PREVENTING NEGATIVE VALUES WHEN FORECASTING NON-NEGATIVE TIME SERIES VARIABLES

Abstract

Long run time series variables forecasting is of special importance to academics and professionals alike. In this paper, the disadvantage of the “Natural Logarithm” transformation that prevents generating negative values in the forecast horizon is discussed. An alternative technique that does not suffer from the disadvantage of the “Natural Logarithm” transformation is presented. Both methods have been applied for forecasting the average USD deposits rate offered in the Lebanese retail banking industry.

Keywords

• forecasting
• time series
• arima
• garch
• monte carlo simulation