Structural vector-autoregressive models are potentially very useful tools for guiding both macro- and microeconomic policy. In this paper, we present a recently developed method for exploiting non-Gaussianity in the data for estimating such models, with the aim of capturing the causal structure underlying the data, and show how the method can be applied to both microeconomic data (processes of firm growth and firm performance) as well as macroeconomic data (effects of monetary policy).