| 1 |
/* A simple purely backward model, used for testing both “simul_backward” and |
| 2 |
perfect foresight simulations, consisting of: |
| 3 |
- a VAR(3) with one variable in level, one in log and one log-differentiated; |
| 4 |
- two other variables that depend on the contemporaneous values of the VAR(3), |
| 5 |
and which together for a simultaneous block. |
| 6 |
Those five equations are repeated an arbitrary number of times (n). */ |
| 7 |
|
| 8 |
@#define n = 10 |
| 9 |
|
| 10 |
@#for i in 1:n |
| 11 |
var x@{i}, y@{i}, z@{i}, t@{i}, s@{i}; |
| 12 |
varexo e_x@{i}, e_y@{i}, e_z@{i}, e_t@{i}, e_s@{i}; |
| 13 |
@#endfor |
| 14 |
|
| 15 |
model(use_dll); |
| 16 |
@#for i in 1:n |
| 17 |
x@{i} = 0.1*x@{i}(-1) + 0.2*log(y@{i}(-1)) + 0.3*diff(log(z@{i}(-1))) + e_x@{i}; |
| 18 |
[name = 'y'] |
| 19 |
log(y@{i}) = 0.4*x@{i}(-1) + 0.5*log(y@{i}(-1)) + 0.6*diff(log(z@{i}(-1))) + e_y@{i}; |
| 20 |
[name = 'z'] |
| 21 |
/* NB: we choose 0.5 as steady state for z, since initial value is drawn from |
| 22 |
[0,1] (in the “simul_backward” case) */ |
| 23 |
diff(log(z@{i})) = -0.8*(log(z@{i}(-1)) - log(0.5)) + 0.1*x@{i}(-1) + 0.2*log(y@{i}(-1)) - 0.3*diff(log(z@{i}(-1))) + e_z@{i}; |
| 24 |
|
| 25 |
t@{i} = x@{i} + 2*log(y@{i}) + 3*diff(log(z@{i})) + s@{i} + e_t@{i}; |
| 26 |
s@{i} = 4*x@{i}(-1) + 3*t@{i} + e_s@{i}; |
| 27 |
@#endfor |
| 28 |
end; |