Channels, Lags, and Structural Limits: A Structural VAR Decision-Support Framework for Monetary Policy Analysis in Malaysia

Abstract

This paper develops a computational decision-support framework for monetary policy analysis, providing the first multi-channel structural vector autoregression (SVAR) assessment of monetary policy transmission in Malaysia, spanning the COVID-19 pandemic and post-pandemic inflation surge (2016–2022), using a five-variable recursive Cholesky system: Overnight Policy Rate (OPR), GDP growth, CPI inflation, M2 growth, and nominal USD/MYR exchange rate (NER; T = 84). Monthly data are drawn from the Bank Negara Malaysia (BNM) Monthly Statistical Bulletin and the Department of Statistics Malaysia, and the system is estimated in first differences following unit root pretesting, with the lag order selected by standard information criteria. Block exogeneity tests show that the OPR Granger-causes CPI inflation at the 1% level and GDP growth at the 10% level, while the money supply and exchange rate blocks show no significant response to policy and GDP Granger-causes the OPR, consistent with a systematic reaction function. Impulse response functions (IRFs) with 95% analytic confidence bands and forecast error variance decompositions (FEVDs) with bootstrapped 95% confidence intervals yield three findings: the interest rate channel accounts for 11.0% of GDP variance and 5.5% of CPI variance; money supply and exchange rate channels exhibit limited monetary traction (own-shock dominance: 63.7% and 78.2%); and external disturbances account for over 70% of macroeconomic variance. A commodity price robustness test confirms the price puzzle reflects supply-cost effects rather than misidentification. Episode-specific decompositions for the 2020–2021 pandemic period and the 2021–2022 inflation surge further reveal how transmission effectiveness shifts under different types of external disruption. These findings provide evidence-based guidance for BNM's forward guidance design and macroprudential coordination strategy, supporting probability-based inflation communication, macroprudential complementarity, and domestic bond market deepening. The estimation pipeline—unit root pretesting, information-criterion lag selection, recursive identification, and confidence-band estimation—constitutes a replicable quantitative decision-support tool for central banks in small open economies.