Replication of Huggett (JME 1997)

Huggett uses T=1000. I started with T=150 and it did not converge. So I increased it to T=500 but it still does not converge.

I turned on transpathoptions.graphpricepath=1; and transpathoptions.graphaggvarspath=1; and you can watch what goes wrong. Namely this initial guess leads to a very poor update that throws the tail of the thing off massively and then it never recovers.

I would run the value fn, agent dist, and agg vars ‘path’ commands using PricePath0 (which is essentially doing a single iteration of TransitionPath_Case command). And try see what is going on to get a decent initial guess for PricePath0.