Pijoan-Mas (2006) model

Dear all,

I have recently updated a repo on my github to solve Pijoan-Mas model (RED 2006) which is in essence an Aiyagari model with elastic labor supply. I guess it might be interesting for students who want to start from a relatively simple model and then extend it.
Here is the link: GitHub - aledinola/PijoanMasTaxes

It should work well with the latest version of the toolkit. Warning: as usual, my code could be full of bugs! If you find any, please let me know :slight_smile:

Best,
Alessandro

1 Like

Fantastic!

Exciting milestone, first VFI Toolkit example written by someone other than me :smiley:

1 Like

I added a short pdf with a description of the model.

There seems to be a problem in the computation of some statistics like the earning shares, which is based on the lorenz curve

Some shares are greater than one, which can’t be :slight_smile:
Not really sure what’s going on

I’ll take a look at it later this week, let you know what I find (busy teaching until late in the week)

1 Like

Just pushed to github. Should be fixed now. Somehow the code for Lorenz curve had gotten a typo, this was infecting all the stuff based on lorenz. Thanks for spotting.

1 Like

You’re welcome! I ran again the code and this time the results make sense. They more or less correspond to the version of Pijoan-Mas model denoted as “Model E0” (see Tables 1 and 2 in the paper).
One question: If I do not specify vfoptions and simoptions, the toolkit automatically selects discretization with refinement for the VFI and the Tan improvement for the distribution, is that right?

MOMENTS
Corr(h,z) : 0.020037
CV(h) : 0.202920
Hours : 0.355203
K/Y : 2.989966
w*L/Y : 0.640020
I/Y : 0.248167

CV
CV(Hours) : 0.202920
CV(Earnings): 0.639090
CV(Income) : 0.627450
CV(Wealth) : 1.358534

GINI
Gini(Hours) : 0.095844
Gini(Earnings): 0.319153
Gini(Income) : 0.307965
Gini(Wealth) : 0.641633

CORR
corr(Hours,z) : 0.020037
corr(Wealth,z): 0.511574

SHARES EARNINGS
q1 earnings: 0.074302
q2 earnings: 0.124517
q3 earnings: 0.174612
q4 earnings: 0.228116
q5 earnings: 0.398453

SHARES WEALTH
q1 wealth: 0.000907
q2 wealth: 0.023915
q3 wealth: 0.094677
q4 wealth: 0.231797
q5 wealth: 0.648704

I think everything in the toolkit nowadays uses Tan improvement (everything involving agent distributions, only exception I can think of would be a model with semi-exo but no exo shocks as Tan improvement does not apply to semi-exo shocks).

And yes, infinite horizon with a decision variable will use refinement by default for value function (so Aiyagari 1994 codes won’t use it as there is no decision variable to refine, but Pijoan-Mas 2006 has the labor decision so it will use refinement).

1 Like