The paper proposes the use of panel data unit-root tests to assess market-share instability in order to obtain indications of industry dynamics. The idea is to consider movements in market shares as much more than mere elements of the market structure. In fact, these movements reflect conduct that arises from that market. If shares are mean-reverting, then firm actions have only a temporary effect on shares. On the other hand, if shares are evolving, as signaled by the presence of unit roots, then any gain in shares with respect to the competitors is long term.