Name: Tuneer Bagchi
ID: M230401027
Chapter 10 (Computer Excercise)
C1:
OLS Regression Results
==============================================================================
Dep. Variable: i3 R-squared: 0.668
Model: OLS Adj. R-squared: 0.649
Method: Least Squares F-statistic: 34.84
Date: Fri, 27 Jun 2025 Prob (F-statistic): 1.74e-12
Time: [Link] Log-Likelihood: -107.11
No. Observations: 56 AIC: 222.2
Df Residuals: 52 BIC: 230.3
Df Model: 3
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
const 1.3401 0.417 3.214 0.002 0.504 2.177
inf 0.5848 0.076 7.690 0.000 0.432 0.737
def 0.3612 0.119 3.034 0.004 0.122 0.600
y79 1.6090 0.502 3.207 0.002 0.602 2.616
==============================================================================
Omnibus: 4.473 Durbin-Watson: 0.794
Prob(Omnibus): 0.107 Jarque-Bera (JB): 3.430
Skew: -0.527 Prob(JB): 0.180
Kurtosis: 3.599 Cond. No. 12.3
==============================================================================
Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly
specified.
Conclusion: The equation from 10.15 is:
i3t = 1.73 + 0.606inft + 0.513deft
After the policy change in 1979 and the inclusion of "y79" as dummy variable, we get
i3t = 1.34 + 0.5848inft + 0.36deft + 1.61y79
Here all the coefficients are statistically significant (p-value<0.05). The
coefficient of "y79" is 1.61, which means short-term interest rates increased by
approximately 1.61 percentage points on average in the post-1979 period. It suggest
that the change in monetary policy had a direct and measurable effect on interest
rates, beyond what can be explained by inflation expectations alone.
C2:
i)
OLS Regression Results
====================================================================================
===
Dep. Variable: lchnimp R-squared (uncentered):
0.992
Model: OLS Adj. R-squared (uncentered):
0.992
Method: Least Squares F-statistic:
2187.
Date: Sat, 28 Jun 2025 Prob (F-statistic):
1.23e-126
Time: [Link] Log-Likelihood:
-109.22
No. Observations: 131 AIC:
232.4
Df Residuals: 124 BIC:
252.6
Df Model: 7
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
lchempi -0.7085 1.219 -0.581 0.562 -3.122 1.705
lgas 0.3718 0.296 1.257 0.211 -0.213 0.957
lrtwex 0.0522 0.412 0.127 0.899 -0.763 0.868
befile6 0.0924 0.250 0.370 0.712 -0.402 0.587
affile6 0.0973 0.256 0.380 0.705 -0.410 0.605
afdec6 -0.3508 0.281 -1.247 0.215 -0.908 0.206
t 0.0128 0.004 3.432 0.001 0.005 0.020
==============================================================================
Omnibus: 8.790 Durbin-Watson: 1.556
Prob(Omnibus): 0.012 Jarque-Bera (JB): 9.127
Skew: -0.502 Prob(JB): 0.0104
Kurtosis: 3.814 Cond. No. 2.01e+03
==============================================================================
Notes:
[1] R² is computed without centering (uncentered) since the model does not contain a
constant.
[2] Standard Errors assume that the covariance matrix of the errors is correctly
specified.
[3] The condition number is large, 2.01e+03. This might indicate that there are
strong multicollinearity or other numerical problems.
Conclusion: As none of the varuables other than the trend has p-value less than
0.05, they are not statistically significant.
ii)
For joint significance of all variables except the time trend, we run restricted
model with only "t" and constant and check the F-statistic and p value as below:
"x_restricted = data[['t']]
x_restricted = sm.add_constant(x_restricted)
model_restricted = [Link](y, x_restricted).fit()
f_test_result = model.compare_f_test(model_restricted)
print(f"F-statistic: {f_test_result[0]}, p-value: {f_test_result[1]}")
F-statistic: 0.6508741207201669, p-value: 0.6613525491725993"
Concluson: An F-test comparing the full model (with all variables) to a restricted
model (with only a time trend) yields a p-value of 0.661, which is well above 0.05.
Therefore, we conclude that the group of variables excluding the time trend are not
jointly statistically significant.
iii)
After adding monthly dummy variables:
OLS Regression Results
==============================================================================
Dep. Variable: lchnimp R-squared: 0.411
Model: OLS Adj. R-squared: 0.316
Method: Least Squares F-statistic: 4.334
Date: Sat, 28 Jun 2025 Prob (F-statistic): 6.19e-07
Time: [Link] Log-Likelihood: -103.98
No. Observations: 131 AIC: 246.0
Df Residuals: 112 BIC: 300.6
Df Model: 18
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
const 27.3022 31.397 0.870 0.386 -34.907 89.512
lchempi -0.4516 1.272 -0.355 0.723 -2.971 2.068
lgas -0.8207 1.345 -0.610 0.543 -3.486 1.844
lrtwex -0.1972 0.530 -0.372 0.710 -1.246 0.852
befile6 0.1649 0.257 0.642 0.523 -0.344 0.674
affile6 0.1534 0.272 0.564 0.574 -0.386 0.692
afdec6 -0.2950 0.299 -0.985 0.327 -0.888 0.298
t 0.0123 0.004 3.151 0.002 0.005 0.020
feb -0.3554 0.294 -1.210 0.229 -0.937 0.227
mar 0.0626 0.255 0.245 0.807 -0.442 0.568
apr -0.4406 0.258 -1.705 0.091 -0.953 0.071
may 0.0313 0.259 0.121 0.904 -0.482 0.545
jun -0.2009 0.259 -0.775 0.440 -0.715 0.313
jul 0.0111 0.268 0.041 0.967 -0.521 0.543
aug -0.1271 0.268 -0.475 0.636 -0.658 0.403
sep -0.0752 0.258 -0.291 0.772 -0.587 0.437
oct 0.0798 0.257 0.310 0.757 -0.430 0.589
nov -0.2603 0.253 -1.029 0.306 -0.762 0.241
dec 0.0965 0.262 0.369 0.713 -0.422 0.615
==============================================================================
Omnibus: 9.056 Durbin-Watson: 1.422
Prob(Omnibus): 0.011 Jarque-Bera (JB): 9.116
Skew: -0.543 Prob(JB): 0.0105
Kurtosis: 3.699 Cond. No. 4.90e+04
==============================================================================
Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly
specified.
[2] The condition number is large, 4.9e+04. This might indicate that there are
strong multicollinearity or other numerical problems.
And for testing seasonality,
"x_restricted = data[['lchempi', 'lgas', 'lrtwex', 'befile6', 'affile6', 'afdec6',
't']]
x_restricted = sm.add_constant(x_restricted)
model_restricted = [Link](y, x_restricted).fit()
f_test_seasonality = model.compare_f_test(model_restricted)
print(f"F-statistic: {f_test_seasonality[0]}, p-value: {f_test_seasonality[1]}")
F-statistic: 0.8465567075431214, p-value: 0.5943260249721585"
Conclusion: Including the monthly dummies changes the values of other estimates and
all other variables other than "t" are statistically insignificant (p-value>0.05).
For seasonality, An F-test comparing the model with monthly dummies to the model
without them yields a p-value of 0.594, which is well above the 5% significance
level. This indicates that the monthly dummy variables are not jointly significant,
and there is no statistical evidence of seasonality.
C3:
OLS Regression Results
==============================================================================
Dep. Variable: lprepop R-squared: 0.889
Model: OLS Adj. R-squared: 0.876
Method: Least Squares F-statistic: 66.23
Date: Sat, 28 Jun 2025 Prob (F-statistic): 2.68e-15
Time: [Link] Log-Likelihood: 78.659
No. Observations: 38 AIC: -147.3
Df Residuals: 33 BIC: -139.1
Df Model: 4
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
const -6.6634 1.258 -5.298 0.000 -9.223 -4.104
lmincov -0.2123 0.040 -5.286 0.000 -0.294 -0.131
lusgnp 0.4860 0.222 2.190 0.036 0.034 0.938
t -0.0267 0.005 -5.763 0.000 -0.036 -0.017
lprgnp 0.2852 0.080 3.544 0.001 0.121 0.449
==============================================================================
Omnibus: 2.155 Durbin-Watson: 1.014
Prob(Omnibus): 0.340 Jarque-Bera (JB): 1.744
Skew: 0.521 Prob(JB): 0.418
Kurtosis: 2.872 Cond. No. 5.89e+03
==============================================================================
Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly
specified.
[2] The condition number is large, 5.89e+03. This might indicate that there are
strong multicollinearity or other numerical problems.
Conclusion: The variable log(prgnp) is statistically significant as the p-value =
0.001. Since it’s a log-log model, the coefficient can be interpreted as an
elasticity. Its coefficient (0.2852) implies that the elasticity of PR
employ/population ratio with respect to PR GNP is 0.2852. It means a 1% increase in
real per capita GNP (prgnp) is associated with a 0.285% increase in the
employment-population ratio, holding other factors constant. After including
log(prgnp), the negative effect of log(mincov) on the employment-population ratio
slightly weakens, but remains statistically significant. This means that higher
minimum wage coverage is still associated with lower employment, even after
accounting for changes in real per capita GNP.
