Call Us: US - +1 845 478 5244 | UK - +44 20 7193 7850 | AUS - +61 2 8005 4826

acceleration of real economic growth

Macro time series data such as the ones used in this study tend to exhibit either a deterministic and/or stochastic time trend and are therefore non-stationary. Engle and Granger [1987] have shown that regressing a non-stationary series against another one may lead to spurious results because conventional significance tests are likely to suggest that a relationship exists when, in fact, there is none. Unit root (nonstationarity) test were undertaken by using an Augmented Dickey-Fuller test (ADF) [Dickey-Fuller, 1981] with a constant (representing positive or negative drift) and a deterministic trend (representing a non-stationary mean). Table 3 (part A) below presents the results of running ADF test (one lag) on the variables in logarithmic form with a deterministic trend.8 The results indicate that the null hypothesis of nonstationarity cannot be rejected for any of the variables in level form with a deterministic trend, suggesting that the variables in question do appear to exhibit a non-stationary (time dependent) mean throughout the period under review. In other words, detrending the data by a single trend line will not render the data in question stationary because the trend line itself may be shifting over time [Charemza and Deadman, 1997]. However, most of the variables become stationary in first differences at the five percent level of significance (and in one case at the 10 percent level).