Scribd
Upload
108 views3 pages

ECON 141 Problem Set 4

This document contains the results of four regression models. Model 1 regresses F on S and S^2 and finds that the coefficient on S^2 is equal to the coefficient when regressing the residuals of Model 1 on S^2 in Model 4. Model 2 regresses F on S only. Model 3 regresses S^2 on S. The rest of the document analyzes a regression of the log of expenditures (l_EXP) on the log of income (l_INC) to test for proportionality. It finds proportionality can be rejected both with standard errors and with heteroskedasticity-robust standard errors. A White's test shows heteroskedasticity is present.

Uploaded by

dnlmsgn
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
108 views3 pages

ECON 141 Problem Set 4

This document contains the results of four regression models. Model 1 regresses F on S and S^2 and finds that the coefficient on S^2 is equal to the coefficient when regressing the residuals of Model 1 on S^2 in Model 4. Model 2 regresses F on S only. Model 3 regresses S^2 on S. The rest of the document analyzes a regression of the log of expenditures (l_EXP) on the log of income (l_INC) to test for proportionality. It finds proportionality can be rejected both with standard errors and with heteroskedasticity-robust standard errors. A White's test shows heteroskedasticity is present.

Uploaded by

dnlmsgn
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 3

WONG, Wai Sing Wilson

Econ 141 Section 102


November 18, 2010
Problem Set 4
2.

Model 1: OLS, using observations 1-12


Dependent variable: F
const
S
sq_S
Mean dependent var
Sum squared resid
R-squared
F(2, 9)
Log-likelihood
Schwarz criterion

Coefficient
Std. Error
7.06818
3.01121
0.698626
0.1065
-0.00601149 0.000797501
19.91667
76.39735
0.883703
34.19408
-28.13351
63.72174

t-ratio
2.3473
6.5599
-7.5379

S.D. dependent var


S.E. of regression
Adjusted R-squared
P-value(F)
Akaike criterion
Hannan-Quinn

p-value
0.04350
0.00010
0.00004

**
***
***

7.727852
2.913519
0.857859
0.000062
62.26702
61.72843

Model 2: OLS, using observations 1-12


Dependent variable: F
const
S

Coefficient
25.303
-0.0828671

Std. Error
4.60039
0.062507

t-ratio
5.5002
-1.3257

p-value
0.00026
0.21442

***

Model 3: OLS, using observations 1-12


Dependent variable: sq_S
const
S

Coefficient
-3033.33
130

Std. Error
711.024
9.66092

t-ratio
-4.2661
13.4563

p-value
0.00165
<0.00001

***
***

p-value
<0.00001

***

Model 4: OLS, using observations 1-12


Dependent variable: res_F
res_sq_S
Mean dependent var
Sum squared resid
R-squared
F(1, 11)
Log-likelihood

Coefficient
Std. Error
-0.00601149 0.000721367
0.000000
76.39735
0.863263
69.44664
-28.13351

t-ratio
-8.3335

S.D. dependent var


S.E. of regression
Adjusted R-squared
P-value(F)
Akaike criterion

7.126896
2.635377
0.863263
4.42e-06
58.26702

WONG, Wai Sing Wilson


Econ 141 Section 102
November 18, 2010

Schwarz criterion

58.75193

Hannan-Quinn

58.08749

From the data above, we see that the coefficient for Si2 (sq_S) in Model 1 and the
coefficient for residuals for Si2 (res_sq_S) in Model 4 are equal (-0.00601149). The R2
for Model 1 is 0.883703 and the R2 for Model 4 is 0.863263.
3.

(a)
Model 1: OLS, using observations 1-235
Dependent variable: l_EXP
const
l_INC

Coefficient
0.571425
0.851879

Std. Error
0.137947
0.0202853

t-ratio
4.1424
41.9950

p-value
0.00005
<0.00001

***
***

t(233, 0.025) = 1.970


Variable
const
l_INC

Coefficient
0.571425
0.851879

95 confidence interval
(0.299642, 0.843208)
(0.811913, 0.891845)

Since the 95% confidence interval for the coefficient of l_INC (Xi) does not contain
the value 1, reject the null hypothesis of proportionality at the 5% level.
(b)

Model 2: OLS, using observations 1-235


Dependent variable: l_EXP
Heteroskedasticity-robust standard errors, variant HC1
const
l_INC

Coefficient
0.571425
0.851879

Std. Error
0.171659
0.0256094

t-ratio
3.3288
33.2643

p-value
0.00101
<0.00001

***
***

t(233, 0.025) = 1.970


Variable
const
l_INC

Coefficient
0.571425
0.851879

95 confidence interval
(0.233222, 0.909628)
(0.801423, 0.902334)

Since the 95% confidence interval for the coefficient of l_INC (Xi) does not contain
the value 1, reject the null hypothesis of proportionality at the 5% level with
heteroskedasticity robust standard errors.

(c)

WONG, Wai Sing Wilson


Econ 141 Section 102
November 18, 2010
White's test for heteroskedasticity
OLS, using observations 1-235
Dependent variable: uhat^2
coefficient
std. error
t-ratio
p-value
------------------------------------------------------const
0.844081
0.271418
3.110
0.0021 ***
l_INC
-0.254293
0.0788518
-3.225
0.0014 ***
sq_l_INC
0.0194633
0.00571420
3.406
0.0008 ***
Unadjusted R-squared = 0.101411
Test statistic: TR^2 = 23.831683,
with p-value = P(Chi-square(2) > 23.831683) = 0.000007

Since the p-value of Whites test is less than 0.05, reject null hypothesis of
homoskedastic error terms at the 5% significance level.

You might also like