PerfectForesight chimera

[sbenthall]

A chimera for the PerfectForesight model, "almost the simplest consumption model".

[sbenthall]

https://github.com/econ-ark/DemARK/blob/master/notebooks/Gentle-Intro-To-HARK-PerfForesightCRRA.ipynb

[llorracc]

Seb, I mot sure that the Perfect Foresight model is the best next chimera to work with, because that model is basically the special case where you don’t actually need any of the apparatus of either hark or Dolo/ARK to solve it: We have an analytical solution. My PerfForesightCRRA <http://econ.jhu.edu/people/ccarroll/public/lecturenotes/consumption/PerfForesightCRRA> handout structures the problem formally as a Bellman problem, and might be helpful to you in seeing the logic of the apparatus in a case where it magically simplifies to a linear consumption function. A better choice would probably be the KinkedRConsumerType DemARK example, because it is just one small step beyond the BufferStock chimera.

…

On Tue, Jan 28, 2020 at 4:49 PM Sebastian Benthall ***@***.***> wrote: A chimera for the PerfectForesight model, "almost the simplest consumption model". — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#8?email_source=notifications&email_token=AAKCK7ZOYYVYJF72EVO2FXDRABHXJA5CNFSM4KMUW7SKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IJIOGKQ>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKCK73A3ENJLQOSVSDULXLRABHXJANCNFSM4KMUW7SA> .

-- - Chris Carroll

[sbenthall]

I see your point that PerfForesight is "too simple" to be very interesting as a research output.

From a software perspective, I was trying this out as a "hello world" for writing something in Dolang from scratch.

I've bumped into an issue with this attempt, which I've noted here:
EconForge/dolo.py#177

Were it not for that issue, this chimera would have been trivial. As is, I've either found a bug in dolang or (far more likely) discovered that there's something about writing dolang I don't understand yet.

In either case, I think it was worth taking a shot at.

I'll make an issue for KinkedRConsumerType and work on that one next.

[sbenthall]

#9

[llorracc]

I think it's somewhere in between. Perfect foresight models are SO peculiar that they are kind of pathological in some ways. Dolo may implicitly make assumptions that hold in every case EXCEPT the pathological one of perfect foresight models.

…

On Sat, Feb 1, 2020 at 1:30 AM Sebastian Benthall ***@***.***> wrote: #9 <#9> — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#8?email_source=notifications&email_token=AAKCK722P4RKILAUTXZ32VDRAS7A7A5CNFSM4KMUW7SKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEKQNYUA#issuecomment-580967504>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKCK757Y6ZR55S6CLLD7KDRAS7A7ANCNFSM4KMUW7SA> .

-- - Chris Carroll

[sbenthall]

Closing this issue because, for all the reasons @llorracc says, it's too peculiar.