Parametric and semi-parametric regression estimation of the effect of the United States soybean quality attributes on export pric

This research was motivated by the failure of current soybean grades to incorporate oil and protein information, an ongoing debate about inefficiency of current grades and standards, and conflicting findings from previous studies.;The theoretical framework for this study was based on the Ladd and Martin input characteristics demand model. The implicit prices for quality characteristics were estimated from U.S. soybean export prices by differentiating FOB export prices with respect to each of the quality characteristics. The quality-price data set was created by matching common information from FGIS and CENSUS data sets. The hedonic models were based on 1,952 individual export observations (loadings) that took place between 1989 and 1994 and tested using both parametric and nonparametric estimation techniques. Given the objectives of this research, a semi-parametric model that combined parametric and nonparametric approaches, was considered as a suitable analytical tool. The smoothing spline technique was chosen to estimate the nonparametric part of hedonic regression in this study.;Statistically significant estimates were found for the CBOT one-week moving average price, port of origin, three Asian importers (Japan, Taiwan and Korea), heat damage, total damaged kernels, foreign material and splits. Estimations based on the subset of 982 observations confirmed the significance of oil content and protein.;Results of both parametric and nonparametric regression estimations confirmed that some of the quality variables had a nonlinear relationship with export price. The study identified two possible sources of non-linearity of implicit price. The first was related to different degrees of competition in the international market and second was linked to approximation of chemical composition based on the relationship with certain grade characteristics.