Interpretation of Policy Functions

Dear all,

I am simulating a DSGE model that is represented by a system of nonlinear expectational difference equations using Dynare. A solution to such a system is called a set of policy functions which are approximated by using perturbation techniques. According to theory, a policy function maps state variables of the model into choices of the control variables. Is this reasoning correct?

As an example, I get the following policy function for the control variable L (represents labor demand):

L = 70 + ... + 3.81 \tau_{t-1}

\tau_{t-1} represents a tax rate and takes on values between 0 and 1. Do I interpret the coefficient of 3.81 in a similar way as if the policy function was simply a linear regression setup? And does the interpretation of the coefficient depend on the scaling of \tau_{t-1}? So if I were to increase the tax rate from t-1 by 1 percentage point, by how much does L change?

Thank you!

Best, Matthias

  1. Yes, that interpretation is correct, with the difference to OLS being that there is no potential issue with joint determination.
  2. Similar to regression equations, the interpretation of the numbers depends on the measurement, e.g. elasticities if everything is in logs.

Thank you!