David Murphy is a PhD Candidate at Cornell’s Dyson School and is on the job market this year
An excellent July 30th article in the New York Times highlights the impact of climate change and population pressures on soil degradation in Sub-Saharan Africa (SSA), and the effect it will likely have on increasing poverty and food insecurity. One obstacle to helping repair soil nutrient levels is the lack of information among farmers as to the nutrient status of their soils (Marenya et al., 2008; Berazneva et al., 2016). This can lead to inefficient application of fertilizers, reducing profitability and hindering improvements in soil fertility (Wopereis et al., 2006; Vanluawe et al., 2011).
In my job market paper, I investigate whether individualized soil tests and fertilizer recommendations can be cost-effective methods to reverse soil degradation. In collaboration with the International Institute of Tropical Agriculture (IITA), we collected and tested soils from 550 smallholder fields in western Kenya and IITA developed individualized fertilizer recommendations for each farm. We then visited the smallholders and separately provided the household head and his/her spouse with the information from the tests and recommendations. To determine the effect of this information transfer, we conducted incentive-compatible experimental auctions for organic and inorganic fertilizers with the individuals, both before and after they received the soil test information and recommendations. More details about our experimental design are included in an earlier blog post.
The main results, found using triple difference estimations (where the three differences are: difference in auction round, difference between treatment and control, and difference between those who received a recommendation to use a fertilizer (or if in the control group, would have received the recommendation if they had been randomly selected to be in the treatment) and those who did not). are shown in the tables below. Recommendations to use a nitrogen fertilizer had large impacts on farmer willingness to pay (WTP) for DAP (diammonium phosphate, a common inorganic fertilizer in western Kenya). Column I of Table 1 shows that recommendations led to an increase in farmer WTP of 60.8 Kenyan Shillings (KSh), or 24% of the mean price for the quantity auctioned compared to the counterfactual group. In an earlier blog post, I discussed the differences that we found in the distributions of bid changes between gender. Because of this, In Column II, we show quadruple difference results that assess the difference between men and women. Results remain positive for both men and women, though the impact is statistically insignificant for women compared to men in the counterfactual group.
Results for organic inputs are more nuanced and depend greatly on the gender of the respondent. In Column I of Table 2 we see that the recommendation to use organic inputs such as animal manure or compost increased respondent WTP for auctioned organic inputs by about 16 KSh compared to the counterfactual group, with weak statistical significance. When we add the gender difference, we find that the recommendations had no overall effect compared to the counterfactual. For women however, we find a stronger impact: the recommendations increased WTP for organic inputs by 34 KSh compared to men in the counterfactual group, with statistical significance at the 5% level (10% when using the Wild bootstrap). We believe that these results are likely due to the differential levels of access to organic inputs between men and women on the household farm. Our data show that women use substantially less organic inputs on plots that they manage compared to men. As we see in the literature, men often allocate scarce resources towards plots that they manage at the expense of female-managed plots (Udry et al. 1995). We also find in our data that female-headed households own far less livestock than male-headed households, decreasing the availability of organic material for agricultural use. In our auctions, we create a temporary market where organic inputs are potentially available for purchase. As a result, it is plausible that women, who usually have less access to organics than men, may bid higher for organic inputs than men after learning that they should be used to improve soil fertility.
After setting all the regressors at their means, we determine the predicted WTP for DAP after the recommendations. The following figure shows the predicted WTP for 1KG of DAP at the baseline, after receiving a recommendation to avoid spending resources on DAP fertilizer use, and after receiving a recommendation to use DAP. Vertical red lines represent 95% confidence intervals, and the horizontal dotted line is the market price. The analysis shows the baseline bid to be about 57.0 KSh. This drops to 7.3 KSh after a respondent learns that their soil is likely unresponsive to DAP, and increases to 91.6 KSh after hearing a recommendation to use this input: a 61% increase from the baseline. This provides clear evidence of the significant impact of the recommendations, as well as demonstrating that on average, farmers changed their bids in the direction of the recommendation.
Using these results from the experimental auctions, we conducted a cost-benefit analysis to measure the projected financial impact on farmers of the soil tests. We have two groups of farmers in this sample: those who are advised to increase DAP use (78.12% of the sample) and those who are advised that their soil is likely unresponsive to DAP. Using a comprehensive price for the simple soil tests that includes staff and transport cost estimates, we find positive net benefits for both groups. For those who increase DAP use, we find a net benefit for the average farmer of 2,449 KSh per hectare of maize planted. For those who decrease DAP use, the net benefit is 13,758 KSh. (As a standard for comparison, the mean farm size in the sample is about 0.4 hectare, and the median spending per week on food in the sample is 700 KSh.) The higher amount for those who decrease DAP use comes from the avoided expenditure on DAP that is ineffective for their soil. This significant net benefit can then be spent on more productive fertilizers by the farmer.
Soil degradation is one of the most serious constraints to long-term growth prospects in SSA (FAO, 2015). My job market paper shows that heterogeneity in soil nutrient levels, even at very small scales, means that simple recommendations that focus on only increasing inorganic fertilizer usage to improve soil fertility can lead to loss in investments for farmers, many of whom have unresponsive soils to such fertilizer types. Increasing information to farmers through targeted fertilizer recommendations using individualized soil tests can therefore be one strategy to increase farmer optimization of inputs, increasing soil fertility, crop yields, and alleviating poverty in rural areas of SSA.
Special thanks to David Lee, Dries Roobroeck, Janice Thies, Jura Liaukonyte, Johannes Lehmann, Edmundo Barrios, and James Otieno for valuable input and support for this research. I also thank Ray Weil and Cheryl Palm for the use of the SoilDoc soil testing system. Funding for this study was provided by the Atkinson Center for a Sustainable Future, a U.S. Borlaug Fellowship, and several grants and fellowships from Cornell University. Other major support for this research came from a USDA National Institutes for Food and Agriculture grant, from the International Institute of Tropical Agriculture, and from the World Agroforestry Centre.