THE BEHAVIOUR OF MONETARY AGGREGATES IN SINGAPORE

Abstract

This paper aims to model the behaviour of Singapore's monetary aggregates through adopting two commonly-used approach: the component approach and aggregation approach. These approaches will couple with time-series (ARIMA) and regression modelling techniques to achieve four parsimonious models for each of the three money measures: MI, M2 and M3. The relative abilities of the approaches and modelling techniques will be compared. The postulations or common beliefs that the component approach and time-series technique yield better forecasts for the multipliers over that of the aggregation approach and regression technique respectively will be examined. The three money multipliers will also be tested for stability and subsequently, using cointegration analysis to examine whether these structural breaks, if any, are just temporary.

Using one-step ahead ex-post predictions for the period of 1990 to 1994, it was found that all the models obtained are as competent in that all offer rather high predictive power with error margin of between 1 to 5%. There is also no evidence to support the supposition that the component approach is superior to the -aggregation approach, though regression models (error-correction) does provide better estimates than the time-series model. For M2 and M3, the best selected model belongs to the regression model (error-correction) of the aggregation approach with a predictive accuracy of up to 98%. For Ml, it is the behavioural regression model of the component approach that offers the least error margin of about 3%. Among the models, M2 yields the best estimates. It was also found that there exist structural break in all the multipliers. However, these structural breaks are nevertheless just temporary, with all the multipliers displaying a tendency towards a long-run equilibrium. Another finding inferred from this study offers alternative policy option to monetary targeting using the determinants of the components of multipliers, such as interest rates, rather than using monetary base alone.