Although the ordinary least square (OLS) approach treats regression relationship as deterministic whilst time series
contain stochastic trends, the approach is preferred to other estimation techniques amongst the Maximum Likelihood (ML) and Method of Moments (MoM). This is because the least square approach produces
estimates which are unbiased,
efficient and14
consistent as stated in the Gauss-Markov theorem of the classical
linear regression model (CLRM). E-views 7 statistical software package was used.
3.4 Unit root testsNon-stationary data
result in spurious regression, therefore testing for data stationarity is essential. There are several unit root tests which
include the Dickey-Fuller test,
the Augmented Dickey-Fuller (ADF) and the Phillip-Perron test but the ADF test was used in this study because it corrects for serial autocorrelation in the residuals (Wooldridge, 2008). If the absolute ADF test statistic is greater than the critical values then the series under consideration is said to be stationary. Series found to contain unit roots are differenced until they are stationary.
Share with your friends: