Author/Authors :
Mourad, Mahmoud Lebanese University - Faculty of Economic Sciences and Business Administration, Lebanon
Abstract :
This paper treats the analysis of the seasonal integration of the time series that cover the Lebanese treasury in monthly data on the period 1998-2005 (96 observations). The seasonal integration procedure proposed by Franses and the ADF procedure proposed by Dickey-Fuller have been used to identify the type of seasonality and trend (stochastic or deterministic) . The choice between two models : the first M1 is first order and seasonally differenced, it requires the filter (1- B) ( 1- B^12) , the second M2 considers the variable in first difference with deterministic seasonality (it is labeled the FDSD model). The comparison of the forecasting quality between the M1 and M2 models encourages the M2 models (weak value of the MAPE criterion). A cointegration Analysis Using the Engle-Granger Procedure has succeeded to nine relations of bivariate cointegration and by consequence the existence of a long-term economic relationship from which, the ECM epresentation permits to integrate the fluctuations of short-term can be informed. The presence of a cointegration relationship between the Total cash in and the Total cash out offers to the Lebanese government an useful information on the future evolution showing an important deficit of the Lebanese treasury ( close to USD 2 billion by year).
