World Firm Size Distribution and A Test of Gibrat’s Law
Sungyong Chang
Columbia Business School, Columbia University
This paper explores world firm size distributions and firm growth processes. There have been disputes on what would be the underlying mechanism that generates skewed firm size distributions. Some scholars believe that the random stochastic growth process, Gibrat’s law, generates skewed firm size distributions in the real world. However, other scholars believe that systematic factors affect firm growth rates and firm size distributions. This paper investigates the world firm size distribution and tests Gibrat’s law by analyzing the Orbis database which covers over 44 million firms around the world. The first finding is that the world firms size distributions follow the Pareto distribution, also known as a power-law distribution. The second finding is that there exist (1) size-growth rate correlations and (2) serial correlations in firm growth rates. Both are small and negative, but statistically significant. To investigate the effects of size-growth rate correlations and serial correlations in firm growth rates on firm size distributions, we build a computational model based on Gibrat’s law. The results from the computational model show that the random stochastic growth process, Gibrat’s law, generates the Pareto distributed firm sizes, even under the empirical size-growth rate correlations and serial correlations in firm growth rates. Finally, we discuss the role of randomness in generating systematic patterns at the macro level.
Figure 1. World Firm Size Distribution in 2011
Sungyong Chang 