Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Abstract In this paper, we investigate whether the U.S. broiler companies used public information sharing mechanisms to collude to reduce output and fix prices. Utilizing natural language processing methods, we converted quarterly earnings call transcripts of publicly traded broiler companies into structured data. We identified six keywords that could be used as signals: cut; balance; constrain; discipline; reduction; and adjustment. Our results show statistically significant and mildly elastic negative relationship between the keywords signals and three different broiler production precursors. For example, a 1% increase in the aggregated signal is associated with 1.5% reduction in the broiler-type hatchery supply flocks. The results are corroborated when using LASSO estimation to identify the most important elements in the vector of collusion signals.