Files in this item



application/pdfNORTON-THESIS-2020.pdf (686kB)
(no description provided)PDF


Title:Ambiguity and credence quality: Implications for technology adoption
Author(s):Norton, Benjamin Prescott
Advisor(s):Michelson, Hope
Contributor(s):Manyong, Victor; Winter-Nelson, Alex; Crost, Benjamin
Department / Program:Agr & Consumer Economics
Discipline:Agricultural & Applied Econ
Degree Granting Institution:University of Illinois at Urbana-Champaign
Sub-Saharan Africa
experimental economics
Abstract:Use of fertilizer and hybrid seed remains low in much of Sub-Saharan Africa. A possible contributor to low adoption is that farmers are uncertain about the quality of agricultural inputs available to them. While previous studies have shown that risk and uncertainty preferences are relevant to the decision to adopt a technology, existing research assumes that farmers have homogeneous beliefs about the quality of available inputs. I test this assumption using an incentivized Becker-DeGroot-Marschack auction in Tanzania and examine how farmer beliefs about mineral fertilizer quality in local markets influence their willingness-to-pay. I find that farmers are willing to pay 46% more for fertilizer that was laboratory tested and found to be pure than for untested fertilizer. Farmers who believe that more of the fertilizer for sale in their local market is low in quality are willing to pay a higher premium for laboratory-tested pure quality fertilizer, compared to untested fertilizer. Yet these results present something of a puzzle, given that three rounds of testing of fertilizer for sale in regional markets over five years have demonstrated that the nutrient content of fertilizer for sale in these contexts is consistently at or near advertised levels. Farmers appear to believe that low-quality fertilizer is far more prevalent in proximate markets than it actually is. How have farmers’ incorrect beliefs persisted in equilibrium? I posit two interconnected mechanisms. First, misattribution: Yields are stochastic due to weather and other factors, and when a yield in a particular year is unusually low, farmers misattribute noise as indicative of low-quality fertilizer. Second, farmers experience both risk (uncertainty about whether a bag of fertilizer is bad) and ambiguity (uncertainty about the likelihood a bag of fertilizer is bad), and thus hold multiple priors. I develop a Bayesian learning model that incorporates both misattribution and multiple priors and show that in equilibrium beliefs do not converge to the truth. Supporting the model's findings, I use farmer survey data from Uganda to establish that historic precipitation variability relates to farmers’ fertilizer quality belief distributions. I use the learning model to simulate several policy interventions, and show that subsidies, information campaigns, and plot-specific fertilizer recommendations improve beliefs, but do not cause beliefs to fully converge to the truth. Instead, policy makers should consider programs that address the misattribution problem.
Issue Date:2020-07-21
Rights Information:Copyright 2020 by Benjamin Prescott Norton. All rights reserved.
Date Available in IDEALS:2020-10-07
Date Deposited:2020-08

This item appears in the following Collection(s)

Item Statistics