search

sbenthall / SHARKFin

Simulating Heterogeneous Agents with Finance
7 stars 5 forks source link

Default CRRA and DiscFac for SPARK SHARK? #216

Closed sbenthall closed 1 year ago

sbenthall commented 1 year ago

What shall we use for CRRA and DiscFac?

sbenthall commented 1 year ago

@alanlujan91 @llorracc ?

This is sort of a research design question. I can't tell if the research would benefit from looking at a grid of values or not. (This is the DiscFac and CRRA for the homogeneous consumer population.) (If we did grid over these values, we would need to make sure that the impatience condition was met.)

mesalas commented 1 year ago

That's a good point. Ill make an option for passing CRRA and discount factors into ammps so we can adjust them if needed.

sbenthall commented 1 year ago

These are implemented as command line arguments in SHARKFin: https://github.com/sbenthall/SHARKFin/blob/master/simulate/run_any_simulation.py#L71-L72

I wish I had a better sense of how the parameter grid was being passed through to the command line and AMMPS.

I think that's currently being done in code by @wjt5121 that isn't in a repository? I would like to be able to verify that the two simulations are synced.

sbenthall commented 1 year ago

CRRA as 5.0 and DiscFac as 0.99 are the defaults. These numbers passed the impatience condition with the dividend process calibrated to S&P500. We don't have to rock this boat unless there's a good reason!

sbenthall commented 1 year ago

CRRA is now 3.0 because the adjusted dividend statistics require it.