Three examples | Serial correlation | Heteroskedasticity of unknown form | Both | Working with fits | Time series inputs | Choosing options | References
Package: fgls 0.1.0
Emilio González-Coya
fgls: Feasible GLS for Serial Correlation and Heteroskedasticity
Feasible generalized least squares (FGLS) estimators for linear regressions whose errors are serially correlated, heteroskedastic of unknown form, or both. Serial correlation is handled by the autoregressive filter of Perron and González-Coya (2026) <doi:10.1016/j.jeconom.2026.106346>, estimated from a Durbin regression whose order is chosen by an information criterion, with a variance estimator that stays valid when the regressors are only pre-determined rather than strictly exogenous. Heteroskedasticity is handled by the adaptive Lasso estimate of the skedastic function of González-Coya and Perron (2024) <doi:10.1515/jem-2023-0007>, with a bias-corrected sandwich variance. Both corrections can be combined. Benchmark estimators (ordinary least squares with heteroskedasticity and autocorrelation consistent standard errors, Cochrane-Orcutt, infeasible GLS) and the simulation designs of both papers are included.
Authors:
fgls_0.1.0.tar.gz
fgls_0.1.0.zip(r-4.7-any)fgls_0.1.0.zip(r-4.6-any)
fgls_0.1.0.tar.gz(r-4.7-any)fgls_0.1.0.tar.gz(r-4.6-any)
fgls_0.1.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
fgls/json (API)
| # Install 'fgls' in R: |
| install.packages('fgls', repos = c('https://cran.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/emiliogcs/fgls/issues
Pkgdown/docs site:https://emiliogcs.com
Last updated from:a32f0812ed. Checks:6 OK. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel | OK | 201 | ||
| source / vignettes | OK | 208 | ||
| linux-release | OK | 154 | ||
| windows-devel | OK | 127 | ||
| windows-release | OK | 166 | ||
| wasm-release | OK | 113 |
Exports:adf_testar_rootscochrane_orcuttcompare_fglsdurbin_estimatorfglsfgls_controlhybrid_glsinfeasible_glslasso_fglsmake_dictionarynonexogeneity_shareols_robustserial_fglssimulate_combinedsimulate_heterosimulate_predictivesimulate_serialtest_predeterminedweight_diagnostics
Dependencies:codetoolsforeachglmnetiteratorslatticeMatrixRcppRcppEigenshapesurvival
Getting started with fgls
Last update: 2026-11-07
Started: 2026-11-07
Heteroskedasticity of unknown form (González-Coya and Perron, 2024)
The idea | An example | Candidate covariates | No loss under homoskedasticity | Alternatives and benchmarks | Weights and robustness options | Computation
Last update: 2026-11-07
Started: 2026-11-07
Serial correlation (Perron and González-Coya, 2026)
The problem | The estimator | Table 1 of the paper | Variance and inference | Lag selection | Lagged dependent variables | MA(1) errors | Checking pre-determinedness
Last update: 2026-11-07
Started: 2026-11-07
Serial correlation and heteroskedasticity together
The first-stage correction | No loss under homoskedasticity | Bootstrap cross-check
Last update: 2026-11-07
Started: 2026-11-07
Validation: replicating the papers and agreement with pyfgls
Perron and González-Coya (2026), Table 1 | Perron and González-Coya (2026), Tables 2--5 (selected cells) | González-Coya and Perron (2024), Table 1 | Supplement, Table S.9: F-C and F-CH | Agreement with pyfgls
Last update: 2026-11-07
Started: 2026-11-07
