CONSUMPTION FUNCTION IN SINGAPORE : RATIONAL EXPECTATIONS APPROACH

Abstract

The relationship between consumption and income has been widely accepted as the core feature of macroeconomics for over a quarter of a century. Not until the recent debate, both the theoretical proposition and the econometric model of the consumption function were considered settled. Most economists adhered to either the Life Cycle Hypothesis (LCH) of Ando, Bromberg and Modigliani or the Permanent Income Hypothesis (PIH) of Friedman. Recent developments have raised fundamental issues and controversies on the consumption function. Particularly controversial is the striking result first discovered by Hall (1978). The incorporation of rational expectations into the LC-PIH framework has given rise to the Random Walk Model of Aggregate Consumption (RWMAC). The strong stochastic implication of the Random Walk Model of Aggregate Consumption is that only consumption lagged one period is able to explain this period's consumption. All other variables which are conventionally believed to affect consumption should have no explanatory power to affect this period's consumption. We propose to test the above model developed by Hall (1978) using Singapore as a case study. Our results clearly indicate that Singapore's data do not support the RWMAC. The “surprise only" test also shows that both anticipated and unanticipated income matter. That is, consumers do take into account anticipated and unanticipated changes in income. Besides, real interest rate is also statistically significant. To generate both anticipated and unanticipated income, we attempted to use Akaike's Statistical Specification Procedure (ASSP) in deriving our atheoretical income generating equation (IGE). More specifically, our ASSP made use of two statistical methods proposed by Akaike, namely the Akaike Final Prediction Error (FPE) and the Akaike's Information Criterion (AIC). ASSP assisted us in determining statistically our atheoretical IGE. ASSP also enables us to reconcile our statistically derived IGE with economic theory. These two procedures appear to have overcome some of the difficulties which are present in Barro's Two-Stage (BTS) Method and Granger's Multivariate Specification Procedure (GMSP). Although this study is hampered by an incomplete data set, it is hoped that the study will contribute to the literature, at least in Singapore, on consumption and the incorporation of rational expectations into the consumption function.