Markup Accounting

We document new facts of markup dispersion across the firm size distribution. Markup differences between firms of similar size account for over three-fourths of overall dispersion, calling for mechanisms that affect markups beyond differences in firm productivity and market concentration. To study these mechanisms, we develop an analytical oligopolistic competition model of variable markups that accounts for the observed joint distribution of markups and firm size, including the large mass of small firms with high markups and the presence of large firms with small markups. The key ingredient is the introduction of firm-specific demand elasticity shifters that account for differences in markups between firms of similar size. We apply this model to Indian and Colombian manufacturing, and US public firms and find consistent results throughout. Without demand elasticity shifters, equilibrium markups collapse to those of standard models of oligopolistic or monopolistic competition; in these cases the model captures less than half of the observed variation in markups and counterfactually assigns most of the variation to differences between firms of different size. The model further implies a negative correlation between marginal costs and markups, but also a positive correlation between market shares and marginal costs, indicating that large firms are not necessarily the most productive. In aggregate, the model implies at least 5 times larger efficiency losses from markup dispersion relative to standard models of heterogeneous markups.