Abstract
Methodological controversies on the relative merits of estimating time-series models with alternative approaches exist. However, applied researchers may eventually opt for an easy to use and computationally less demanding approach. This article shows that the LSE/Hendry approach, known as GETS, is promising when it is implemented with PcGets. To illustrate we report the GETS and the Johansen co-integration-based ECM equations, with PcGets, of the demand for money for Fiji.
Notes
1 Microfit, TSP (Citation1999) and EViews (Citation2004) were used to find that the variables viz., real money, real GDP and the nominal rate of interest are non-stationary in levels but stationary in first differences. The ADF and Phillips–Perron tests from Microfit, the ADF(WS) test of Pantula et al . (Citation1994) from TSP and the Elliott–Rothenberg–Stock test from EViews were used. One limitation in PcGets is that it has no unit root tests.
2 There are two alternative methods of estimating the GETS versions of equations with PcGets. First, as in the user manual, the long-run relationship can be estimated by running the regressions in the levels of the variables and deriving the static equilibrium equation. The one-period residuals of this equilibrium relationship, with an adjustment to the intercept, are used in the second stage to estimate the short run ECM model. An alternative approach, used by us, is to retain the one-period lagged values of the dependent and independent variables, by treating them as fixed variables and then follow the same procedure as earlier.