Contagion, Correlations and Topology : Analysis of the Currency Dynamics in the Latin American Markets (Contagio, Correlaciones Y Topología. Análisis De La Dinámica Cambiaria En Los Mercados Latinoamericanos)

In this paper we introduce a method to describe dynamical patterns for the exchange rate movements in the main Latin American markets and to analyze the contagion phenomena in currency crisis. This method combines Symbolic Time Series Analysis (Daw et. al. 2003) with the nearest neighbor single linkage clustering algorithm (Mantegna and Stanley 2000). From symbolization of data we obtain a distance between different time series that can be used to construct a Minimal Spanning Tree (MST). Besides, an ultrametric distance is obtained to construct the Hierarchical Tree (HT). These trees are used to detect clusters according to their proximity and to obtain a hierarchical organization. This classification is then used to study the contagion phenomena in Latin American currency crisis and the connections and hierarchy in regional currency markets