This R package covers a large range of semiparametric regression methods with time-varying coefficients using nonparametric kernel smoothing for the estimation.

You can install the released version of tvReg from CRAN with:

install.packages("tvReg")

or the development version from GitHub with:

devtools::install_github("icasas/tvReg")

The five basic functions in this package are tvLM(), tvAR(), tvSURE(), tvPLM(), tvVAR() and tvIRF(). Moreover, this package provides the confint(), fitted(), forecast(), plot(), predict(), print(), resid() and summary() methods adapted to the class attributes of the tvReg. In addition, it includes bandwidth selection methods, time-varying variance-covariance estimators and four estimation procedures: the time-varying ordinary least squares, which are implemented in the tvOLS() methods, the time-varying generalised least squares for a list of equations, which is implemented in the tvGLS() methods, time-varying pooled and random effects estimators for panel data, which are implemented in the tvRE() and the time-varying fixed effects estimator, which is implemente in the tvFE().

Details on the theory and applications to finance and macroeconomics can be found in Casas, Isabel and Fernandez-Casal, 2019, and in the package vignette https://icasas.github.io/tvReg/articles/tvReg.html.

Casas, Isabel and Fernandez-Casal, Ruben, tvReg: Time-varying Coefficient Linear Regression for Single and Multi-Equations in R (April 1, 2019). Available at SSRN:https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3363526.