A semi-annual econometric model of the United States feed grain-livestock economy
No Thumbnail Available
Authors
Meeting name
Sponsors
Date
Journal Title
Format
Thesis
Subject
Abstract
The forecasting performance of agricultural commodity models has been critically evaluated following the extreme agricultural commodity price fluctuations of the early and mid 1970's. One approach to improving intermediate run forecasting models involves the use of multicommodity simultaneous systems. Of the major domestic commodities, the feed and livestock subsectors form interdependent and integral parts of the U.S. agricultural economy. A feed-livestock sectoral model was formulated to permit analysis of the predictive value of a large scale econometric system for agriculture. Review of existing agricultural sectoral models points to a reliance on complex structural relationships to portray market operations. Structural specifications typically admit nonlinearities and rely on inconsistent, curve-fitting estimators. Essentially these models have few advancements vis-a-vis the numerous individual commodity models already in existence other than the explicit simultaneity incorporated into sectoral linkages. As an alternative approach a linear, fairly simple structural specification was adopted in an effort to avoid over modeling areas of structural imprecision. The available theory to support structural specifications in such large scale models is limited due to problems of aggregation, specification and choice of sampling period. For this reason, the model was recognized as an approximation and simple linear representations were employed as opposed to those which would, for example, recognize homogeneity restrictions from firm and consumer theory. Several commodity, subsectoral, and sectoral linkages were derived using underlying economic theory. Crop price determination mechanisms which recognized the non-perishable nature of the commodity and discontinuous production process were developed. Particular attention was directed toward integrating relevant physical information into the model structure. Because of the pronounced seasonality of crop and livestock production, a semi-annual observation period was adopted. Based on equal divisions of the feed crop year, the semi-annual observation period enhanced the possibility for representing the signaling mechanism between the subsectors and allowed the specification to capture stronger linkages between the livestock and feed grain sector. Several theoretical properties of simultaneous equation techniques were discussed as a preliminary to the selection of estimation method for the model. The iterated instrumental variables (IIV) estimator was chosen for application because it remains operational even when the number of predetermined variables exceeds the number of observations. It also assures consistent estimators for the structural and reduced form parameters. In structural and reduced form comparisons the IIV method demonstrated superiority over traditional inconsistent least squares techniques. Most large scale econometric models use the ordinary least squares estimation method. Because the skeletal model developed was recognized as inherently an approximation of the complex operation of the feed livestock sector, certain forecast improvement methods were developed and analyzed. Under the topic of updating, mixed estimation, varying parameter and Kalman filtering techniques were discussed and linked as members of the generalized least squares methods family. Adaptive regression was singled out for predictive interval forecast evaluation. As a second approach to the recalibration problem several composite forecasting schemes also were developed as approaches to forecast improvement. The incorporation of specialist’s information appeared particularly attractive in the limited example applications. The commodities included in the structural representation were beef, pork, broilers, eggs and milk in the livestock subsector and corn, sorghum, soybeans, and soybean meal in the crop subsector. Forecast evaluation within the sample period showed that the average weighted (by market shares) fit of the reduced form accounted for ninety percent of the variance in the prices and output levels of these commodities. Evaluation of the model’s ex ante predictive accuracy during the 1978 crop year supported its validity as a useful forecasting tool. The results established indicate that the approaches investigated provide powerful alternatives to current model building procedures.
Table of Contents
DOI
PubMed ID
Degree
Ph. D.
Thesis Department
Rights
OpenAccess.
License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.