Power laws in cities population, financial markets and internet sites (scaling in systems with a variable number of components)

We study a few dynamical systems composed of many components whose sizes evolve according to multiplicative stochastic rules. We compare them with respect to the emergence of power laws in the size distribution of their components. We show that the details specifying and enforcing the smallest size of the components are crucial as well as the rules for creating new components. In particular, a growing system with a fixed number of components and a fixed smallest component size does not converge to a power law. We present a new model with variable number of components that converges to a power law for a very wide range of parameters. In a very large subset of this range, one obtains for the exponent a the special value I specific for the city populations distribution. We discuss the conditions in which a can take different values. In the case of the stock market, the distribution of the investors' wealth is related to the ratio between the new capital invested in stock and the rate of increase of the stook index. (C) 2000 Elsevier Science B.V. All rights reserved.