26/07/2024, 08:39 Untitled16.

ipynb - Colab

a.)

1
2
3 import numpy as np
4 import pandas as pd
5 import matplotlib.pyplot as plt
6
7 # Sample data (replace with actual data from the file)
8
9 df = pd.read_excel("Test6.xlsx")
10 print(df.head())
11
12 # Handling 'NA' values (replace with appropriate method)
13 df.fillna(method='bfill', inplace=True)
14
15 # Create time series plots
16 plt.figure(figsize=(12, 8))
17
18 # CPI of Euro area and USA
19 plt.subplot(2, 2, 1)
20 plt.plot(df['YYYY-MM'], df['CPI_EUR'], label='Euro Area')
21 plt.plot(df['YYYY-MM'], df['CPI_USA'], label='USA')
22 plt.title('CPI')
23 plt.legend()
24
25 # Log(CPI) of Euro area and USA
26 plt.subplot(2, 2, 2)
27 plt.plot(df['YYYY-MM'], df['LOGPEUR'], label='Euro Area')
28 plt.plot(df['YYYY-MM'], df['LOGPUSA'], label='USA')
29 plt.title('Log(CPI)')
30 plt.legend()
31
32 # Monthly inflation (DP) of Euro area and USA
33 plt.subplot(2, 2, 3)
34 plt.plot(df['YYYY-MM'], df['DPEUR'], label='Euro Area')
35 plt.plot(df['YYYY-MM'], df['DPUSA'], label='USA')
36 plt.title('Monthly Inflation (DP)')
37 plt.legend()
38
39 plt.tight_layout()
40 plt.show()
41
42 # Conclusions (replace with your own analysis based on the plots)
43 print("Conclusions from the plots:")
44 print("- Both CPI and log(CPI) show similar trends for Euro area and USA.")
45 print("- Monthly inflation (DP) seems to be more volatile.")
46 print("- Further analysis is needed to determine the predictive power of US inflation on Euro ar
47

https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 1/6
26/07/2024, 08:39 Untitled16.ipynb - Colab

YYYY-MM TREND CPI_EUR CPI_USA LOGPEUR LOGPUSA DPEUR DPUSA

0 2000M01 1 105.1 107.6 4.654912 4.678421 NaN NaN
1 2000M02 2 105.4 108.3 4.657763 4.684905 0.002850 0.006485
2 2000M03 3 105.8 109.1 4.661551 4.692265 0.003788 0.007360
3 2000M04 4 105.9 109.2 4.662495 4.693181 0.000945 0.000916
4 2000M05 5 106.0 109.3 4.663439 4.694096 0.000944 0.000915

Conclusions from the plots:

- Both CPI and log(CPI) show similar trends for Euro area and USA.
- Monthly inflation (DP) seems to be more volatile.
- Further analysis is needed to determine the predictive power of US inflation on Euro area in

b.)

https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 2/6
26/07/2024, 08:39 Untitled16.ipynb - Colab
1
2
3 from statsmodels.tsa.stattools import adfuller
4
5 # Perform ADF test for Euro area log(CPI)
6 result_eur = adfuller(df['LOGPEUR'], regression='ct', autolag='AIC', maxlag=3)
7 print("ADF Test for Euro Area Log(CPI):")
8 print("t-value of Coefficient of log(CPIt-1):", result_eur[0])
9 print("Critical t-value:", result_eur[4]['5%'])
10 if result_eur[0] < result_eur[4]['5%']:
11 print("Conclusion: Reject null hypothesis, time series is stationary.")
12 else:
13 print("Conclusion: Fail to reject null hypothesis, time series is non-stationary.")
14
15 print("\n")
16
17 # Perform ADF test for USA log(CPI)
18 result_usa = adfuller(df['LOGPUSA'], regression='ct', autolag='AIC', maxlag=3)
19 print("ADF Test for USA Log(CPI):")
20 print("t-value of Coefficient of log(CPIt-1):", result_usa[0])
21 print("Critical t-value:", result_usa[4]['5%'])
22 if result_usa[0] < result_usa[4]['5%']:
23 print("Conclusion: Reject null hypothesis, time series is stationary.")
24 else:
25 print("Conclusion: Fail to reject null hypothesis, time series is non-stationary.")
26

ADF Test for Euro Area Log(CPI):

t-value of Coefficient of log(CPIt-1): -3.7763913501233537
Critical t-value: -3.4418722820477714
Conclusion: Reject null hypothesis, time series is stationary.

ADF Test for USA Log(CPI):

t-value of Coefficient of log(CPIt-1): -2.8249430667019966
Critical t-value: -3.442098279332953
Conclusion: Fail to reject null hypothesis, time series is non-stationary.

c.) Finding correlations first

1
2 import statsmodels.api as sm
3
4 # Perform OLS regression
5 X = sm.add_constant(df['LOGPUSA']) # Add a constant to the independent variable
6 model = sm.OLS(df['LOGPEUR'], X)
7 results = model.fit()
8
9 # Calculate residuals
10 residuals = results.resid
11
12 # Perform ADF test on residuals
13 result_residuals = adfuller(residuals, regression='ct', autolag='AIC', maxlag=3)
14 print("\nADF Test for Residuals:")
15 print("t-value of Coefficient of Residuals(t-1):", result_residuals[0])
16 if result_residuals[0] < -3.8:
17 print("Conclusion: Reject null hypothesis, residuals are stationary. Series are cointegrated.
18 else:
19 print("Conclusion: Fail to reject null hypothesis, residuals are non-stationary. Series are n
20

ADF Test for Residuals:

t-value of Coefficient of Residuals(t-1): -3.4244571111192053
Conclusion: Fail to reject null hypothesis, residuals are non-stationary. Series are not coint

https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 3/6
26/07/2024, 08:39 Untitled16.ipynb - Colab

1
2 import pandas as pd
3 import matplotlib.pyplot as plt
4 from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
5
6 # Calculate and plot sample autocorrelations
7 plt.figure(figsize=(12, 4))
8 plot_acf(df['DPEUR'].dropna(), lags=20)
9 plt.title('Sample Autocorrelations of DPEUR')
10 plt.show()
11
12 # Calculate and plot sample partial autocorrelations
13 plt.figure(figsize=(12, 4))
14 plot_pacf(df['DPEUR'].dropna(), lags=20)
15 plt.title('Sample Partial Autocorrelations of DPEUR')
16 plt.show()
17
18 # Fit the AR model
19 X = sm.add_constant(pd.concat([df['DPEUR'].shift(6), df['DPEUR'].shift(12)], axis=1).dropna())
20 y = df['DPEUR'].iloc[12:] # Adjust index to match lagged variables
21 ar_model = sm.OLS(y, X)
22 ar_results = ar_model.fit()
23
24 # Print the model summary
25 print(ar_results.summary())
26

https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 4/6
26/07/2024, 08:39 Untitled16.ipynb - Colab

<Figure size 1200x400 with 0 Axes>

<Figure size 1200x400 with 0 Axes>

OLS Regression Results

==============================================================================
Dep. Variable: DPEUR R-squared: 0.494
Model: OLS Adj. R-squared: 0.486
Method: Least Squares F-statistic: 63.03
Date: Fri, 26 Jul 2024 Prob (F-statistic): 8.03e-20
Time: 03:08:03 Log-Likelihood: 600.92
No. Observations: 132 AIC: -1196.
Df Residuals: 129 BIC: -1187.
Df Model: 2
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
const 0.0002 0.000 0.883 0.379 -0.000 0.001
DPEUR 0.2130 0.071 3.020 0.003 0.073 0.353
DPEUR 0.6544 0.076 8.593 0.000 0.504 0.805
==============================================================================
Omnibus: 11.666 Durbin-Watson: 1.600
Prob(Omnibus): 0.003 Jarque-Bera (JB): 22.704
https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 5/6
26/07/2024, 08:39
qUntitled16.ipynb - Colab
Skew: -0.328 Prob(JB): 1.17e-05
Kurtosis: 4.923 Cond. No. 389.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

d.)

1 # Fit the AR model

2 X = sm.add_constant(pd.concat([df['DPEUR'].shift(6), df['DPEUR'].shift(12), df["DPUSA"].shift(1
3 y = df['DPEUR'].iloc[12:] # Adjust index to match lagged variables
4 d_model = sm.OLS(y, X)
5 d_results = d_model.fit()
6
7 # Print the model summary
8 print(d_results.summary())

OLS Regression Results

==============================================================================
Dep. Variable: DPEUR R-squared: 0.616
Model: OLS Adj. R-squared: 0.601
Method: Least Squares F-statistic: 40.41
Date: Fri, 26 Jul 2024 Prob (F-statistic): 1.25e-24
Time: 03:08:03 Log-Likelihood: 619.08
No. Observations: 132 AIC: -1226.
Df Residuals: 126 BIC: -1209.
Df Model: 5
Covariance Type: nonrobust
==============================================================================
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
const 0.0003 0.000 1.175 0.242 -0.000 0.001
DPEUR 0.2081 0.071 2.913 0.004 0.067 0.349
DPEUR 0.6957 0.078 8.907 0.000 0.541 0.850
DPUSA 0.2270 0.049 4.606 0.000 0.130 0.325
DPUSA -0.0519 0.054 -0.967 0.335 -0.158 0.054
DPUSA -0.2451 0.053 -4.635 0.000 -0.350 -0.140
==============================================================================
Omnibus: 10.325 Durbin-Watson: 1.980
Prob(Omnibus): 0.006 Jarque-Bera (JB): 14.416
Skew: 0.421 Prob(JB): 0.000741
Kurtosis: 4.382 Cond. No. 488.
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

The p-value of DPUSA with lag 6 is very high, hence we can reject it and fit the model with remaining parameters

1 # Fit the AR model

2 X = sm.add_constant(pd.concat([df['DPEUR'].shift(6), df['DPEUR'].shift(12), df["DPUSA"].shift(1
3 y = df['DPEUR'].iloc[12:] # Adjust index to match lagged variables
4 d_model = sm.OLS(y, X)
5 d_results = d_model.fit()
6
7 # Print the model summary
8 print(d_results.summary())

OLS Regression Results

==============================================================================
Dep. Variable: DPEUR R-squared: 0.613
Model: OLS Adj. R-squared: 0.601
Method: Least Squares F-statistic: 50.31
Date: Fri, 26 Jul 2024 Prob (F-statistic): 2.60e-25

https://colab.research.google.com/drive/1leKs6__GCNSCY7yCsn-SaZt6j2crI5dr?authuser=0#scrollTo=m0EbDkQf-Vog&printMode=true 6/6