FAIRR : Functional Approximation of Impulse Responses in R

FAIRR is an R package designed to estimate symmetric and asymmetric impulse response functions using the FAIR methodology (R. Barnichon, C. Matthes, 2018, Journal of Monetary Economics), using Gaussian basis functions :

$$\psi_h = \sum^K_{k=1} a_k \exp\left(\frac{(h-b)^2}{c^2}\right)$$

Or exponential basis functions :

$$\psi_h = a \exp(-\ell h)$$

The estimation is performed via Stan and users must therefore have a C++ compiler installed. If you encounter errors related to "Rtools" or "C++14", please ensure your toolchain is properly configured for Stan by following the RStan Getting Started guide.

Update

1.1.0 : Added support for exponential basis functions. Changed the function names : Gaussian basis FAIR_G(K) can be estimated with varfairG() if you want to get orthogonal shocks from a VAR(p) estimated automatically and with fairG() if you already have a series of exogenous shocks. For exponential decay basis function FAIR_E(1), use fairE(), which requires shocks to already be exogenous or orthogonal.

1.1.1 : Fixed documentation wording to include the exponential basis function.

1.1.2 : Fixed final output list for asymmetric exponential basis FAIR (wrong extract command).

Plan for 1.2.0 (non-binding) : Adding up to K = 4 on Gaussian basis IRF's, adding a Gamma distribution basis function.

Installation

devtools::install_github("gauthiersr/FAIRR")
library(FAIRR)

References

Barnichon, R., & Matthes, C. (2018). "Functional Approximation of Impulse Responses." Journal of Monetary Economics, 99, 41-55.

Stan Development Team. "RStan: the R interface to Stan." mc-stan.org