DSGE models are used for analyzing policy and the sources of business cycles. A competing approach uses VARs that are partially identified using, for example, narrative shock measures and are often viewed as imposing fewer restrictions on the data. Narrative shocks are identified non-structurally through information external to particular models. This uses non-structural narrative shock measures to inform the structural estimation of DSGE models. Since fiscal policy has received much recent attention but the foundations of the fiscal side of DSGE models are less well studied than their monetary building block, fiscal DSGE models are a particularly promising application. Preliminary results from a standard medium-scale DSGE model support this argument: Structurally identified monetary shocks line up well with narrative measures, whereas government spending shocks do not. Extending the model to include distortionary taxes and more general fiscal policy processes, I find that model implied labor tax shocks line up well with narrative tax shocks. Including different narrative shock measures affects parameter identification and implied measures such as fiscal multipliers.
