Module: EC5606

Academic year: 2024-25

Assessment Type: Project

Submission Deadline: 9th August 2025, 12 noon UKtime, to be submitted via WISEflow

Economic Growth Regression Analysis

In this project assignment, students are required to conduct an empirical analysis of economic growth using regression techniques for the assigned country.

Submission Requirements

Students must submit the following files via WISEflow before the deadline, 9th of August, at

12:00 noon:

1.   Upload the report in pdf format for tasks in Part 2 and 3 of this brief using upload button right under “1. Paper” in WISEflow, see the red highlights below. The report should have (1) a maximum of twelve A4 pages, (2) the font Times New Roman size 12, (3) 2.54cm (1 inch) top, bottom left and right margins, (4) 1.5 line spacing.

2.   Upload the data file required in Part 1 and the software file used to complete the tasks in Part 2 and 3 using the upload button under Appendix Material in “1. Paper” in WISEflow, see the purple highlight below.

Project Brief

Part 1: Data Collection and Software Analysis (25%)

1.   Each student is allocated a different country to complete the empirical analysis for the project. Students can find the assigned country in the file “Country Allocation.xlsx” under Assessment in Brightspace.

For your assigned country, collect the following variables over the period 1961-2020 from the World Development Indicators website:

https://databank.worldbank.org/source/world-development-indicators

1)   “GDP per capita growth (annual %)” (henceforth: GDPgr)

2)    “GDP per capita (constant 2015 US$)” (henceforth: GDPpc)

3)    “Gross fixed capital formation (% of GDP)” (henceforth: GFC)

4)    “Inflation, consumer prices (annual %)” (henceforth: INFL)

5)    “Population growth (annual %)” (henceforth: POPgr)

6)    “General government final consumption expenditure (% of GDP) ” (henceforth: GOV)

7)    “Imports of goods and services (% of GDP)” (henceforth: IMP)

8)    “Exports of goods and services (% of GDP)” (henceforth: EXP)

Download the data as an Excel file with a .xlsx extension. Convert the data so that each variable should be a column vector, with the first-row entry being the variable name. The first column should be the observation number. The submitted excel data file counts to 10% of the final mark.

2.   Import the dataset into R using RStudio or into Eviews.

1)  Create a variable “Trade openness” (henceforth: OPEN) as OPEN=IMP+EXP.

2)  Create a variable “Initial GDP per capita” (henceforth: INIT) as INIT=ln(GDPpc), where “ln” denotes the natural logarithm.

3)  If there are any missing values, replace them with the nearest available observation after them. You may need to do this step several times consecutively.

Part 2 and 3 in this brief contain tasks that you need to perform in R or Eviews. The software code/analysis for all required tasks in Part 2 and 3 must be saved, reported/submitted as follows. IfRStudio is used, (1) report the used R codes in the Appendix of the project report by simply presenting all used commands with explanation comments for each line, if necessary; (2) submit a R code file (*.r) that replicates all reported results in the project report. If Eviews is used,    submit    an    Eviews    workfile    (*.wf1)     file    containing    all    adopted    objects (variables/regressions/tables/graphs)  corresponding  to  the  reported  results  in  your  project report. Please name the objects appropriately so that they can be found easily. The submitted software file (*.r or *.wf1 file) counts to 15% of the mark.

Part 2: Basic Growth Regression Analysis (35%)

3.   [10%] Preliminary Analysis:

1)    Create a graph that includes the time-series plots of all variables. Discuss the graph briefly. (5%)

2)    Create a table of summary statistics for the variables including the mean, standard deviation, minimum and maximum. Discuss your observations briefly. (5%)

Carefully report the results into a table using professional standards. Copying and pasting outputs directly from the software results in zero marks for this task. This principle follows for all task about tables.

4.    [15%] Estimate six growth regressions with GDPgr (economic growth per capita) as the dependent variable and the following independent variables in each regression:

Regression 1: GFC

Regression 2: GFC, POPgr

Regression 3: GFC, POPgr, INFL

Regression 4: GFC, POPgr, INFL, GOV

Regression 5: GFC, POPgr, INFL, GOV, OPEN

Regression 6: first lag ofINIT, GFC, POPgr, INFL, GOV, OPEN

1)    Typeset the above regressions in Microsoft Word using the equation editor carefully. Explain the variables in the equation briefly. (5%)

2)   Report the OLS estimation results for the above regression in a table using professional  standards  carefully.  The  table  should  contain  the  estimated coefficients with the statistical significance indicated by stars,  R2 , the adjusted R2 , and the F-statistic with the statistical significance indicated by stars. (10%)

Regressions must be reported side by side and statistical significance should be denoted  using  significance  stars  (*  next  to  coefficients  with  the   10% significance level, ** next to coefficients with the 5% significance level, *** next to coefficients with the  1% significance level, for two-sided t-tests of the null hypothesis that the coefficient is equal to zero). Note: use these principles in all tasks for reporting regression results below. Pasted software output will not be marked.

5.   [10%] Considering results from Regression 6:

1)    Interpret the estimated coefficients. Are the signs what you expected? Explain. (5%)

2)   Discuss the statistical significance of your estimates in the context of the economic growth. (5%)

Part 3: Extensions and Robustness (40%)

6.   [10%] Some countries’ economic growth may be affected by the Global Financial Crisis of 2007-2009. To investigate the potential impact of this event on your assigned country, amend Regression 5 from Part 2 accordingly using dummy variables. Report the regression results in a table with the same requirements in Task 4.2) and discuss the results briefly.

7.   [10%] The estimation results may be affected by autocorrelations in the residuals.

1)    Test for autocorrelation in the residuals of Regressions 5 and 6. Comment on the results. (5%)

2)   Alter Regressions 5 and 6 to deal with autocorrelation and report the results in a table. Comment on the differences between the new results and the original. (5%)

8.   [10%] Consider the possibility of nonlinearity, in the form of quadratic terms, in one or more of the variables in Regression 5. Amend Regression 5 accordingly and report the estimation  results  of that  regression  in  a  table.  Interpret  the  results.  Are  the  results reasonable?

9.   [10%] Growth regressions are frequently plagued by “simultaneity” which is a form of endogeneity. Simultaneity means that regressors are correlated with the errors at the same period, causing endogeneity.  For example, consider a sudden hike in the price of oil in a year. On the one hand, this is a shock in the error term, which affects the dependent variable economic growth, possibly reducing it. On the other hand, government spending decisions in the same year may also be affected by the price hike. Thus, the price hike can be correlated with the regressor. In this case, the error term is correlated to the regressor in the same period, which is the definition of endogeneity.

Use the time structure of the data to estimate a model based on Regression 6 that is not suffering from endogeneity. Comment on the differences between the results from this model and those from the original one of Task 4.

Key References

1.   Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A Contribution to the Empirics of Economic Growth. Quarterly Journal of Economics, 107(2), 407-437.

2.   Barro, R. J. (1991). Economic Growth in a Cross  Section of Countries. Quarterly Journal of Economics, 106(2), 407-443.

3.   Levine, R., & Renelt, D. (1992). A  Sensitivity Analysis of Cross-Country Growth Regressions. American Economic Review, 82(4), 942-963.

4.   Acemoglu,  D.,  Johnson,  S.,  &  Robinson,  J.  A.  (2001).  The  Colonial  Origins  of Comparative Development: An Empirical Investigation. American Economic Review, 91(5), 1369-1401.

5.   Islam, N. (1995). Growth Empirics: A Panel Data Approach. Quarterly Journal of Economics, 110(4), 1127-1170.

6.   Sala-i-Martin, X. (1997). I Just Ran Two Million Regressions. American Economic Review, 87(2), 178-183.

Advice

•     Start data collection early because preparing data can be time-consuming.

•    Work with the R/Eviews software early as it takes time to learn. Do not stay behind.

•    There is a plethora of sites and videos on how to do things on.

Frequently Asked Questions:

Question: Can I add some results in the appendix, but comment on them on the main text?

Answer: No.

Question: Can go above the12-page limit?

Answer: No.

Question: I did what the questions requested and unfortunately the econometric results are bad, e.g., al the signs are wrong and the variables not statistically significant. Will I fail the project?

Answer: No. It is the quality of the econometric analysis to be marked, and not the research outcome per se. Use econometric theory to justify your decisions. The results sometimes will be good sometimes bad.

Question: I know you told us that we cannot exceed the  12-page limit, but can I add just one more extra page? Or reduce the size of the font? (or anything similar?)

Answer: No.