Reduced form vector directional quantiles

MONTES ROJAS, GABRIEL VICTORIO

Posadas

Conferencia; LI Reunión Anual de la Asociación Argentina de Economía Política; 2016

Asociación Argentina de Economía Política

This paper develops a reduced form multivariate quantile model using the directional quantiles framework. Quantile regression equations are constructed for a fixed orthonormal basis in which each basis component represents a directional quantile corresponding to a particular endogenous variable. The reduced form vector directional quantiles correspond to the xed point given by the solution to a collection of quantile indexes for each directional quantile model. The model thus delivers a map from the unit ball with dimension given by the number of endogenous variables to the parameter space that correspond to the effect of weakly exogenous or pre-determined variables on the endogenous variables vector. We study monotonicity properties and asymptotic properties of the vector coecients. The estimator is applied to study the interdependence among countries in the European sovereign CDS market.