Abstract
M.Comm.
The aim of this study is the econometric improvement of the stochastic equations of an
econometric model with specific reference made to the explanation and incorporation of
stationarity and cointegration testing. The study is based on an existing macroeconometric
forecasting model. The focus of the study is not on the improvement of the
specification of individual equations per se, but rather on the econometric improvement
thereof, therefore changes to the specification of individual equations have only been
made in cases where test results strongly recommended it.
The RAU-model had previously been exposed to neither structural stability-, stationarity-,
nor cointegration testing and therefore both the explanation and implementation of
these tests have been included in the study. It is, however, important to note that the
main purpose of both stationarity and co-integration testing is not to substitute nonstationary
data with data which is proven to be stationary, but rather to identify nonstationary
and non-cointegrationary data for future improvement and enhancement of
the RAU model.
Following the completion of the abovementioned tests, parameters have been estimated
for the individual equations of the three sectors of the RAU-model (i.e. the Real-,
Balance of payments-, and the Monetary sectors). Thereafter the results have been
evaluated on the basis of the economic-, statistic-, and econometric evaluation criteria.
In cases where econometric inconsistencies arose from the violation of the assumptions
underlying the econometric tests, appropriate transformation processes have been
applied in an attempt to resolve the problem. Thereafter, tests have been carried out to
determine the forecasting ability of the model as well as to compare the model results
with the a priori results.
In general, the aim of the study, to econometrically improve the stochastic equations of
the RAU model, has been achieved on the basis of overall better regression- and
evaluation results that have been obtained. Following the completion of the study, a
new approach to econometric modelbuilding, which makes provision for the inclusion of
both stationarity- and cointegration testing, is proposed.