使用ivreg2时如何将第一阶段估计结果导出至LaTeX？

问题说明 #

使用ivreg2做2SLS估计时，需要在LaTeX输出中同时展示：

• 第二阶段内生变量foreign的2SLS系数
• 第一阶段工具变量mpgx的系数及标准误
• 第一阶段IV F统计量
  但原代码无法提取第一阶段的系数和标准误。

修正后的可复现代码 #

* 安装依赖包
ssc install estout, replace
ssc install ivreg2, replace

sysuse auto, clear
eststo clear

gen rep = (rep78 >= 4)
summ mpg
gen mpgx = (mpgx >= `r(mean)')
summ trunk
gen trunkx = (trunk >= `r(mean)')

* 分组回归并提取第一阶段结果
eststo: ivreg2 price weight (foreign = mpgx) if rep == 1, savefirst first
estadd scalar fiv = e(F)
tempname first1
estimates `first1' = e(first)
estadd scalar first_b_mpgx = _b[mpgx]
estadd scalar first_se_mpgx = _se[mpgx]

eststo: ivreg2 price weight (foreign = mpgx) if rep == 0, savefirst first
estadd scalar fiv = e(F)
tempname first2
estimates `first2' = e(first)
estadd scalar first_b_mpgx = _b[mpgx]
estadd scalar first_se_mpgx = _se[mpgx]

eststo: ivreg2 price weight (foreign = mpgx) if trunkx == 1, savefirst first
estadd scalar fiv = e(F)
tempname first3
estimates `first3' = e(first)
estadd scalar first_b_mpgx = _b[mpgx]
estadd scalar first_se_mpgx = _se[mpgx]

eststo: ivreg2 price weight (foreign = mpgx) if trunkx == 0, savefirst first
estadd scalar fiv = e(F)
tempname first4
estimates `first4' = e(first)
estadd scalar first_b_mpgx = _b[mpgx]
estadd scalar first_se_mpgx = _se[mpgx]

* 导出LaTeX表格
esttab using "iv_test.tex", replace   ///
    b(3) se(3) star(* 0.10 ** 0.05 *** 0.01) ///
    label booktabs nonotes nomtitle collabels(none)  ///
    keep(foreign) /// 仅保留第二阶段内生变量系数
    addlines( ///
        "First-stage mpgx coef." @first_b_mpgx @first_b_mpgx @first_b_mpgx @first_b_mpgx ///
        "First-stage mpgx se." @first_se_mpgx @first_se_mpgx @first_se_mpgx @first_se_mpgx ///
    ) ///
    scalars("fiv IV F-stat")  ///
    mgroups("Repair" "Trunk", pattern(1 0 1 0) ///
        prefix(\multicolumn{@span}{c}{) suffix(}) span erepeat(\cmidrule(lr){@span}))

关键改动说明 #

1. 提取第一阶段结果：利用ivreg2的savefirst选项将第一阶段估计结果存储在e(first)中，通过临时命名调用该结果，提取工具变量mpgx的系数和标准误，用estadd scalar将其加入当前估计对象。
2. LaTeX表格插入第一阶段行：使用esttab的addlines参数，将第一阶段的系数和标准误作为独立行插入表格，与第二阶段系数保持对齐格式。
3. 保留核心需求：继续用keep(foreign)仅展示第二阶段内生变量的2SLS系数，同时保留IV F统计量的输出。

内容的提问来源于stack exchange，提问作者aimalik