Bilbiie 2020, Monetary Policy and Heterogeneity: An Analytical Frameowor

Dear all, I have to run the TANK model of the paper in the title where the probability of becoming hand-to-mouth (call it s) depends on tomorrow aggregate output: s(Y_{t+1}).
I have written s among the parameters (in Dynare) of the baseline model where s was exogenous; but now that s is endogenous, how can I write the fact that s depends on tomorrow aggregate activity?

Thank you very much in advance.

. Basically my problem is that now I should write “s” depending on y(+1) but I cannot write it as s(y(+1)) otherwise the following error appears: “ERROR: tankcw1.mod: line 24, cols 8-15: Symbol s is being treated as if it were a function (i.e., takes an argument that is not an integer).”
How could I write it?

You cannot work with implicit definitions of functions. You need to specifiy what the function s() is.

Thank you. I specify s=kappay(+1) and in Dynare given that the model is in log-linear terms, I’ve written the log-linearized function for s as: s-0.5= kappay(+1) ,given that the steady state for s is 0.5 and for y is zero.
Now, my problem is that I find the same impulse responses for different calibration of kappa… if the risk is counter-cyclical (kappa greater than zero in my model), I should expect that government spending crowds-in private consumption while if risk is pro-cyclical ( values for kappa less than zero), I should expect that government spending crowds-out private consumption. However, in both cases I find that government spending crowds-out private consumption even for k=6 which is a high value as you can see in attached.
Thank you very much!
tankcw1.mod (1.3 KB)

I get a BK violation with the attached file. Did you check whether kappa even affects the decision rules?

1)As regards BK violation, it depends on the value of k; for example with k=1, k=0.80, k= -0.40, k=-0.20, the BK conditions are satisfied and irfs are displayed. Sorry for having attached the file with the wrong value of k…
2) k affects for example the consumption rule via s and via 1-s; therefore, my model should work once I’ve written the equation for s as: s-0.5=kappa*y(+1).
If you try first with counter-cyclical risk i.e. k=0.80 for example and then you try with pro-cyclical risk i.e. k=-0.40 for example, you get the same IRFs in response to a government shock… and this is very strange

From what I can see, the s indeed does not affect the solution for the model variables. s moves differently, but no other variable does.