Skip to content

Fix crash in IV estimation when demeaning does not converge #639

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
adamaltmejd wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Fix crash in IV estimation when demeaning does not converge #639

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 21 additions & 19 deletions R/estimation.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1349,32 +1349,33 @@ feols = function(fml, data, vcov, weights, offset, subset, split, fsplit, split.
res_first_stage[[iv_lhs_names[i]]] = my_res

# Checking errors
error_endo_no_variation = my_res$ssr < 1e-10
error_inst_no_expl = abs(my_res$ssr - my_res$ssr_no_inst) < 1e-10
# NaN ssr arises from demeaning non-convergence; isTRUE guards against NA in the if()
error_endo_no_variation = isTRUE(my_res$ssr < 1e-10)
error_inst_no_expl = isTRUE(abs(my_res$ssr - my_res$ssr_no_inst) < 1e-10)
if(error_endo_no_variation || error_inst_no_expl){
Comment on lines +1352 to 1355
# we keep the information on this first stage => useful for reporting

if(error_endo_no_variation){
msg = sma("[IV error] The endogenous variable {bq ? iv_lhs_names[i]} is ",
"fully explained by the exogenous variables and the instruments. ",
"Please revise the model.")

} else if(error_inst_no_expl){
msg = sma("[IV error] The instruments have 0 explanatory power ",
msg = sma("[IV error] The instruments have 0 explanatory power ",
"for the endogenous variable {bq ? iv_lhs_names[i]}. ",
"Please revise the model.")
}


if(IN_MULTI){
stack_multi_notes(msg)
return(fixest_NA_results_IV(env, res_first_stage, msg))

} else {
mema("[IV error] Problematic 1st stage results:")
my_res$is_iv = FALSE
print(my_res)

stopi("{msg}\nFor information, above are reported the results of the 1st stage estimation.")
}

Expand Down Expand Up @@ -1513,33 +1514,34 @@ feols = function(fml, data, vcov, weights, offset, subset, split, fsplit, split.
res_first_stage[[iv_lhs_names[i]]] = my_res

# Checking errors
error_endo_no_variation = my_res$ssr < 1e-10
error_inst_no_expl = abs(my_res$ssr - my_res$ssr_no_inst) < 1e-10
# NaN ssr arises from demeaning non-convergence; isTRUE guards against NA in the if()
error_endo_no_variation = isTRUE(my_res$ssr < 1e-10)
error_inst_no_expl = isTRUE(abs(my_res$ssr - my_res$ssr_no_inst) < 1e-10)
if(error_endo_no_variation || error_inst_no_expl){
# we keep the information on this first stage => useful for reporting

if(error_endo_no_variation){
msg = sma("[IV error] The endogenous variable {bq ? iv_lhs_names[i]} is ",
"fully explained by the exogenous variables and the instruments. ",
"Please revise the model.")

} else if(error_inst_no_expl){
msg = sma("[IV error] The instruments have 0 explanatory power ",
msg = sma("[IV error] The instruments have 0 explanatory power ",
"for the endogenous variable {bq ? iv_lhs_names[i]}. ",
"Please revise the model.")
}


if(IN_MULTI){
stack_multi_notes(msg)
return(fixest_NA_results_IV(env, res_first_stage, msg))

} else {

mema("[IV error] Problematic 1st stage results:")
my_res$is_iv = FALSE
print(my_res)

stopi("{msg}\nFor information, above are reported the results of the 1st stage estimation.")
}

Expand Down
Loading