Codes

I design open-source Python statistical packages for central banks and economists in general, in the area of forecasting and financial modeling. One of my most recent packages, based on Adrian et al. (AER 2019) on estimating Growth-at-Risk, is used by more than twenty central banks in the world.

The packages are freely available through Github. Please cite the companion working paper when using the tools.

Disclaimer: Reuse of these tools does not imply any endorsement of the research and/or product. Any research presented should not be reported as representing the views of the IMF, its Executive Board, or member governments.

Growth at Risk: Density Forecasting via Quantile Regressions and Parametric Fit (with Changchun Wang)

• Github repo

• Python package and Excel tool to estimate the Growth-at-Risk model, which is a density forecasting model based on quantile regressions and parametric fit. Changchun Wang (also at the IMF) designed the Excel-Python interface

  • Developped at the International Monetary Fund, from the seminal paper of Adrian et al. (2019) Vulnerable Growth, American Economic Review. I have incorporated a significant number of improvements to make GaR applicable to a wide range of countries, including emerging markets and countries with data limitations.

  • Used by more than twenty central banks in the world (either the Excel tool developped by Changchun or the pure Python scripts I wrote)

  • Please cite our working paper presenting the toolkit and applications Growth-at-Risk: Concept and Application in IMF Country Surveillance (2019), IMF Working Paper No. 19/36

Distributional GaRCH model to design VaR-based FX Interventions for Central Banks

• Github repo

• Python module to estimate conditional densities from a GaRCH model and design VaR-based intervention FX interventions areas for central banks

  • The paper uses a Python package that I have written, DistGARCH, also available on my Github page, with the public FX intervention data from the Banco Mexico. DistGARCH is based on the ARCH package of Kevin Sheppard.

  • You can use the code for non-commercial applications, providing that you cite the IMF Working Paper Lafarguette, R. and Veyrune, R. (2020) “Foreign Exchange Interventions Rules for Central Banks: A Risk-Based Framework”, IMF Working Paper

Quantile Local Projections

• Github repo

• Python module to estimate quantile local projections, and produce useful outputs for economists

  • Based on the QuantileReg package from statsmodels and my conditional quantile sampling module (https://github.com/romainlafarguette/cqsampling)

  • The quantile uncrossing part is based on either:

    • Chernozhukov et al. (2010) Quantile and Probability Curves Without Crossing, Econometrica
    • Schmidt and Zhu (2016), Quantile Spacings: A Simple Method for the Joint Estimation of Multiple Quantiles Without Crossing

Conditional Density Projection via Quantile Regressions, Resampling and Multifit Models

• Github repo

• Python module to project a conditional density using quantile regressions, resampling and different density fit strategies (Resampling, Kernel, Gaussian Mixtures, etc.)

• Working paper coming out soon

Robust Density with Over-Parametrized Models

• Github repo

• Estimate a Conditional Skew Normal using robust estimators (Theil-Sen and Firth Logistic Regressions) and an over-parametrized model.

• Useful for small and/or noisy observational samples

• Working paper to arrive soon

Partial Least Squares Wrapper for Data Reduction based on Scikit

• Github repo

• A wrapper based on Scikit to use a Partial Least Square estimator to conduct data reduction.

• Includes new functionalities such as variable influence in the projection computation, selection on top contributors and a few charting tools

Granular Instrumental Variables from Gabaix and Koijen (2020)

• Github repo

• Python module to estimate the Granular Instrumental Variables from Gabaix and Koijen (2020)

• The panel regressions are based on linearmodels from Kevin Sheppard

Quantile Spacing from Schmidt and Zhu (2016)

• Github repo
• Python module to estimate Quantile Spacing from Schmidt and Zhu (2016)

Cluster Analysis Wrapper with Performance Metrics and Visualization Tools

• Github repo
• Python wrapper to conduct data clustering, with performance metrics and visualization tools. Based on Scikit