Date of Award

2011

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Environmental Science

First Advisor

Kurkalova, Lyubov

Abstract

This study presents an alternative method of empirical estimation of a logit model that uses the full information on the attributes of agents and only aggregate measures of agents' choices in an application to the estimation of the costs of the adoption of conservation tillage by Iowa farmers. The methodology treats the aggregated data as an expected value-the group average of individual probabilities of choosing conservation tillage subject to measurement error. The study then adapts the new methodology to the distinctive case where the researcher not only has the group expected value but also knows the magnitude of the within group standard deviation of the individual outcomes. The study derives the maximum likelihood estimator consistent with the data structure described above. The approach is illustrated in a Monte Carlo analysis and is further verified in applications to two real-life datasets on farmer tillage choices. All three applications show that the new method performs well. When applied to previously not studied 2002 and 2004 data the model estimates average subsidy payments required to entice farmers to use conservation tillage to range from $13/acre to $18/acre for corn and soybean farmers with a sample standard error of 0.18. In general, the proposed method could be especially attractive in estimation of binary choice models in the disciplines in which reliable aggregated choice data are routinely available. It contributes to the development of econometrically sound approaches to modeling of imprecisely measured (or noisy), aggregated data. Such models are of continuous interest in agricultural and natural resource economics.

Share

COinS