The Money-prices Nexus for Malaysia: New Empirical Evidence from the Time-varying Cointegration and Causality Tests

Abstract

The main objective of this study is to empirically re-investigate the money-prices nexus for Malaysia through the Johansen multivariate cointegration and the modified Wald (MWALD) causality techniques. This study covered the monthly dataset from 1971:M1 to 2008:M11. The Johansen cointegration test suggests that the variables under investigation are co-move in the long run. Furthermore, the MWALD causality test shows a bidirectional causal relationship between money supply (M2) and aggregate prices, meaning that both the monetarist's and also the structuralists' views are vindicated in the Malaysian economy. However, the time-varying cointegration and causality tests indicate that the cointegrating and also the causal relationships are not stable over the analysis period. These results suggest that inflation in Malaysia is not purely a monetary phenomenon. Therefore, implementing a tighter monetary policy may not be an effective macro-economic instrument in managing the inflationary behaviour in the Malaysian economy.

Keywords:

• Cointegration
• inflation
• money
• MWALD test
• rolling regression

JEL CLASSIFICATION:

• C22
• E31
• E51

Acknowledgements

The author would like to thank the anonymous referees. The author would also like to thank Muhamed Zulkhibri Abdul Majid from the Bank Negara Malaysia (Central Bank of Malaysia) for his sincere contribution to this research. All remaining flaws are the responsibility of the author and the usual disclaimer applies.

Notes

1. Nevertheless, as the order of integration for the variables under investigation are non-uniform that is either I(0) or I(1) process, this study has also employed the bounds testing approach to cointegration developed by Pesaran et al. (2001) to re-affirm the existence of a potential long run equilibrium relationship. The results of the bounds testing approach reveals that the variables under investigation (lnM2 _t , lnY _t , lnP _t , lnEX _t ) are cointegrated at the 1% significant level. The bounds testing to cointegration result is corroborated to the Johansen multivariate cointegration test reported in Tables 4 and 5. To conserve space, the estimation results for bounds testing approach to cointegration are not reported here, however it is available upon request from the author.