| asymptotic_var | R Documentation |
Computes the asymptotic bias and variance-covariance matrix of \hat\beta
under Theorem 2.3 of Tugan (2021). These quantities are used by
irf_bands to construct parametric confidence bands.
asymptotic_var(fit)
fit |
An object of class |
The function computes three components:
The Hessian D_{F,\Lambda} (Eq. A.4), which accounts
for factor estimation uncertainty via a two-term formula.
The sandwich variance \Omega, accumulated
unit by unit.
Two bias terms: B_\Psi (from factor loading
estimation) and B_\gamma (HAC serial correction with bandwidth
\bar G = \lfloor T^{1/3} \rceil).
Notes on MATLAB replication: This implementation deviates from the
original Asymptotic_Distribution_of_beta.m in two places, following
the paper rather than the code:
B_\gamma: the MATLAB accumulation (line 189) uses only the
final value of the loop variable g rather than summing over
g = 1, \ldots, \bar G as in Eq. (2.56). Corrected here.
\Omega: MATLAB uses
\Gamma\,\mathrm{diag}(u)^2\,\Gamma^\top, which drops the
within-period cross-variable terms u_{t,n} u_{t,m} present in
Eq. (2.65). This function computes the per-period outer products of
Eq. (2.65); in simulations this gives (weakly) better
confidence-interval coverage.
A list with:
Bias vector for \hat\beta, of the same length as
fit$beta.
Asymptotic variance-covariance matrix of \hat\beta.
Tugan, M. (2021). Panel VAR models with interactive fixed effects. Econometrics Journal, 24, 225–246. \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1093/ectj/utaa021")}
sim <- sim_pvarife(n_units = 30, n_time = 20, n_vars = 2,
n_lags = 1, n_factors = 1, seed = 1)
fit <- pvarife(sim$y, n_lags = 1, n_factors = 1, n_out = 5, n_in = 3)
avar <- asymptotic_var(fit)
cat("Bias:", avar$bias, "\n")
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.