An abstract data modelling approach to the problems of econometric measurement, and system definition and analysis, is proposed. The approach is based on an extension of classical ideas of mathematical systems theories, developed for the purpose of laying formal foundations for a theory of information systems. In the proposed approach, an economic system is defined by the measurement data, and can be meaningfully discussed and, with the help of computing tools, also processed, even before statistical redundancy reductions have been made. A distinctive feature of the formalism is the separation of local and global analyses of the system variables.