We propose to exploit stochastic volatility for statistical identification of structural vector autoregressive models (SV-SVAR). We discuss full and partial identification of the model and develop efficient Expectation Maximization algorithms for Maximum Likelihood inference. Simulation evidence...