Usefulness of DSGE results

Dear Dynare users,

I have a question about the usefulness of DSGE results in practical macroeconomic forecasting. I’m working as a forecaster in an institute. I’m not a DSGE expert but I often see DSGE results. DSGE results are usually reported as a % deviations from the initial/constant pre-reform steady state level. Suppose that the steady state of the model is calibrated using the data 2015-2019. Also suppose that the GDP level in the steady state is constrained, i.e. derived from definition equation given other values. Also suppose that there is a reform planned and a DSGE modeller estimates that the next year the GDP will be 0,2 % higher relative to the pre-reform steady state level thanks to this reform.

My question is: how to use this result in my forecasts for the next year? First, I need to estimate what the GDP would be without the reform, e.g. I estimate 3 % growth yoy, and then I just need to add 0,2 % to my estimate, i.e. 3 + 0,2 = 3,2 %?

Thanks for your answer.


That depends on whether you think the two effects are additive.

Thank you for your answer prof Pfeifer. I’m a bit confused because DSGE results are reported as a percentage deviations from SS. I would be grateful to you for giving me more guidelines how to use DSGE results in forecasting in practice.


To interpret such results, we would need the measurement equations. The measurement equations tell you how the observed data relates to states in the model.

Thank you for the answer. If I understand correctly the result that the next year GDP level is estimated to be 0.2% higher as in the steady state due to reform is completely inuseful to me as a forecaster. This is just a theoretical result?

Let me reiterate my point: You have information about two effect: the reform effect of 0.2% and a counterfactual of 3% relative to the old steady state. The problem is that you did not state what the 0.2% refers to. Is it that the reform moves the steady state up by 0.2%? Or is it a temporary effect relative to steady state? Similarly, it’s not clear how the two effects interact. You could think of situations where they two effects are simply additive. For example, think about private demand and public demand just being additive. But you could also think about interaction effects. For example, a tax reform may make the economy more efficient and shift up the steady state. In that case, if you predicted to be 3% above the old steady state, you would expect to be 3% above the new steady state.