HOW TO RUN ECONOMETRIC RESEARCH ON YOUR COMPUTER

HOW TO RUN ECONOMETRIC RESEARCH ON YOUR COMPUTER

BY AGBIOKORO TIMOTHY CHITUA

(2009)

I

nstall any of these econometric software; E-View, SPSS, PcGive etc on your computer. E-View is mostly recommended because of its simplicity and easy presentation.

ORDINARY LEAST SQUARE (OLS) METHODOLOGY

Step One: Power your Personal Computer (PC) and allow the system to boot. After booting, locate the E-View programme through the Start-button on your desktop and click on it or create a short-cut to the E-View (already installed programme) and place it on your desktop so that you could easily click on the platform when the need arises.

Step Two: When the E-View platform is opened, click on file, go to new and trace a sub folder; workfile and click. A box titled “workfile range” will appear, showing annual, semi-annual, quarterly, and monthly, with a ‘start date’ and ‘end date’ option. If your data is an annual time series data, click on annual, then enter the range of your data, i.e. 1970-2006 depending on the range of your data and click OK. Then go back to the E-View platform and click on Quick and go to Empty Group (Edit Series) and enter your data on the excel spread sheet that appeared. Highlight each column entirely and name the data you wish to enter the figures yearly then press ‘Enter key’ on your keyboard.

Step Three: Go back to the Quick button and click on ‘Estimate Equation’ and an equation specification box appears, then state your dependent variable, space it, followed by the constant term and other regressors and click OK. Then your regression result will appear.

UNIT ROOT TEST

Go to the quick button, and click on ‘Series Statistics’, a sub folder will appear. Trace it and click on Unit Root Test. A box of series name will appear then enter the series name (variable) that you want to test for the presence of unit root. A unit root test box will appear, choose between using the ‘Augmented Dickey-Fuller’ method or the ‘Philips-Perron’ method and choose ‘trend’ or ‘trend and intercept’, then test for the presence of unit root in your data at level. The result will appear. Compare the ‘ADF Test-Statistic’ and the calculated ‘Critical Value’ given your significance level. Your data is only stationary at level or differencing stage, when your critical value is greater than the ADF test statistic. If it is not stationary at level, test at first differencing and second differencing.

GRANGER CASAULITY METHOD

Go to the Quick button, click on Group Statistics and trace a sub-folder ‘Granger Causality Test’ and click on it. A dialogue box of list of series, Groups, and/or series expression will appear. Then enter your variables in their order of stationarity, and then a box of lag specification will appear, choose two, and click OK. Compare the calculated F-Statistics and the tabulated F-statistics to interpret your result. If your motive is to find out the causality between the regresand and each of the regressors, then it is advisable to enter the regressand and each of the regressors independently.

JOHANSEN COINTEGRATION ANALYSIS

Go to the Quick button, click on Group Statistic and trace a sub folder, ‘Cointegration test’ and click on it. A list of series, Groups or series expression will appear, enter the series name accordingly to their stationarity level or differencing and click OK. The five set of assumptions of Johansen’s Cointegration test will appear, then make a choice among the assumptions or click to summarize the five set of assumptions. Then click OK, the result will be displayed. Look at your hypothesized no of CE(s) and choose your rejection or acceptance region given your significant level.

ERROR CORRECTION MODEL

Go to the Quick button; trace it to ‘Generate Series’ to generate the error series, when you click on it, a dialogue box of ‘Generate Series by Equation’ will appear, then enter your error correction coefficient i.e. Ecm(-1) = resid, and click OK. Go back to the ‘Quick button’ and trace it to ‘Estimate Equation’, ‘Equation Specification’ box will appear. Enter your model there according to their level of stationarity, the dependent variable first, followed by the constant and the rest of the regressors including the error correction term and click OK. The result will appear. A negative and statistically significant error correction coefficient, are necessary conditions for the error correction method to uphold the result of your Cointegration result (Granger Representative Theory).