Cross-Sectional Firm Dynamics: Theory and Empirical Results from the Chemical Sector

The literature on the dynamics of industry structure has generated very few robust empirical predictions. The paper shows that if one particular assumption is made about the pay off structure of an investment project, three well known games, a modified 'Grab the Dollar' game, an auction and a stochastic race all have identical (robust) empirical predictions. The assumed structure of pay offs applies particularly well to the chemical sector. The theoretical implication is that small firms are more likely to install new capacity than their larger competitors, leading to a tendency of firm sizes to converge. We show that the conventional empirical approach using discrete choice models is inappropriate in this context, due to Galton's Fallacy. Using a novel method to study empirical firm dynamics, by analysing dynamically evolving cross-section distributions and by that exploiting time series and cross-section information more fully than standard cross-section regressions, we show that there is a strong tendency of firm sizes to converge in the chemicals sector.