loglinear model and Stoch_simul(.,replic=integer)

[pjg]

Hi all, two questions:

1. if I run a log-linear model, i.e. EulerEq: exp(c)^-sigma=beta*(1+r(+1))*exp(c(+1))^-sigma, and so on...after simulate the model, to get the consumption I need to redefine the variables as exp(x) right?. So consumption in the example above is exp(c). But also I need to apply exp( ) to all output presented as MOMENTS OF SIMULATED VARIABLES and POLICY AND TRANSITION FUNCTIONS and others, right?.

2. When I used the default option Stoch_simul;, I am curious if the simulated series presented are just one draw or is the average of a bigger number of simulations?. If I define Stoch_simul(order_2, irf=20,replic=100) the simulated series presented are the average of 100 draws or the 100 draws are only to build the IRFs. The user guide says:

replic = INTEGER: number of simulated series used to compute the
IRFs (default = 1 if order = 1, and 50 otherwise).

for that reason my confusion. thanks in advance!

[jpfeifer]

1. The way you enter your Euler equation, you are performing a log-transformation, i.e. c is actually log(C). So if you want to back out the level of consumption, you need exp(c). For the moments, this is not so easy, because you std(exp(c)) is not equal to exp(std(c)). What you can do is define an auxiliary variable

Code: Select all

c_level=exp(c);

Dynare will then provide the moments for this variable which is the level. My question is why you use the log-transformation in the first place if you are interested in levels and not in percentage deviations from steady state.
2. At first order, there is no point in using replications to generate IRFs, because due to linearity, every simulation will be the same. Hence, the default is one. For order>1, the location in the state space matters and the IRFs are the average over replic simulated series. The default at second order is 50.