I am replicating a model that compares two supporting policies for the renewable energy sector. The model contains a fossil fuel and a renewable sector. To determine which policy is better, the authors simulate the prices of fossil fuel energy and renewable energy and examine under which policy price parity is reached sooner.

The model is estimated and the model is intensive form, meaning none of the model equations contain a trend. The model equations were log-linearized. The observation equations on the other hand contain a trend that is calibrated with 1.6. An example of an observation equation is:

\Delta\log(C_t) = \Gamma + 100 * (c_t - c_{t-1}){\lim}_{x\rightarrow\infty}

Where \Gamma is the annual trend growth rate common to the observation variables considered in the model.

I am wondering how to simulate the two prices to attain the goal of the model of examining the period when price parity is reached. The authors draw 200’000 shocks for each period and then use the mean of each period for that period. But there is no information on where the simulation starts.

a) Let’s say the simulation starts at the steady state. The shocks will be approximately zero for each period, thus I think the simulation would just roughly line out the steady state for the simulation period, which seems unusable.

b) It seems more purposeful to start the simulation off steady state, say at the point which describes the real economy right now. But then the question is what values to use for capital and technology to initiate the simulation.

c) One other way I could imagine is computing a model without any supporting policy and start the simulation for the other models from its steady state. Is this how this problem is usually approached?

Thank you for any help with this!