A Direct Test of Sample Selection in the Cattle Procurement Market

Abstract This study tests for sample selection using data from a feedlot in Oklahoma by analyzing the distribution of cattle quality between alternative marketing arrangements and cash markets. Unlike previous studies employing Heckman and Roy models to detect sample selection, we directly examine the empirical distributions of cattle quality variables derived from heteroskedastic probit, residual, truncated, and generalized least squares regressions. Nonparametric procedures are used to identify differences in the empirical distribution functions of quality variables between the two markets. The results find strong evidence of sample selection in cattle procurement for the analyzed dataset. Initiatives promoting greater disclosure of quality-related information for cattle sold in the cash market are recommended to mitigate this issue.