Scribd
Upload
46 views3 pages

X E X Z π π Z X X V E X Z E X Z E V Z Since E V Z Z are independent, E X Z E X Z E Y Z β β X

The document describes a two-step process to estimate the coefficient β1. First, regress the endogenous variable Xi on the instrumental variable Zi to obtain fitted values Xi*. Second, regress the outcome variable Yi on the fitted values Xi* to obtain an estimate of β1 that is unbiased by the endogeneity of Xi.

Uploaded by

Naren Playthatfunkymusic
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
46 views3 pages

X E X Z π π Z X X V E X Z E X Z E V Z Since E V Z Z are independent, E X Z E X Z E Y Z β β X

The document describes a two-step process to estimate the coefficient β1. First, regress the endogenous variable Xi on the instrumental variable Zi to obtain fitted values Xi*. Second, regress the outcome variable Yi on the fitted values Xi* to obtain an estimate of β1 that is unbiased by the endogeneity of Xi.

Uploaded by

Naren Playthatfunkymusic
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Estimation of 1 can be done in 2 steps:

1. Regress Xi* on Zi to obtain fitted values


^ ^ ( X i|Z i )= ^ 0 + ^ 1 Z i
X i= E
This result is a consequence of
X i= X i +V i

E ( X i|Z i )=E ( X i|Zi ) + E ( V i|Z i )

Since E ( V i|Zi ) =0 because V iZ i are independent , E ( X i|Z i )= E ( X i|Z i )


2. Regress Yi on XXI to obtain 1
E ( Y i|Z i ) = ^ 0+ ^1 ^
^ X1

Call:
lm(formula = log(wage) ~ educ + exper + I(exper^2) + IQ + black +
smsa, data = data1)

Residuals:
Min 1Q Median 3Q Max
-1.54043 -0.21708 0.02305 0.24191 1.36615

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 4.4327611 0.1044557 42.437 < 2e-16 ***
educ 0.0700292 0.0049672 14.098 < 2e-16 ***
exper 0.0951279 0.0097012 9.806 < 2e-16 ***
I(exper^2) -0.0027266 0.0005016 -5.436 6.11e-08 ***
IQ 0.0025287 0.0006817 3.709 0.000214 ***
black -0.1614085 0.0260646 -6.193 7.17e-10 ***
smsa 0.1629938 0.0190604 8.551 < 2e-16 ***
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1

Residual standard error: 0.3697 on 1993 degrees of freedom


Multiple R-squared: 0.2178, Adjusted R-squared: 0.2154
F-statistic: 92.49 on 6 and 1993 DF, p-value: < 2.2e-16
nearc2i = 1 if lived near a 2-year college in 1966, 0 otherwise
Correlated with educ and exogenous to ability

nearc4i = 1 if lived near a 4-year college in 1966, 0 otherwise


Correlated and exogenous

smsa66i = 1 if lived in a metropolitan area in 1966, 0 otherwise


Correlated cus metropolis got a lot of college with education. Exogenous
Uncorrelated cus suburb got colleges

momdad14i = 1 if lived with both parents at age 14, 0 otherwise


Correlated cus mom/dad make you study. Exogenous
Call:
lm(formula = educ ~ exper + I(exper^2) + IQ + black + smsa +
nearc2 + nearc4 + smsa66 + momdad14, data = data1)

Residuals:
Min 1Q Median 3Q Max
-4.6734 -1.1721 -0.1662 1.0241 6.0924

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 11.410385 0.398366 28.643 < 2e-16 ***
exper -0.468911 0.042259 -11.096 < 2e-16 ***
I(exper^2) 0.010015 0.002241 4.468 8.32e-06 ***
IQ 0.048908 0.002857 17.118 < 2e-16 ***
black 0.174456 0.119790 1.456 0.145453
smsa 0.150258 0.109622 1.371 0.170625
nearc2 0.082788 0.077368 1.070 0.284724
nearc4 0.295311 0.090295 3.270 0.001092 **
smsa66 -0.099730 0.109200 -0.913 0.361209
momdad14 0.358955 0.102653 3.497 0.000481 ***
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1

Residual standard error: 1.658 on 1990 degrees of freedom


Multiple R-squared: 0.4701, Adjusted R-squared: 0.4677
F-statistic: 196.1 on 9 and 1990 DF, p-value: < 2.2e-16
Call:
ivreg(formula = log(wage) ~ educ + exper + I(exper^2) + IQ +
black + smsa | exper + I(exper^2) + IQ + black + smsa + nearc2 +
nearc4 + smsa66 + momdad14, data = data1)

Residuals:
Min 1Q Median 3Q Max
-1.62027 -0.24702 0.02063 0.27968 1.25371

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.1196420 0.6028675 5.175 2.51e-07 ***
educ 0.1811357 0.0503506 3.597 0.000329 ***
exper 0.1480115 0.0261742 5.655 1.78e-08 ***
I(exper^2) -0.0038752 0.0007631 -5.078 4.17e-07 ***
IQ -0.0029904 0.0026002 -1.150 0.250245
black -0.1712082 0.0294855 -5.807 7.41e-09 ***
smsa 0.1411551 0.0234788 6.012 2.17e-09 ***
---
Signif. codes:
0 *** 0.001 ** 0.01 * 0.05 . 0.1 1

Residual standard error: 0.4135 on 1993 degrees of freedom


Multiple R-Squared: 0.02143, Adjusted R-squared: 0.01849
Wald test: 49.61 on 6 and 1993 DF, p-value: < 2.2e-16

You might also like