From 85e22cf623e30893d7d1158637b1fe8824c180b1 Mon Sep 17 00:00:00 2001 From: thomaswiemann Date: Tue, 28 Nov 2023 20:27:32 -0600 Subject: [PATCH] adds arxiv --- R/civ.R | 4 ++-- README.Rmd | 9 +++++---- README.md | 35 +++++++++++++---------------------- vignettes/civ.Rmd | 2 +- vignettes/civ.Rmd.txt | 2 +- 5 files changed, 22 insertions(+), 30 deletions(-) diff --git a/R/civ.R b/R/civ.R index 0accc6c..dc36823 100644 --- a/R/civ.R +++ b/R/civ.R @@ -31,7 +31,7 @@ #' Fox J, Kleiber C, Zeileis A (2023). "ivreg: Instrumental-Variables Regression #' by '2SLS', '2SM', or '2SMM', with Diagnostics". R package. #' -#' Wiemann T (2023). "Optimal Categorical Instruments." +#' Wiemann T (2023). "Optimal Categorical Instruments." https://arxiv.org/abs/2311.17021 #' #' @examples #' # Simulate data from a simple IV model with 800 observations @@ -89,7 +89,7 @@ civ <- function(y, D, Z, X = NULL, K = 2) { #' Fox J, Kleiber C, Zeileis A (2023). "ivreg: Instrumental-Variables Regression #' by '2SLS', '2SM', or '2SMM', with Diagnostics". R package. #' -#' Wiemann T (2023). "Optimal Categorical Instruments." +#' Wiemann T (2023). "Optimal Categorical Instruments." https://arxiv.org/abs/2311.17021 #' #' @export #' diff --git a/README.Rmd b/README.Rmd index 0eaab0f..1dd064b 100644 --- a/README.Rmd +++ b/README.Rmd @@ -20,9 +20,9 @@ knitr::opts_chunk$set( [![CodeFactor](https://www.codefactor.io/repository/github/thomaswiemann/civ/badge)](https://www.codefactor.io/repository/github/thomaswiemann/civ) -``civ`` is an implementation of the categorical instrumental variable estimator proposed by Wiemann (2023; [arxiv:123](https://)). ``civ`` allows for optimal instrumental variable estimation in settings with relatively few observations per category as seen in many economic applications including judge IV designs. To obtain valid inference in these challenging empirical settings, ``civ`` leverages a regularization assumption that implies existence of a latent categorical variable with fixed finite support achieving the same first stage fit as the observed instrument. +``civ`` is an implementation of the categorical instrumental variable estimator proposed by Wiemann (2023; [arxiv:2311.17021](https://arxiv.org/abs/2311.17021)). ``civ`` allows for optimal instrumental variable estimation in settings with relatively few observations per category as seen in many economic applications including judge IV designs. To obtain valid inference in these challenging empirical settings, ``civ`` leverages a regularization assumption that implies existence of a latent categorical variable with fixed finite support achieving the same first stage fit as the observed instrument. -See the corresponding working paper [Optimal Categorical Instrumental Variables](https://) for further discussion and theoretical details. +See the corresponding working paper [Optimal Categorical Instrumental Variables](https://arxiv.org/abs/2311.17021) for further discussion and theoretical details. ## Installation @@ -142,7 +142,7 @@ To provide some evidence for the practical benefits of CIV over alternative esti Key takeaways from the results: TSLS, post-Lasso IV, and random forest-based IV are heavily biased. -For extensive finite-sample comparisons, see Wiemann (2023; [arxiv:123](https://)). +For extensive finite-sample comparisons, see Wiemann (2023). ### TSLS @@ -189,6 +189,7 @@ c(Estimate = ranger_res[2, 1], "Std. Error" = ranger_res[2, 2], ``` # References -Wiemann T (2023). "Optimal Categorical Instruments." Belloni A, Chen D, Chernozhukov V, Hansen C (2018). “Sparse Models and Methods for Optimal Instruments With an Application to Eminent Domain.” Econometrica, 80(6), 2369-2429. + +Wiemann T (2023). "Optimal Categorical Instruments." https://arxiv.org/abs/2311.17021 diff --git a/README.md b/README.md index 59a87de..5a59083 100644 --- a/README.md +++ b/README.md @@ -11,17 +11,19 @@ `civ` is an implementation of the categorical instrumental variable -estimator proposed by Wiemann (2023; [arxiv:123](https://)). `civ` -allows for optimal instrumental variable estimation in settings with -relatively few observations per category as seen in many economic -applications including judge IV designs. To obtain valid inference in -these challenging empirical settings, `civ` leverages a regularization +estimator proposed by Wiemann (2023; +[arxiv:2311.17021](https://arxiv.org/abs/2311.17021)). `civ` allows for +optimal instrumental variable estimation in settings with relatively few +observations per category as seen in many economic applications +including judge IV designs. To obtain valid inference in these +challenging empirical settings, `civ` leverages a regularization assumption that implies existence of a latent categorical variable with fixed finite support achieving the same first stage fit as the observed instrument. See the corresponding working paper [Optimal Categorical Instrumental -Variables](https://) for further discussion and theoretical details. +Variables](https://arxiv.org/abs/2311.17021) for further discussion and +theoretical details. ## Installation @@ -112,17 +114,6 @@ robust standard errors. See also `?civ` and `?summary.civ` for details. # Load dependencies library(civ) library(AER) -#> Loading required package: car -#> Loading required package: carData -#> Loading required package: lmtest -#> Loading required package: zoo -#> -#> Attaching package: 'zoo' -#> The following objects are masked from 'package:base': -#> -#> as.Date, as.Date.numeric -#> Loading required package: sandwich -#> Loading required package: survival # Compute CIV with K=2 and conduct inference civ_fit <- civ(y = y, D = D, Z = Z, X = as.factor(X), K = 2) @@ -196,7 +187,7 @@ c(Estimate = civ_res_2[2, 1], "Std. Error" = civ_res_2[2, 2], ``` Since correct classification requires very few observations per -instrument, CIV is nearly identical to the infeasible orcle estimator +instrument, CIV is nearly identical to the infeasible oracle estimator that presumes knowledge of the low-dimensional optimal instrument Z0: ``` r @@ -222,8 +213,7 @@ forests in the first stage. Key takeaways from the results: TSLS, post-Lasso IV, and random forest-based IV are heavily biased. -For extensive finite-sample comparisons, see Wiemann (2023; -[arxiv:123](https://)). +For extensive finite-sample comparisons, see Wiemann (2023). ### TSLS @@ -283,8 +273,9 @@ c(Estimate = ranger_res[2, 1], "Std. Error" = ranger_res[2, 2], # References -Wiemann T (2023). “Optimal Categorical Instruments.” - Belloni A, Chen D, Chernozhukov V, Hansen C (2018). “Sparse Models and Methods for Optimal Instruments With an Application to Eminent Domain.” Econometrica, 80(6), 2369-2429. + +Wiemann T (2023). “Optimal Categorical Instruments.” + diff --git a/vignettes/civ.Rmd b/vignettes/civ.Rmd index 12401d4..4bcbb4a 100644 --- a/vignettes/civ.Rmd +++ b/vignettes/civ.Rmd @@ -88,4 +88,4 @@ t(head(civ_fit$kcmeans_fit$cluster_map[, c(1, 4)], 10)) ``` # References -Wiemann T (2023). "Optimal Categorical Instruments." +Wiemann T (2023). "Optimal Categorical Instruments." https://arxiv.org/abs/2311.17021 diff --git a/vignettes/civ.Rmd.txt b/vignettes/civ.Rmd.txt index e1cd30b..8463248 100644 --- a/vignettes/civ.Rmd.txt +++ b/vignettes/civ.Rmd.txt @@ -67,4 +67,4 @@ t(head(civ_fit$kcmeans_fit$cluster_map[, c(1, 4)], 10)) ``` # References -Wiemann T (2023). "Optimal Categorical Instruments." +Wiemann T (2023). "Optimal Categorical Instruments." https://arxiv.org/abs/2311.17021