Dynamique non-linéaire des marchés boursiers du G7 : une application des modèles STAR

One of the debates of topicality is focused around the study of the structure of stock prices dependency. The search of a specification of this dependency and a representation of underlying dynamics has been the subject of several empirical studies. These studies were primarily centered on the definition of the functional form of the process underlying the dynamics of prices. The idea consists to adapt the nonlinear oscillations theory to economic phenomena since this type of modelling coincide with the evolution of economic and financial time series. The reflexion was essentially led within the time framework and few studies are related to the specification of nonlinearity through tests of market efficiency, taking into account the heterogeneity of agents and the asymmetry due to transaction costs. The answer is not obvious because the approaches linarity-nonlinearity, efficiency-inefficiency, dependence-independence are not mutually exclusive. This paper explores the stock market dynamics. For this purpose, we test the nonlinear contribution of the STAR models in modelling stock prices dynamics. To do so, we test the independence assumption versus the nonlinear dependence assumption. In particular, we propose to model the time series of stock prices in the framework of Smooth Transition Autoregressive Models (STAR). They allow to capture the nonlinearity component, by using some recent statistical and econometric tools, and to analyze the short-run dynamics of prices.