I'm currently estimating local projection regressions and a commonly known issue is needing to correct for autocorrelated residuals. Hence, I'm using the Newey-West HAC estimator for standard errors.
For another part of this same project, I will need to use bootstrapped standard errors, and I'm currently testing the wild dependent bootstrap Davidson and Monticini (2014).
Specifically, their paper proposes the wild standard bootstrap technique as the bootstrap version of the Newey-West estimator (just as the wild bootstrap is the bootstrap equivalent of the Eicker-White method). My question is that in finite sample, should we expect the wild bootstrap SEs to converge to (or at least approximately agree with) the Newey-West SEs as the number of bootstrap replications approaches infinity? Assuming the same size is fairly large (> 500).
