ECON7350: Applied Econometrics for Macroeconomics and Finance
Research Project 1
Due date: 27 April 2020, 11:59pm
Instruction
Answer all questions following a similar format to the solutions of your tutorial ques- tions and clearly label all your answers. We suggest that you attempt to solve all empirical questions using Stata. However, you should be able to use another statisti- cal software (e.g. Eviews or Python) without a problem. If you do choose to work with an alternative software, please note that support for software-specific issues from the course coordinator and tutors may be very limited.
Please upload your report via the “Turnitin” submission link (in the “Assessment / Research Project 1” folder). Please note that hard copies will not be accepted. At the moment, the due date is 11:59 PM on 27 April 2020, but please check BlackBoard regularly for announcements regarding any changes to this. Your report should be a write-up of your answers (in PDF format, single-spaced, and in 12 font size)1.
You are allowed to work on this assignment in groups, i.e., you can discuss how to answer the questions with your classmate(s). However, this is not a group assignment, which means that you must answer all the questions in your own words and submit your report separately. The marking system will check the similarity, and UQ’s student integrity and misconduct policies on plagiarism strictly apply.
Questions
The dataset project1data.csv for this research project is from Ganegodage and Rambaldi (2014), “Economic consequences of war: Evidence from Sri Lanka”, Journal of Asian Economics. The following variables are included in the dataset:
• year is the period 1961-2014.
• lnyt is the log of GDP per labor unit.
• lnkt is the log of physical capital stock per labor unit.
• lnot is the log of openness (i.e., the ratio of (Exports+Imports) to GDP).
• lnwt is the log of the War Index.
• d0 is to represent the period with most inward looking economic policies.
1You do not need to attach your Stata do-file and log-file, or any other specific software output. 1
• d01 is to represent the economic slow down of 2001 due to electricity crisis com- bined with intensity in the war.
• d77 is a dummy for the period 1978-2008 when more open economic policies were pursued.
(a) (2 points) Load the data to Stata and declare the data as time series using year. Draw the time series plot of {lnkt}. Comment on the stationarity of {lnkt}.
(b) (2 points) Compute and plot the ACF and PACF of {lnkt}. Comment on your findings. Hint: Focus on the implication for the ARMA model specification.
(c) (2 points) Use AIC and BIC to select an ARIMA(p, d, q) model. Estimate the AR and MA parameters and report the estimated model.
(d) (3 points) Draw a time series plot of the residuals you obtain via estimating the ARIMA model selected in (c). Run the Ljung-Box test (at significance level ↵ = 5%) for the white noise hypothesis and report test results. Do you think the ARIMA is adequate? Explain your answer.
(e) (2 points) Test if processes {lnkt}, {lnyt}, {lnwt} and {lnot} have unit roots. For each process, implement the Augmented Dickey-Fuller (ADF) tests to regression equations (4.20)–(4.22) in Enders pp. 206. Report your test results (i.e., test statis- tics, critical values and/or p-values, etc.) as well as rejection decision.
(f) (2 points) Use BIC to choose a best fitting ARDL(p, q, l, m) model of the form:
a(L)lnyt = a0 + 0t + 1D01 + 2D0 + 3D77 + ✓(L)lnkt + (L)lnwt + (L)lnot + ✏t
Report the estimated modelHint: Just try models with max{p, q, l, m} = 3.
(g) (2 points) Write out the ECM representation of the ARDL model estimated in
Part (f). Comment on your findings.
(h) (3 points) Use the Engle-Granger testto test if (lnkt, lnyt, lnwt, lnot) are cointe- grated. Report your test results and rejection decision. Explain your answer. Hint: In Stata, it helps to install the egranger package and use its egranger command.
(i) (2 points) Estimate a first differences model of the form:
lnyt = 0 +⌘1D01 +⌘2D0 +⌘3D77 + 1 lnkt + 2 lnwt + 3 lnot +et
where is the difference operator, i.e. xt = (1 L)xt = xt xt 1. Report the results for the estimated model.
2