I have been reading some QUEST model-based simulation papers (long-run effects). However, sometimes it is not clear how the simulation is done. May I kindly ask how many ways to simulate long-run effects?

**Approach one**

let say you have some equation (1-\tau)x_t = m_t + .... where \tau measures some policy, say tax or tariff, etc. You can change the value of \tau and the values of the new steady-state will be the long-run effect, right? No transitions here.

**Approach two**

let say you have some equation (1-\tau_t)x_t = m_t + .... where \tau_t is a time-varying policy variable, say tax or tariff, etc. You can use a deterministic version of the model to simulate a change in \tau_t from old steady-state to new steady-state (if permanent shock). So values of the new steady-state variables are long-run effects of the policy, yeah?

**Approach three**

let say you have some equation (1-\tau^x_t)x_t + (1-\tau^y_t)y_t= m_t + .... The policy question is whether government should financed some project using \tau^x_t or \tau^y_t? Which policy has biggest benefit in the long-run? I guess you would compare new steady-state values for cases (\tau^x_t=0, \tau^y_t \neq 0) and (\tau^x_t\neq0, \tau^y_t = 0), right?

Any other ways to do policy counterfactual simulations?