Dear all,
I wonder if it is necessary to use the vec(bootstrap) option after ivpoisson cfunction command. I am aware of the fact that we must correct the variance from the first-stage as described by Wooldridge (2010, section 18.5, pg 743). However, the ivpoisson help file (https://www.stata.com/manuals/rivpoisson.pdf) give us the following:
"Control-function (CF) estimators can be used to account for endogenous covariates. As Wooldridge (2010, sec. 18.5) describes, CF estimators assume a certain structural relationship between the endogenous covariates and the exogenous covariates and use functions of first-stage parameter estimates to control for the endogeneity in the second stage. Wooldridge (2010, sec. 18.5) notes that the VCE of the second-stage estimator must be adjusted to account for estimates from the first stage. ivpoisson cfunction solves this problem by stacking the moment conditions that define each stage and applying a single GMM estimator."
So, is bootstrapping really necessary when using ivpoisson cfunction ?
Thank you in advance!
Kind regards,
Mauricio.
I wonder if it is necessary to use the vec(bootstrap) option after ivpoisson cfunction command. I am aware of the fact that we must correct the variance from the first-stage as described by Wooldridge (2010, section 18.5, pg 743). However, the ivpoisson help file (https://www.stata.com/manuals/rivpoisson.pdf) give us the following:
"Control-function (CF) estimators can be used to account for endogenous covariates. As Wooldridge (2010, sec. 18.5) describes, CF estimators assume a certain structural relationship between the endogenous covariates and the exogenous covariates and use functions of first-stage parameter estimates to control for the endogeneity in the second stage. Wooldridge (2010, sec. 18.5) notes that the VCE of the second-stage estimator must be adjusted to account for estimates from the first stage. ivpoisson cfunction solves this problem by stacking the moment conditions that define each stage and applying a single GMM estimator."
So, is bootstrapping really necessary when using ivpoisson cfunction ?
Thank you in advance!
Kind regards,
Mauricio.
Comment