Research

Drought-Tolerant Maize in East Africa

Improved seed technology has the potential to improve quality of life for farmers in developing countries through increased yields and greater resilience to adverse climate events. However, adoption of improved seed technology is low, especially in Eastern and Southern Africa. In one paper, I address the puzzling disconnect between promoted advantages and low adoption rates. I identify hurdles to adopting drought-tolerant maize because of challenges to learning and updating. This work is part of a larger randomized controlled trial in Tanzania and Mozambique evaluating the complementarity of index-based insurance with drought tolerant maize seed. This project is important in rural East Africa, as many farmers neither have access to improved seed nor have the financial tools to consider risky new approaches. My research shows that the cultivation of drought-tolerant maize can stabilize household productivity and reduce food insecurity; however, those who could most benefit from the technology might be the slowest to adopt it—indicating the potential for effective policies to create large gains in welfare for farmers.

My AMA Innovation Lab Spotlight covers the partnerships I developed to scale bundled index insurance and drought-tolerant maize in Tanzania.

Drought-Tolerant Maize Response to Precipitation: Evaluating On-Farm Trial Data with High Resolution Climate Data

Drought is a primary cause of crop failure, which has severe economic and welfare consequences for farmers in Eastern and Southern Africa. Improved crop varieties, such as drought-tolerant (DT) maize, can provide substantial improvements in yield level and variance. This paper assesses the performance of DT maize in On-Farm trials to identify the advantage of the DT trait under different drought conditions. I find significant heterogeneity in returns from the DT maize seed technology. This is important because variation in realized advantages might moderate the estimated large benefit from the technology.

This analysis first evaluates DT maize in comparison with other varieties without that trait. I use On-Farm Trial data from CIMMYT (The International Maize and Wheat Improvement Center) and high-resolution precipitation data.  DT maize indeed has a significant 10% yield gain, on average. Further, DT maize is less sensitive to drought than comparison varieties, demonstrating a significant protective effect. I then investigate how the DT trait advantage varies from more and less controlled environments. There are demonstrated high returns to the technology in the very-controlled On-Station Trials. There is heterogeneity in the advantages in the less-controlled On-Farm Trials. I further analyze this heterogeneity using a quantile regression approach. For low-performing farms, those with poor soils and agronomic practices, the protective effect is reduced in magnitude and statistical significance.

My AAEA 2017 poster on drought-tolerant maize response to precipitation in East Africa uses on-farm trial data from CIMMYT.

The Impact of Availability Bias in Drought Expectations on Adoption of Drought-Tolerant Maize

Traditional varieties of maize frequently fail due to drought pressure in East Africa, where maize is the dominant staple. Yet farmers are slow to adopt new drought-tolerant (DT) varieties that can improve household welfare through increased yield stability and income security. A farmer’s adoption decision is based on his subjective beliefs about the likelihood of drought and the crop response to drought stress. Farmer’s expectations of drought likelihood depend on experience and cognitive biases. Early droughts and very severe droughts can mask the DT trait advantage, complicating experience with the seed technology. I develop a model of the adoption decision to show the role of information in expectations, giving insight into the limited uptake. Then, combining rich household survey data with high-resolution climate and yield data, I evaluate adoption behavior against simulations. I find that the comprehension of the DT maize characteristics is not the primary driver of adoption. I also find that a rosier expectation of the yield advantage or a more pessimistic expectation of future rainfall lead to higher adoption. However, between different comprehension frameworks there is negligible difference. When households are split by productivity, it is the lowest productivity households whose behavior is best predicted by expectations. The nuance of the DT technology merits particular attention with regards to what regions it is promoted in, even if farmers comprehension is not the largest barrier to adoption.

I presented in the AAEA 2018 Risk and Uncertainty Lightning Session.

Health Savings Behaviors in Burkina Faso: Results from a Framed Field Experiment

This paper explores household asset allocation under risk using the portfolio choice framework. Using simulations and data from a framed field experiment conducted in Burkina Faso, I find that the introduction of the risk-free savings instrument increases welfare for those at the bottom of the wealth distribution. In arid agricultural settings, households often have little access to resources to respond to negative shocks. Allocation strategies are important for household asset accumulation and policy for development of financial tools and institutions.

CGAP Working Paper link, and our working paper: Resilience and health shocks.

Subjective Expectations of Rice Price in Madagascar and Agricultural Productivity

For agricultural households, every economic decision bears some amount of risk, or the possibility of an adverse outcome. Decision makers use available information, including knowledge about risk, to determine their expectations of the future: uncertainty can weigh heavily in decisions. In the case of agricultural households, production and market risk and uncertainty can have real consequences for decisions. Understanding expectation formation and effects is crucial for successful development policy. Using household survey data collected in 12 villages across six communes around Lac Alaotra in Madagascar in 2004, I analyze the relationship between rice price expectations and labor input decision making.

Duke Journal of Economics Vol. XXV, Spring 2013