Abstract:
Farm machinery continues to increase in its importance to the agricultural sector.
Depreciation, the decline in value of a durable asset over time, represents one of the
largest costs of agricultural production. The general objectives of this study were to
update and expand the number of Remaining Value (RV) functions for farm machinery;
estimate and compare the depreciation patterns for Box-Cox (B-C) and Additive-
Exponential (A-E) functional forms; and compare the predictive abilities of the two
models.
Data used in estimation were obtained from 15 years of machinery auction sales.
Based on the hedonic approach, the Box-Cox and Additive-Exponential models were
formulated to include variables for age, usage per year, condition, manufacturer, auction
type and macroeconomic variables. Models were tested for potential correlation and
heteroscedasticity problems. A Mean Absolute Percentage Error (MAPE) method was
used to compare the predictive abilities.
Depreciation functions were estimated for 17 types of machinery in four major
categories: tractors, harvesting equipments, planting and tillage equipment and other
equipment. A series of comparisons were conducted to examine the difference in
depreciation patterns within and between major categories. B-C and A-E functions were
estimated and compared for each data set. The comparison results showed that B-C
always generated higher Log-likelihood values, but was relatively weak in predictive
ability. The predictive ability of the Exponential model was also examined and the
MAPE results showed that the Exponential model generally exhibited an overall best
predictive ability.
