You are on page 1of 15

Econometrics

Group Assignment
Group members:
Akhil S Pillai (A001)
Hemanth Sharma (A018)
Rishabh Chandel (A031)
Sahiba Kaur Arora (A037)
Vipul Tiwari (A053)
Overview of our Assignment
• Title: Statistical analysis of growth in India’s Gross
Domestic Product

• Objective: To study the hypothesis of percentage


growth in India’s Gross Domestic Product and
related explanatory variables
Brief Description
• In the assignment our focus is to understand the relationship of
India’s Gross Domestic Product with variables such as Indices
i.e Nifty 50 Turnover and Price
• The group aims to understand which factors do show relation
with the Gross Domestic Product either presumed or not
• Hence the general hypothesis set in the assignment are,
• H0: GDP is not explained by the explanatory variable tested
• H1: GDP is explained by the explanatory variable tested
Brief Description (contd.)
• The confidence interval for the exercise shall be at 95%. Hence,
the significance level, α (the probability of rejecting the null
hypothesis H0 when it is true) will remain at 5%
• There shall be a concentrated effort to take variables that are
presumed to have relation with GDP and prove it statistically
• We have explored the nuances of Stationarity, Autocorrelation
and Multicollinearity
• The software used for the analysis is Microsoft Excel and
Student version (Lite) of Eviews
• Data is sourced from independent and reliable sources like
Bloomberg and Multilateral organisations (World Bank, ADB,
Kfw etc.)
Introduction to our Analysis
• General hypotheses
• H0: GDP is not explained by the explanatory variable
tested
• H1: GDP is explained by the explanatory variable tested
• Independent Variables
• Nifty 50 Turnover (Percentage Change)
• Nifty 50 Price (Percentage Change)
• Dependent Variable
• GDP Growth (Percentage Change)
Steps Involved
• Step 1: Check for Stationarity?
• If yes, continue
• If no, difference series and check again for stationarity and
repeat
• Step 2: Check independent variables for correlation
• If yes, continue
• If no, change variables
• Step 3: Run the equation
• Step 4: Results and interpretation
• Step 5: Run additional checks for autocorrelation and
multi-collinearity
Models
• Model 1: Regression of Growth in India’s Gross
Domestic Product (% change) against Nifty 50
Turnover (% change) and Price (% change)
• Model 2: Regression of Growth in India’s Gross
Domestic Product (% change) against differentiated
Nifty 50 Turnover (% change) and Nifty 50 Price (%
change)
• Model 3: Regression of Growth in India’s Gross
Domestic Product (% change) against Nifty 50 Price
(% change)
Model 1 Output
• Equation: ‘gdp_growth__annual___ c turnover____change_
price____change_’
Model 1 Interpretation
• Equation: ‘gdp_growth__annual___ c turnover____change_
price____change_’
Interpretation
As p value is not less than 0.05, we fail to reject the Null Hypothesis (H0). Therefore,Turnover
Turnover (% change) (%change) does not impact GDP(% change)
As p value is less than 0.05, we reject the Null Hypothesis (H0). Therefore,Price (%change) does
Price (% change) impact GDP(% change)
Only 38.73% of the explanation of GDP(% change) is given by Turnover (%change) and Price
R-square (%change)
It penalises the model for additional regressors(independent variables). Therefore, it explains
that only 27.60% of the explanation of GDP(% change) is given by Turnover (%change) and Price
Adjusted R-Square (%change)
It gives the overall goodness of fit of the model. As p value is not less than 0.05, we fail to reject
the Null hypothesis and conclude that overall the model is not statistically fit to interpret Annual
Prob(F-statistic) GDP growth percentage
Durbin Watson
Statistic It is close to 2; hence shows that the variables do not have correlation
It is used to compare models by giving overall goodness of it. The model with lower AIC is better
Akaike Info Criterion compared to other.
Variance Inflation If Centered VIF > 10, multicollinearity exists. As, Centered VIF for both Turnover (%change) and
Factors Price (%change)is less than 10. We can conclude there is no multicollinearity.
Model 2 Output
• Equation: ‘gdp_growth__annual___ c d(turnover____change_)
price____change_’
Model 2 Interpretation
• Equation: ‘gdp_growth__annual___ c d(turnover____change_)
price____change_’
Interpretation
As p value is not less than 0.05, we fail to reject the Null Hypothesis (H0). Therefore,Turnover
Turnover (% change) (%change) does not impact GDP(% change)
As p value is less than 0.05, we reject the Null Hypothesis (H0). Therefore,Price (%change) does
Price (% change) impact GDP(% change)
51.45% of the explanation of GDP(% change) is given by Turnover (%change) and Price
R-square (%change)

It penalises the model for additional regressors(independent variables). Therefore, 41.74%% of


Adjusted R-Square the explanation of GDP(% change) is given by Turnover (%change) and Price (%change)
It gives the overall goodness of fit of the model. As p value is less than 0.05, we reject the Null
hypothesis and conclude that overall the model is statistically fit to interpret Annual GDP growth
Prob(F-statistic) percentage
Durbin Watson
Statistic It is close to 2; hence shows that the variables do not have correlation
It is used to compare models by giving overall goodness of it. The model with lower AIC is better
compared to other. AIC in Model 2< Model 1. Therefore, Model 2 is better than Model 1 as it is
Akaike Info Criterion a better fit
Variance Inflation If Centered VIF > 10, multicollinearity exists. As, Centered VIF for both Turnover (%change) and
Factors Price (%change)is less than 10. We can conclude there is no multicollinearity.
Model 3 Output
• Equation: ‘gdp_growth__annual___ c price____change_’
Model 3 Interpretation
• Equation: ‘gdp_growth__annual___ c price____change_’

Interpretation
As p value is less than 0.05, we reject the Null Hypothesis (H0). Therefore,Price (%change) does
Price (% change) impact GDP(% change)

R-square 38.59% of the explanation of GDP(% change) is given by Price (%change)

It penalises the model for additional regressors (independent variables). Therefore, 33.48% of
Adjusted R-Square the explanation of GDP(% change) is given by and Price (%change)
It gives the overall goodness of fit of the model. As p value is less than 0.05, we reject the Null
hypothesis and conclude that overall the model is statistically fit to interpret Annual GDP growth
Prob(F-statistic) percentage
Durbin Watson
Statistic It is close to 2; hence shows that the variables do not have correlation
It is used to compare models by giving overall goodness of it. The model with lower AIC is better
compared to other. AIC in Model 3< Model 1 and Model 2. Therefore,Model 2 is better than
Akaike Info Criterion Model 1 as it is a better fit
Variance Inflation If Centered VIF > 10, multicollinearity exists. As, Centered VIF for both Turnover (%change) and
Factors Price (%change)is less than 10. We can conclude there is no multicollinearity.
Actual GDP growth versus Fitted GDP growth*
• Equation: ‘gdp_growth__annual___ c price____change_’
• Fitted GDP growth (annual %) = c + b1*Price (% change)
• GDP growth (annual %) is given

Relationship between GDP growth rate and Fitted GDP growth rate

GDP growth (annual %) Fitted GDP growth (annual %)

12
9.266368377
Percentage Change

10 10.25996306
9.385035553
9.2848246169.2639647599.801360337 7.148243955
8 7.922943418 7.907197732 7.26329445 8.02068007
8.01005265
8.739453261 8.479783897
5.463071103 7.607419788 6.6383638
7.860381476 8.172933512
8.280302522 7.505220233
6 7.383959019 6.386106401 6.930733311
6.298658948
5.456387552 7.107034368
4 3.890957062
2
0
2003.0 2004.0 2005.0 2006.0 2007.0 2008.0 2009.0 2010.0 2011.0 2012.0 2013.0 2014.0 2015.0 2016.0
Year

*Note: Chart made from output as per Model 3


Thank You

Submitted By:
Akhil S Pillai (A001)
Hemanth Sharma (A018)
Rishabh Chandel (A031)
Sahiba Kaur Arora (A037)
Vipul Tiwari (A053)

You might also like