Missing agricultural price data: an application of mixed estimation

Abstract

This article shows the convenience of combining the Country–Product– Dummy (CPD) model and the Theil–Goldberger (TG) mixed estimator to obtain better estimates of missing prices than those obtained by marginal mean imputation. We use the TG estimator to combine aggregate price data for regions of Ecuador with producer-level data (the sample data) to fill-in the missing price observations. Our results show better price estimates for predicting missing data than estimates obtained from using the CPD model on the sample data only. Through this approach, missing data can be replaced with economically meaningful data.

Notes

¹In order to test for the robustness of the mixed estimator, we incorporated some of the sample variance to the weight of the extraneous data so as to balance the importance of the latter. We introduced a parameter ψ ∊ [0,1] such that.

In this way, we would compensate for the lack of idiosyncratic variation in the aggregated price data, hence reducing its weight in the mixed estimator. Our findings indicate that for values of ψ greater than 0, the coefficients of variation of the residuals are always smaller than those obtained by using only the sample data (ordinary least squares estimation). However, results turn out to be best, in terms of this same criterion, when ψ = 0, i.e. the regular TG estimator.