This dissertation proposes three new methodologies in empirical production economics for assessing technical change, production risks, and technological frontiers. Each methodology is demonstrated with an application to Maryland dairy operations, with an emphasis on comparing production technologies between confinement and management-intensive grazing (MIG) dairy systems. The rapid decline of small to medium scale dairies has made the study of alternative dairy production like MIG politically and socially important.

The first essay develops a regression-based approach to the Malmquist productivity index (MPI) decomposition that attributes production heterogeneity to technical change (i.e., shifts of technological frontiers) and technical efficiency change (i.e., shifts of technical efficiency). Unlike the conventional, producer-level decomposition measures, the proposed method obtains sample-level decomposition measures, for which the researcher can fully utilize unbalanced panel data and control for the influence of potentially-confounding non-production factors. The results find 1.3% and 0.6% annual technical change during 1995-2009 for confinement and grazers respectively. For both dairy systems, farm ownership and off-farm income are positively and negatively associated with technical efficiency respectively.

The second essay considers an empirical application of the state-contingent (SC) approach to production risks. In the context of agricultural production, uncertainty in the SC framework is defined over distinct weather events or market conditions, for which the producer is assumed to prepare a portfolio of SC production outcomes. This study shows how production data over multiple years can be regarded as multiple draws from the contingent states of nature, by which SC technologies can be approximated by Data Envelopment Analysis (DEA). The results suggest that optimal production decisions for a moderate-to-maximally risk-averse producer have become riskier for the confinement system and less risky for the grazing system.

The third essay proposes a refinement of the DEA frontier approximation by integrating the concepts of technical, allocative, and scale inefficiencies. Technology is estimated in the form of a weighted-average of the benchmarking-frontiers that are associated with these inefficiency concepts. In the current dataset, the proposed method finds 7.5% to 9.2% higher mean-technical efficiency than the standard practice, indicating the increased discriminatory power in efficiency analysis.