Dear Johannes, dear Dynare team,
thank you very much for this new release. I am particularly excited about the new possibility of estimating models with occasionally binding constraints, both for my research but also for the DSGE model we are developping at the National Bank of Belgium for forecasting and policy simulations. Our main interest is taking the effective lower bound into account during estimations.
In order to familiarize myself with the new feature, after having a look at the RBC example with irreversible investment you provided, I have tried to estimate a small NK model on Euro Area data for GDP-deflator inflation, GDP and the short term interest rate. The mode finder works fine in the absence of the OBC, and I use that mode as a starting point for the mode-finder with the ELB constraint imposed, using the piece wise Kalman-Filter. However, it seems that the mode-finding algorithm runs immediately runs into problems (see the attached .log file). I am wondering whether this is simply due to the particularities of the model/ the data, or whether I have made a mistake in the way I have written down the OBC. The lines related to monetary policy and the OBC are in model_obc_small_lowkappa_c.mod, lines 201-204, and 246-248. The constraint is supposed to keep the “endogenous” component rnot above or equal to 0.25/4, approximately, ((consterr+constepinf) is the steady state quarterly nominal interest rate). Does the constraint I have coded achieve that goal…? When I use the simulation mode instead of the estimation mode, it looks as if the constraint is imposed correctly.
For occbin_setup, see line 295, and line 302 for the estimation command. Note that for the estimation I commented out occbin_solver. The estimation wouldn’t run otherwise.
As always, your help would be highly appreciated!
exampleOBC.zip (2.1 MB)