The popular Value-at-Risk of an institution provides a measure of its own risk. However, in many cases it is of interest to know the measure of the contribution of an institution to the systemic risk, based on the Conditional Value-at-Risk. In this paper we compare the estimation of such measure according to two different methods. The former is based on the quantile regression, the latter on copula functions. In both cases, the heteroskedasticity of the time series is explicity taken into account through a GARCH structure. Moreover, the comparison is also made across different distributional assumptions.