According to " A Guide to Specifying Observation Equations for the Estimation of DSGE Models", we should be setting the standard deviation of technology shocks to equal around 0.007. What about non-technology shocks such as monetary or preference shocks? What is the ideal standard deviation to set them to?
The numbers in my Guide are set to fit an AR1-process to TFP. For monetary policy or preference shocks, take a look at literature with estimated models like Smets/Wouters (2007) or Peter Ireland’s work.