212ECN

MAY 2016
ROUGH ANSWERS

Coventry University

Faculty of Business, Environment and Society

212ECN

Introduction to Econometrics

Instructions to candidates

Time allowed: 3 hours

This is a closed book examination

Answer any 3 questions out of 4

All questions carry equal marks

Non programmable calculators may be used

Statistical tables are provided

You may take this exam paper away at the end of the examination:
please keep it in a safe place for future reference

Continued…
 212ECN

Question 1

Table I
Model 1: Logit, using observations 1-100
Dependent variable: PC
Standard errors based on Hessian

Coefficient Std. Error z Slope*

const 0.170395 0.383668 0.4441
SEX 0.199162 0.447693 0.4449 0.0485263
CAMPUS 1.15939 0.665621 1.7418 0.28131
WORK −0.984154 0.580572 -1.6951 −0.221213
LECTURES −0.483189 0.226262 -2.1355 −0.117857

Mean dependent var 0.430000 S.D. dependent var 0.497570

McFadden R-squared 0.091215 Adjusted R-squared 0.018042
Log-likelihood −62.09863 Akaike criterion 134.1973
Schwarz criterion 147.2231 Hannan-Quinn 139.4690
*
Evaluated at the mean
Number of cases 'correctly predicted' = 72 (72.0%)
f(beta'x) at mean of independent vars = 0.498
Likelihood ratio test: Chi-square(4) = 12.4657 [0.0142]

Predicted
0 1
Actual

0 49 8
1 20 23

Data represents a random selection of 100 students collected by Christopher

Lemmon in 1994.

Variables are:
PC = 1 if student owns personal computer, = 0 otherwise,
SEX = 1 if male, = 0 otherwise,
CAMPUS = 1 if live on campus, = 0 otherwise,
WORK = 1 if work at least 20 hours a week, = 0 otherwise,
LECTURES – average number of lectures missed per week.

Page 1 Continued…
 212ECN

Table II
Model 2: Logit, using observations 1-100
Dependent variable: PC
Standard errors based on Hessian
Coefficient Std. Error z Slope*
const −0.281851 0.201989 -1.3954

Mean dependent var 0.430000 S.D. dependent var 0.497570

McFadden R-squared 0.000000 Adjusted R-squared NA
Log-likelihood −68.33149 Akaike criterion 138.6630
Schwarz criterion 141.2682 Hannan-Quinn 139.7173
*
Evaluated at the mean
Number of cases 'correctly predicted' = 57 (57.0%)
f(beta'x) at mean of independent vars = 0.498

Predicted
0 1
Actual

0 57 0
1 43 0

Page 2 Continued…
 212ECN

i. Estimate by how much the probability of owning personal computer

changes based on Model 1 if we assume that a student is a female,
leaving outside the campus, working 22 hours a week and her average
number of lectures missed per week changes from 0 to 1. Please refer
to Table I.
[25 marks]
for LECTURES = 0:
0.170395 + 0.199162 * 0 + 1.15939 * 0 – 0.984154 * 1 – 0.483189 * 0 =
= -0.813759
D1 = Prob( PC = 1 ) = 1 / ( 1 + e-(-0.813759) ) = 0.307090

for LECTURES = 1:
0.170395 + 0.199162 * 0 + 1.15939 * 0 – 0.984154 * 1 – 0.483189 * 1 =
= -1.296948
D2 = Prob( PC = 1 ) = 1 / ( 1 + e-(-1.296948) ) = 0.214679

Absolute change:
D2 - D1 = 0.214679 - 0.307090 = -0.092411 = -9.2411 percentage point
(not %)

Relative change:
(D2 - D1) / D1 = (0.214679 - 0.307090) / 0.307090 = -0.092411 / 0.307090
= -0.300925 = -30.0925%

ii. Is the coefficient on LECTURES in Model 1 statistically significant?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 1% significance level; calculate z
statistic and present your decision. Please refer to Table I
[25 marks]
𝐻0 : 𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 = 0
𝐻1 : 𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 ≠ 0

Critical value (two-tailed test):

z1% = 2.58 (using tables)

Calculated value:

𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 − 𝛽𝐻0 −0.483189 − 0

𝑧= = = −2.1355
𝑆𝐸(𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 ) 0.226262

Page 3 Continued…
 212ECN

Because -2.58 < -2.1355 so we can’t reject H0. This means that coefficient on
LECTURES in Model 1 is not statistically significant at 1% significance level
(99% confidence level) as the coefficient is not statistically different from
zero.

iii. Assess Model 1 based on measures of fit, namely McFadden R-

squared and Count R2.

This requires you to: explain what the McFadden R-squared and Count
R2 are, show how they are calculated and make a comment on the
values. Please refer to Table I and II.
[10 marks]
2
McFadden R = 1 – (logLUnRest / logLRest) = 1 – (-62.09863 / -68.33149) =
0.091215 = 9.1215% (rather poor fit)

Count R2 = number of correct predictions / total number of observation =

= (49 + 23) / 100 = 72% (rather good fit)

iv. Test whether the coefficients in Model 1 are jointly significant at the 1%
significance level.

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 1% significance level; show the
calculated likelihood ratio statistic and present your decision. Please
refer to Table I.
[25 marks]
𝐻0 : 𝛽𝑆𝐸𝑋 = 0 ∧ 𝛽𝐶𝐴𝑀𝑃𝑈𝑆 = 0 ∧ 𝛽𝑊𝑂𝑅𝐾 = 0 ∧ 𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 = 0
(all coefficients are simultaneously equal zero)
𝐻1 : 𝛽𝑆𝐸𝑋 ≠ 0 ∨ 𝛽𝐶𝐴𝑀𝑃𝑈𝑆 ≠ 0 ∨ 𝛽𝑊𝑂𝑅𝐾 ≠ 0 ∨ 𝛽𝐿𝐸𝐶𝑇𝑈𝑅𝐸𝑆 ≠ 0
(not all coefficients are simultaneously equal zero)

Critical value (one-tailed test):

χ2 (k)1% = χ2 (4)1% = 13.28

Calculated value:
LR = 2 * (logLUnRest - logLRest) = 2 * (-62.09863 – (-68.33149)) =
= 12.46572

Because 12.46572 < 13.28 so we can’t reject H0. This means that Model 1 is
not statistically significant at 1% significance level (99% confidence level) as
all the coefficients are simultaneously equal zero.

v. What are advantages and disadvantages of the linear probability

model?
[15 marks]
simplicity
OLS

̶ fitted probabilities can be below 0 and above 1

̶ heteroskedasticity of variances of residuals

Page 4 Continued…
 212ECN

̶ non-normality of residuals
̶ linear relationship → Pi = P(yi = 1|x) increases linearly with x,
that is, the marginal effect of x remains constant throughout

Page 5 Continued…
 212ECN

Question 2
The data are taken from Whittington, Alm, and Peters (1990) and represent
time series with annual observations from 1917 to 1984. The variables are:
FERTILITY – number of children born per 1000 women in US,
EXEMPTION – average real dollar value of personal tax exemption on
dependents in US (in US dollars),
PILL – availability of birth control pill in US, 1 if year 1963-1984, 0
otherwise,
WW2 – US involvement in World War II, 1 if 1941-1945, 0 otherwise.

Note: l_X - natural logarithm of X.

Table III
Model 3: OLS, using observations 1917-1984 (T = 68)
Dependent variable: l_FERTILITY

Coefficient Std. Error t-ratio p-value

const 4.27343 0.105896 40.3551 <0.00001 ***
l_EXEMPTION 0.0861726 0.0253429 3.4003 0.00116 ***
PILL −0.352682 0.0443791 -7.9470 <0.00001 ***
WW2 −0.228181 0.0761064 -2.9982 0.00386 ***

Mean dependent var 4.521353 S.D. dependent var 0.208403

Sum squared resid 1.460845 S.E. of regression 0.151082
R-squared 0.497982 Adjusted R-squared 0.474450
F(3, 64) 21.16181 P-value(F) 1.23e-09
Log-likelihood 34.08894 Akaike criterion −60.17788
Schwarz criterion −51.29985 Hannan-Quinn −56.66013
rho 0.881668 Durbin-Watson 0.195481

Table IV
Model 4: OLS, using observations 1917-1984 (T = 68)
Dependent variable: l_FERTILITY

Coefficient Std. Error t-ratio p-value

const 4.35679 0.108278 40.2371 <0.00001 ***
l_EXEMPTION 0.0603242 0.0252535 2.3887 0.01982 **
PILL −0.310992 0.0446594 -6.9636 <0.00001 ***

Mean dependent var 4.521353 S.D. dependent var 0.208403

Sum squared resid 1.666027 S.E. of regression 0.160097
R-squared 0.427471 Adjusted R-squared 0.409855
F(2, 65) 24.26570 P-value(F) 1.34e-08
Log-likelihood 29.62042 Akaike criterion −53.24084
Schwarz criterion −46.58232 Hannan-Quinn −50.60253
rho 0.913444 Durbin-Watson 0.141084

Page 6 Continued…
 212ECN

i. Is the coefficient on l_EXEMPTION in Model 3 in Table III statistically

significant?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 1% significance level; calculate t
statistic and present your decision.
[25 marks]
𝐻0 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 = 0
𝐻1 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ≠ 0

Critical value (two-tailed test):

t(n-k-1)1% = t(68-3-1)1% = t(64)1% = 2.58 or 2.75 (using tables)

Calculated value:

𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 − 𝛽𝐻0 0.0861726 − 0

𝑡= = = 3.400
𝑆𝐸(𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ) 0.0253429

Because 2.58 < 3.400 so we reject H0. This means that coefficient on
l_EXEMPTION in Model 3 is statistically significant at 1% significance level
(99% confidence level) as the coefficient is statistically different from zero.

ii. Is Model 4 in Table IV statistically significant?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 5% significance level; calculate F
statistic and present your decision.
[25 marks]
𝐻0 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 = 0 ∧ 𝛽𝑃𝐼𝐿𝐿 = 0
(all coefficients are simultaneously equal zero)
𝐻1 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ≠ 0 ∨ 𝛽𝑃𝐼𝐿𝐿 ≠ 0
(not all coefficients are simultaneously equal zero)

Critical value (one-tailed test):

F(k, n-k-1)5% = F(2, 68-2-1)5% = F(2, 65)5% = 3.15 or 3.07 (using
tables)

Calculated value:
𝑅 2 /𝑘 0.427471/2
𝐹= = = 24.26568
(1 − 𝑅 2 )/(𝑛 − 𝑘 − 1) (1 − 0.427471)/65

Because 3.15 < 24.26568 so we reject H0. This means that Model 4 is
statistically significant at 5% significance level (95% confidence level) as all
the coefficients are not simultaneously equal zero.

iii. State how you would test for the omission of the WW2 in Model 3 in
Table III using an F-statistic. Please refer to Table III and IV.

This requires you to: state the null and alternative hypotheses used,

Page 7 Continued…
 212ECN

determine a critical value at the 5% significance level; calculate F

statistic and present your decision.
[25 marks]
𝐻0 : 𝛽𝑊𝑊2 = 0 (can omit WW2)
𝐻1 : 𝛽𝑊𝑊2 ≠ 0 (can’t omit WW2)

Critical value (one-tailed test):

F(r, n-kUnRest-1)5% = F(1, 68-3-1)5% = F(1, 64)5% = 4.00 or 3.92 (using
tables)

Calculated value:

(𝑅𝑆𝑆𝑅𝑒𝑠𝑡 − 𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 )/𝑟 (1.666027 − 1.460845)/1

𝐹= = = 8.9891
𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 /(𝑛 − 𝑘𝑈𝑛𝑅𝑒𝑠𝑡 − 1) 1.460845/64

Because 4.00 < 8.9891 so we reject H0. This means that test is statistically
significant at 5% significance level (95% confidence level) as the coefficient
is statistically different from zero. It suggests that WW2 can’t be omitted.

iv. Interpret the value of the parameter on l_EXEMPTION variable in the

Model 3 in Table III.
[10 marks]
If the average real dollar value of personal tax exemption increases by
1%, on average, the number of children born per 1000 women
increases by 0.0861726% keeping other variables constant (ceteris
paribus).

v. Are the signs of the l_EXEMPTION, PILL and WW2 estimated

coefficients in the Model 3 in Table III in line with the expectations
based on the theory? Explain your answer.
[15 marks]
Sign for l_EXEMPTION is positive as expected.

Sign for PILL is negative as expected.

Sign for WW2 is negative as expected.

Page 8 Continued…
 212ECN

Question 3

Table V presents the Ramsey’s RESET test of the Model 3 in Table III in
Question 2.

Table VI presents the White test of the Model 3 in Table III in Question 2.

Table V

Auxiliary regression for RESET specification test

OLS, using observations 1917-1984 (T = 68)
Dependent variable: l_FERTILITY

coefficient std. error t-ratio p-value

-----------------------------------------------------------
const 15101.3 2958.73 5.104 3.40e-06 ***
l_EXEMPTION 470.608 92.1501 5.107 3.36e-06 ***
PILL −1926.15 377.181 −5.107 3.36e-06 ***
WW2 −1246.36 244.055 −5.107 3.36e-06 ***
yhat^2 −1205.78 236.440 −5.100 3.45e-06 ***
yhat^3 88.7032 17.4169 5.093 3.54e-06 ***

Sum squared resid = 1.025807

Test statistic: F = 13.146874,

with p-value = P(F(2,62) > 13.1469) = 1.74e-005

Table VI

White's test for heteroskedasticity

OLS, using observations 1917-1984 (T = 68)
Dependent variable: uhat^2

coefficient std. error t-ratio p-value

-------------------------------------------------------------
const −0.231524 0.120789 −1.917 0.0600 *
l_EXEMPTION 0.144596 0.0611289 2.365 0.0212 **
PILL −0.496512 0.155354 −3.196 0.0022 ***
WW2 −0.271128 0.161139 −1.683 0.0976 *
sq_l_EXEMPTION −0.0192378 0.00738373 −2.605 0.0115 **
X2_X3 0.104139 0.0320051 3.254 0.0019 ***
X2_X4 0.0530424 0.0317673 1.670 0.1001

Unadjusted R-squared = 0.342045

Test statistic: TR^2 = 23.259037,

with p-value = P(Chi-square(6) > 23.259037) = 0.000714

Page 9 Continued…
 212ECN

i. Explain the nature of the multicollinearity problem in regression

analysis and state what consequences it has for OLS estimation.

Explain in detail how we can test for the problem and how a regression
can be corrected if multicollinearity is found.
[50 marks]
Nature
The word collinear describes a linear correlation between two
independent variables, and multicollinearity indicates that more than
two independent variables are involved.
Perfect multicollinearity - an extreme case, two variables might be a
linear combination of each other: x1 = 3x2
Imperfect multicollinearity occurs when two or more RHS variables
are imperfectly linearly related: x1i = 3x2i + ui

Impact
Perfect multicollinearity is rare and normally the result of an error on
the user’s part
β1 & β2 - indeterminable
SE(β1) & SE(β2) = infinite

If there is less than perfect multicollinearity there is not too much of

a problem
The estimates are still BLUE
Estimates will remain unbiased
Extreme multicollinearity can lead to:
Instability
Changing the data or model slightly can lead to
massive swings in t ratios, coefficients etc
Over “Acceptance” of the Null hypothesis
The standard errors are over-estimated
So t-scores are understated
This means that the t ratios are unreliable

Detection

Page 10 Continued…
 212ECN

There is no generally accepted, true statistical tests for

multicollinearity
Most used are:
1. Often a significant F statistic (high adj. R2) for the regression along
with insignificant t ratios suggest that there might be severe
multicollinearity
The results are both significant and insignificant
Perhaps removing one of the explanatory variables might be
helpful (Be careful!)
2. Looking at a subset of the data and seeing if the coefficients are
stable is another approach
3. High correlation between RHS variables
This works when there are no dummy variables
Be wary if you suspect that there is multicollinearity based
on combinations of more than 2 variables
High correlations are sufficient but not a necessary
condition for existence of multicollinearity because it can
exist even though simple correlations are low.
How high is high? Simple correlation is high if it causes
unacceptably large variances in the coefficient estimates in
which we are interested.
4. High variance inflations factors (VIF)
Suppose this is our regression:
y = β0 + β1x1 + β2x2 +…+ βkxk + ε
The variance inflation factor (VIF) is calculated from two
steps:
1. Run an OLS regression that has Xi as a function of all the
other explanatory variables in the equation: For i = 1, this
equation would be:
X1 = α1 + α2X2 + α3X3 + … + αKXK + v (1)
where v is a classical stochastic error term
2. Calculate the variance inflation factor for:

Page 11 Continued…
 212ECN

(2)
Where Ri2 is the unadjusted R2 from step one
There are no critical tables for VIF but just a rule of thumb.
If VIFi > 10 then it suggests multicollinearity problem as Ri2 >
0.9
VIF is a sufficient, not necessary, test for multicollinearity
because
Thus, whether variance of estimated coefficient is large or
small will depend on three factors
And not only VIF
So you might have large variances of estimated coefficients
in equation that has no large VIFs

Causes of multicollinearity
There are many reasons why we might find multicollinearity
Definitions of variables
e.g. nominal series reflect common trend due to changes in
price index
An identity relationship included
Too small a sample with too little variability
Too many dummies (dummy variable trap)
Entry errors

Solutions
1. Do nothing
A remedy should be considered only if the consequences
cause insignificant t ratios or unreliable estimated
coefficients!
The deletion of a multicollinear variable that belongs in a
equation will cause specification bias (omitting a relevant
variable).

Page 12 Continued…
 212ECN

2. Drop a redundant variable

Redundant variables are two or more variables that
measuring essentially the same thing e.g. disposable income
and GDP both are measuring income.
3. Increase the sample size
In particular change the sample periodicity
1. Move from annual to quarterly data or to weekly
where possible or appropriate
4. Respecify the model
Introducing lags or different transformations of the data can
reduce the impact of the problem
However remember that lags may bring multicollinearity as
they reduce sample size and lags of variable may have very
high correlation with each other.
5. Use restrictions on the coefficients
This can remove or offset the impact of the troublesome
variables
6. Factor Analysis and Principal Component Analysis
These methods orthogonalise the data
Components are created that are linear combinations of the
variables. These are used in the regression

ii. Undertake the Ramsey’s RESET test.

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 5% significance level; calculate F
statistic and present your decision. Please refer to Table V.
[25 marks]
𝐻0 : 𝛽𝑦ℎ𝑎𝑡^2 = 0 ∧ 𝛽𝑦ℎ𝑎𝑡^3 = 0 (no misspecification of Model 3)
𝐻1 : 𝛽𝑦ℎ𝑎𝑡^2 ≠ 0 ∨ 𝛽𝑦ℎ𝑎𝑡^3 ≠ 0 (misspecification of Model 3)

Critical value (one-tailed test):

F(r, n-kUnRest-1)5% = F(2, 68-5-1)5% = F(2, 62)5% = 3.15 or 3.07 (using tables)

Calculated value:
(𝑅𝑆𝑆𝑅𝑒𝑠𝑡 − 𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 )/𝑟 (1.460845 − 1.025807)/2
𝐹= = = 13.1469
𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 /(𝑛 − 𝑘𝑈𝑛𝑅𝑒𝑠𝑡 − 1) 1.025807/62

Because 3.15 < 13.1469 so we reject H0. This means that the specification of

Page 13 Continued…
 212ECN

Model 3 is not correct at 5% significance level (95% confidence level) as both

of the coefficients are not simultaneously equal zero.

iii. Undertake the White test.

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 5% significance level; calculate the
White statistic and present your decision. Please refer to Table VI.
[25 marks]
𝐻0 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 = 0 ∧ 𝛽𝑃𝐼𝐿𝐿 = 0 ∧ 𝛽𝑊𝑊2 = 0 ∧ 𝛽𝑠𝑞_𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 = 0 ∧
𝛽𝑋2_𝑋3 = 0 ∧ 𝛽𝑋2_𝑋4 = 0 (homoskedasticity)
𝐻0 : 𝛽𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ≠ 0 ∨ 𝛽𝑃𝐼𝐿𝐿 ≠ 0 ∨ 𝛽𝑊𝑊2 ≠ 0 ∨ 𝛽𝑠𝑞_𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ≠ 0 ∨
𝛽𝑋2_𝑋3 ≠ 0 ∨ 𝛽𝑋2_𝑋4 ≠ 0 (heteroskedasticity)

Critical value (one-tailed test):

χ2(r)5% = χ2(6)5% = 12.59 (using tables)

Calculated value:
𝑊ℎ𝑖𝑡𝑒 ′ 𝑡𝑒𝑠𝑡 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = 𝑇𝑅 2 = 68 ∗ 0.342045 = 23.25906

Because 12.59 < 23.25906 so we reject H0. This means that residuals from
Model 3 are heteroskedastic at 5% significance level (95% confidence level)
as not all the coefficients are simultaneously equal zero.

Page 14 Continued…
 212ECN

Question 4

Table VII presents the summary statistics of the residuals from the Model 3 in
Table III in Question 2.

Table VIII presents the Chow test of the Model 3 in Table III in Question 2.

Table IX presents the Breusch-Godfrey test of the Model 3 in Table III in

Question 2.

Table VII

Summary Statistics, using the observations 1917 - 1984

for the variable uhat3 (68 valid observations)
Mean Median Minimum Maximum
-1.95922e-016 0.00133344 -0.282646 0.327800
Std. Dev. C.V. Skewness Ex. kurtosis
0.147661 undefined 0.130942 -0.667137
5% Perc. 95% Perc. IQ range Missing obs.
-0.237752 0.264607 0.220602 0

Table VIII
Augmented regression for Chow test
OLS, using observations 1917-1984 (T = 68)
Dependent variable: l_FERTILITY
Omitted due to exact collinearity: sd_PILL

coefficient std. error t-ratio p-value

----------------------------------------------------------------
const 4.29154 0.104730 40.98 1.29e-046 ***
l_EXEMPTION 0.0817263 0.0250704 3.260 0.0018 ***
PILL −0.314391 0.0476021 −6.605 1.03e-08 ***
WW2 −0.223639 0.0749385 −2.984 0.0041 ***
splitdum −0.268888 2.10857 −0.1275 0.8989
sd_l_EXEMPTION 0.0248868 0.454504 0.05476 0.9565

Mean dependent var 4.521353 S.D. dependent var 0.208403

Sum squared resid 1.370414 S.E. of regression 0.148672
R-squared 0.529058 Adjusted R-squared 0.491079
F(5, 62) 13.93022 P-value(F) 4.02e-09
Log-likelihood 36.26160 Akaike criterion −60.52320
Schwarz criterion −47.20616 Hannan-Quinn −55.24658
rho 0.869511 Durbin-Watson 0.210712

Chow test for structural break at observation 1980

F(2, 62) = 2.04562 with p-value 0.1379

Page 15 Continued…
 212ECN

Table IX

Breusch-Godfrey test for autocorrelation up to order 2

OLS, using observations 1917-1984 (T = 68)
Dependent variable: uhat

coefficient std. error t-ratio p-value

--------------------------------------------------------------
const 0.00222214 0.0511315 0.04346 0.9655
l_EXEMPTION −0.00133500 0.0122323 −0.1091 0.9134
PILL −0.000351478 0.0214273 −0.01640 0.9870
WW2 0.0316666 0.0372191 0.8508 0.3981
uhat_1 0.951970 0.126778 7.509 2.82e-010 ***
uhat_2 −0.0760711 0.128657 −0.5913 0.5565

Unadjusted R-squared = 0.774634

Test statistic: TR^2 = 52.675086,

with p-value = P(Chi-square(2) > 52.6751) = 3.65e-012

Page 16 Continued…
 212ECN

i. Are the residuals from the Model 3 normally distributed?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 1% significance level; calculate Jarque-
Bera statistic and present your decision. Please refer to Table VII.
[25 marks]
H0 : skewnessuhat3 = 0 ∧ excess kurtosisuhat3 = 0
(residuals are normally distributed)
H1 : skewnessuhat3 ≠ 0 ∨ excess kurtosisuhat3 ≠ 0
(residuals are not normally distributed)

Critical value (one-tailed test):

χ2 (r)1% = χ2 (2)1% = 9.21 (using tables)

Calculated value:
𝑛 2
(𝑘 − 3)2 68 2
(−0.667137)2
𝐽 − 𝐵 = (𝑠𝑘 + )= (0.130942 + )
6 4 6 4
= 1.455356

Because 1.455356 < 9.21 so we can’t reject H0. This means that residuals
are normally distributed at 1% significance level (99% confidence level)
as skewness and excess kurtosis are jointly equal 0.

ii. Is there a structural break in Model 3?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 5% significance level; calculate Chow
statistic and present your decision. Please refer to Table VIII.
[25 marks]
𝐻0 : 𝛽𝑠𝑝𝑙𝑖𝑡𝑑𝑢𝑚 = 0 ∧ 𝛽𝑠𝑑_𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 = 0
(no structural break in Model 3)
𝐻1 : 𝛽𝑠𝑝𝑙𝑖𝑑𝑢𝑚 ≠ 0 ∨ 𝛽𝑠𝑑_𝑙_𝐸𝑋𝐸𝑀𝑃𝑇𝐼𝑂𝑁 ≠ 0
(structural break in Model 3)

Critical value (one-tailed test):

F(r, n-kUnRest-1)5% = F(2, 68-5-1)5% = F(2, 62)5% = 3.15 or 3.07 (using tables)

Calculated value:
(𝑅𝑆𝑆𝑅𝑒𝑠𝑡 − 𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 )/𝑟 (1.460845 − 1.370414)/2
𝐹= =
𝑅𝑆𝑆𝑈𝑛𝑅𝑒𝑠𝑡 /(𝑛 − 𝑘𝑈𝑛𝑅𝑒𝑠𝑡 − 1) 1.370414/62
= 2.045631

Because 2.045631 < 3.15 so we can’t reject H0. This means that there is no
structural break at observation 1980 in Model 3 at 5% significance level (95%
confidence level) as all of the coefficients are simultaneously equal zero.

iii. Explain briefly how to compare R2 between linear and log-log model.
[15 marks]
Let assume that we have two models:

Page 17 Continued…
 212ECN

lnY = β0 + β1lnX1 + β2lnX2 + ε (1)

Y = α0 + α1X1 + α2X2 + ε (2)
If you were to compare the two R2 between model 1 and model 2 you may
proceed as follows:
̂
1. Obtain fitted values from model 1 i.e. 𝑙𝑛𝑌
̂
2. Take antilog of these values i.e. 𝑒𝑥𝑝(𝑙𝑛𝑌)
3. Compute R2 as the squared coefficient of correlation between these
2
antilog values and actual Y i.e. 𝑅𝑐𝑜𝑚𝑝𝑎𝑟𝑎𝑏𝑙𝑒 =
2
̂ , 𝑌)]
[𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛(𝑒𝑥𝑝(𝑙𝑛𝑌)
4. This R is comparable to R2 of the linear model 2
2

iv. What are the three components of a constant estimate?

[10 marks]
An estimate of constant has at least three components:
1. the true constant (β0 or α)
2. the constant impact of any specification errors (an omitted variable, for
example)
3. the mean of ε for the correctly specified equation (if not equal to zero)

v. Does Model 3 exhibit autocorrelation up to order 2?

This requires you to: state the null and alternative hypotheses used,
determine a critical value at the 5% significance level; calculate
Breusch-Godfrey statistic and present your decision. Please refer to
Table IX.
[25 marks]
𝐻0 : 𝛽𝑢ℎ𝑎𝑡3_1 = 0 ∧ 𝛽𝑢ℎ𝑎𝑡3_2 = 0 (no autocorrelation up to order 2)
𝐻0 : 𝛽𝑢ℎ𝑎𝑡3_1 ≠ 0 ∨ 𝛽𝑢ℎ𝑎𝑡3_2 ≠ 0 (autocorrelation up to order 2)

Critical value (one-tailed test):

χ2(r)5% = χ2(2)5% = 5.99 (using tables)

Calculated value:
𝐵𝑟𝑒𝑢𝑠𝑐ℎ − 𝐺𝑜𝑑𝑓𝑟𝑒𝑦 𝑡𝑒𝑠𝑡 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = 𝑇𝑅 2 = 68 ∗ 0.774634 = 52.67511
or
𝐵𝑟𝑒𝑢𝑠𝑐ℎ − 𝐺𝑜𝑑𝑓𝑟𝑒𝑦 𝑡𝑒𝑠𝑡 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = (𝑇 − 2) ∗ 𝑅 2 = 66 ∗ 0.774634
= 51.12584

Because 5.99 < 52.67511 so we reject H0. This means that there is
autocorrelation up to order 2 in Model 3 at 5% significance level (95%
confidence level) as all the coefficients on lagged residuals are not
simultaneously equal zero.

Page 18 Continued…
 212ECN

Statistical tables
Student t F
Z Durbin-Watson
Χ2 QLR

Student t

df – 1 tailed 0.10 0.05 0.025 0.01 0.005

df – 2 tailed 0.2 0.1 0.05 0.02 0.01
10 1.37 1.81 2.23 2.76 3.17
11 1.36 1.80 2.20 2.72 3.11
12 1.36 1.78 2.18 2.68 3.05
13 1.35 1.77 2.16 2.65 3.01
14 1.35 1.76 2.14 2.62 2.98
15 1.34 1.75 2.13 2.60 2.95
16 1.34 1.75 2.12 2.58 2.92
17 1.33 1.74 2.11 2.57 2.90
18 1.33 1.73 2.10 2.55 2.88
19 1.33 1.73 2.09 2.54 2.86
20 1.33 1.72 2.09 2.53 2.85
21 1.32 1.72 2.08 2.52 2.83
22 1.32 1.72 2.07 2.51 2.82
23 1.32 1.71 2.07 2.50 2.81
24 1.32 1.71 2.06 2.49 2.80
25 1.32 1.71 2.06 2.49 2.79
26 1.31 1.71 2.06 2.48 2.78
28 1.31 1.70 2.05 2.47 2.76
30 1.31 1.70 2.04 2.46 2.75
infinity 1.28 1.64 1.96 2.33 2.58

Cumulative standardized normal distribution table is shown on the next page

Page 19 Continued…
 212ECN

Cumulative standardized normal distribution

z 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

0 0.50000 0.50399 0.50798 0.51197 0.51595 0.51994 0.52392 0.52790 0.53188 0.53586
0.1 0.53983 0.54380 0.54776 0.55172 0.55567 0.55962 0.56356 0.56749 0.57142 0.57535
0.2 0.57926 0.58317 0.58706 0.59095 0.59483 0.59871 0.60257 0.60642 0.61026 0.61409
0.3 0.61791 0.62172 0.62552 0.62930 0.63307 0.63683 0.64058 0.64431 0.64803 0.65173
0.4 0.65542 0.65910 0.66276 0.66640 0.67003 0.67364 0.67724 0.68082 0.68439 0.68793
0.5 0.69146 0.69497 0.69847 0.70194 0.70540 0.70884 0.71226 0.71566 0.71904 0.72240
0.6 0.72575 0.72907 0.73237 0.73565 0.73891 0.74215 0.74537 0.74857 0.75175 0.75490
0.7 0.75804 0.76115 0.76424 0.76730 0.77035 0.77337 0.77637 0.77935 0.78230 0.78524
0.8 0.78814 0.79103 0.79389 0.79673 0.79955 0.80234 0.80511 0.80785 0.81057 0.81327
0.9 0.81594 0.81859 0.82121 0.82381 0.82639 0.82894 0.83147 0.83398 0.83646 0.83891
1 0.84134 0.84375 0.84614 0.84849 0.85083 0.85314 0.85543 0.85769 0.85993 0.86214
1.1 0.86433 0.86650 0.86864 0.87076 0.87286 0.87493 0.87698 0.87900 0.88100 0.88298
1.2 0.88493 0.88686 0.88877 0.89065 0.89251 0.89435 0.89617 0.89796 0.89973 0.90147
1.3 0.90320 0.90490 0.90658 0.90824 0.90988 0.91149 0.91309 0.91466 0.91621 0.91774
1.4 0.91924 0.92073 0.92220 0.92364 0.92507 0.92647 0.92785 0.92922 0.93056 0.93189
1.5 0.93319 0.93448 0.93574 0.93699 0.93822 0.93943 0.94062 0.94179 0.94295 0.94408
1.6 0.94520 0.94630 0.94738 0.94845 0.94950 0.95053 0.95154 0.95254 0.95352 0.95449
1.7 0.95543 0.95637 0.95728 0.95818 0.95907 0.95994 0.96080 0.96164 0.96246 0.96327
1.8 0.96407 0.96485 0.96562 0.96638 0.96712 0.96784 0.96856 0.96926 0.96995 0.97062
1.9 0.97128 0.97193 0.97257 0.97320 0.97381 0.97441 0.97500 0.97558 0.97615 0.97670
2 0.97725 0.97778 0.97831 0.97882 0.97932 0.97982 0.98030 0.98077 0.98124 0.98169
2.1 0.98214 0.98257 0.98300 0.98341 0.98382 0.98422 0.98461 0.98500 0.98537 0.98574
2.2 0.98610 0.98645 0.98679 0.98713 0.98745 0.98778 0.98809 0.98840 0.98870 0.98899
2.3 0.98928 0.98956 0.98983 0.99010 0.99036 0.99061 0.99086 0.99111 0.99134 0.99158
2.4 0.99180 0.99202 0.99224 0.99245 0.99266 0.99286 0.99305 0.99324 0.99343 0.99361
2.5 0.99379 0.99396 0.99413 0.99430 0.99446 0.99461 0.99477 0.99492 0.99506 0.99520
2.6 0.99534 0.99547 0.99560 0.99573 0.99585 0.99598 0.99609 0.99621 0.99632 0.99643
2.7 0.99653 0.99664 0.99674 0.99683 0.99693 0.99702 0.99711 0.99720 0.99728 0.99736
2.8 0.99744 0.99752 0.99760 0.99767 0.99774 0.99781 0.99788 0.99795 0.99801 0.99807
2.9 0.99813 0.99819 0.99825 0.99831 0.99836 0.99841 0.99846 0.99851 0.99856 0.99861
3 0.99865 0.99869 0.99874 0.99878 0.99882 0.99886 0.99889 0.99893 0.99896 0.99900
3.1 0.99903 0.99906 0.99910 0.99913 0.99916 0.99918 0.99921 0.99924 0.99926 0.99929
3.2 0.99931 0.99934 0.99936 0.99938 0.99940 0.99942 0.99944 0.99946 0.99948 0.99950

Chi-Squared table is shown on the next page

Page 20 Continued…
 212ECN

Right tail areas for the Chi-Squared distribution

df \ area area = .100 .050 .025 .010 .005

df = 1 2.71 3.84 5.02 6.63 7.88
2 4.61 5.99 7.38 9.21 10.60
3 6.25 7.81 9.35 11.34 12.84
4 7.78 9.49 11.14 13.28 14.86
5 9.24 11.07 12.83 15.09 16.75
6 10.64 12.59 14.45 16.81 18.55
7 12.02 14.07 16.01 18.48 20.28
8 13.36 15.51 17.53 20.09 21.95
9 14.68 16.92 19.02 21.67 23.59
10 15.99 18.31 20.48 23.21 25.19
11 17.28 19.68 21.92 24.72 26.76
12 18.55 21.03 23.34 26.22 28.30
13 19.81 22.36 24.74 27.69 29.82
14 21.06 23.68 26.12 29.14 31.32
15 22.31 25.00 27.49 30.58 32.80
16 23.54 26.30 28.85 32.00 34.27
17 24.77 27.59 30.19 33.41 35.72
18 25.99 28.87 31.53 34.81 37.16
19 27.20 30.14 32.85 36.19 38.58
20 28.41 31.41 34.17 37.57 40.00
25 34.38 37.65 40.65 44.31 46.93
30 40.26 43.77 46.98 50.89 53.67
35 46.06 49.80 53.20 57.34 60.27

F table is shown on the next page

Page 21 Continued…
 212ECN

Upper 5% points on the F distribution

df2 \ df1 df1 = 1 2 3 4 5 6 7 8 9 10 12 15 20 30 40 60 120 inf

df2 = 1 161.5 199.5 215.7 224.6 230.2 234.0 236.8 238.9 240.5 241.9 243.9 245.9 248.0 250.1 251.1 252.2 253.3 254.3
2 18.51 19.00 19.16 19.25 19.30 19.33 19.35 19.37 19.38 19.34 19.41 19.42 19.45 19.46 19.47 19.48 19.49 19.50
3 10.12 9.55 9.28 9.12 9.01 8.94 8.89 8.85 8.81 8.79 8.74 8.70 8.66 8.62 8.59 8.57 8.55 8.53
4 7.71 6.94 6.59 6.39 6.26 6.16 6.09 6.04 6.00 5.96 5.91 5.86 5.80 5.75 5.72 5.69 5.66 5.63
5 6.61 5.79 5.41 5.19 5.05 4.95 4.88 4.82 4.77 4.74 4.68 4.62 4.56 4.50 4.46 4.43 4.40 4.37
6 5.99 5.14 4.76 4.53 4.39 4.28 4.21 4.15 4.10 4.06 4.00 3.94 3.87 3.81 3.77 3.74 3.70 3.67
7 5.59 4.74 4.35 4.12 3.97 3.87 3.79 3.73 3.68 3.64 3.57 3.51 3.44 3.38 3.34 3.30 3.27 3.23
8 5.32 4.46 4.07 3.84 3.69 3.58 3.50 3.44 3.39 3.35 3.28 3.22 3.15 3.08 3.04 3.01 2.97 2.93
9 5.12 4.26 3.86 3.63 3.48 3.37 3.29 3.23 3.18 3.14 3.07 3.01 2.94 2.86 2.83 2.79 2.75 2.71
10 4.96 4.10 3.71 3.48 3.33 3.22 3.14 3.07 3.02 2.98 2.91 2.85 2.77 2.70 2.66 2.62 2.58 2.54
11 4.84 3.98 3.59 3.36 3.20 3.09 3.01 2.95 2.90 2.85 2.79 2.72 2.65 2.57 2.53 2.49 2.45 2.40
12 4.75 3.89 3.49 3.26 3.11 3.00 2.91 2.85 2.80 2.75 2.69 2.62 2.54 2.47 2.43 2.38 2.34 2.30
13 4.67 3.81 3.41 3.18 3.03 2.92 2.83 2.77 2.71 2.67 2.60 2.53 2.46 2.38 2.34 2.30 2.25 2.21
14 4.60 3.74 3.34 3.11 2.96 2.85 2.76 2.70 2.65 2.60 2.53 2.46 2.39 2.31 2.27 2.22 2.18 2.13
15 4.54 3.68 3.29 3.06 2.90 2.79 2.71 2.64 2.59 2.54 2.48 2.40 2.33 2.25 2.20 2.16 2.11 2.07
16 4.49 3.63 3.24 3.01 2.85 2.74 2.66 2.59 2.54 2.49 2.42 2.35 2.28 2.19 2.15 2.11 2.06 2.01
17 4.45 3.59 3.20 2.96 2.81 2.70 2.61 2.55 2.49 2.45 2.38 2.31 2.23 2.15 2.10 2.06 2.01 1.96
18 4.41 3.55 3.16 2.93 2.77 2.66 2.58 2.51 2.46 2.41 2.34 2.27 2.19 2.11 2.06 2.02 1.97 1.92
19 4.38 3.52 3.13 2.90 2.74 2.63 2.54 2.48 2.42 2.38 2.31 2.23 2.16 2.07 2.03 1.98 1.93 1.88
20 4.35 3.49 3.10 2.87 2.71 2.60 2.51 2.45 2.39 2.35 2.28 2.20 2.12 2.04 1.99 1.95 1.90 1.84
21 4.32 3.47 3.07 2.84 2.68 2.57 2.49 2.42 2.37 2.32 2.25 2.18 2.10 2.01 1.96 1.92 1.87 1.81
23 4.28 3.42 3.03 2.80 2.64 2.53 2.44 2.37 2.32 2.27 2.20 2.13 2.05 1.96 1.91 1.86 1.81 1.76
26 4.23 3.37 2.98 2.74 2.59 2.47 2.39 2.32 2.27 2.22 2.15 2.07 1.99 1.90 1.85 1.80 1.75 1.69
27 4.21 3.35 2.96 2.73 2.57 2.46 2.37 2.31 2.25 2.20 2.13 2.06 1.97 1.88 1.84 1.79 1.73 1.67
28 4.20 3.34 2.95 2.71 2.56 2.45 2.36 2.29 2.24 2.19 2.12 2.04 1.96 1.87 1.82 1.77 1.71 1.65
29 4.18 3.33 2.93 2.70 2.55 2.43 2.35 2.28 2.22 2.18 2.10 2.03 1.94 1.85 1.81 1.75 1.70 1.64
30 4.17 3.32 2.92 2.69 2.53 2.42 2.33 2.27 2.21 2.16 2.09 2.01 1.93 1.84 1.79 1.74 1.68 1.62
40 4.08 3.23 2.84 2.61 2.45 2.34 2.25 2.18 2.12 2.08 2.00 1.92 1.84 1.74 1.69 1.64 1.58 1.51
60 4.00 3.15 2.76 2.53 2.37 2.25 2.17 2.10 2.04 1.99 1.92 1.84 1.75 1.65 1.59 1.53 1.47 1.39
120 3.92 3.07 2.68 2.45 2.29 2.18 2.09 2.02 1.96 1.91 1.83 1.75 1.66 1.55 1.50 1.43 1.35 1.25
inf 3.84 3.00 2.60 2.37 2.21 2.10 2.01 1.94 1.88 1.83 1.75 1.67 1.57 1.46 1.39 1.32 1.22 1.00

Durbin-Watson table is shown on the next page

Page 22 Continued…
 212ECN

Durbin-Watson distribution: 5% significance points of dl and du in one-

tailed test

k=1 k=2 k=3 k=4 k=5 k=6

n dl du dl du dl du dl du dl du dl du
15 1.08 1.32 0.95 1.54 0.82 1.75 0.69 1.97 0.56 2.21 0.50 2.39
16 1.10 1.37 0.98 1.54 0.86 1.73 0.74 1.93 0.62 2.15 0.55 2.12
17 1.13 1.38 1.02 1.54 0.90 1.71 0.78 1.90 0.67 2.10 0.60 2.26
18 1.16 1.39 1.05 1.53 0.93 1.69 0.82 1.87 0.71 2.06 0.65 2.20
19 1.18 1.40 1.08 1.53 0.97 1.68 0.86 1.85 0.75 2.02 0.69 2.16
20 1.20 1.41 1.10 1.54 1.00 1.68 0.90 1.83 0.79 1.99 0.73 2.12
21 1.22 1.42 1.13 1.54 1.03 1.67 0.93 1.81 0.83 1.96 0.77 2.09
22 1.24 1.43 1.15 1.54 1.05 1.66 0.96 1.80 0.86 1.94 0.80 2.06
23 1.26 1.44 1.17 1.54 1.08 1.66 0.99 1.79 0.90 1.92 0.84 2.04
24 1.27 1.45 1.19 1.55 1.10 1.66 1.01 1.78 0.93 1.90 0.87 2.01
25 1.29 1.45 1.21 1.55 1.12 1.66 1.04 1.77 0.95 1.89 0.90 1.99
26 1.30 1.46 1.22 1.55 1.14 1.65 1.06 1.76 0.98 1.88 0.93 1.97
27 1.32 1.47 1.24 1.56 1.16 1.65 1.08 1.76 1.01 1.86 0.95 1.96
28 1.33 1.48 1.26 1.56 1.18 1.65 1.10 1.75 1.03 1.85 0.96 1.94
29 1.34 1.48 1.27 1.56 1.20 1.65 1.12 1.74 1.05 1.84 1.00 1.93
30 1.35 1.49 1.28 1.57 1.21 1.65 1.14 1.74 1.07 1.83 1.02 1.92
35 1.40 1.52 1.34 1.58 1.28 1.65 1.22 1.73 1.16 1.80 1.10 1.88
40 1.44 1.54 1.39 1.60 1.34 1.66 1.29 1.72 1.23 1.79 1.18 1.85
50 1.50 1.59 1.46 1.63 1.42 1.67 1.38 1.72 1.34 1.77 1.29 1.82
60 1.55 1.62 1.51 1.65 1.48 1.69 1.44 1.73 1.41 1.77 1.33 1.81
70 1.58 1.64 1.55 1.67 1.52 1.70 1.49 1.74 1.46 1.77 1.43 1.80
80 1.61 1.66 1.59 1.69 1.56 1.72 1.53 1.74 1.51 1.77 1.48 1.80
90 1.63 1.68 1.61 1.70 1.59 1.73 1.57 1.75 1.54 1.78 1.51 1.80
100 1.65 1.69 1.63 1.72 1.61 1.74 1.59 1.76 1.57 1.78 1.55 1.80
200 1.76 1.78 1.75 1.79 1.74 1.80 1.73 1.81 1.71 1.82 1.71 1.83

QLR table is shown on the next page

Page 23 Continued…
 212ECN

Critical values of the QLR statistic with 15% trimming

Significance level
Number of restrictions (r)
10% 5% 1%
1 7.12 8.68 12.16
2 5.00 5.86 7.78
3 4.09 4.71 6.02
4 3.59 4.09 5.12
5 3.26 3.66 4.53
6 3.02 3.37 4.12
7 2.84 3.15 3.82
8 2.69 2.98 3.57
9 2.58 2.84 3.38
10 2.48 2.71 3.23
11 2.40 2.62 3.09
12 2.33 2.54 2.97
13 2.27 2.46 2.87
14 2.21 2.40 2.78
15 2.16 2.34 2.71
16 2.12 2.29 2.64
17 2.08 2.25 2.58
18 2.05 2.20 2.53
19 2.01 2.17 2.48
20 1.99 2.13 2.43
Source: Andrews (2003)

Page 24 End