P2-1 Aggregation Issues in Dynamic E ciency Analysis: Applications in Agriculture
PhD student: Ioannis Skevas
Thesis Committee: Prof. Bernhard Brümmer, Prof. Thomas Kneib, Prof. Grigorios Emvalomatis
Efficiency analysis has been the subject of several studies mainly from a static point of view. However, static efficiency measures ignore the adjustment of quasi-fixed inputs to their long-run levels and the time interdependence of production decisions. Given farms' production decisions made in the past and the changing economic conditions, a farm may find itself in a situation in which it is not making optimal use of resources. In order to become efficient, a farm has to adjust the factors impacting its short run efficiency. However, high adjustment costs may force a farm to remain inefficient in the short run, which implies that there may be persistence of inefficiency from one period to the other.
Surprisingly, little attention has been given to the dynamic nature of inefficiency potentially due to the computational complexity of such models. Bayesian methods appear to be ideal for dealing with such complex models. Markov Chain Monte Carlo (MCMC) methods and in particular the Gibbs sampling and the Metropolis-Hastings algorithms along with data augmentation can provide parameter inferences as well as inferences on technical inefficiency of the decision-making units. The dynamic nature of efficiency is captured by specifying an autoregressive structure on firm-specific technical efficiency. However, previous studies on dynamic efficiency ignore the fact that farms may face different adjustment costs and as a result, heterogeneity of inefficiency persistence is not taken into account. The project will deal with the following issues:
1. Heterogeneity of inefficiency persistence
2. Factors that explain variability in inefficiency persistence
3. Dynamic cost minimization and profit maximization objectives of decision-making units
Using data from the Farm Accountancy Data Network (FADN) and focusing on specialized dairy farms, dynamic Stochastic Frontier Analysis (SFA) will be applied for the first two issues, while, Seemingly Unrelated Regression (SUR) will be used for the third one.
back to overview of all projects