Panel unit root tests of firm size and its growth

Abstract

This paper investigates Gibrat's law by using a panel unit root test, as a panel unit root can increase power in contrast to a conventional individual ADF test. At first this paper uses the panel unit root test to testify Gibrat's law under independent and identical distribution, with the test results rejecting the null hypothesis of Gibrat's law. Independent and identical distributions are not reasonable in a real situation. Any firm in a given industry may have some correlation with other firms. Moreover, the limiting distribution of Im, Pesaran, and Shin (IPS) statistic is invalid and will produce a large distortion. This paper applies the Taylor and Sarno (1998) MADF test to deal with a cross-sectional correlation problem and study the issue. This paper finds that the conclusion is not the same.