Abstract

We study the difficulty that can be caused by firms ’ use of specialized language, or code, when organizations merge, and whether participants ’ valuations of mergers take this difficulty into account. The experiments use a picture-naming paradigm. Subjects are grouped into “firms,” separately at two universities, and create code words to describe a series of pictures from their own college campus, rapidly and clearly. The codes are a component of “organizational culture, ” which can be created rapidly and has efficiency properties. After creating a shared code, subjects from two firms are placed into a merged group, based on payments they demand to join that group, and must name a mixture of pictures from both campuses. The bids both allow self-selection and enable a test of whether subjects who join the merged group underestimate the difficulty of code integration. Performance clearly decreases in the merged firm because of differences in code and unfamiliarity with pictures of the other campus. Guesses by all subjects about post-merger performance, and the values of low bidders revealed by Vickrey auctions and inferred from first-price auction bids, underestimated the difficulty of merger on average. Inferred values overall indicate fairly accurate guesses about the difficulty of integration, but the lowest values (corresponding to those who actually join the merger) are too

Citations