Consumption equivalence across different models

Stochastic simulations
1

I’m currently trying to numerically compute consumption equivalence across two regimes (tax and no tax).

Using the code from Born and Pfeifer (2018) as reference, I’ve set up a function to calculate the CE factor:

function outvalue=get_consumption_equivalent_unconditional_welfare(par_value_lambda_ce)

global base_oo base_options base_m permtax_oo permtax_options permtax_m

if par_value_lambda_ce>1 
    outvalue=1e5+par_value_lambda_ce^2;
end

set_param_value('lambda_ce',par_value_lambda_ce) %set consumption equivalent lambda
var_list_ = {'UTIL','V_CE'};
[info1,oo1] = stoch_simul(base_m, base_options, base_oo, var_list_); %get decision rules and moments
[info2,oo2] = stoch_simul(permtax_m, permtax_options, permtax_oo, var_list_); %get decision rules and moments
if info1(1)
    outvalue=1e5+par_value_lambda_ce^2;
    return;
end
if info2(1)
    outvalue=1e5+par_value_lambda_ce^2;
    return;
end
betta = base_m.params(strmatch('betta',base_m.param_names,'exact'));
outvalue = 1/(1-betta)*oo2.mean(strmatch('UTIL',var_list_,'exact')) - oo1.mean(strmatch('V_CE',var_list_,'exact'));

But unfortunately I get the following error:

Dot indexing is not supported for variables of this type.

Error in stoch_simul (line 21)
if isequal(options_.order,0)

Error in get_consumption_equivalent_unconditional_welfare (line 11)
[info1,oo1] = stoch_simul(base_m, base_options, base_oo, var_list_); %get decision rules and moments

Error in csolve (line 60)
    f0=feval(FUN,x);

My initial guess was that stoch_simul cannot treat base_options as a structure when it is declared as a global object. But then I had a look back at the reference code and saw that options_ was also declared as a global object:

Any assistance or suggestions would be appreciated.

2

Can you please provide the full set of files that generate the error.

3

Thanks Johannes. I’ve DM’d the necessary files.
(will share results here once complete)

4

The problem comes from the use of global variables in your code. I amended the original example code to eliminate the globals.

5

@jpfeifer I’m getting the same error when I try stoch_simul after running Thomas Winberry’s toolbox (both the HA model hosted on your GitHub and the KhanThomas model which I updated similarly). I’m guessing it’s a similar issue since globals are used – do you know what updates would be needed to make this possible? Thanks

6

Without seeing the files it is impossible to tell.

7

So for example if I run dynamics.m from your github repo and then try to run something else afterwards like stoch_simul(periods=1000), I get an error as described in the OP (dot indexing not supported..).

8

Where exactly did you put which command?

stoch_simul(periods=1000)

is not a valid Matlab command. It only works within a mod-file. Otherwise, you need to call the Matlab function stoch_simul withs its correct syntax.

1 Like
9

Thanks, I misremembered being able to. For anyone else who happens to stumble upon this thread from searching the error – the correct post-estimation call is stoch_simul(M_, options_, oo_, var_list_)

10

No, in Dynare 6 the correct call is:

[info, oo_, options_, M_] = stoch_simul(M_, options_, oo_, var_list)

Otherwise, you won’t retrieve the output.

1 Like