AN ASSESSMENT OF THE FORECASTING PERFORMANCE OF STRUCTURAL MACROECONOMIC MODELS OF THE SINGAPORE ECONOMY

Abstract

The forecasting performance of mainstream structural macroeconometric models has been on the wane since the 1970s when they show a general inability to produce reliable forecasts. Yet, such structural macro-models can be very useful, for both ex-post and ex-ante forecasting and policy simulation and analysis. Pure ex-ante forecasts generated by such models have often been taken into account by both the government and private sector in anticipation and planning for the future. The former uses it to "determine" in advance the possible future course of the economy and in turn, to adopt appropriate economic policies. Whilst the latter take into account the forecasts generated for their future investment plans. Given that the predictions produced by such models can have such important uses and economy wide implications, it is thus essential to conduct an assessment on how well they mimic actual future economic outcomes. The key objectives of this thesis are: Firstly, to access recent structural forecasting models constructed for the Singapore economy in terms of their theoretical and empirical relevance pertaining to the economy. Secondly, to assess the quantitative ex-ante forecasting performance of such models in terms of how well they forecast the actual outcomes and how well these forecasts compare with those produced by other methods of forecasting.

In Chapter 2, past developments and the current "state of the art" structural macroeconometric modelling will be reviewed.

In Chapter 3, exposition and assessment of empirical relevance of recent structural macro-models constructed for the Singapore economy will be conducted. This is to be followed by the assessment of the ex-ante forecasts generated by such models in terms of how well they mimic actual outcomes.

In Chapter 4, a comparison will be made between such macro-models and time-series models to provide insights into their relative suitability as pure ex-ante forecasting tools both in terms of their inherent characteristics and forecasting performance.

Overall, there has been general paucity of such models for the Singapore economy and the forecasting performance of such structural models constructed for the economy remains to be improved. Time-series forecasting models pose a serious alternative to such models for the pure purpose of ex-ante forecasting in the Singapore context.