Essays on conservation adoption and discrete choice modeling

Abstract

This dissertation examines advances in applied discrete choice econometrics in applied settings and conservation practice adoptions by Kansas farmers. The research contributes to the literature by examining the use of discrete choice models to more deeply examine adoption of conservation practices and the choice of crop rotations in Kansas. In addition, a method for examining the proper functional specification of logistic regression models is explored. The first essay aims to examine landscape, climatic, socio-economic and farm factors affecting choice of crop rotations by farm managers in dryland cropping systems. A particular emphasis is place on the role, insurance products (such as RA-CRC (Revenue Assurance/Crop Revenue Coverage) and ACRE (Average Crop Revenue Election)), as well as marketing options, and characteristics of farming operations. This paper models the joint adoption of crop rotations using a multinomial modeling framework which is used to estimate the probabilities of adopting different crop rotations. The data used for this paper was obtained from a mail survey in 2011 examining Kansas farmers’ land use decisions and consisted of an eight-page survey with 46 questions, leading to more than 400 distinct variables. The purpose of the second essay is to examine and analyze the adoption of conservation practices, no-till, cover crops and use of crediting of nutrients from manure, by Kansas farmers from both a joint and conditional perspective. This study develops a modeling framework that can analyze conditional adoption and examine farmers’ joint and conditional adoption decisions. Estimates calculated from the model will allow for an assessment of the linkages between the adoption of different conservation practices, as well as the socio-economic factors that affect the likelihood of adopting conservation practices given other conservation practices have already been adopted on-farm. The third essay aims to develop a robust test to examine the functional form of predictor/ index function in the logistic regression models as misspecified models can lead to biased and inconsistent estimates, and consequently inappropriate inferences. An Orthogonal Polynomial RESET test is developed to assess proper functional form for different functional form assumptions of the predictor/ index function, as well as provide guidance on the use of the test in applied logistic regression modeling. Monte Carlo Simulations are used to assess the viability of the test and compare it to similar tests found in the literature.