This is a group assignment where you are allowed to work in groups of 1-4 other students. All group members will receive the same marks for the assignment.  You must submit an electronic copy of your  assignment in Canvas in pdf, doc or docx format along with your R-code. Hard copies will not be accepted. Show your calculations (if any) as well as answering the questions in clear full sentences. The number of   tables, graphs, calculations given in parentheses after each question are a guide.

What determines the share of population who cannot afford a healthy diet?

For this home assignment you will be required to model the share of population worldwide who cannot afford a healthy diet.

Please use the file: Nutrition_data.Rdata (World Bank Database – 2017 values). Please read the description  at the end of this document to understand the variables. In this home assignment we are going to model the share of the population whose food budget is below the cost of a healthy diet (dependent variable).

QUESTION 1

Please model the share of the population whose food budget is below the cost of a healthy diet (who cannot afford a healthy diet) using R:

a)   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) DO NOT USE regional dummies in this specification!

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

•    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.

Use OLS standard errors.

(Subtotal: 10 marks) 1 Table [regression output] & Explanations, 6 scatter plots

b) Interpret the coefficients on 5 explanatory variables. Describe if the coefficients are elasticities or semi-elasticities, or simple level coefficients.

(5 marks)

c) Interpret the statistical significance of these coefficients using the p-values OR the t-stat. (5 marks)

d)   Test for heteroscedasticity in R using the Breusch-Pagan test and copy below the results. Interpret the results of the Breusch Pagan test.

(2 marks) 1 Table & Explanation

e)   Present the results from a) using HAC robust errors! Did any of the standard errors change significantly?

(3 marks) 1 Table & Explanations

f)   Explain MLR 4, and whether this is likely to hold in your model? Can you identify any omitted variables? (3 marks)

g)   Please add regional dummies (Region) to the regression you estimated in section (a), and interpret the coefficients on the regional dummies, along with their levels of significance.   (3 marks) New regression table, with dummy interpretation.

Test the joint significance of the regional dummies in the model! You can either perform the test        manually and show your work down below, OR do it in R and copy the solution code and output here.

(2 marks for the output & 2 mark for the interpretation)

(Subtotal: 7 marks)

h)   Explain the basic concept of an instrumental variable estimation, and when to use it. (2 marks)

. i) Explain the practical/ policy relevance of your results. Describe a minimum of 2 policy proposals [

each 2-3 sentences] aiming at reducing the share of population that cannot afford a healthy diet!

(3 marks)

Assignment total: 40 marks

VARIABLE DESCRIPTIONS: Source: www.worldbank.org