Title: Testing for and measuring serial dependence by neural networks
Authors: Jinu Lee - King's College London (United Kingdom) [presenting]
Abstract: Testing serial dependence is central to much of time series econometrics. The focus is on the generalisation of an autocorrelation function to test for and measure serially dependent processes by using neural networks based approximations. Simulations find that the suggested nonparametric method shows good power properties and has the potential to measure nonlinear associations compared to some popular tests and measures. An application to US stock returns illustrates the usefulness of the proposed tests and measures for nonlinear dependences.