VALUACIÓN DE ACTIVOS FINANCIEROS INCORPORANDO MOMENTOS ESTOCÁSTICOS, NO LINEALIDAD Y VARIABILIDAD TIEMPO DE PARÁMETROS

Thesis

Summary *

Classical financial models assume the normality of the probability distribution used in the valuation of assets. This implies that series of asset returns are expected to be homoscedastic and leptokurtic. In fact, real-world series of return show the following features: i) volatility clustering, ii) little or no autocorrelation, iii) dependence among squared values and iv) heavy tails, among others. These are properties that arise when, for instance, the conditional distribution of the data generation process changes in time, particularly its volatility. When volatility is seen as a stochastic process itself, a model able to describe and predict present and future values can be fitted. Among the many alternative ways in which this has been done in the literature, we can distinguish the nonlinear models of the ARCH family specifying the behavior of the variance of the series. Following that lead, we study and model herein the behavior of asset returns in emerging countries in order to estimate the volatility of their markets. The importance of this estimation lies in that it can be used, among other things, to detect the institutions that generate larger spillover effects in a financial system, particularly in an adverse context, in which a limitation of systemic risks and the costs of financial crises are sought, and the financial system is strengthened against external shocks. Information provided by the agent in SIGEVA