Only a “Longish” Tail

The long tail hypothesis claims that increased Internet usage will enhance the purchase of obscure items so that, in aggregate, these obscure items will become an important, perhaps primary portion of the market. We test this hypothesis with two analyses. First, using unusually rich data on book sales, we examine the share of niche titles as Internet retailing grew from obscurity to ubiquity. Using relative measurements of concentration, as has been common, provides results contrary to the hypothesis. But we also conclude that it is inappropriate to use relative measures of concentration with these data. Using absolute measures of concentration provides results consistent with the hypothesis’ sign but not with the magnitude. Second, we use the 2011 bankruptcy of the Borders bookstore chain and the resulting sudden shutdown of all its store locations, as a natural experiment. This analysis finds a shift in sales away from the top successful titles. But the major beneficiaries are not the most obscure titles but instead some quite successful titles and some mildly obscure titles. The former is contrary to the hypothesis, and the latter supports the direction but not the size of the hypothesis. Both analyses support, at best, an enfeebled version of the long tail where the share of relatively obscure titles changes from one unimportant small value to a somewhat greater but still unimportant value. Thus, the use in the title of “longish” as a diminutive form of “long.”