E., Mekonnen, D. A., Odendo, M., Miiro, R., & Nkuba, J. (2012). Impact of farmer field schools on
agricultural productivity and poverty in East Africa. World Development, 40, 402–413.
Dawson, N., Martin, A., & Sikor, T. (2016). Green revolution in Sub-Saharan Africa: Implications of imposed innovation for
the wellbeing of rural smallholders. World Development, 78, 204–218.
de Valença, A. W., Bake, A., Brouwer, I. D., & Giller, K. E. (2017). Agronomic biofortification of crops to fight hidden hunger
in sub-Saharan Africa. Global Food Security, 12, 8–14.
Dixon, J., Garrity, D., Boffa, J.-M., Williams, T., Amede, T., Auricht, C., … Mburathi, G. (Eds.). (2017). Farming systems and food
security in Africa: Priorities for science and policy under global change. London and New York: Routledge.
Donald, S. G., & Lang, K. (2007). Inference with difference-in-differences and other panel data. The Review of Economics and
Statistics, 89, 221–233.
Doss, C., Meinzen-Dick, R., Quisumbing, A., & Theis, S. (2017). Women in agriculture: Four myths. Global Food Security, 16,
69–74. https://doi.org/10.1016/j.gfs.2017.1010.1001
EAC. (2015). EAC Facts and Figures Report. Arusha, Tanzania: East African Community.
Feder, G., Just, R. E., & Zilberman, D. (1985). Adoption of Agricultural Innovations in Developing Countries: A Survey. Eco-
nomic Development and Cultural Change, 33, 255–298.
Ferede, T., Ayenew, A. B., & Hanjra, M.A. (2013). Agroecology matters: Impacts of climate change on agriculture and its
implications for food security in Ethiopia. Retrieved from http://www.oecd.org/env/cc/TADELE_AGROECOLOGY%
20MATTERS%20(revised).pdf
Fermont, A. M., van Asten, P. J. A., Tittonell, P., van Wijk, M. T., & Giller, K. E. (2009). Closing the cassava yield gap: An analy-
sis from smallholder farms in East Africa. Field Crops Research, 112, 24–36.
Foster, A. D., & Rosenzweig, M. R. (2010). Microeconomics of technology adoption. Annual Review of Economics, 2,
395–424.
Friis-Hansen, E., & Duveskog, D. (2012). The empowerment route to well-being: An analysis of farmer field schools in East
Africa. World Development, 40, 414–427.
Geda, A., & Seid, E. H. (2015). The potential for internal trade and regional integration in Africa. Journal of African Trade, 2,
19–50.
Ghiron, L., Shillingi, L., Kabiswa, C., Ogonda, G., Omimo, A., Ntabona, A., … Fajans, P. (2014). Beginning with sustainable scale
up in mind: initial results from a population, health and environment project in East Africa. Reproductive Health Matters,
22, 84–92.
Gitonga, Z. M., De Groote, H., Kassie, M., & Tefera, T. (2013). Impact of metal silos on households' maize storage, storage
losses and food security: An application of a propensity score matching. Food Policy, 43, 44–55.
Gunaratna, N. S., Groote, H. D., Nestel, P., Pixley, K. V., & McCabe, G. P. (2010). A meta-analysis of community-based stud-
ies on quality protein maize. Food Policy, 35, 202–210.
Habyarimana, J. B., & Nkunzimana, T. (2017). Policy reforms and rural livelihoods sustainability: Challenges and opportuni-
ties — Empirical evidence from the adoption of the Land Use Consolidation (LUC) policy in Rwanda. African Development
Review, 29, 96–108.
Hammond, J., Fraval, S., van Etten, J., Suchini, J. G., Mercado, L., Pagella, T., … van Wijk, M. T. (2017). The Rural Household
Multi-Indicator Survey (RHoMIS) for rapid characterisation of households to inform climate smart agriculture interven-
tions: Description and applications in East Africa and Central America. Agricultural Systems, 151, 225–233.
Hanjra, M. A., Ferede, T., & Gutta, D. G. (2009a). Pathways to breaking the poverty trap in Ethiopia: Investments in agricul-
tural water, education, and markets. Agricultural Water Management, 96, 1596–1604.
Hanjra, M. A., Ferede, T., & Gutta, D. G. (2009b). Reducing poverty in sub-Saharan Africa through investments in water and
other priorities. Agricultural Water Management, 96, 1062–1070.
Hanjra, M. A., & Gichuki, F. (2008). Investments in agricultural water management for poverty reduction in Africa: Case stud-
ies of Limpopo, Nile, and Volta river basins. Natural Resources Forum, 32, 185–202.
Hanjra, M. A., Noble, A., Langan, S., & Lautze, J. (2017). Feeding the 10 billion within the sustainable development goals
framework. In I. J. Gordon, H. H. T. Prins, & G. R. Squire (Eds.), Food production and nature conservation: Conflicts and
solutions (pp. 15–40). London, UK & New York: Earthscan (Routledge).
Heckman, J., Ichimura, H., Smith, J., & Todd, P. (1998). Characterizing selection bias using experimental data. Econometrica,
66, 1017–1098.
32 WARINDA ET AL.
Heggum, C. (2016). Policy schemes and trade in dairy products: Codex alimentarius. Encyclopedia of Dairy Sciences, 2011,
312–321.
Hisali, E., Birungi, P., & Buyinza, F. (2011). Adaptation to climate change in Uganda: Evidence from micro level data. Global
Environmental Change, 21, 1245–1261.
HLPE. (2015). Water for food security and nutrition. A report by the High Level Panel of Experts on Food Security and
Nutrition of the Committee on World Food Security, Rome 2015.
Isaacs, K. B., Snapp, S. S., Chung, K., & Waldman, K. B. (2016). Assessing the value of diverse cropping systems under a new
agricultural policy environment in Rwanda. Food Security, 8, 491–506.
Jarvis, A., Ramirez-Villegas, J., Herrera Campo, B. V., & Navarro-Racines, C. (2012). Is cassava the answer to African climate
change adaptation? Tropical Plant Biology, 5, 9–29.
Jayne, T. S., Chamberlin, J., & Headey, D. D. (2014). Land pressures, the evolution of farming systems, and development
strategies in Africa: A synthesis. Food Policy, 48, 1–17.
Jayne, T. S., Mason, N. M., Burke, W. J., & Ariga, J. (2018). Review: Taking stock of Africa's second-generation agricultural
input subsidy programs. Food Policy, 75, 1–14.
Johnson, N. L., Kovarik, C., Meinzen-Dick, R., Njuki, J., & Quisumbing, A. (2016). Gender, assets, and agricultural develop-
ment: Lessons from eight projects. World Development, 83, 295–311.
Kaitibie, S., Omore, A., Rich, K., & Kristjanson, P. (2010). Kenyan dairy policy change: Influence pathways and economic
impacts. World Development, 38, 1494–1505.
Kristjanson, P., Neufeldt, H., Gassner, A., Mango, J., Kyazze, F. B., Desta, S., … Coe, R. (2012). Are food insecure smallholder
households making changes in their farming practices? Evidence from East Africa. Food Security, 4, 381–397.
Langyintuo, A. S., Mwangi, W., Diallo, A. O., MacRobert, J., Dixon, J., & Bänziger, M. (2010). Challenges of the maize seed
industry in eastern and southern Africa: A compelling case for private–public intervention to promote growth. Food Pol-
icy, 35, 323–331.
Latif, S., & Müller, J. (2015). Potential of cassava leaves in human nutrition: A review. Trends in Food Science & Technology,
44, 147–158.
Littnerova, S., Jarkovsky, J., Parenica, J., Pavlik, T., Spinar, J., & Dusek, L. (2013). Why to use propensity score in observa-
tional studies? Case study based on data from the Czech clinical database AHEAD 2006–09. Cor et Vasa, 55,
e383–e390.
Lynum, J., Beintema, N., Roseboom, J., & Badiane, O. (Eds.). (2016). Agricultural research in Africa: Investing in future harvests.
Washington, DC: IFPRI.
Mayoux, L., & Chambers, R. (2005). Reversing the paradigm: Quantification, participatory methods and pro-poor impact
assessment. Journal of International Development, 17, 271–298.
Mendola, M. (2007). Agricultural technology adoption and poverty reduction: A propensity-score matching analysis for rural
Bangladesh. Food Policy, 32, 372–393.
Mogues, T., Fan, S., & Benin, S. (2015). Public investments in and for agriculture. European Journal of Development Research,
27, 337–352.
Mottet, A., Henderson, B., Opio, C., Falcucci, A., Tempio, G., Silvestri, S., … Gerber, P. J. (2017). Climate change mitigation
and productivity gains in livestock supply chains: insights from regional case studies. Regional Environmental Change, 17,
129–141.
Murekezi, A., Oparinde, A., & Birol, E. (2017). Consumer market segments for biofortified iron beans in Rwanda: Evidence
from a hedonic testing study. Food Policy, 66, 35–49.
Mwongera, C., Shikuku, K. M., Twyman, J., Läderach, P., Ampaire, E., Van Asten, P., … Winowiecki, L. A. (2017). Climate
smart agriculture rapid appraisal (CSA-RA): A tool for prioritizing context-specific climate smart agriculture technologies.
Agricultural Systems, 151, 192–203.
Nieves, J. J., Stevens, F. R., Gaughan, A. E., Linard, C., Sorichetta, A., Hornby, G., … Tatem, A. J. (2017). Examining the corre-
lates and drivers of human population distributions across low- and middle-income countries. Journal of the Royal Society
Interface, 14, 0401–0419.
Notenbaert, A., Pfeifer, C., Silvestri, S., & Herrero, M. (2017). Targeting, out-scaling and prioritising climate-smart interven-
tions in agricultural systems: Lessons from applying a generic framework to the livestock sector in sub-Saharan Africa.
Orenstein, D. E., & Shach-Pinsley, D. (2017). A comparative framework for assessing sustainability initiatives at the regional
scale. World Development, 98, 245–256.
Paul, B. K., Frelat, R., Birnholz, C., Ebong, C., Gahigi, A., Groot, J. C. J., … van Wijk, M. T. (2017). Agricultural intensification
scenarios, household food availability and greenhouse gas emissions in Rwanda: Ex-ante impacts and trade-offs. Agricul-
tural Systems, 163, 16–26. https://doi.org/10.1016/j.agsy.2017.1002.1007
Pingali, P. L. (2012). Green revolution: Impacts, limits, and the path ahead. Proceedings of the National Academy of Sciences,
109, 12302–12308.
WARINDA ET AL. 33
Rawlins, R., Pimkina, S., Barrett, C. B., Pedersen, S., & Wydick, B. (2014). Got milk? The impact of Heifer International's live-
stock donation programs in Rwanda on nutritional outcomes. Food Policy, 44, 202–213.
REMA. (2015). Rwanda State of Environment Outlook Report 2015, Greening agriculture with resource efficient, low carbon and
climate resilient practices. Kigali, Rwanda: Rwanda Environment Management Authority (REMA), Government of
Rwanda.
Rockström, J., Williams, J., Daily, G., Noble, A., Matthews, N., Gordon, L., … Smith, J. (2017). Sustainable intensification of
agriculture for human prosperity and global sustainability. Ambio, 46, 4–17.
Rohrbach, D. D., Minde, I. J., & Howard, J. (2003). Looking beyond national boundaries: Regional harmonization of seed poli-
cies, laws and regulations. Food Policy, 28, 317–333.
Rushemuka, N. P., Bizoza, R. A., Mowo, J. G., & Bock, L. (2014). Farmers' soil knowledge for effective participatory integrated
watershed management in Rwanda: Toward soil-specific fertility management and farmers' judgmental fertilizer use.
Agriculture, Ecosystems & Environment, 183, 145–159.
Sadoff, C. W., Hall, J. W., Grey, D., Aerts, J. C. J. H., Ait-Kadi, M., Brown, C., … Wiberg, D. (2015). Securing water, sustaining
growth: Report of the GWP/OECD Task Force on Water Security and Sustainable Growth, University of Oxford, UK,
180 pp.
Shikuku, K. M., Winowiecki, L., Twyman, J., Eitzinger, A., Perez, J. G., Mwongera, C., & Läderach, P. (2017). Smallholder
farmers' attitudes and determinants of adaptation to climate risks in East Africa. Climate Risk Management, 16, 234–245.
Shimwela, M. M., Ploetz, R. C., Beed, F. D., Jones, J. B., Blackburn, J. K., Mkulila, S. I., & van Bruggen, A. H. C. (2016). Banana
xanthomonas wilt continues to spread in Tanzania despite an intensive symptomatic plant removal campaign: an
impending socio-economic and ecological disaster. Food Security, 8, 939–951.
Shin, W., Kim, Y., & Sohn, H.-S. (2017). Do different implementing partnerships lead to different project outcomes? Evidence
from the World Bank project-level evaluation data. World Development, 95, 268–284.
Stevenson, J. R., Villoria, N., Byerlee, D., Kelley, T., & Maredia, M. (2013). Green Revolution research saved an estimated
18 to 27 million hectares from being brought into agricultural production. Proceedings of the National Academy of Sci-
ences of the United States of America, 110, 8363–8368.
Stuart, E. (2010). Matching methods for causal inference: A review and a look forward. Statistical Science, 25, 1–21.
Stuart, E. A., Huskamp, H., Duckworth, K., Simmons, J., Song, Z., Chernew, M., & Barry, C. L. (2014). Using propensity scores
in difference-in-differences models to estimate the effects of a policy change. Health Services and Outcomes Research
Methodology, 14, 166–182.
Tachie-Obeng, E., Akponikpè, P. B. I., & Adiku, S. (2013). Considering effective adaptation options to impacts of climate
change for maize production in Ghana. Environmental Development, 5, 131–145.
Thoemmes, F. (2012). Propensity score matching in SPSS. Retrieved from http://sourceforge.net/projects/psmspss/files/
Timmer, C. P. (2017). Food security, structural transformation, markets and government policy. Asia & the Pacific Policy Stud-
ies, 4, 4–19.
Tittonell, P., Muriuki, A., Shepherd, K. D., Mugendi, D., Kaizzi, K. C., Okeyo, J., … Vanlauwe, B. (2010). The diversity of rural
livelihoods and their influence on soil fertility in agricultural systems of East Africa – A typology of smallholder farms.
UN. (2016). Progress towards the sustainable development goals: Report of the secretary-general. New York, NY: United
Nations.
USAID. (2013). 2013 USAID office of food for peace food security country framework for Burundi (FY 2014–FY 2019).
Washington, DC: FHI 360/FANTA.
van Asten, P. J. A., Wairegi, L. W. I., Mukasa, D., & Uringi, N. O. (2011). Agronomic and economic benefits of coffee–banana
intercropping in Uganda's smallholder farming systems. Agricultural Systems, 104, 326–334.
Waha, K., Müller, C., Bondeau, A., Dietrich, J. P., Kurukulasuriya, P., Heinke, J., & Lotze-Campen, H. (2013). Adaptation to cli-
mate change through the choice of cropping system and sowing date in sub-Saharan Africa. Global Environmental
Change, 23, 130–143.
Walker, T., Ryan, J., & Kelley, T. (2010). Impact assessment of policy-oriented international agricultural research: Evidence
and insights from case studies. World Development, 38, 1453–1461.
Wichelns, D. (2015). Achieving water and food security in 2050: Outlook, policies, and investments. Agriculture, 5, 188–220.
Williams, T. O. (2015). Reconciling food and water security objectives of MENA and sub-Saharan Africa: Is there a role for
large-scale agricultural investments? Food Security, 7, 1199–1209.
Williams, T. P. (2017). The political economy of primary education: Lessons from Rwanda. World Development, 96, 550–561.
Wood, S., & Chamberlin, J. (2003). Enhancing the role of spatial analysis in strategic impact assessment: Improving data reso-
lution for regional studies. Quarterly Journal of International Agriculture, 42, 167–187.
World Bank. (2008). World Development Report 2008: Agriculture for development. Washington, DC: The World Bank.
34 WARINDA ET AL.
How to cite this article: Warinda E, Nyariki DM, Wambua S, Muasya RM, Hanjra MA. Sustainable
development in East Africa: impact evaluation of regional agricultural development projects in Burundi,
Kenya, Rwanda, Tanzania, and Uganda. Nat Resour Forum. 2020;44:3–39. https://doi.org/10.
1111/1477-8947.12191
WARINDA ET AL. 35
TABLE A1 Regional projects sampled across domains in Eastern Africa
Project BU KN RW TZ UG Domain
1. Accelerated uptake & utilization of soil fertility management practices X x x LLL
2. Scaling up integrated soil fertility management for improved livelihoods X x x x LLL
3. Applying tissue culture to improve access to cassava and sweet potato x X x x x HHH
clean planting materials for farmers
4. Enhancing utilization of quality seed potato by small-scale farmers x X x HLH
5. Scaling up technologies in orange fleshed sweet potato using X x x HLL
agricultural innovation system to address food and nutrition security
6. Transfer and dissemination of proven and emerging agricultural X x x x HHH
technologies in orange fleshed sweet potato
7. Enhancing competitiveness of snap beans for domestic/export markets X x x x HLL
8. Promotion of snap bean integrated crop management technologies for X x x HHH
improved livelihoods
9. Utilization of bean innovations for food security and improved x x x HHH
livelihoods
10. Development and dissemination of normal and nutritionally enhanced x X x x x HLH
highland maize varieties
11. Development and dissemination of quality protein maize X x x LLL
agro-enterprises for improved household income
12. Increasing sorghum utilization and marketability through food product X x x HLL
diversification
13. Sustainable intensification of sorghum-legume system to improve X x x LLL
livelihood and adaptation to climate change in semi-arid areas of
Eastern and Central Africa
14. Crop-livestock integration for sustainable management of natural x X x HHH
resources
15. Harnessing crop-livestock integration to build livelihoods resilience x X x x HLH
16. Tick and tick-borne diseases prevention and control x X x x HHH
17. Enhanced management of banana Xanthomonas wilt for sustainable x X x x x HLL
banana productivity
18. Integrated management of cassava brown streak disease and cassava x X x x x LLL
mosaic disease for enhanced productivity and utilization of cassava
19. Efficient use of crop residues: animal feed versus conservation X x x HLH
agriculture
20. Innovative approaches for mainstreaming integrated natural resource X x x LLL
management in agricultural research and development institutions
21. Promoting sustainable natural resource management through X x x LLL
effective governance and farmer market linkages
22. Exploiting market opportunities for value add dairy and meat product X x x x HLH
23. Farmer in the driver's seat: farmer empowerment for innovation in x X x HHH
smallholder agriculture
Note: Domain = Agricultural Development Domains, based on Agricultural Potential, Market Access, and Population
Density, respectively where H stands for high and L for low.
Abbreviations: BU, Burundi; KN, Kenya; RW, Rwanda; TZ, Tanzania; UG, Uganda.
Source: Authors.
36 WARINDA ET AL.
FIGURE A1 Distribution of Propensity Scores. Source: Based on survey data (2015)
F I G U R E A 2 Distribution of
Propensity Scores. Source: Based on
survey data (2015)
WARINDA ET AL. 37
F I G U R E A 3 Standardized differences
before matching of covariates. Source: Based
on survey data (2015)
38 WARINDA ET AL.
F I G U R E A 4 Dotplot of standardized
mean differences (Cohen's d) for all covariates
before and after matching. A1 = type of
respondent (beneficiary/non-beneficiary);
B1 = Gender; B4 = Age; B7 = Education;
B20 = Hires laborers; C1 = member of farmer
organization; D53 = access to credit facilities;
E3 = land size; E11 = land under technologies;
E16 = type of farming (monocropping or
mixed cropping); G1 = crop productivity.
Source: Based on survey data (2015)
WARINDA ET AL. 39
F I G U R E A 5 Lineplot of standardized
differences before and after matching.
Source: Based on survey data (2015)

Impact_Eval_RP

Uploaded by

Copyright:

Available Formats

Impact_Eval_RP

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Impact_Eval_RP

Uploaded by

Copyright:

Available Formats

Received: 29 July 2017 Revised: 27 January 2019 Accepted: 6 March 2019

Sustainable development in East Africa: impact

Enock Warinda1 | Dickson M Nyariki2 | Stephen Wambua3 |

Nat Resour Forum. 2020;44:3–39. wileyonlinelibrary.com/journal/narf © 2020 United Nations 3

management technologies increased the potato yields from

beneficiaries, capacity building, economic benefits, food security,

3 | DATA AND METHODS

3.1 | Study region

3.2 | Data collection and analysis

Economic indicator Burundi Kenya Rwanda Tanzania Uganda Overall

Source: EAC (2015) and others.

eðxÞ = pðz = 1jxÞ ð1Þ

We estimate the propensity score from the logistic regression as:

3.3 | Impact evaluation approach

3.4 | Impact evaluation model specification

We determine value added in agriculture for beneficiaries as:

Costsα,i = FCα,i + VCα,i ð12Þ

We determine value added in agriculture for non-beneficiaries as:

ðΔÞ = VAα −VAβ ð15Þ

4 | RESULTS AND DISCUSSION

4.1 | Economic benefits of regional projects

4.1.1 | Dynamics of income from regional projects

TABLE 2 Framework for calculating Difference-in-Differences (DiD)

Pre (2010) Post (2015) Post-pre difference (2015–2010) = (ϕ^1 )

Treatment – Control (T-C) Difference (ϕ^2 Þ Y RP NP

Source: Author's conceptualization.

TABLE 3 Logistic regression predicting increase in farm incomes

95% C.I. for Exp)B)

B S.E. Wald df Sig. Exp(B) Lower Upper

Source: Authors, based on survey data (2015).

FIGURE 1 Propensity scores based on farm incomes for matched respondents

F I G U R E 3 Average post-pre differences in farm incomes (US$/household/year) between respondents. Source:

T A B L E 4 Average post-pre differences in farm incomes at household level (US$/household/year) between

Burundi Kenya Rwanda Tanzania Uganda

Abbreviations: Ben, Beneficiaries; NBen, Non-beneficiaries.

4.1.2 | Gender and income dynamics

F I G U R E 4 Average post-pre differences in non-farm incomes (US$/household/year) between respondents.

TABLE 5 Total income DiD between beneficiaries and non-beneficiaries (US$/household/year)

Post-pre difference (2015–2010) = (ϕ^1 )

Burundi Kenya Rwanda Tanzania Uganda

Source: Own calculation.

4.1.3 | Comparison between projects and agriculture innovations

Burundi Kenya Rwanda Tanzania Uganda

4.1.4 | Dynamics of value productivity for selected crops

4.1.5 | Sustainable land and water management

The percentage yield impact from adoption of a technology is measured as follows:

4.1.6 | Dynamics of agricultural value addition

TABLE 7 Value productivity per hectare for food commodities (US$/ha/year)

4.2 | Capacity building benefits of regional projects

4.2.1 | Dynamics of partnerships for enhancing productivity

4.2.2 | Satisfaction with membership in organizations and technologies

Kenya Rwanda Tanzania Uganda Burundi Post-pre difference

Abbreviations: Ben, Beneficiaries; NBen, Non-beneficiaries.

TABLE 9 Estimates of average annual value added (US$/household/year)

Abbreviations: Ben, Beneficiaries; NBen, Non-beneficiaries.

FIGURE 11 Average value added (US$/household/year). Source: Own calculation

4.2.3 | Participation in potential future projects

4.3 | Benefits in agriculture policy and planning

T A B L E 1 0 Level of satisfaction with

Source: Survey data (2015).