It is quite common that public policies take as benchmark mean values of relevant policy variables (average firm size, average productivity, average income, etc.). In such case the policy maker introduces incentives that should change individual behaviour in order to increase average values of the policy relevant variables. When this kind of policies are unsuccessful it is thought that the incentives are badly designed so that the entire problem is considered a problem of "mechanism design". In this paper the Authors maintain that policy makers should always take into account that incentives have an impact not only on individual behaviour but also on the distribution and law of motion of the aggregate phenomena that may be different from the sum of micro behaviours. Three different cases of proportional growth phenomena (firm's size growth, income distribuition and research production) are analyzed, in the paper, through simulations. Some tentative general principles for public policies are identified.