EMPIRICAL EXERCISE
INSTRUCTIONS: This exercise is to help deepen your understanding of the concepts and tools you learnt during the course. It will amount to 30% of your continuous assessment mark. The mark will be awarded on the basis of the spirit in which you have attempted the exercises. It is not necessary that your answers be absolutely correct but that you have demonstrated a genuine effort in answering the assigned empirical exercises. You are expected to analyse data EXCEL software of your choice and write up an empirical report explaining your work and the interpretation of your results.
The empirical project should be at least 10 pages (not including the title page, bibliography and up to two pages of tables and/or graphs). The project must be word processed and double spaced using 12 point font (use 10 point for tables) and margins of at least an inch at top/bottom/left/right. Projects which violate these constraints will be penalized by 5 points (out of the 30 in total!).
EMPIRICAL PROJECT QUESTION:
Visit the UNCTAD Statistics page
http://unctadstat.unctad.org/wds/TableViewer/tableView.aspx
and download annual GDP data from 1970 to 2014 for any two of the following countries: UK and France; Australia and Canada. Perform the follow tasks on your chosen pair of countries.
A. Time series properties of GDP series
A1. Plot the levels of GDP of your two chosen countries against time and explain whether each of the time series is stationary or not.
A2. Plot the sample autocorrelation function of the time series and comment on them.
B. Testing for Unit Root in Time Series
B1. Following an appropriate model selection procedure, conduct Dickey Fuller (DF) and/or augmented Dickey-Fuller (ADF) t-type tests for the time series of your chosen countries(x and y), and explain your findings. You should also use the Dickey-Fuller 1, 2, and 3 F- type tests to support your findings. (N. B. You need to place in the appendix the actual regression model(s) which was (were) estimated to obtain your test statistics.)
B2. Discuss the main limitations of these unit root test procedures.
C. Estimating a long-run cointegration equation and testing for cointegration
C1. Using the full sample of observations estimate by OLS the cointegration regression equation of y on a constant and x.
C2. From the results of your unit root tests in section B is there any possibility of finding a cointegration relationship between country x and y?
C3. Using the Engel-Granger technique test for cointegration between country x and y. Explain your findings.
C4. Explain what is meant by "spurious regression" and discuss its main consequences.
D. Estimating a dynamic error-correction model using the second step in the Engle-Granger two-step procedure
D1. Using the largest sample of observations possible, set up and then estimate by OLS a general dynamic model in error correction form, as suggested by the Engle-Granger two-step procedure. Call the residuals from C1 "ECMt" (t=1,..,T) and use these as estimates of the disequilibrium errors (the error correction term). Include five lags of Yt and Xt in the initial, unrestricted dynamic model, which should therefore be of the form:
In all subsequent regressions, keep the same sample period as in step D1.
D2. Comment on the meaning, interpretation, plausibility and statistical significance of the parameter or its OLS estimate.
D3. Carry out a set of misspecification tests to assess the statistical adequacy of the final, restricted model and discuss the results.
D4. Discuss the main advantages and disadvantages of the Engle-Granger two-step procedure for testing for co-integration. Outline an alternative procedure that you might use to test for the number of cointegrating relationships among a set of time series.
Attachment:- Assignment File.rar