Simulation code cannot allow for *age* varying interest rates

[Mv77]

Some time ago, @wdu9 highlighted an issue with the simulation code that prevents age-varying interest rates from being used.

For age varying interest rates, the fix is very simple as @mnwhite and I suggested in that thread:

• Add Rfree (and Rboro?) to time_vary.
• Edit get_Rfree() appropriately.

But the issue turned out to be that Will wanted not age-varying but time-varying interest rates. This is much trickier and the discussion spurred plans to separate time and age in HARK and to make the default rate of return risky.

Agree with the plans or not, there is still a very easy fix for age-varying rates, which we don't have yet (this PR does not achieve that, but something else).

An external user emailed me yesterday requesting this feature. I don't have the time to fix it myself in the next two weeks.

[alanlujan91]

There are still some blocking issues in #1108 that I am trying to work on this week. That PR adds agent type IndShockRiskyAssetConsumerType with an age_varying interest rate, which essentially merges the functionality of IndShockConsumerType and PortfolioConsumerType.

@Mv77 did the external user request this feature for KinkedRConsumerType (since you mentioned Rboro)? The current changes I'm working on would not affect this type, since it does not use Rfree but instead uses Rboro and Rsave.

[mnwhite]

If you want interest rates to vary over absolute time, regardless of the age of individual agents, this can already be done with HARK. Just use MarkovConsumerType and discretize Rfree into some set of states it can take on, and then specify transition probabilities among them, which will be the same for all ages in the lifecycle. For simulation, there's some boolean attribute that indicates whether the discrete state is idiosyncratic or commonly shared; just set it to shared.

…

On Tue, Apr 12, 2022 at 11:30 AM alanlujan91 ***@***.***> wrote: There are still some blocking issues in #1108 <#1108> that I am trying to work on this week. That PR adds agent type IndShockRiskyAssetConsumerType with an *age_varying* interest rate, which essentially merges the functionality of IndShockConsumerType and PortfolioConsumerType. @Mv77 <https://github.com/Mv77> did the external user request this feature for KinkedRConsumerType (since you mentioned Rboro)? The current changes I'm working on would not affect this type, since it does not use Rfree but instead uses Rboro and Rsave. — Reply to this email directly, view it on GitHub <#1132 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADKRAFP3JP35I4DPZDACTPLVEWJK3ANCNFSM5THUCW3A> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[alanlujan91]

@Mv77 @wdu9 does time/age varying Rfree in simulation works now or is this still outstanding?

[Mv77]

I believe your recent changes should have fixed this, right?

Rfree can now be a list and goes into time_vary? If so, this has been addressed.