Money, Output and Causality: The Case of Singapore

Abstract

I. Introduction A well-known area of debate in economics literature has been the precise relationship between money and income. This article addresses and tests for any empirical relationship between money and income in Singapore. Other things being equal, if there is a significant and causal positive relationship from money to income (monetarist model), then it may be advisable to execute an expansionary monetary policy. To date, empirical studies between money and income often require the assumption of stationarity for the underlying data. As such, a more rigorous statistical approach of cointegration and error-correction model is used as does not depend on this assumption. The article is divided into the five sections. In Section II, a review of literature on the relationship between money and income policy in Singapore and other countries is provided. The research methodology of cointegration and error-correction model is introduced in Section III. Section IV is devoted to the empirical findings. Discussion and several concluding remarks are offered in the final section. II. Literature Review This section discusses the empirical-based studies available from United States, UK, Malaysia and Singapore.(1) In the United States, Sims (1972) using Granger's concept presented an innovative statistical method, and concluded that the causality was undirectional from money (M) to income (Y). However, eight years later, Sims (1980) found the opposite results: the supply of money did not play an important role in explaining real output fluctuations. These findings cast doubt on what monetarists had come to regard as a well-supported and fundamental macroeconomic fact. Hafer (1982), and Devan and Rangazar (1987) challenged Sims' findings. Both studies found that there was unidirectional causality from M to Y. No reverse causation from Y to M was found. Devan and Rangazar stressed the importance of selecting the appropriate lag lengths and the role of explanatory variables used in the model. On the former issue, they questioned the short lag length of four quarters used in the Sims' model. They pointed out that the monetary effects on output require longer lags, sometimes up to 15 quarters. With regard to the use of explanatory variables, Devan and Rangazar noted that as money is the only variable used in Sims' model, the result would be due solely to variations in money; independent of correlation with all other variables. They added that the Fed's use of interest rates in stimulating output implies that changes in GDP could also be due to the manipulation of interest rates and not to the stock of money only. In a more recent study, Ahmed (1993) noted that it is not easy to isolate the direction of causality between money and output. Empirical evidence in itself is contradicting -- whether money supply is responsible for output changes or output is responsible for variations in money supply is correct some of the time. Despite these contradictory phenomena, he added that it is important to study the direction of causality for formulating an active or passive monetary policy. Ahmed further noted that while money might not be the key explanatory factor for GDP fluctuations, and hence good monetary policy might not eliminate business cycles, monetary policy can help to reduce the severity of the cycles. Studies done in the UK are not as extensive as in the United States. Williams, Goodhart and Gowland (1976) applied Sims' statistical methodology and concluded that there was some evidence of a reverse causality running from nominal income to money. Putman and Wilford (1978) tried to explain the different findings in the United States and UK. They hypothesized that the dissimilarity was due to the fact that the United States is a reserve currency nation, while the UK is a non-reserve currency country. From their analysis, they concluded that a reserve currency nation could better control its money supply, and hence can advocate an active monetary policy for the economy. …