Scribd
Upload
13 views

EViews Practical 2 Answers

The document provides answers to practical exercises from an EViews course. It includes: 1) Estimating an energy demand function and interpreting the estimated coefficients. Higher GDP increases energy demand by 0.998% and higher energy prices decrease demand by 0.333%. 2) Using dummy variables to estimate the effect of education on salary in a regression. College increases salary by $3,144 and graduate degrees by $2,996 compared to high school. The regression explains 95.25% of salary variation. 3) Testing all coefficients for statistical significance and the overall regression, finding significance in all cases. Heteroscedasticity is also tested for and corrected.

Uploaded by

k7dgxxhff9
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
13 views

EViews Practical 2 Answers

The document provides answers to practical exercises from an EViews course. It includes: 1) Estimating an energy demand function and interpreting the estimated coefficients. Higher GDP increases energy demand by 0.998% and higher energy prices decrease demand by 0.333%. 2) Using dummy variables to estimate the effect of education on salary in a regression. College increases salary by $3,144 and graduate degrees by $2,996 compared to high school. The regression explains 95.25% of salary variation. 3) Testing all coefficients for statistical significance and the overall regression, finding significance in all cases. Heteroscedasticity is also tested for and corrected.

Uploaded by

k7dgxxhff9
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

EViews Practical 2 2023: Answers

Practical exercises:

1. Estimate the energy demand function on p143 in log format – use EViews file Table 5_3.
 Write down the estimated equation
 Interpret estimated coefficients
 Evaluate the estimated regression
o a priori
o 1st order (individual coefficients as well as overall regression)
Dependent Variable: LOG(ENERGYDEMAND)
Method: Least Squares
Date: 09/20/16 Time: 10:24
Sample: 1960 1982
Included observations: 23

Variable Coefficient Std. Error t-Statistic Prob.

C 1.554555 0.089630 17.34419 0.0000


LOG(REALGDP) 0.997797 0.019008 52.49368 0.0000
LOG(ENERGYPRICE) -0.333169 0.024179 -13.77902 0.0000

R-squared 0.994196 Mean dependent var 4.412381


Adjusted R-squared 0.993615 S.D. dependent var 0.224157
S.E. of regression 0.017911 Akaike info criterion -5.085675
Sum squared resid 0.006416 Schwarz criterion -4.937567
Log likelihood 61.48526 Hannan-Quinn criter. -5.048426
F-statistic 1712.859 Durbin-Watson stat 0.805202
Prob(F-statistic) 0.000000

log(energydemand) = b0 + b1log(realgdp) + b2log(energyprice)

log(energydemand) = 1.55 + 0.998log(realgdp) – 0.333 log(energyprice)

for every 1% increase in real GDP the demand for energy increases with 0.998% - holding
everything else constant (almost unit elasticity)

for every 1% increase in the price of energy the demand for energy decreases with 0.333% -
holding everything else constant

Both of the estimated coefficients display the correct expected sign, expect higher demand
when production / income increases and expect lower demand when prices increase

H0: b1 = 0; H1: b1 ≠ 0; and H0: b2 = 0; H1: b2 ≠ 0; probability of both estimated coefficients is very
low (0.000%); reject Ho and conclude that estimated b1 and b2 are both statistically significant.

R2 : 0.9936, thus 99.36% of the variation in energy demand is explained by the regression

H0: b1 = b2 = 0; H1: b1 ≠ b2 ≠ 0; probability of calculated F-statistic is very low (0.000%); reject Ho


and conclude that estimated b1 and b2 do differ from 0 and that the regression overall is
statistically significant.

*********************

2. Do Problem 6.27 on p.214 of G&P (Use Excel data: Table 6_15)


a) Look at the data, specifically the Education variable. How would you interpret the estimated
coefficient of Education?

Since the values do not represent years, but categories of schooling, it will be impossible to
interpret the coefficient (if Education is included in its current format.
b) Consider the use of dummy variables to replace the Education variable. How many
dummies do you need?

With 3 categories, we need two dummy variables. In Table 6_15A high school education is
used as the base – with dummies then for college and graduate school.

c) Create an EViews file with the adjusted data in Table 6_15A.

d) Regress Salary on Experience, Management and the Education dummies.


 Write down the estimated equation
 Interpret estimated coefficients
 Evaluate the estimated regression
o a priori
o 1st order (individual coefficients as well as overall regression
o Now also test for heteroscedasticity (White) and correct for it – regardless of
whether it is present in the original regression or not.

SALARY = b0 + b1EXPERIENCE + b2MANAGMENT + b3DUMCOLL + b4DUMGRAD

SALARY = 8036 + 546EXPERIENCE + 6884MANAGMENT + 3144DUMCOLL + 2996DUMGRAD

Dependent Variable: SALARY


Method: Least Squares
Date: 09/23/16 Time: 14:52
Sample: 1 46
Included observations: 46

Variable Coefficient Std. Error t-Statistic Prob.

C 8035.598 386.6893 20.78050 0.0000


EXPERIENCE 546.1840 30.51919 17.89641 0.0000
MANAGEMENT 6883.531 313.9190 21.92773 0.0000
DUMCOLL 3144.035 361.9683 8.685941 0.0000
DUMGRAD 2996.210 411.7527 7.276723 0.0000

R-squared 0.956767 Mean dependent var 17270.20


Adjusted R-squared 0.952549 S.D. dependent var 4716.632
S.E. of regression 1027.437 Akaike info criterion 16.80984
Sum squared resid 43280719 Schwarz criterion 17.00861
Log likelihood -381.6264 Hannan-Quinn criter. 16.88430
F-statistic 226.8359 Durbin-Watson stat 2.236925
Prob(F-statistic) 0.000000

for every 1 year of additional experience, the salary of employees increase with 546 US$ -
holding everything else constant

Managers receive 6884 US$ more than non-managers - holding everything else constant

Employees with college education receive 3144 US$ more than employees with only high-
school education - holding everything else constant
Employees with graduate school education receive 2996 US$ more than employees with only
high-school education - holding everything else constant

Positive coefficients of EXPERIENCE and MANAGEMENT are expected

Positive coefficients of two dummies are expected – any additional qualification after school
should raise your income

All the probabilities are 0.000% - therefore al the estimated coefficients are statistically significant

R2 : 0.9525, thus 95.25% of the variation in salaries is explained by the regression

H0: b1 = b2 = b3 = b4 =0; H1: b1 ≠ b2 ≠ b3 ≠ b4 ≠ 0; probability of calculated F-statistic is very low


(0.000%); reject Ho and conclude that estimated b1 ,b2 , b3 and b4 do differ from 0 and that the
regression overall is statistically significant.

You might also like