UC Berkeley

Growing population, changing climate, and human development will require sustainable agricultural intensification in sub-Saharan Africa -- a region where most people still live in rural areas, and rural poverty remains severe. While the 20th century saw massive increases in agricultural productivity over most of the world, sub-Saharan Africa was largely bypassed. In many respects, catching up in the 21st century poses more difficult challenges than were faced in the 20th -- with the novel challenges of climate change, soil degradation, and the closing agricultural frontier being chief among them. Solutions are required that profitably improve productivity, strengthen and/or rebuild soil fertility, and do so within the limits imposed by a warming and carbon-constrained world. This will be needed if the region is to shake off the stagnation that has characterized it for the past century, and contribute to solving to the global challenges posed by the coming century.

This dissertation focuses on one potential solution -- biochar -- and follows it from agronomic efficacy, to preliminary economic analysis, to rigorous trial in the field. As such, it seeks to be an example of the interdisciplinary approach that I argue is needed to guide the development, deployment, and scaling of solutions to problems in the environment/development space.

The first chapter is a meta-analysis of crop yield response to biochar. Using data from 84 studies, I (and co-authors) employ meta-analytical, missing data, and semiparametric statistical methods to explain heterogeneity in crop yield responses across different soils, biochars, and agricultural management factors, and then estimate potential changes in yield across different soil environments globally. We find that soil cation exchange capacity and organic carbon were strong predictors of yield response, with low cation exchange and low carbon associated with positive response. We also find that yield response increases over time since initial application, compared to non-biochar controls. High reported soil clay content and low soil pH were weaker predictors of higher yield response. No biochar parameters in our dataset -- biochar pH, percentage carbon content, or temperature of pyrolysis -- were significant predictors of yield impacts. Projecting our fitted model onto a global soil database, we find the largest potential increases in areas with highly weathered soils, such as those characterizing much of the humid tropics. Richer soils characterizing much of the world's important agricultural areas appear to be less likely to benefit from biochar.

The second chapter is a preliminary economic analysis of biochar's potential in two contexts -- rural western Kenya, and northern Vietnam. Using recall-based datasets from smallholder farmers, I (and co-authors) estimate yields as a function of biochar and fertilizer use. We find an positive association between biochar use and average yields in Kenya, but no correlation in Vietnam. We then use these estimates to calculate optimal input mixes under hypothetical biochar and carbon prices, given heterogeneity in response both to biochar and fertilizer, and heterogeneous budget constraints. In Kenya, we find that biochar is more-likely-than-not to be profitable to adopt for 23\% of our sample if unsubsidized and available at its current sale price of around \$188/ton, while a hypothetical carbon subsidy of \$100/ton CO$_2$e increases this proportion to 47\%, though these proportions are not different from zero at 95\% confidence. Because of limited short-term complementarity between biochar and inorganic fertilizer, we estimate that biochar adoption would change profits little, given budget constraints for agricultural inputs. We conclude that carbon subsidies may have a marginal impact on biochar's profitability in Western Kenya, but that further research is needed to improve the precision of these estimates, extend them to account for any longer-term changes in soil characteristics that might impact biochar's profitability, and account for any potential biases stemming from time-varying variables that not measured or modeled in the context of this study.

The third chapter reports the results of a Kenyan field experiment on adoption and impact of biochar, which was motivated by the encouraging findings of the previous two studies. In addition to technical efficacy, I sought to determine what mix of policies might most effectively speed biochar dissemination, given the slow pace of technological change in African agriculture over the past several decades. I randomly assigned prices, demonstrations, and risk-free trials. Yields increased by 37\% and 50\% in the first two seasons, and response to inorganic fertilizer improved. However, uptake was 2.6\% and 10\% respectively. Farmers were highly price-sensitive. Social network effects were marginally significant, but positive at low penetration and negative at high penetration. Given uptake well below the social optimum, subsidies for biochar appear justified from a social cost/benefit standpoint.

The dissertation closes with a short discussion of lessons learned, and ways forward for further applied research.