portfolio consumer model with recursive block definition working with generic monte carlo

sbenthall commented 3 weeks ago

This PR:

introduces a new 'new model', an unnormalized idiosyncratic shock model with (optional) portfolio share based on SolvingMicroDSOPS notes
an RBlock which can combine several DBlocks into one block.
passing smoke tests for running this model in the ATMC simulator

The model in consumer.py is an attempt to demonstrate how a value, such as R, can be either calibrated directly or set by a dynamic block (in this case, the portfolio block) while reusing the same consumption block data.

It also shows the use of a tick block to set $k_{t+1} = a_t$

[x] Tests for new functionality/models or Tests to reproduce the bug-fix in code.
[x] Updated documentation of features that add new functionality.
[x] Update CHANGELOG.md with major/minor changes.

sbenthall commented 3 weeks ago

This PR builds on #1414, which should be merged first

sbenthall commented 3 weeks ago

this is really work towards #1374

sbenthall commented 1 week ago

TODO: Allow nested RBlocks

sbenthall commented 1 week ago

TODO: Demo in a test the way that the consumption DBlock can be used with or without the RBlock

sbenthall commented 6 days ago

I believe this PR is ready. Requesting final review from @alanlujan91 or @mnwhite

This PR now supports recursively defined models, and demonstrates how a model can be composed of pre-made blocks.

Specifically, it shows how the portfolio consumption model can be built from the basic consumer model by adding one more block. This block makes the R rate of return an endogenous state that's a function of the risky return and stigma, overriding the default calibration of this value.

sbenthall commented 5 days ago

Added normalized version of consumption block

econ-ark / HARK

portfolio consumer model with recursive block definition working with generic monte carlo #1417