|
Activity Number:
|
573
- Simulation and Stochastic Bayesian Modeling
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
|
Abstract #306995
|
Presentation
|
|
Title:
|
Bayesian Stochastic Frontier Models for Productivity Index
|
|
Author(s):
|
Ehsan Soofi* and Jessie Nouri
|
|
Companies:
|
Univ of Wisconsin-Milwaukee and University of Wisconsin-Milwaukee
|
|
Keywords:
|
Bayesian Modeling;
Dirichlet process prior;
Stochastic frontier model;
Labor productivity index
|
|
Abstract:
|
The labor productivity index defined as the ratio of output to the labor input is used for comparing the efficiency of various production lines within a firm. We modify the stochastic frontier model for studying this index. The modification includes a shift in the coefficient of the labor input in the Cobb-Douglas production function, a factor to adjust for the complexity of the product in the frontier, and combining some existing specifications for the inefficiency in terms of line characteristics and time effect. We develop various Bayesian models using data collected at a major manufacturing firm for a panel of production lines through time. Bayesian models with and without time effect are estimated using various prior distributions, including Dirichlet process prior. Production lines are ranked by comparing the posterior distributions of the efficiency term of the lines. These posteriors depict global and local stochastic dominances between the efficiencies of the production lines included in this study. The results provide insights for the management decision.
|
Authors who are presenting talks have a * after their name.
