sequence-jacobian

Sequence-Space Jacobian Methods for Solving and Estimating Heterogeneous Agent Models


License
MIT
Install
pip install sequence-jacobian==1.0.0

Documentation

Sequence-Space Jacobian

Interactive guide to Auclert, Bardóczy, Rognlie, Straub (2019): "Using the Sequence-Space Jacobian to Solve and Estimate Heterogeneous-Agent Models".

Click here to download all files as a zip. Note: major update on July 26, 2019.

1. Resources

1.1 RBC notebook

Warm-up. Get familiar with solving models in sequence space using our tools. If you don't have Python, just start by reading the html version. If you do, we recommend downloading our code and running the Jupyter notebook directly on your computer.

1.2. Krusell-Smith notebook

The first example. A comprehensive tutorial in the context of a simple, well-known HA model. Shows how to compute the Jacobian both "by hand" and with our automated tools. Also shows how to calculate second moments and the likelihood function.

1.3. One-asset HANK notebook

The second example. Generalizes to a more complex model, with a focus on our automated tools to streamline the workflow. Introduces our winding number criterion for local determinacy.

1.4. Two-asset HANK notebook

The third example. Showcases the workflow for solving a state-of-the-art HANK model where households hold liquid and illiquid assets, and there are sticky prices, sticky wages, and capital adjustment costs on the production side. Introduces the concept of solved blocks.

1.5. HA Jacobian notebook

Inside the black box. A step-by-step examination of our fake news algorithm to compute Jacobians of HA blocks.

2. Setting up Python

To install a full distribution of Python, with all of the packages and tools you will need to run our code, download the latest Python 3 Anaconda distribution. Note: make sure you choose the installer for Python version 3. Once you install Anaconda, you will be able to play with the notebooks we provided. Just open a terminal, change directory to the folder with notebooks, and type jupyter notebook. This will launch the notebook dashboard in your default browser. Click on a notebook to get started.

For more information on Jupyter notebooks, check out the official quick start guide. If you'd like to learn more about Python, the QuantEcon lectures of Tom Sargent and John Stachurski are a great place to start.