Phase Space Reconstruction from Econommic Time Series Data: Improving Models of Complex Real-World Dynamic Systems

摘要: Failure of economic models to anticipate the global financial crisis illustrates need for modeling better capture complex real-world dynamics. Conventional models—in which variables evolve toward equilibria or fluctuate about in response exogenous random shocks—are ill-equipped portray dynamics may cycle aperiodically along low-dimensional ‘strange attractors’. We present a method developed physics literature—‘phase space reconstruction’—that reconstructs strange attractors dynamical systems using time series data on single variable. Phase reconstruction provides pictures that can guide model specification.