Gibrat’s law for cities, growth regressions and sample size

This paper uses un-truncated city population data from six countries (the United States, Spain, Italy, France, England and Japan) to illustrate how parametric growth regressions can lead to biased results when testing for Gibrat’s law in city size distributions. The OLS results show non-monotonic behaviours depending on the sample size. Moreover, it is possible to find a critical sample size from which we reject Gibrat’s law