OLSvsIV_second_step: Testing interface to second_step_OLS_IV class.

Description Usage Arguments Details Value Examples

Description

Testing interface to second_step_OLS_IV class.

Usage

1
OLSvsIV_second_step(tau, pi_sq, size = 0.05, inc = 0.005, n_sim = 1000L)

Arguments

tau

Controls degree of endogeneity of OLS.

pi_sq

First-stage R-squared (strengh of instruments).

size

One minus the norminal coverage probability of the intervals.

inc

Step size for grid over which width is minimized.

n_sim

Number of simulation draws.

Details

This is a testing interface to the C++ class that is use, among other things, as the second-step in the two-step simulation based confidence interval construction from Section 4.3 of the paper, based on drawing from the limit experiment. In this implementation it treats all population parameters that enter the limit as know with the exception of tau. (That is, it abstracts from sampling uncertainty.)

Value

List with two elements: simulation-based two-sided confidence interval for the post-FMSC estimator evaluated at tau and the same for the moment average estimator.

Examples

1
as.data.frame(OLSvsIV_second_step(tau = 3, pi_sq = 0.1))

fditraglia/fmscr documentation built on May 16, 2019, 12:10 p.m.