occbin bug: varables with trend are not properly handled in switching equations
Assume a model with variables with trend:
var bg y (deflator=y0);
var t;
with parameters
parameters eta deftar
eta=0.01;
deftar=0.03;
I a variable with a trend appears in an equation that switches:
[name = 'debt stabilization rule', bind = 'bg0']
t=0;
[name = 'debt stabilization rule', relax = 'bg0']
t = eta*( (bg-bg(-1))/y - deftar);
...
steady_state_model;
t=0;
y=1;
bg=2.4;
end;
then, the switches are treated BEFORE handling the trends, which means that in static_resid the first difference bg-bg(-1)
is first evaluated to be ZERO (which is wrong, since the trend is not considered there), so that in steady state we have a residual equation like
resid() = t_ss - eta*(-deftar);
which provides a non-zero residual.
I fear a similar issue pops up in the dynamic file.