QUESTION 1

Please model the drivers of GDPpc using R:

Include a minimum of 5 (five) explanatory variables in the regression equation and provide a scatter plot of your dependent and independent variables (5 scatter plots). (0.5 x 5 marks)

We suggest that you take the natural log of the dependent variable!

When modeling, explain each of your functional form specification choices with respect to:

• The economic or common sense behind the model – why do you pick this variable? (0.5 x 5 marks)
• Multicollinearity – are the independent variables multicollinear? (0.5 x 5 marks)
• Functional form specification- potential nonlinear relationships, eg: log-linear or quadratic relationships. Explain why you use a level or logarithmic form of a variable. (0.5 x 5 marks)

in writing. You will be graded on model accuracy in this section. Pay attention to appropriate degrees of freedom!

Use OLS standard errors.

• (Subtotal: 10 marks) 1 Table [regression output] & Explanations, 6 scatter plots
• Interpret the coefficients on 5 explanatory variables. Describe if the coefficients are elasticities semi-elasticities, or simple level coefficients.
• Interpret the statistical significance of these coefficients using the p-values OR the t-stat.
• Test for heteroscedasticity in R using the Breusch-Pagan test and copy below the results. Interpret the results of the Breusch Pagan test.
  1 Table & Explanation
• Present the results from a) using HAC robust errors! Did any of the standard errors change significantly?
  (3 marks) 1 Table & Explanations

 

• Explain The Gauss Markov Assumptions (2.5 marks), and whether these are likely to hold in your model? (2.5 marks)
  (Subtotal: 5 marks)
• Explain the basic concept of an instrumental variable estimation, and when to use it. (2 marks)

What are the criteria for a good instrument? (2 marks)

Can any of the variables in the dataset be in a reverse causal relationship with the dependent variable? If yes, identify please. (2 marks)

What could be used as an instrument to remedy this? (1 mark)

(Subtotal: 7 marks)

• Present a functioning R code reproducing your results below!