GRANGER TYPE CAUSALITY FOR NONLINEAR DATA
Gan Ohama · Bulletin of informatics and cybernetics · 2002
The causality proposed by Granger (1969) and several tests for it are often used in economic science to assess relationships among more than two factors of interest. However, the Granger causality presumes linearity and has limitation to analyze the relationship of "nonlinear" factors. In this paper we extend the Granger causality to nonlinear relationships and propose methods for analysis of data by the causality; the efficiency of the methods is also investigated.