Revista Colombiana de Estadística
versão impressa ISSN 0120-1751
GALLON, SANTIAGO e GOMEZ, KAROLL. Nonparametric Time Series Analysis of the Conditional Mean and Volatility Functions for the COP/USD Exchange Rate Returns. Rev.Colomb.Estad. [online]. 2010, vol.33, n.1, pp. 25-41. ISSN 0120-1751.
The modeling and estimation of the conditional volatility associated with a stochastic process usually have been based on parametric ARCH-type and stochastic volatility models. These time series models are very powerful in representing the dynamic stochastic properties of the data generating process only if the parametric functions are correctly specified. The nonparametric approach acquires importance as a complementary and flexible method to explore these properties without imposing particular functional forms on the conditional moments of process. This paper presents an application of nonparametric time series methods to estimate the conditional volatility function of the COP/USD exchange rate returns. Additionally, we estimate the conditional mean function under this approach.
Palavras-chave : Nonparametric regression; Local polynomial regression; Nonlinear time series; Variance function estimation; Autoregressive conditional heteroscedasticity; Time series analysis.