You are on page 1of 6

University of Tartu

Econometric Analysis
2017-18
Exercises
Lecturer: M. Hakan Eratalay

24/11/2017

Question 1.
Suppose that we have random variables z, y and x such that:

E(z) =? V ar(z) =?
E(y) =? V ar(y) = 50
E(zy) = 1500
x=z y
E(x) = 5 V ar(x) =?
E(x2 ) = 80 Cov(z; y) = 1000

Fill in the missing points marked with "?". Also …nd LP (yj1; z):

Question 2.
Here is the Stata regression output of a multiple linear regression estimated by OLS. As
you can see some results are missing and they are marked with "?". Fill in these missing
points. Show your calculations explicitly.

1
a) Are the coe¢ cients signi…cant at 5%? At 1%?
b) Are the coe¢ cients of the explanatory variables jointly signi…cant at 5%? At 1%?
State the null hypothesis in this case.
c) What percentage of the total variation in y is explained by these explanatory variables?

Question 3.
Consider the multiple regression model:

y= 0 + 1 x1 + 2 x2 + 3 x3 +u

We want to test the null hypothesis that H0 : 1 0:5 3 = 1:


a) Let ^ 1 and ^ 3 be the OLS estimators of 1 and 3 . Write V ar( ^ 1 0:5 ^ 3 ) in terms
of the variances and covariances of ^ 1 and ^ 3 .
b) Write the t statistic for testing H0 = 1 0:5 3 = 1: Specify clearly what is in the
numerator and denominator of the t-statistic.
c) De…ne a new parameter 1 = 1 0:5 3 such that ^1 = ^ 1 0:5 ^ 3 . Rewrite the model
in a new form which has coe¢ cients 0 ; 1 ; 2 and 3 that would allow you to directly obtain
^1 and its standard error.

Question 4.
Consider the following model.
2
log(wage) = 0 + 1 educ + 2 educ:meduc + 3 exper + 4 exper + 5 tenure (1)
6 married + 7 married:educ + 8 IQ + 9 IQ:educ +u (2)

where log(wage) is logarithm of wages, educ is years of education, meduc is years of


mothers’education, exper is years of experience and tenure is the years of experience in the
current working place.

2
a) What is the partial e¤ect of educ? What is the partial e¤ect of exper ? What is the
optimal level of exper ? Write the answers in terms of parameters.
b) Based on the table of correlations, do we expect multicollinearity problem? Comment
on the signi…cance of the variables and the overall signi…cance of the equation.
**For the rest of the question, we use the data WAGE2.RAW from Wooldridge data
sources. Attached are STATA estimation outputs for di¤erent equations. Answer the rest
of the parts using these outputs. NOTE!: every STATA estimation output belongs to a
certain model, you should be careful about which one you are using. You should state which
estimation results you use.**
c) What is the partial e¤ect of educ evaluated at the sample average of the variables in
the estimation results of equation (1)? Test if this partial e¤ect is signi…cant.
d) What is the partial e¤ect of exper evaluated at the sample average of the variables in
the estimation results of equation (1)? Test if this partial e¤ect is signi…cant.
e) What is the partial e¤ect of being married evaluated at the sample average of the
variables in the estimation results of equation (1)? Test if this partial e¤ect is signi…cant.
f) Test the null hypothesis that partial e¤ect of IQ on log(wage) is zero for any level of
educ:
g) What is the estimated value of optimum level of experience in equation (1)?

Stata outputs for Question 4.


Table 1
. correlate educ educmeduc exper exper2 tenure married marriededuc IQ IQeduc
(obs=857)

educ educme~c exper exper2 tenure married marrie~c IQ IQeduc

educ 1.0000
educmeduc 0.7252 1.0000
exper -0.4544 -0.3381 1.0000
exper2 -0.4626 -0.3484 0.9761 1.0000
tenure -0.0361 -0.0257 0.2769 0.2948 1.0000
married -0.0551 -0.0413 0.1095 0.1183 0.0807 1.0000
marriededuc 0.3676 0.2658 -0.0947 -0.0910 0.0476 0.8967 1.0000
IQ 0.5171 0.4700 -0.2250 -0.2271 0.0343 -0.0126 0.1999 1.0000
IQeduc 0.8890 0.6937 -0.3932 -0.3992 -0.0016 -0.0430 0.3299 0.8452 1.0000

Table 2

3
. regress lwage educ educmeduc exper exper2 tenure married marriededuc IQ IQeduc

Source SS df MS Number of obs = 857


F(9, 847) = 25.21
Model 31.5615063 9 3.50683403 Prob > F = 0.0000
Residual 117.799534 847 .139078553 R-squared = 0.2113
Adj R-squared = 0.2029
Total 149.36104 856 .174487197 Root MSE = .37293

lwage Coef. Std. Err. t P>|t| [95% Conf. Interval]

educ .0235397 .0472254 0.50 0.618 -.0691528 .1162323


educmeduc .0007537 .0003556 2.12 0.034 .0000557 .0014517
exper .0185516 .0138155 1.34 0.180 -.008565 .0456682
exper2 -.0000682 .0005893 -0.12 0.908 -.0012249 .0010884
tenure .0096936 .0026914 3.60 0.000 .0044109 .0149763
married -.040839 .2623614 -0.16 0.876 -.5557936 .4741157
marriededuc .0163588 .0186315 0.88 0.380 -.0202106 .0529282
IQ .0043411 .0054994 0.79 0.430 -.0064529 .0151352
IQeduc .0000714 .0004096 0.17 0.862 -.0007327 .0008754
_cons 5.387566 .6257581 8.61 0.000 4.159347 6.615784

Table 3, where educdmeduc = educ (meduc meduc); dmarriededuc = (married


married) educ; dIQeduc = (IQ IQ) educ
. regress lwage educ educdmeduc exper exper2 tenure married dmarriededuc IQ dIQeduc

Source SS df MS Number of obs = 857


F(9, 847) = 25.21
Model 31.5615063 9 3.50683403 Prob > F = 0.0000
Residual 117.799534 847 .139078553 R-squared = 0.2113
Adj R-squared = 0.2029
Total 149.36104 856 .174487197 Root MSE = .37293

lwage Coef. Std. Err. t P>|t| [95% Conf. Interval]

educ .0533751 .0079379 6.72 0.000 .0377949 .0689554


educdmeduc .0007537 .0003556 2.12 0.034 .0000557 .0014517
exper .0185516 .0138155 1.34 0.180 -.008565 .0456682
exper2 -.0000682 .0005893 -0.12 0.908 -.0012249 .0010884
tenure .0096936 .0026914 3.60 0.000 .0044109 .0149763
married -.040839 .2623614 -0.16 0.876 -.5557937 .4741156
dmarriededuc .0163588 .0186315 0.88 0.380 -.0202106 .0529282
IQ .0043411 .0054994 0.79 0.430 -.0064529 .0151352
dIQeduc .0000714 .0004096 0.17 0.862 -.0007327 .0008754
_cons 5.387566 .6257581 8.61 0.000 4.159348 6.615784

4
Table 4, where dexper2 = (exper exper )2
. regress lwage educ educmeduc exper dexper2 tenure married marriededuc IQ IQeduc

Source SS df MS Number of obs = 857


F(9, 847) = 25.21
Model 31.5615063 9 3.50683403 Prob > F = 0.0000
Residual 117.799534 847 .139078553 R-squared = 0.2113
Adj R-squared = 0.2029
Total 149.36104 856 .174487197 Root MSE = .37293

lwage Coef. Std. Err. t P>|t| [95% Conf. Interval]

educ .0235397 .0472254 0.50 0.618 -.0691528 .1162323


educmeduc .0007537 .0003556 2.12 0.034 .0000557 .0014517
exper .016974 .0035468 4.79 0.000 .0100125 .0239355
dexper2 -.0000682 .0005893 -0.12 0.908 -.0012249 .0010884
tenure .0096936 .0026914 3.60 0.000 .0044109 .0149763
married -.040839 .2623614 -0.16 0.876 -.5557936 .4741157
marriededuc .0163588 .0186315 0.88 0.380 -.0202106 .0529282
IQ .0043411 .0054994 0.79 0.430 -.0064529 .0151352
IQeduc .0000714 .0004096 0.17 0.862 -.0007327 .0008754
_cons 5.396685 .6233544 8.66 0.000 4.173184 6.620185

Table 5, where deduc = educ educ; marrieddeduc = married deduc:


. regress lwage educ educmeduc exper exper2 tenure married marrieddeduc IQ IQeduc

Source SS df MS Number of obs = 857


F(9, 847) = 25.21
Model 31.5615063 9 3.50683403 Prob > F = 0.0000
Residual 117.799534 847 .139078553 R-squared = 0.2113
Adj R-squared = 0.2029
Total 149.36104 856 .174487197 Root MSE = .37293

lwage Coef. Std. Err. t P>|t| [95% Conf. Interval]

educ .0235398 .0472254 0.50 0.618 -.0691528 .1162323


educmeduc .0007537 .0003556 2.12 0.034 .0000557 .0014517
exper .0185516 .0138155 1.34 0.180 -.008565 .0456682
exper2 -.0000682 .0005893 -0.12 0.908 -.0012249 .0010884
tenure .0096936 .0026914 3.60 0.000 .0044109 .0149763
married .1795139 .0423196 4.24 0.000 .0964502 .2625776
marrieddeduc .0163588 .0186315 0.88 0.380 -.0202106 .0529282
IQ .0043411 .0054994 0.79 0.430 -.0064529 .0151352
IQeduc .0000714 .0004096 0.17 0.862 -.0007327 .0008754
_cons 5.387566 .6257581 8.61 0.000 4.159347 6.615784

5
Table 6
. regress lwage educ educmeduc exper exper2 tenure married marriededuc

Source SS df MS Number of obs = 857


F(7, 849) = 27.66
Model 27.7375918 7 3.96251311 Prob > F = 0.0000
Residual 121.623449 849 .143254945 R-squared = 0.1857
Adj R-squared = 0.1790
Total 149.36104 856 .174487197 Root MSE = .37849

lwage Coef. Std. Err. t P>|t| [95% Conf. Interval]

educ .0475315 .0189945 2.50 0.013 .0102497 .0848133


educmeduc .0010545 .0003561 2.96 0.003 .0003555 .0017534
exper .0178488 .0140207 1.27 0.203 -.0096704 .045368
exper2 -.0000391 .000598 -0.07 0.948 -.0012127 .0011346
tenure .0104853 .0027271 3.84 0.000 .0051326 .015838
married .0079402 .2660011 0.03 0.976 -.5141567 .5300372
marriededuc .0130503 .0188938 0.69 0.490 -.0240337 .0501344
_cons 5.553735 .2702858 20.55 0.000 5.023228 6.084241

You might also like