Skip to content

also known as Euler Equation Iteration (Moll), Time Iteration (Sargent), based on Kaplan's codes

Notifications You must be signed in to change notification settings

MCai416/Bellman-Policy-Iteration

Repository files navigation

Bellman-Policy-Iteration

also known as Euler Equation Iteration (Moll), Time Iteration (Sargent), based on Kaplan's codes, written using Python 3

There are 3+1 .py files, which has slightly different models

EEI_AG_IID looks at the consumption/savings policy (supply side of the capital market) of an income fluctuation model, which is the capital supply side of the Aiyagari model, with iid income distribution, allows savings with a constant rate of return, code written based on Kaplan's Matlab codes

EEI_AG_Markov modifies the iid model to a finite state Markov process, which allows for responses to be conditional on past states, more lags are also possible by making a larget matrix, but it takes a lot of computational power. For example, the iid setting takes 4 seconds to compute, while Markov process takes more than 15 seconds. Meanwhile there isn't a lot of information added using Markov process.

EEI_GM_Deter looks at the classical social planner's deterministic growth model, which is different from the income fluctuation model, you can compare the results between policy iteration in EEI_GM_Deter.png and bisection/shooting algorithm in the "Discrete Ramsey" repository. The picture results have the same axes settings.

Lastly, although irrelevant, VFI_IID is the python code for Bellman Value Function Iteration under the same setup as EEI_IID. This algorithm is based on the contraction mapping of Bellman Operator. Although it is the easiest in theory, the actual performance is horrific.

About

also known as Euler Equation Iteration (Moll), Time Iteration (Sargent), based on Kaplan's codes

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages