Effects of Irrigation on Smallholder Welfare: Evidence from Haiti

Effects of Irrigation on Smallholder Welfare: Evidence from Haiti

Food insecurity and hunger remain thorny challenges in Haiti, resulting from the combination of political instability, environmental degradation, poverty and natural disasters (Perez-Escamilla et al., 2009; Ivers et al., 2010; Sample, 2016). Crop yields are low even compared to the neighboring Caribbean countries resulting in low food availability. Little food production is compounded by rural poverty to limit food access for roughly 50% of Haiti’s population (USAID, 2017). Declining yields in the major staple crops, maize and rice (Ray et al., 2013) do not bode well for future food security.

The adoption literature in developing countries is replete with studies of divisible technologies such as high-yielding seed varieties and fertilizer (see Kaliba, Verkuijl, and Mwangi (2000), Gine and Yang (2009), and Duo, Kremer, and Robinson (2011)). Less is known about indivisible technologies, such as irrigation. Efficient use of indivisible technologies relies on group spillovers. In developing settings where most farmers operate farmlands under 1 hectare in size[1], not every household will be able to afford to own an irrigation pump that can irrigate ten times that land area. Economically attractive alternatives to allow smallholders access to irrigation include collective ownership (farmer groups) or market-based solutions involving private operators (perhaps large farmers who can share the technology with other farmers). Understanding how this latter arrangement may work, its effect on neighboring farmers, and which economic constraints may limit its potential is the primary goal of this study.

To improve food security in targeted geographical areas, the Haitian Ministry of Agriculture (MARNDR) is implementing a comprehensive agricultural support project titled Projet de Reinforcement des Services Publics Agricoles (RESEPAG II)[2], with the support of the World Bank group (WBG) and the Global Agriculture and Food Security Program (GAFSP). Together with other principal investigators at the World Bank, we proposed to add an intervention on irrigation. Starting in early spring of 2018, farmers listed on a national farmers’ registry will apply for co-financing towards various inputs and technologies, including irrigation pumps. This study focuses on the irrigation pumps component.

Research Question

In many developing countries, expanding irrigation holds the potential to substantially increase agricultural yields, value and provide resilience to rain shortfalls. Because most irrigation schemes are large and placed in chosen locations, little causal evidence exists on the effect of irrigation on household welfare. Further, identifying the causal effect of irrigation on household outcomes is complicated by confounding attributes of the location and farmers involved. Thus, little causal evidence is available. This project is exploring the diffusion and the effects of access to irrigation services by smallholders on agricultural productivity, household food security, crop diversification and rural poverty. It also seeks to examine the interaction of irrigation with existing market access.

We propose to answer the following specific questions:

(I) What are the effects of enabling access to irrigation services on the recipients?

(ii) What factors affect the adoption of irrigation services by these farmers’ neighbors and how does access to irrigation services affect their outcomes?

(iii) How do market access and other economic characteristics impact these effects?

These effects and their underlying mechanisms will be investigated based on the following theory of change.

Research Design and Identification Strategy

With colleagues at the World Bank, we have designed an RCT to distribute irrigation pumps to randomly selected farmers in randomly selected farm clusters. These clusters are created based on the geographical distribution of the farm plots of the eligible applicants for the pumps. Each treated cluster has about 500 meters in radius and a distance of 2 Km is maintained between centroids of any treated cluster and its corresponding control cluster.

All farmers on the national registry who express interest in the project are considered eligible. Pumps will be allocated to about 120 randomly selected clusters. One irrigation pump will be randomly allocated to one of the applicants per cluster. Recipients are expected to become local providers of irrigation services within their respective clusters.

The intervention is anticipated to help alleviate poverty among smallholder farmers through various mechanisms such as allowing for the production of high-value crops (e.g., horticulture) in the dry season and better market participation by the households. Data will be collected both at baseline (beginning of the rainy season) and endline (end of the dry season) so that a Difference-in-Differences method can be employed to estimate the causal effects of access to irrigation services. An additional module on social networks is also planned.


Berdegue, Julio A and Ricardo Fuentealba (2011). “Latin America: The state of smallholders in agriculture". IFAD conference on new directions for smallholder agriculture. Vol. 24, p. 25.
Duo, Esther, Michael Kremer, and Jonathan Robinson (2011). \Nudging farmers to use fertilizer: Theory and experimental evidence from Kenya". The American Economic Review 101.6, pp. 2350-2390.
Gine, Xavier and Dean Yang (2009). “Insurance, credit, and technology adoption: Field experimental evidence from Malawi". Journal of development Economics 89.1, pp. 1--11.
Ivers, Louise C, Yuchiao Chang, J Gregory Jerome, and Kenneth A Freedberg (2010).
“Food assistance is associated with improved body mass index, food security and attendance at clinic in an HIV program in central Haiti: a prospective observational cohort study". AIDS research and therapy 7.1, p. 33.
Jayne, Thomas S, Takashi Yamano, Michael T Weber, David Tschirley, Rui Benfica,
Antony Chapoto, and Ballard Zulu (2003). “Smallholder income and land distribution in Africa: implications for poverty reduction strategies". Food policy 28.3, pp. 253-275.
Kaliba, Aloyce RM, Hugo Verkuijl, and Wilfred Mwangi (2000). “Factors affecting adoption of improved maize seeds and use of inorganic fertilizer for maize production in the intermediate and lowland zones of Tanzania". Journal of Agricultural and Applied Economics 32.1, pp. 35-47.
Nagayets, Oksana (2005). “Small farms: current status and key trends". The future of small farms, 2005 Jun 26:355.
Perez-Escamilla, Rafael, Michael Dessalines, Mousson Finnigan, Helena Pachon, Amber Hromi-Fiedler, and Nishang Gupta (2009). “Household food insecurity is associated with childhood malaria in rural Haiti". The Journal of nutrition 139.11, pp. 2132-2138.
Ray, Deepak K, Nathaniel D Mueller, Paul C West, and Jonathan A Foley (2013). “Yield trends are insufficient to double global crop production by 2050". PloS one 8.6, e66428.
Sample, Drew (2016). \GHI Haiti: In Aftermath of Hurricane Matthew, Haiti Facing “Alarming" Hunger Levels Higher Than Any Other Country in the Americas". url: https://www.ifpri.org/news-release/ghi-haiti-aftermath-hurricane-matthew-haiti-facing-%E2%80%9Calarming%E2%80%9D-hunger-levels-higher-any.
USAID (2017). “Haiti: Agriculture & Food Security Fact Sheet". url: https://www.usaid.gov/sites/default/_les/documents/1862/FINAL Food Security March 2017.pdf.

[1] See Jayne et al. (2003), Nagayets (2005), and Berdegue and Fuentealba (2011).
[2] Project for the Strengthening of Agricultural Public Services.
ADM Institute Project on Postharvest Loss Prevention in India

ADM Institute Project on Postharvest Loss Prevention in India

The ADM Institute Postharvest Loss Prevention project is based in the state of Bihar in India. Bihar is one of the poorest states in India with an annual per capita income of around USD 250. Majority of the population in Bihar is rural and around 90 percent of the cropped area is under food grains (rice, wheat, maize). In a randomized control design, we work with 4000 farmers in 80 villages of 5 districts of Bihar. Around 83 percent of the farmers in the region are smallholders and own less than 1 hectare of farmland.

Map of Study Area in the state of Bihar, India

Goals of the Project

The ADM Institute Postharvest Loss Prevention project in Bihar, India aims to improve postharvest technology and management of grain quality among smallholder farmers to improve agricultural incomes, household welfare and food security. On-farm volume losses reduce the quantity of grain available for sale and consumption. Value losses in grain quality due to improper postharvest management result in low market prices, loss of nutritional value, food safety and health risks. We use a combination of randomized control trials, physical and survey measurements to evaluate PHL technology and estimate economic value of reducing PHL. Preliminary estimates show that farmers in Bihar are potentially losing up to US$23.7 per hectare (~10% of average per capita income) due to PHL.

Four significant characteristics differentiate this research from previous studies examining postharvest losses in India.

  • Sub-optimal postharvest management leads to potential negative outcomes for farmers not only through physical quantity losses but also from value lost due to poor grain quality characteristics. Most studies have focused only on characterizing the physical losses along the postharvest supply chain. This research goes a step further and offers an assessment of the economic value lost to smallholder farmer attributable to postharvest management of grains.
  • We recognize that understanding postharvest technology adoption and the resulting value to farmers needs to account for behavioral and contextual factors. Competing forces ranging from the farmers’ characteristics, village market conditions that determine incentives for grain quality, government policies affecting the grain market and other contextual factors mediate postharvest practices among farmers as well as their welfare outcomes.
  • The study design also lets us examine the factors affecting technology adoption among farmers, and how that adoption changes other postharvest behavior. This approach will help formulate specific programs for postharvest technology and practices that take advantage of driving factors and alleviate barriers that may exist.
  • We also examine the health costs of poor postharvest management and improper storage techniques that compromise the food safety standards and result in widespread aflatoxin contamination of food grains.

Project Interventions

We use a randomized control design to evaluate the impact of providing either PHL technology, information or both to farmers in Bihar. Using an auction design, we provide farmers GrainPro hermetically sealed bags to examine the impact of storage behavior, storage duration, grain quality, grain prices and food safety measures.

  • Demonstration and training of farmers on PHL issues in villages
  • 2500 subsidized hermetic grain storage bags (based on willingness to pay) distributed across 44 villages in 5 districts and 1500 free bags in 32 villages
  • Information intervention for 1100 farmers in 22 villages on aflatoxin contamination, health risks and safe storage practices to minimize contamination
  • Testing of grain samples for aflatoxin in collaboration with IFPRI and ICRISAT
  • Demonstration of STR dryers on local university campuses (RAU and BAU) and in the ADMI village
  • Measuring traders’ and end-users’ preferences for various grain quality attributes so that PHL grain improvements match market demand
  • Measuring health implications for farmers with visible symptoms of aflatoxicosis

Preliminary Findings

  • Using a field experiment, we estimate farmers are willing to pay an average US$0.40 per bag, more than three times what they currently spend on traditional jute bags ($0.12 per bag)
  • The demand increased by more than 20% when additional health and food safety information was provided
  • We find a high level of demand even from landless and marginal farmers
  • Demand is driven largely by need to store for home food consumption
  • Farmers using hermetic bags report a large decline in incidence of rodent, fungus and pest damage
  • 14% reduction in storage losses when using hermetic bags
  • 19% increase in likelihood of storing grain for food consumption
  • The bags also reduce the need for hazardous fumigants used during storage
  • Overall, 72% of farmers reported being highly satisfied with hermetic bags and 62% report they would recommend the bags to their peers
  • None of the farmers were aware of the risks of mycotoxins in grains in baseline
  • Initial aflatoxin tests of farmers’ stored grain reveal high incidence of contamination: 36% in traditional bags tested positive, while none of the samples stored in hermetic bags had any aflatoxin contamination
  • Also impacts prices received in the markets – farmers receive low prices because many high-value buyers such as large feed manufacturers and exporters may refuse to procure due to food safety risks

Project Status: Ongoing