Loading Packages
if (!require("pacman")) install.packages("pacman")
## Loading required package: pacman
pacman::p_load(tidyverse, hdm, stargazer, broom, knitr)
Linear Model
data("GrowthData")
y <- GrowthData$Outcome
d <- GrowthData$gdpsh465
x <- as.matrix(GrowthData %>% select(-c("Outcome", "gdpsh465", "intercept")))
varnames <- colnames(GrowthData)
xnames <- varnames[-c(1, 2, 3)] # names of X variables
dandxnames <- varnames[-c(1, 2)] # names of D and X variables
fmla <- as.formula(paste("Outcome ~ ", paste(dandxnames, collapse = "+")))
ols <- lm(fmla, GrowthData)
partial_out <- rlassoEffect(x, y, d, method = "partialling out")
double_selection <- rlassoEffect(x, y, d, method = "double selection")
bind_rows(
tidy(ols) %>% mutate(method = "full reg via ols") %>% filter(term == "gdpsh465") %>%
select(method, estimate, std.error),
tibble(
method = "via post-lasso",
estimate = summary(partial_out)[[1]][1,1],
std.error = summary(partial_out)[[1]][1,2]
),
tibble(
method = "partial reg via double selection",
estimate = summary(double_selection)[[1]][1,1],
std.error = summary(double_selection)[[1]][1,2]
)
) %>%
kable(digits = 3, format = "html")
|
method
|
estimate
|
std.error
|
|
full reg via ols
|
-0.009
|
0.030
|
|
via post-lasso
|
-0.050
|
0.014
|
|
partial reg via double selection
|
-0.050
|
0.016
|
IV Model
data("AJR")
y <- AJR$GDP
d <- AJR$Exprop
z <- AJR$logMort
x <- model.matrix(~-1 + (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)
lasso_IV <- rlassoIV(x, d, y, z, select.Z = FALSE, select.X = TRUE)
summary(lasso_IV)
## [1] "Estimation and significance testing of the effect of target variables in the IV regression model"
## coeff. se. t-value p-value
## d1 0.8450 0.2699 3.131 0.00174 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
confint(lasso_IV)
## 2.5 % 97.5 %
## d1 0.3159812 1.374072
rY <- lm(GDP ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
rD <- lm(Exprop ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
rZ <- lm(logMort ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
summary(tsls(rY ~ rD | rZ, intercept = FALSE))
## [1] "Estimates and Significance Testing from from tsls"
## Estimate Std. Error t value p value
## rD 1.267 1.731 0.732 0.464
rY <- rlasso(GDP ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
rD <- rlasso(Exprop ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
rZ <- rlasso(logMort ~ (Latitude + Latitude2 + Africa + Asia + Namer + Samer)^2, data = AJR)$res
summary(tsls(rY ~ rD | rZ, intercept = FALSE))
## [1] "Estimates and Significance Testing from from tsls"
## Estimate Std. Error t value p value
## rD 0.8450 0.2699 3.131 0.00174 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1