Abstract
U.S. dairy production has been consolidating into large-scale confinement operations. Large numbers of small- to medium-scale dairies have disappeared in the last two decades, and many more are disappearing. This article analyzes small- to medium-scale dairy operations in Maryland during 1995–2009 for changes in technology and efficiency through a novel two-stage DEA approach to examine productivity changes. Conventional confinement dairy operations and management-intensive grazing dairies are analyzed separately. The results show that both dairy systems have become more productive on the technological frontiers, yet the rate of technical change for graziers was less than half the rate for confinement.
Highlights
U.S dairy production has been consolidating into large-scale confinement operations
The estimates of linear time trends indicate, on average, technological gap ratio (TGR) grew 1.21 percent per year and technical efficiency (TE) declined À0.56 percent per year for confinement operations. This translates into the Malmquist Productivity Index (MPI) decomposition of a 0.65 percent annual growth of MPI that consists of 1.21 percent technical change (TC) and À0.56 percent technical efficiency change (TEC)
The estimates for graziers translate into a 1.22 percent annual decline of MPI that decomposes to 0.59 percent TC and À1.81 percent TEC
Summary
Consider a model of milk production yit in which producer i ∈ I 1⁄4 f1; ::; Ig chooses L-dimensional inputs xit in time t ∈ T 1⁄4 f1; :::; Tg. Rþ is the technological frontier in time t, exp(Àuit) is multiplicative technical efficiency TEit ∈ (0, 1], and exp(vit) is a random error. One shortcoming of the method is that its deterministic estimation (i.e., exp(vit) 1⁄4 0 in (1)) can be sensitive to extreme data points and measurement errors. Another is that the method presumes a balanced panel data structure in calculating the producer-level MPI decomposition. To accommodate the use of unbalanced panel data, producer-level calculations of the MPI decomposition are replaced with a sample-level regression analysis of a frontier and efficiency. We derive statistical inferences based on the approach developed by Kneip, Simar and Wilson (2015) for the class of second-stage analysis estimators on DEA efficiency measures
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
