NUST Business School

ECO-215 – Fundamentals of Econometrics

Assignment:03

Submitted to:
Dr . Mohsin Ali
Submitted by:
Muhammad Saad Aslam

BSTHM 2K22

Date: 20th May, 2024

To show that the change in 𝑦∗ due to the change in 𝑧∗ equals the long-run propensity (LRP)
times the change in 𝑧∗, we need to derive the equilibrium values for 𝑦𝑡 and 𝑧𝑡 , and then
show how changes in 𝑧∗ translate into changes in 𝑦∗ .
The given second-order Finite Distributed Lag (FDL) model is:
𝑦𝑡=𝛼0+𝛿0𝑧𝑡+𝛿1𝑧𝑡−1+𝛿2𝑧𝑡−2+𝑢𝑡

Step 1: Determine the equilibrium values 𝑦∗ and 𝑧∗

In equilibrium, all variables are constant over time, meaning:

𝑧𝑡=𝑧𝑡−1=𝑧𝑡−2=𝑧∗

𝑦𝑡=𝑦

Substituting these equilibrium values into the model, we get: 𝑦∗=𝛼0+𝛿0𝑧∗+𝛿1𝑧∗+𝛿2𝑧∗+𝑢𝑡

Since in equilibrium the error term 𝑢𝑡 is typically assumed to be zero (because equilibrium
implies no random shocks), we simplify to:

𝑦∗=𝛼0+𝛿0𝑧∗+𝛿1𝑧∗+𝛿2𝑧∗

Step 2: Combine terms and solve for 𝑦∗

Combine the terms involving 𝑧∗:

𝑦∗=𝛼0+(𝛿0+𝛿1+𝛿2)𝑧∗
Step 3: Identify the Long Run Propensity (LRP)

The Long Run Propensity (LRP) measures the total effect on 𝑦∗ of a permanent one-unit
change in 𝑧∗. From the equation above, we see that the coefficient on 𝑧∗ is:

𝛿0+𝛿1+𝛿2

Therefore, the LRP is: 𝐿𝑅𝑃=𝛿0+𝛿1+𝛿2

Step 4: Show that the change in 𝑦∗ due to the change in 𝑧∗ equals the LRP times the
change in 𝑧∗
Consider a change in the equilibrium value of 𝑧, denoted as Δ𝑧∗. The corresponding change
in 𝑦∗, denoted as Δ𝑦∗, can be derived from the equilibrium equation: 𝑦∗=𝛼0+(𝛿0+𝛿1+𝛿2)𝑧∗

If 𝑧∗ changes to 𝑧∗+Δ𝑧∗, the new equilibrium value of 𝑦 becomes: 𝑦𝑛𝑒𝑤∗=𝛼0+(𝛿0+𝛿1+𝛿2)

(𝑧∗+Δ𝑧∗)

𝑦𝑛𝑒𝑤∗=𝛼0+(𝛿0+𝛿1+𝛿2)𝑧∗+(𝛿0+𝛿1+𝛿2)Δ𝑧

The change in 𝑦∗ is:

Δ𝑦∗=𝑦𝑛𝑒𝑤∗−𝑦∗

Δ𝑦∗=[𝛼0+(𝛿0+𝛿1+𝛿2)𝑧∗+(𝛿0+𝛿1+𝛿2)Δ𝑧∗]−[𝛼0+(𝛿0+𝛿1+𝛿2)𝑧∗]

Δ𝑦∗=(𝛿0+𝛿1+𝛿2)Δ𝑧∗

This shows that the change in 𝑦∗y∗ due to a change in 𝑧∗ is:

]Δ𝑦∗=(𝛿0+𝛿1+𝛿2)Δ𝑧∗
Conclusion

Thus, the change in the equilibrium value of 𝑦y due to a change in the equilibrium value of 𝑧
is equal to the long-run propensity (LRP) times the change in 𝑧:

Δ𝑦∗=𝐿𝑅𝑃×Δ𝑧∗

Where 𝐿𝑅𝑃=𝛿0+𝛿1+𝛿2

TASK 2
The general fertility rate (gfr) is the number of children born to every 1,000 women of
childbearing age. For the years 1913 through 1984, the equation this model.

𝑔𝑓𝑟𝑡 = 𝛼0 + 𝛿1 𝑝𝑒𝑡 + 𝛿2𝑝𝑒𝑡−1 + 𝛿3 𝑝𝑒𝑡−2 + 𝛽4𝑤𝑤2𝑡 + 𝛽5𝑝𝑖𝑙𝑙𝑡 + 𝑢𝑡

Where gfr in terms of the average real dollar value of the personal tax exemption (pe)
and twobinary variables. The variable ww2 takes on the value unity during the years
1941 through 1945,when the United States was involved in World War II. The variable
pill is unity from 1963onward, when the birth control pill was made available for
contraception
[i]. Use FERTIL3 data to Estimate the above equation. Interprets the estimates.
[ii].Find the Long run propensity and test it to see whether it is significant.
[iii]. Considering the new model with time as a linear dependent

𝑔𝑓𝑟𝑡 = 𝛼0 + 𝛿1 𝑝𝑒𝑡 + 𝛽2𝑤𝑤2𝑡 + 𝛽3𝑝𝑖𝑙𝑙𝑡 + 𝛾0𝑡 + 𝑢𝑡

How do you compare this estimate with the standard static model presented in the
slides?
What difference do you draw while adding the time as a linear dependent variable?
[iv]. Considering the model with time as quadratic dependent

𝑔𝑓𝑟𝑡 = 𝛼0 + 𝛿1 𝑝𝑒𝑡 + 𝛽2𝑤𝑤2𝑡 + 𝛽3𝑝𝑖𝑙𝑙𝑡 + 𝛾0𝑡 + 𝛾1 𝑡^2 + 𝑢𝑡

How do you compare this estimate with the standard static model presented in slides
and the model in [iii] part? What difference do you draw while adding the time as a
quadratic dependent variable?
Part 1

Here's the regression equation with the coefficient values included:

gfr𝑡=95.8705+0.0726718×pe𝑡−0.0057796×pe𝑡_1+0.0338268×pe𝑡_2−22.1265×ww2𝑡−
31.30499×pill𝑡+𝑢𝑡

N=70 , R^2=0.4986

Interpreting the Coefficients

1. Intercept (𝛼0=95.8705):

 The intercept represents the estimated general fertility rate when all
independent variables are zero. This includes the baseline level of fertility
rate without the influence of personal tax exemption, World War II, and
the availability of the birth control pill. However, since binary variables
cannot be zero during the specified periods, the interpretation of the
intercept alone is not practically meaningful without considering the
variables' context.

2. Coefficient of pe𝑡 (𝛿1=0.0726718):

  This coefficient indicates that a one-unit increase in the average real
dollar value of the personal tax exemption in year t is associated with an
increase of approximately 0.073 in the general fertility rate, holding other
factors constant. This positive relationship suggests that higher personal
tax exemptions may encourage higher fertility rates.

3. Coefficient of pe𝑡−1 (𝛿2=−0.0057796):

 This coefficient indicates that a one-unit increase in the average real

dollar value of the personal tax exemption in the previous year (𝑡−1) is
associated with a decrease of approximately 0.006 in the general fertility
rate, holding other factors constant. This negative relationship is very
small and suggests a minor lagged effect of tax exemption on fertility
rates.

4. Coefficient of pe𝑡−2(𝛿3=0.0338268):

 This coefficient indicates that a one-unit increase in the average real

dollar value of the personal tax exemption two years prior (𝑡−2) is
associated with an increase of approximately 0.034 in the general fertility
rate, holding other factors constant. This positive relationship suggests
that tax exemptions may have a delayed positive effect on fertility rates.

5. Coefficient of ww2𝑡(𝛽4=−22.1265):

 This coefficient indicates that during the years 1941 through 1945 (when
ww2𝑡=1), the general fertility rate was, on average, approximately 22.13
points lower than in other years, holding other factors constant. This
negative relationship reflects the impact of World War II on fertility rates.

6. Coefficient of pillt(𝛽5=−31.30499):

 This coefficient indicates that from 1963 onward (when pill), the general
fertility rate was, on average, approximately 31.30 points lower than in
previous years, holding other factors constant. This negative relationship
reflects the significant impact of the availability of the birth control pill on
reducing fertility rates.

Part 2
The new regression model is:
gfr𝑡=95.8705+0.1007191⋅pe𝑡−0.0057796⋅𝑥𝑡+0.0338268⋅𝑦𝑡−22.1265⋅ww2𝑡−3
1.30499⋅pill𝑡+ut
N=70 ,R^2=0.4986
where:
 𝑥𝑡=pe𝑡−1−pe𝑡
 𝑦𝑡=pe𝑡−2−pe𝑡
Interpretation of Coefficients
 pe: The coefficient for pe(0.1007191) indicates that a one-unit increase in
the current personal tax exemption (pe)is associated with an increase of
approximately 0.1007 in the general fertility rate (gfr). This effect is
statistically significant with a t-value of 3.38 and a p-value of 0.001,
indicating strong evidence against the null hypothesis (no effect).
 x: The coefficient for x (-0.0057796) suggests that the effect of the
difference between the lagged value of pe (pe𝑡−1) and the current pepe is
 essentially zero, as indicated by the t-value of -0.04 and a p-value of
0.970. This coefficient is not statistically significant.
 y: The coefficient for y (0.0338268) suggests that the effect of the
difference between the two-period lagged value of pe (pe𝑡−2) and the
current pepe is also close to zero, with a t-value of 0.27 and a p-value of
0.790. This coefficient is not statistically significant.
 ww2: The coefficient for the World War II dummy variable (ww2) is -
22.1265, indicating that during the years 1941-1945, the general fertility
rate (gfr) was, on average, 22.1265 units lower compared to other years.
This effect is statistically significant with a t-value of -2.06 and a p-value
of 0.043.
 pill: The coefficient for the birth control pill dummy variable (pill) is -
31.30499, suggesting that from 1963 onward, the general fertility rate
(gfr) was, on average, 31.305 units lower. This effect is highly
statistically significant with a t-value of -7.86 and a p-value of 0.000.
Testing the Significance of pepe
From the regression output, the coefficient for pepe is 0.1007191with a standard
error of 0.0298027.
The hypothesis tests are as follows:
 Null Hypothesis (𝐻0): δ1=0
 Alternative Hypothesis (𝐻1): 𝛿1≠0
The t-statistic for testing 𝛿1δ1 is:
𝑡=Coefficient\Standard Error
=0.1007191\0.0298027
=3.38
The corresponding p-value is 0.001.
Conclusion:
 t-value: 3.38
 p-value: 0.001
Since, the t value is greater than the critical value(1.96) and the p-value is 0.001,
which is less than 0.05(the common significance level), we reject the null
hypothesis 𝐻0.
This indicates that the coefficient for pe that is LRP is statistically significant.

PART 3
Comparison of the Standard Static Model and the New Model with Time as
a Linear Variable
Standard Static Model:
gfr𝑡=95.8705+0.0726718⋅pe𝑡−0.0057796⋅pe𝑡−1+0.0338268⋅pe𝑡−2−22.1265⋅w
w2𝑡−31.30499⋅pill𝑡+𝑢𝑡
New Model with Time as a Linear Variable:
gfr𝑡=111.7694+0.2788778⋅pe𝑡−35.59228⋅ww2𝑡+0.9974479⋅pill𝑡−1.149872⋅𝑡+𝑢
𝑡
Key Differences and Interpretations:
1. Significance of Variables:
 In the standard static model, only the variables ww2 and pill are
statistically significant.
 In the new model, pe, ww2, and t are statistically significant, while
pill is not.
2. Coefficients Comparison:
 Personal Tax Exemption (pe):
  Standard Model: Coefficient = 0.0726718 (Not significant)
 New Model: Coefficient = 0.2788778 (Significant, p-value <
0.000)
 Interpretation: The effect of pe on the general fertility rate
becomes more pronounced and statistically significant when
the time variable is included. This suggests that over time,
the impact of the personal tax exemption on fertility rates
has become clearer.
 World War II (ww2):
 Standard Model: Coefficient = -22.1265 (Significant, p-
value = 0.043)
 New Model: Coefficient = -35.59228 (Significant, p-value <
0.000)
 Interpretation: The negative impact of World War II on
fertility rates is stronger when accounting for time,
indicating that the war had a more profound effect when
considering temporal trends.
 Birth Control Pill (pill):
 Standard Model: Coefficient = -31.30499 (Significant, p-
value < 0.000)
 New Model: Coefficient = 0.9974479 (Not significant, p-
value = 0.874)
 Interpretation: The significant negative effect of the
introduction of the birth control pill on fertility rates seen in
the static model is not observed in the model that includes
time, suggesting that the time trend captures much of the
pill's impact.
 Time (t):
 New Model: Coefficient = -1.149872 (Significant, p-value <
0.000)
  Interpretation: The negative coefficient for time indicates a
general declining trend in fertility rates over the years 1913
to 1984, reflecting broader social and economic changes.
3. Model Fit:
 R-squared:
 Standard Model: 0.4986
 New Model: 0.6622
 Interpretation: The model fit improves significantly when
time is included, indicating that time accounts for a
substantial portion of the variation in fertility rates.
Conclusion:
Including time as a linear variable in the model reveals several important
insights:
 The impact of personal tax exemptions on fertility rates is more
significant and pronounced over time.
 The general trend in fertility rates is declining over the period studied.
 The previously significant effect of the birth control pill in the static
model is not observed when time is included, suggesting temporal factors
play a crucial role.

Part 4
Comparison of Models: Static, Linear Time, and Quadratic Time

1. Standard Static Model:

gfr𝑡=95.8705+0.0726718⋅pe𝑡−0.0057796⋅pe𝑡−1+0.0338268⋅pe𝑡−2−22.1265⋅w
w2𝑡−31.30499⋅pill𝑡+𝑢𝑡

2. Model with Time as Linear Variable:

gfr𝑡=111.7694+0.2788778⋅pe𝑡−35.59228⋅ww2𝑡+0.9974479⋅pill𝑡−1.14987
2⋅𝑡+𝑢𝑡

3. Model with Time as Quadratic Variable:

gfr𝑡=124.0919+0.3478126⋅pe𝑡−35.88028⋅ww2𝑡−10.11972⋅pill𝑡−2.531426⋅𝑡+0.
0196126⋅𝑡2+𝑢𝑡
Key Differences and Interpretations:
Coefficients and Significance:
1. Personal Tax Exemption (pe):
 Standard Model: 0.0726718 (not significant, 𝑝=0.565)
 Linear Time Model: 0.2788778 (significant, 𝑝<0.000)
 Quadratic Time Model: 0.3478126 (significant, 𝑝<0.000)
 Interpretation: The effect of personal tax exemption becomes more
pronounced and statistically significant as the model includes
linear and then quadratic time trends, suggesting that the impact of
pe is better captured with time considerations.
2. World War II (ww2):
 Standard Model: −22.1265 (significant, 𝑝=0.043)
 Linear Time Model: −35.59228 (significant, 𝑝<0.000)
 Quadratic Time Model: −35.88028(significant, 𝑝<0.000)
 Interpretation: The negative impact of World War II on fertility
rates becomes stronger and remains significant across models,
indicating a consistent and substantial decrease in fertility rates
during the war years.
3. Birth Control Pill (pill):
 Standard Model: −31.30499 (significant, 𝑝<0.000)
  Linear Time Model: 0.9974479 (not significant, 𝑝=0.874)
 Quadratic Time Model: −10.11972 (not significant, 𝑝=0.115)
 Interpretation: The initially significant negative effect of the birth
control pill observed in the static model disappears when time is
considered, suggesting the influence of the pill might be
overshadowed by broader temporal trends.
4. Time (t):
 Linear Time Model: −1.149872(significant, 𝑝<0.000)
 Quadratic Time Model: −2.531426 (significant, 𝑝<0.000)
 Interpretation: There is a significant negative trend in fertility rates
over time, and this effect is even more pronounced in the quadratic
model.
5. Quadratic Time (tsq):
 Quadratic Time Model: 0.0196126(significant, 𝑝<0.000)
 Interpretation: The positive coefficient for 𝑡2 indicates a
decelerating decline in fertility rates over time, suggesting that the
decline slows down in later years.
Model Fit:
 Standard Model R-squared: 0.4986
 Linear Time Model R-squared: 0.6622
 Quadratic Time Model R-squared: 0.7267
Interpretation: The model fit improves significantly when time is included, and
even more so when quadratic time is added. This indicates that time trends (both
linear and quadratic) capture a substantial portion of the variation in fertility
rates.

Conclusion:
Adding time as a linear variable improves the model fit and reveals significant
trends in fertility rates over time. The impact of personal tax exemption
becomes clearer and more significant. Introducing a quadratic term for time
further enhances the model, capturing a decelerating decline in fertility rates,
suggesting that fertility rates decrease over time but the rate of decline slows
down. The inclusion of temporal variables demonstrates that the static model
does not fully capture the dynamic nature of fertility rates across the studied
period.
To determine whether autocorrelation exists in your regression model, you can
use the Durbin-Watson (DW) statistic. The formula 𝑑=2(1−𝜌) relates the
Durbin-Watson statistic to the autocorrelation coefficient 𝜌. In this context, ρ is
the coefficient of 𝑢_1 in the regression of u on 𝑢1.
Using the formula 𝑑=2(1−𝜌):
D=2-2p
D=2-2(0.875)
D=0.23

Interpretation of the Durbin-Watson Statistic

The DW statistic ranges from 0 to 4:
 𝑑=2 indicates no autocorrelation.
 𝑑<2 indicates positive autocorrelation.
 𝑑>2 indicates negative autocorrelation.
In this case:
 𝑑=0.23 is significantly less than 2, indicating a strong positive
autocorrelation in the residuals.
PART 2
To determine whether autocorrelation exists in your regression model, you can
use the Durbin-Watson (DW) statistic. The formula 𝑑=2(1−𝜌) relates the
Durbin-Watson statistic to the autocorrelation coefficient 𝜌. In this context, ρ is
the coefficient of 𝑢_1 in the regression of u on 𝑢1.
Using the formula 𝑑=2(1−𝜌):
D=2-2p
D=2-2(0.9317138)
D=0.14
Interpretation of the Durbin-Watson Statistic
The DW statistic ranges from 0 to 4:
 𝑑=2 indicates no autocorrelation.
 𝑑<2 indicates positive autocorrelation.
 𝑑>2 indicates negative autocorrelation.
In this case:
 𝑑=0.14 is significantly less than 2, indicating a strong positive
autocorrelation in the residuals.