Economic Quarterly

This article documents the large changes in the size distribution and size dynamics of banks from 1960 to 2005. The authors find that neither the lognormal distribution nor the Pareto distribution fit the entire distribution, though they each do better with subsets of the distribution. Gibrat's Law is tested and found to hold in the 1960s and 1970s, but not afterwards. Gross entry rates are found to be surprisingly robust over the entire period: there is no trend in the fraction of banks that comprise new entries. Finally, assuming that trends in size dynamics from 2000–2005 continue, a forecast of the number of banks predicts that the rapid decline in number of banks will slow down and the number of banks will level off.