Distinguishing Random And Deterministic Systems

This paper discusses tests on time series for the presence of low dimensional deterministic chaos. Empirical applications on U.S. business cycle data are reviewed. Two tasks are necessary to test a time series for (low dimensional) deterministic chaos: (1) dimension must be calculated and shown to be low relative to comparable (psuedo) random numbers; (2) Lyapunov exponents and other measures of local instability must be calculated and show local instability relative to a useful standard. The evidence for low dimensional deterministic chaos in U.S. postwar II GNP data is weak (English). Brock: University of Wisconsin, Madison, Wisconsin. Journal of Economic Literature Classification Numbers: 023, 211, 213