Limited-Information Maximum Likelihood (LIML)

Solves the LIML eigenvalue problem natively (k-class with the minimum-eigenvalue kappa).

Usage

morie_iv_liml(
  data,
  outcome,
  endogenous,
  instruments,
  exogenous = NULL,
  robust = TRUE,
  alpha = 0.05
)

Arguments

data: Data frame.
outcome: Name of the outcome column.
endogenous: Character vector of endogenous regressor names.
instruments: Character vector of excluded-instrument names.
exogenous: Optional character vector of exogenous covariate names.
robust: Logical; if TRUE use HC1 robust standard errors.
alpha: Significance level for confidence intervals.

Value

A named list with elements coefficients, std_errors, t_stats, p_values, ci_lower, ci_upper, variable_names, n_obs, method, details.

Examples

set.seed(8)
n <- 1000
z <- rbinom(n, 1, 0.5); u <- rnorm(n)
d <- 0.2 + 0.6 * z + 0.4 * u + 0.3 * rnorm(n)
y <- 0.5 + 1.5 * d + u + rnorm(n)
df <- data.frame(y, d, z)
res <- morie_iv_liml(df, outcome = "y", endogenous = "d", instruments = "z")
res$coefficients["d"]
#>        d 
#> 1.564552