Diversification Discount or Premium? New Evidence from Bits Establishment-Level Data

This paper examines whether the finding of a diversification discount in U.S. stock markets is only a data artifact. Segment data may give rise to biased estimates of the value effect of diversification because segments are defined inconsistently across firms, and that inconsistency does not occur at random. I use a new establishment-level database that covers the whole U.S. economy (BITS) to construct business units that are more consistently and objectively defined across firms, and thus more comparable. Using a common methodological approach on a sample of firms which exhibit a diversification discount according to segment data, I find that, when BITS data are used, diversified firms actually trade at a significant average premium. The premium is robust to variations in the method, sample, business unit definition, and measures of excess value and diversification used.