Example-16-5-3sls.R
# 3SLS of whole system (run Example-16-5-systemfit-prep.R first!)
summary(systemfit(eq.system,inst=instrum,data=oursample,method="3SLS"))Chapter 16
Example-16-5-ivreg.R
library(AER)
data(mroz, package='wooldridge')
oursample <- subset(mroz,!is.na(wage))
# 2SLS regressions
summary( ivreg(hours~log(wage)+educ+age+kidslt6+nwifeinc
|educ+age+kidslt6+nwifeinc+exper+I(exper^2), data=oursample))
summary( ivreg(log(wage)~hours+educ+exper+I(exper^2)
|educ+age+kidslt6+nwifeinc+exper+I(exper^2), data=oursample))Example-16-5-systemfit-prep.R
library(systemfit)
data(mroz, package='wooldridge')
oursample <- subset(mroz,!is.na(wage))
# Define system of equations and instruments
eq.hrs <- hours ~ log(wage)+educ+age+kidslt6+nwifeinc
eq.wage <- log(wage)~ hours +educ+exper+I(exper^2)
eq.system<- list(eq.hrs, eq.wage)
instrum <- ~educ+age+kidslt6+nwifeinc+exper+I(exper^2)Example-16-5-systemfit.R
# 2SLS of whole system (run Example-16-5-systemfit-prep.R first!)
summary(systemfit(eq.system,inst=instrum,data=oursample,method="2SLS"))Example-16-5-2SLS.py
import wooldridge as woo
import numpy as np
import pandas as pd
import linearmodels.iv as iv
mroz = woo.dataWoo('mroz')
# restrict to non-missing wage observations:
mroz = mroz.dropna(subset=['lwage'])
# 2SLS regressions:
reg_iv1 = iv.IV2SLS.from_formula(
'hours ~ 1 + educ + age + kidslt6 + nwifeinc +'
'[np.log(wage) ~ exper + I(exper**2)]', data=mroz)
results_iv1 = reg_iv1.fit(cov_type='unadjusted', debiased=True)
reg_iv2 = iv.IV2SLS.from_formula(
'np.log(wage) ~ 1 + educ + exper + I(exper**2) +'
'[hours ~ age + kidslt6 + nwifeinc]', data=mroz)
results_iv2 = reg_iv2.fit(cov_type='unadjusted', debiased=True)
# print results:
table_iv1 = pd.DataFrame({'b': round(results_iv1.params, 4),
'se': round(results_iv1.std_errors, 4),
't': round(results_iv1.tstats, 4),
'pval': round(results_iv1.pvalues, 4)})
print(f'table_iv1: \n{table_iv1}\n')
table_iv2 = pd.DataFrame({'b': round(results_iv2.params, 4),
'se': round(results_iv2.std_errors, 4),
't': round(results_iv2.tstats, 4),
'pval': round(results_iv2.pvalues, 4)})
print(f'table_iv2: \n{table_iv2}\n')
cor_u1u2 = np.corrcoef(results_iv1.resids, results_iv2.resids)[0, 1]
print(f'cor_u1u2: {cor_u1u2}\n')Example-16-5-3SLS.py
import wooldridge as woo
import numpy as np
import linearmodels.system as iv3
mroz = woo.dataWoo('mroz')
# restrict to non-missing wage observations:
mroz = mroz.dropna(subset=['lwage'])
# 3SLS regressions:
formula = {'eq1': 'hours ~ 1 + educ + age + kidslt6 + nwifeinc +'
'[np.log(wage) ~ exper+I(exper**2)]',
'eq2': 'np.log(wage) ~ 1 + educ + exper + I(exper**2) +'
'[hours ~ age + kidslt6 + nwifeinc]'}
reg_3sls = iv3.IV3SLS.from_formula(formula, data=mroz)
results_3sls = reg_3sls.fit(cov_type='unadjusted', debiased=True)
print(f'results_3sls: \n{results_3sls}\n')Example-16-5-2SLS-1.jl
using WooldridgeDatasets, GLM, DataFrames, Econometrics, Statistics
mroz_wm = DataFrame(wooldridge("mroz"))
# restrict to non-missing wage observations:
mroz = mroz_wm[.!ismissing.(mroz_wm.wage), :]
# 2SLS regressions:
reg_iv1 = fit(EconometricModel,
@formula(hours ~ educ + age + kidslt6 + nwifeinc +
(log(wage) ~ exper + (exper^2))), mroz)
table_iv1 = coeftable(reg_iv1)
println("table_iv1: \n$table_iv1\n")
reg_iv2 = fit(EconometricModel,
@formula(log(wage) ~ educ + exper + (exper^2) +
(hours ~ age + kidslt6 + nwifeinc)), mroz)
table_iv2 = coeftable(reg_iv2)
println("table_iv2: \n$table_iv2\n")
cor_u1u2 = cor(residuals(reg_iv1), residuals(reg_iv2))
println("cor_u1u2 =$cor_u1u2")Example-16-5-2SLS-2.jl
using WooldridgeDatasets, GLM, DataFrames, PyCall
include("../03/getMats.jl")
# install Python's linearmodels with: using Conda; Conda.add("linearmodels")
iv = pyimport("linearmodels.iv")
mroz_wm = DataFrame(wooldridge("mroz"))
# restrict to non-missing wage observations:
mroz = mroz_wm[.!ismissing.(mroz_wm.wage), :]
# prepare for equation 1:
f1 = @formula(hours ~ 1 + educ + age + kidslt6 + nwifeinc)
yexog = getMats(f1, mroz)
y_eq1 = yexog[1]
exog_mat_eq1 = yexog[2]
f2 = @formula(1 ~ 0 + log(wage))
endo_mat_eq1 = getMats(f2, mroz)[2]
f3 = @formula(1 ~ 0 + exper + (exper^2))
iv_mat_eq1 = getMats(f3, mroz)[2]
# prepare for equation 2:
f1 = @formula(log(wage) ~ 1 + educ + exper + (exper^2))
yexog = getMats(f1, mroz)
y_eq2 = yexog[1]
exog_mat_eq2 = yexog[2]
f2 = @formula(1 ~ 0 + hours)
endo_mat_eq2 = getMats(f2, mroz)[2]
f3 = @formula(1 ~ 0 + age + kidslt6 + nwifeinc)
iv_mat_eq2 = getMats(f3, mroz)[2]
# use Python's linearmodels:
reg_iv1 = iv.IV2SLS(y_eq1, exog_mat_eq1, endo_mat_eq1, iv_mat_eq1)
results_iv1 = reg_iv1.fit(cov_type="unadjusted", debiased=true)
println("results_iv1: \n$results_iv1\n")
reg_iv2 = iv.IV2SLS(y_eq2, exog_mat_eq2, endo_mat_eq2, iv_mat_eq2)
results_iv2 = reg_iv2.fit(cov_type="unadjusted", debiased=true)
println("results_iv2: \n$results_iv2")Example-16-5-3SLS.jl
using WooldridgeDatasets, GLM, DataFrames, PyCall
include("../03/getMats.jl")
iv3 = pyimport("linearmodels.system")
mroz_wm = DataFrame(wooldridge("mroz"))
# restrict to non-missing wage observations:
mroz = mroz_wm[.!ismissing.(mroz_wm.wage), :]
# prepare for equation 1:
f1 = @formula(hours ~ 1 + educ + age + kidslt6 + nwifeinc)
yexog = getMats(f1, mroz)
y_eq1 = yexog[1]
exog_mat_eq1 = yexog[2]
f2 = @formula(1 ~ 0 + log(wage))
endo_mat_eq1 = getMats(f2, mroz)[2]
f3 = @formula(1 ~ 0 + exper + (exper^2))
iv_mat_eq1 = getMats(f3, mroz)[2]
# prepare for equation 2:
f1 = @formula(log(wage) ~ 1 + educ + exper + (exper^2))
yexog = getMats(f1, mroz)
y_eq2 = yexog[1]
exog_mat_eq2 = yexog[2]
f2 = @formula(1 ~ 0 + hours)
endo_mat_eq2 = getMats(f2, mroz)[2]
f3 = @formula(1 ~ 0 + age + kidslt6 + nwifeinc)
iv_mat_eq2 = getMats(f3, mroz)[2]
# use Python's linearmodels:
reg_3sls = iv3.IV3SLS(Dict([
("eq1", (y_eq1, exog_mat_eq1, endo_mat_eq1, iv_mat_eq1)),
("eq2", (y_eq2, exog_mat_eq2, endo_mat_eq2, iv_mat_eq2))]))
results_3sls = reg_3sls.fit(cov_type="unadjusted", debiased=true)
println("results_3sls: \n$results_3sls")