This paper compares the power in small samples of different tests for conditional heteroscedasticity. Two new tests, based on neural networks, are proposed: the main interest in them arises from the fact that they do not require the exact specification of the conditional variance under the...

In this paper, new noncausality tests built on a general nonlinear framework are proposed and their performance studied by a Monte Carlo experiment and a variety of nonlinear artificial series. Two of these test are based on a Taylor expansion of the nonlinear model around a given point in a...