The debate on resale price maintenance (RPM) for books is probably one of the oldest competition policy debates in Europe. Due to limitations in data availability and empirical methods, relevant economic analysis has so far been restricted to theoretical considerations. This dissertation comprises essays that contribute methods and results towards an empirical assessment of the welfare effects of RPM for books. Particular focus is on two notions that reoccur regularly in the debate: the potential for anti-competitive behavior (tacit collusion) of publishers, and uncertainty about title-specific demand. First, I look at pricing patterns in a comprehensive cross-section dataset on the German book industry. In particular, I consider the phenomenon that in many market segments, a large share of titles has the same 'focal' price, like 20 or 25. My empirical analysis aims to distinguish between collusion-based and uncertainty-based explanations for focal pricing and RPM. The evidence is largely inconsistent with publisher collusion but consistent with an explanation where titles with highly uncertain demand have below-average ('fair') focal prices that are in addition fixed through RPM. Second, I argue that social influence, or word of mouth, is the crucial unpredictable factor that underlies demand uncertainty from the publisher perspective. I also interpret a quantification of demand uncertainty proposed by Denis Diderot in 1763 in this respect. Then, I discuss ways to identify the effect of word of mouth in scanner data of a title's weekly sales. I propose a parametric approach based on a structural theoretical model, as well as a semiparametric alternative, and apply these methods to a representative sample of German novels released in 2003. Both approaches yield results that are consistent with Diderot's claim that only one out of ten published titles is a profitable success. Finally, I discuss my results and their limitations in the context of the policy debate.